diff options
| author | YurenHao0426 <Blackhao0426@gmail.com> | 2026-07-22 05:51:35 -0500 |
|---|---|---|
| committer | YurenHao0426 <Blackhao0426@gmail.com> | 2026-07-22 05:51:35 -0500 |
| commit | d31edf53ee47cfdc0bb7e66a3148dc75e072381e (patch) | |
| tree | b04f195fce576d6572c108998ea4f162e2218e30 | |
| parent | b4458cdb3533f7b342f3ff3c8e3582962181d09f (diff) | |
results: close oral B signature screen
40 files changed, 9704 insertions, 0 deletions
@@ -12,6 +12,13 @@ No oral-B result may be generated before this protocol is committed. Native accept-bar baselines may continue in parallel, but the oral-B confirmation may not be inspected until C4 is closed. +**Development status: failed on 2026-07-22.** The complete 36-run frozen grid +was generated from clean revision `b4458cd` and no confirmation seed was +touched. All 12 candidates failed the sign-inversion requirement in all three +task environments. No candidate is eligible, so the confirmation specified +below is prohibited. Detailed negative results are retained in `RESULTS.md` +and `results/bci_dev_selection.json`. + ## What the biological experiment establishes The paper motivates four experimental conditions for a vectorized dendritic @@ -106,6 +113,9 @@ no run or seed may be removed. A result file must record the source revision, dirty state, task/model seed, full dynamics, phase masks, scalar reward observations, forward-equivalent work, and peak memory. +This section remains the preregistered confirmation that would have followed +an eligible development result; it was not run after the failed screen. + ### B1. Innovation identification and network predictability - Mean absolute per-cell `corr(r_i, h_i)` is at most 0.10. @@ -120,6 +120,56 @@ result therefore depends on subtracting the soma-predictable component, not mere normalizing an oversized update. This is the most direct evidence that the Harnett-specific innovation operation is algorithmically necessary. +## Continuous-BCI Harnett-signature screen (development only) + +The oral-B protocol was committed before any result and then run as a complete +12-candidate by three-environment grid at clean revision `b4458cd`. It crossed +scalar error versus error-plus-within-episode velocity, online-control gains +`0`, `0.1`, and `0.3`, and forward rates `0.01` and `0.03`. Every result paired +intact, fixed-vectorizer, online-lesion, plasticity-lesion, and both-lesion +conditions on identical continuous trajectories. Confirmation environment +seeds 10--15 were never evaluated. + +| feedback | kappa | eta | worst final success | worst sign-inversion | eligible | +|:--|--:|--:|--:|--:|:--:| +| error | 0.0 | 0.01 | 0.156 | -0.0038 | no | +| error | 0.0 | 0.03 | 0.988 | -0.0018 | no | +| error | 0.1 | 0.01 | 0.141 | -0.0035 | no | +| error | 0.1 | 0.03 | 0.988 | -0.0018 | no | +| error | 0.3 | 0.01 | 0.102 | -0.0032 | no | +| error | 0.3 | 0.03 | 0.984 | -0.0017 | no | +| error + velocity | 0.0 | 0.01 | 0.258 | -0.0089 | no | +| error + velocity | 0.0 | 0.03 | 0.996 | -0.0071 | no | +| error + velocity | 0.1 | 0.01 | 0.219 | -0.0087 | no | +| error + velocity | 0.1 | 0.03 | 0.996 | -0.0069 | no | +| error + velocity | 0.3 | 0.01 | 0.156 | -0.0085 | no | +| error + velocity | 0.3 | 0.03 | 0.992 | -0.0068 | no | + +The screen demonstrates a sharp dissociation. At `eta=0.03`, all structural +variants solve the task (worst-environment final success 98.4--99.6%). Raw +apical activity is almost completely soma-correlated (mean absolute +correlation `0.999`), whereas residualization removes most of that dependence +(raw-minus-residual correlation gap `0.880`). Surrounding-network activity +decodes residual sign at mean balanced accuracy `0.727`, and early residuals +strongly predict later activity changes (`r=0.860` on average). Plasticity +lesion eliminates learning gain in every run. + +These positive signatures do **not** pass oral B. All 36 trajectories have a +negative causal-role sign-inversion index (range `-0.0089` to `-0.0011`), and +role-aligned residuals are more associated with instantaneous error magnitude +than with signed error velocity in every run. Acute removal of online control +after intact training drops success by only `1.03` points on average (maximum +`5.08`), so the desired-velocity interpretation is not supported. Across the +36 runs, the frozen per-run checks failed 36 times for sign inversion, 30 for +the strict residual-soma threshold, 6 for the fixed-vectorizer margin, and 5 +for learning gain. The preregistered selector therefore returns +`failed_no_eligible_candidate`, and oral-B confirmation is prohibited. + +This negative result prevents a misleading inference: successful local +learning, outcome decoding, and residual decorrelation do not imply the +cell-role-specific error-derivative signature observed by Harnett and +colleagues. + ## Frozen endogenous mixed-traffic confirmation (MNIST, d3/w256, 15 epochs) The 95-run C1 panel crossed five paired model seeds, soma and endogenous top-down traffic, two @@ -20,6 +20,14 @@ were still running and before any continuous-BCI result was generated. Oral-B co sequenced after C4; implementation and CPU-only mechanics may be prepared while the accept jobs occupy both authorized GPUs. +**Oral-B development status: failed.** The complete frozen 12-candidate by three-task-seed screen +found no eligible mechanism. High-learning-rate variants learned the BCI reliably, and several +innovation/decoder signatures were strong, but every run had the wrong P+/P- error-change +sign-inversion and error magnitude remained more predictive than temporal error change. Per the +preregistered stop rule, environment seeds 10--15 and all confirmation model seeds remain +untouched. This falsifies the current model's claimed emergence of the Harnett derivative signature; +it cannot be repaired by reporting only its successful learning or decoder results. + ## Frozen accept claims and gates ### C1. Innovation is necessary under naturally mixed apical traffic diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t0_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t0_m0.json new file mode 100644 index 0000000..d7d3f88 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p0_eta0p01_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.005815509706735611, + "final_success": 0.15625 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005845028907060623, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11806441470980644 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.00628449022769928, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1347898654639721 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12462, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25866.875, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.0, + 0.078125, + 0.25 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.012438163161277771, + "final_success": 0.15625, + "gain_over_fixed_vectorizer_final": 0.15625, + "late_success": 0.109375, + "learning_gain": 0.109375, + "training_wall_s": 0.2139432206749916 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.0, + 0.078125, + 0.25 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005870215594768524, + "final_success": 0.15625, + "late_success": 0.109375, + "learning_gain": 0.109375, + "training_wall_s": 0.14631011337041855 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005850266665220261, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11745921894907951 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.005773719400167465, + "final_success": 0.15625 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 465.48828125, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24899, + "causal_role_sign_inversion_index": -0.002889420693915609, + "decoder_distance_residual_corr": 0.1635916912105782, + "early_residual_late_activity_change_corr": 0.7615588041523232, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.15625, + "mean_abs_raw_soma_corr": 0.9981404052506477, + "mean_abs_residual_soma_corr": 0.09531960677321405, + "raw_minus_residual_abs_soma_corr": 0.9028207984774337, + "residual_minus_soma_outcome_acc": 0.052314814814814925, + "residual_outcome_decoder_balanced_acc": 0.8652777777777778, + "role_aligned_error_cv_corr": 0.1742078753401823, + "role_aligned_velocity_cv_corr": 0.058675878359523426, + "soma_outcome_decoder_balanced_acc": 0.8129629629629629, + "surrounding_event_decoder_balanced_acc": 0.8528148618289165, + "velocity_minus_error_abs_cv_corr": -0.11553199698065889 + }, + "wall_s": 1.7787450328469276 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t1_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t1_m0.json new file mode 100644 index 0000000..fbaed51 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p0_eta0p01_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.005752060562372208, + "final_success": 0.24609375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005753610283136368, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11722038313746452 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006162047386169434, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13734329119324684 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12440, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25865.5, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.046875, + 0.03125, + 0.078125, + 0.234375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.012927185744047165, + "final_success": 0.24609375, + "gain_over_fixed_vectorizer_final": 0.24609375, + "late_success": 0.11458333333333333, + "learning_gain": 0.11458333333333333, + "training_wall_s": 0.21087270230054855 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.046875, + 0.03125, + 0.078125, + 0.234375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005787212401628494, + "final_success": 0.24609375, + "late_success": 0.11458333333333333, + "learning_gain": 0.11458333333333333, + "training_wall_s": 0.14558686688542366 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005747467279434204, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11722684279084206 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.00584450364112854, + "final_success": 0.24609375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 452.93359375, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24846, + "causal_role_sign_inversion_index": -0.0034216490242629308, + "decoder_distance_residual_corr": 0.17255662648040054, + "early_residual_late_activity_change_corr": 0.7856749549547628, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.24609375, + "mean_abs_raw_soma_corr": 0.9981378802098648, + "mean_abs_residual_soma_corr": 0.09744459076406656, + "raw_minus_residual_abs_soma_corr": 0.9006932894457983, + "residual_minus_soma_outcome_acc": 0.04761904761904767, + "residual_outcome_decoder_balanced_acc": 0.8251089727773666, + "role_aligned_error_cv_corr": 0.2115077317996372, + "role_aligned_velocity_cv_corr": 0.062104886596436586, + "soma_outcome_decoder_balanced_acc": 0.7774899251583189, + "surrounding_event_decoder_balanced_acc": 0.8489709008826907, + "velocity_minus_error_abs_cv_corr": -0.1494028452032006 + }, + "wall_s": 1.8259844444692135 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t2_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t2_m0.json new file mode 100644 index 0000000..3523147 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p0_eta0p01_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.006470393389463425, + "final_success": 0.2578125 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006441768258810043, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12513814866542816 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006478250026702881, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13588760793209076 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12406, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25863.375, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.09375, + 0.109375, + 0.203125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013307977467775345, + "final_success": 0.2578125, + "gain_over_fixed_vectorizer_final": 0.2578125, + "late_success": 0.13541666666666666, + "learning_gain": 0.13541666666666666, + "training_wall_s": 0.19746039435267448 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.09375, + 0.109375, + 0.203125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005952619016170502, + "final_success": 0.2578125, + "late_success": 0.13541666666666666, + "learning_gain": 0.13541666666666666, + "training_wall_s": 0.1530621200799942 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006377402693033218, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12573878094553947 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006344422698020935, + "final_success": 0.2578125 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 458.7890625, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24766, + "causal_role_sign_inversion_index": -0.0037756110210441854, + "decoder_distance_residual_corr": 0.17070923696941476, + "early_residual_late_activity_change_corr": 0.7697777875800323, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.2578125, + "mean_abs_raw_soma_corr": 0.9981452876066104, + "mean_abs_residual_soma_corr": 0.09642710577736702, + "raw_minus_residual_abs_soma_corr": 0.9017181818292433, + "residual_minus_soma_outcome_acc": 0.03755980861244024, + "residual_outcome_decoder_balanced_acc": 0.7900318979266348, + "role_aligned_error_cv_corr": 0.24167193742893306, + "role_aligned_velocity_cv_corr": 0.07116272406812102, + "soma_outcome_decoder_balanced_acc": 0.7524720893141945, + "surrounding_event_decoder_balanced_acc": 0.8450570302676448, + "velocity_minus_error_abs_cv_corr": -0.17050921336081204 + }, + "wall_s": 1.816775631159544 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t0_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t0_m0.json new file mode 100644 index 0000000..84f0271 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p0_eta0p03_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.006037332117557526, + "final_success": 0.98828125 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005989357829093933, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1180933378636837 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006277143955230713, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1363091617822647 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 10392, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25737.5, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0625, + 0.15625, + 0.34375, + 0.609375, + 0.78125, + 0.84375, + 0.984375, + 0.96875 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.01281517744064331, + "final_success": 0.98828125, + "gain_over_fixed_vectorizer_final": 0.98828125, + "late_success": 0.9322916666666666, + "learning_gain": 0.9322916666666666, + "training_wall_s": 0.1956605352461338 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0625, + 0.15625, + 0.34375, + 0.609375, + 0.78125, + 0.84375, + 0.984375, + 0.96875 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0059013329446315765, + "final_success": 0.98828125, + "late_success": 0.9322916666666666, + "learning_gain": 0.9322916666666666, + "training_wall_s": 0.14984595030546188 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.00587974488735199, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11863154545426369 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.00583387166261673, + "final_success": 0.98828125 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 448.4375, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 20358, + "causal_role_sign_inversion_index": -0.0011943816763679907, + "decoder_distance_residual_corr": 0.2907495102089732, + "early_residual_late_activity_change_corr": 0.8547192464278519, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.98828125, + "mean_abs_raw_soma_corr": 0.9983135219329607, + "mean_abs_residual_soma_corr": 0.12304228527851073, + "raw_minus_residual_abs_soma_corr": 0.8752712366544501, + "residual_minus_soma_outcome_acc": 0.17325428194993409, + "residual_outcome_decoder_balanced_acc": 0.6712779973649539, + "role_aligned_error_cv_corr": 0.3200056695631025, + "role_aligned_velocity_cv_corr": 0.022382208433574278, + "soma_outcome_decoder_balanced_acc": 0.4980237154150198, + "surrounding_event_decoder_balanced_acc": 0.7547905430258943, + "velocity_minus_error_abs_cv_corr": -0.2976234611295282 + }, + "wall_s": 1.5829623453319073 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t1_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t1_m0.json new file mode 100644 index 0000000..2e8a523 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p0_eta0p03_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.0059861838817596436, + "final_success": 0.9921875 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005848351866006851, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11784321069717407 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006331808865070343, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13496142998337746 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 10018, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25714.125, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.140625, + 0.296875, + 0.453125, + 0.75, + 0.84375, + 0.921875, + 0.984375, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.012989744544029236, + "final_success": 0.9921875, + "gain_over_fixed_vectorizer_final": 0.9921875, + "late_success": 0.96875, + "learning_gain": 0.96875, + "training_wall_s": 0.19486254826188087 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.140625, + 0.296875, + 0.453125, + 0.75, + 0.84375, + 0.921875, + 0.984375, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005953103303909302, + "final_success": 0.9921875, + "late_success": 0.96875, + "learning_gain": 0.96875, + "training_wall_s": 0.14490199834108353 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006006233394145966, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1180603913962841 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.005836650729179382, + "final_success": 0.9921875 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 436.05859375, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 19571, + "causal_role_sign_inversion_index": -0.00148484176075913, + "decoder_distance_residual_corr": 0.30608797442972474, + "early_residual_late_activity_change_corr": 0.8663677549637265, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.9921875, + "mean_abs_raw_soma_corr": 0.9982721720092196, + "mean_abs_residual_soma_corr": 0.13013920077489838, + "raw_minus_residual_abs_soma_corr": 0.8681329712343212, + "residual_minus_soma_outcome_acc": -0.19488188976377951, + "residual_outcome_decoder_balanced_acc": 0.30118110236220474, + "role_aligned_error_cv_corr": 0.31137017956285623, + "role_aligned_velocity_cv_corr": 0.017094164895422013, + "soma_outcome_decoder_balanced_acc": 0.49606299212598426, + "surrounding_event_decoder_balanced_acc": 0.7478138750830297, + "velocity_minus_error_abs_cv_corr": -0.2942760146674342 + }, + "wall_s": 1.5510678812861443 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t2_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t2_m0.json new file mode 100644 index 0000000..0b6573e --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p0_eta0p03_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.005935214459896088, + "final_success": 0.99609375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005886729806661606, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11934611201286316 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006286874413490295, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13702460005879402 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 9844, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25703.25, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.09375, + 0.3125, + 0.46875, + 0.734375, + 0.921875, + 0.953125, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.012919634580612183, + "final_success": 0.99609375, + "gain_over_fixed_vectorizer_final": 0.99609375, + "late_success": 0.984375, + "learning_gain": 0.984375, + "training_wall_s": 0.2062663845717907 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.09375, + 0.3125, + 0.46875, + 0.734375, + 0.921875, + 0.953125, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005896672606468201, + "final_success": 0.99609375, + "late_success": 0.984375, + "learning_gain": 0.984375, + "training_wall_s": 0.1459806151688099 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005859162658452988, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11906864866614342 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.005825329571962357, + "final_success": 0.99609375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 439.4375, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 19234, + "causal_role_sign_inversion_index": -0.0018247418377719364, + "decoder_distance_residual_corr": 0.3127653078266505, + "early_residual_late_activity_change_corr": 0.8511221863484061, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.99609375, + "mean_abs_raw_soma_corr": 0.9983281606658047, + "mean_abs_residual_soma_corr": 0.1269611717333344, + "raw_minus_residual_abs_soma_corr": 0.8713669889324702, + "residual_minus_soma_outcome_acc": -0.06862745098039214, + "residual_outcome_decoder_balanced_acc": 0.43137254901960786, + "role_aligned_error_cv_corr": 0.34185412074866295, + "role_aligned_velocity_cv_corr": 0.0073624852403260345, + "soma_outcome_decoder_balanced_acc": 0.5, + "surrounding_event_decoder_balanced_acc": 0.7579364323125718, + "velocity_minus_error_abs_cv_corr": -0.33449163550833694 + }, + "wall_s": 1.4827960729599 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t0_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t0_m0.json new file mode 100644 index 0000000..f0f1431 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p1_eta0p01_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.02734375, + "evaluation_wall_s": 0.006328746676445007, + "final_success": 0.11328125 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006226491183042526, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12646709382534027 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006638679653406143, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1355353221297264 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12482, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25868.125, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.0, + 0.046875, + 0.203125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013482116162776947, + "final_success": 0.140625, + "gain_over_fixed_vectorizer_final": 0.140625, + "late_success": 0.08333333333333333, + "learning_gain": 0.08333333333333333, + "training_wall_s": 0.19924890249967575 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.0, + 0.078125, + 0.25 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006369546055793762, + "final_success": 0.15625, + "late_success": 0.109375, + "learning_gain": 0.109375, + "training_wall_s": 0.14983704686164856 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.00644293799996376, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12242982164025307 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006203398108482361, + "final_success": 0.140625 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 459.39453125, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24947, + "causal_role_sign_inversion_index": -0.0027520933435467784, + "decoder_distance_residual_corr": 0.12643558913192807, + "early_residual_late_activity_change_corr": 0.7479467206690867, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.140625, + "mean_abs_raw_soma_corr": 0.9981775255221279, + "mean_abs_residual_soma_corr": 0.09537462446627583, + "raw_minus_residual_abs_soma_corr": 0.902802901055852, + "residual_minus_soma_outcome_acc": 0.048484848484848575, + "residual_outcome_decoder_balanced_acc": 0.8583333333333334, + "role_aligned_error_cv_corr": 0.1401122142068369, + "role_aligned_velocity_cv_corr": 0.047772135146322, + "soma_outcome_decoder_balanced_acc": 0.8098484848484848, + "surrounding_event_decoder_balanced_acc": 0.8543349012235941, + "velocity_minus_error_abs_cv_corr": -0.09234007906051489 + }, + "wall_s": 1.8469149433076382 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t1_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t1_m0.json new file mode 100644 index 0000000..fb3420f --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p1_eta0p01_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0078125, + "evaluation_wall_s": 0.006271615624427795, + "final_success": 0.20703125 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006265535950660706, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1299884244799614 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006897594779729843, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14856457710266113 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12462, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25866.875, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.03125, + 0.015625, + 0.046875, + 0.203125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.01425955444574356, + "final_success": 0.21484375, + "gain_over_fixed_vectorizer_final": 0.21484375, + "late_success": 0.08854166666666667, + "learning_gain": 0.08854166666666667, + "training_wall_s": 0.2111736498773098 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.046875, + 0.03125, + 0.078125, + 0.234375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006494347006082535, + "final_success": 0.24609375, + "late_success": 0.11458333333333333, + "learning_gain": 0.11458333333333333, + "training_wall_s": 0.1607251986861229 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006377235054969788, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12903953716158867 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006248984485864639, + "final_success": 0.21484375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 459.44140625, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24901, + "causal_role_sign_inversion_index": -0.0031658220936501905, + "decoder_distance_residual_corr": 0.1306694795277127, + "early_residual_late_activity_change_corr": 0.7735154433956827, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.21484375, + "mean_abs_raw_soma_corr": 0.9981825080048642, + "mean_abs_residual_soma_corr": 0.0983690856410097, + "raw_minus_residual_abs_soma_corr": 0.8998134223638545, + "residual_minus_soma_outcome_acc": 0.040479421076436095, + "residual_outcome_decoder_balanced_acc": 0.8178199909543193, + "role_aligned_error_cv_corr": 0.17457474287994348, + "role_aligned_velocity_cv_corr": 0.05127432997512734, + "soma_outcome_decoder_balanced_acc": 0.7773405698778832, + "surrounding_event_decoder_balanced_acc": 0.8490868607448736, + "velocity_minus_error_abs_cv_corr": -0.12330041290481614 + }, + "wall_s": 1.7272856831550598 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t2_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t2_m0.json new file mode 100644 index 0000000..a2b9291 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p1_eta0p01_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.01171875, + "evaluation_wall_s": 0.0062831006944179535, + "final_success": 0.20703125 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006323602050542831, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12808040529489517 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006785962730646133, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1476794146001339 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12420, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25864.25, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.078125, + 0.078125, + 0.203125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.012695766985416412, + "final_success": 0.21875, + "gain_over_fixed_vectorizer_final": 0.21875, + "late_success": 0.11979166666666667, + "learning_gain": 0.11979166666666667, + "training_wall_s": 0.21216589584946632 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.09375, + 0.109375, + 0.203125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006391771137714386, + "final_success": 0.2578125, + "late_success": 0.13541666666666666, + "learning_gain": 0.13541666666666666, + "training_wall_s": 0.1598282903432846 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006377819925546646, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1285143420100212 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006217524409294128, + "final_success": 0.21875 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 465.01171875, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24802, + "causal_role_sign_inversion_index": -0.003522021519430374, + "decoder_distance_residual_corr": 0.133514433549035, + "early_residual_late_activity_change_corr": 0.7570375279219594, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.21875, + "mean_abs_raw_soma_corr": 0.9981876130681403, + "mean_abs_residual_soma_corr": 0.09663759251737655, + "raw_minus_residual_abs_soma_corr": 0.9015500205507637, + "residual_minus_soma_outcome_acc": 0.02928571428571425, + "residual_outcome_decoder_balanced_acc": 0.7921428571428571, + "role_aligned_error_cv_corr": 0.2047618850785278, + "role_aligned_velocity_cv_corr": 0.06050276091424267, + "soma_outcome_decoder_balanced_acc": 0.7628571428571429, + "surrounding_event_decoder_balanced_acc": 0.8375909030642447, + "velocity_minus_error_abs_cv_corr": -0.14425912416428513 + }, + "wall_s": 1.77140087261796 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t0_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t0_m0.json new file mode 100644 index 0000000..6305928 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p1_eta0p03_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.00390625, + "evaluation_wall_s": 0.006359763443470001, + "final_success": 0.984375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006387095898389816, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1297815889120102 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006817139685153961, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14793045818805695 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 10458, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25741.625, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0625, + 0.125, + 0.328125, + 0.578125, + 0.78125, + 0.828125, + 0.96875, + 0.96875 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013628389686346054, + "final_success": 0.98828125, + "gain_over_fixed_vectorizer_final": 0.98828125, + "late_success": 0.921875, + "learning_gain": 0.921875, + "training_wall_s": 0.23491241037845612 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0625, + 0.15625, + 0.34375, + 0.609375, + 0.78125, + 0.84375, + 0.984375, + 0.96875 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006462346762418747, + "final_success": 0.98828125, + "late_success": 0.9322916666666666, + "learning_gain": 0.9322916666666666, + "training_wall_s": 0.1623406782746315 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006457965821027756, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13086361438035965 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.00655224546790123, + "final_success": 0.98828125 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 444.1953125, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 20490, + "causal_role_sign_inversion_index": -0.0011604788136277272, + "decoder_distance_residual_corr": 0.2634451027274726, + "early_residual_late_activity_change_corr": 0.8491401775364665, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.98828125, + "mean_abs_raw_soma_corr": 0.998308726752154, + "mean_abs_residual_soma_corr": 0.12190536305577722, + "raw_minus_residual_abs_soma_corr": 0.8764033636963767, + "residual_minus_soma_outcome_acc": 0.18313570487483527, + "residual_outcome_decoder_balanced_acc": 0.6772068511198945, + "role_aligned_error_cv_corr": 0.29823387518228484, + "role_aligned_velocity_cv_corr": 0.03306912778996271, + "soma_outcome_decoder_balanced_acc": 0.49407114624505927, + "surrounding_event_decoder_balanced_acc": 0.7456192711757541, + "velocity_minus_error_abs_cv_corr": -0.2651647473923221 + }, + "wall_s": 1.6134153977036476 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t1_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t1_m0.json new file mode 100644 index 0000000..9430120 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p1_eta0p03_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.00390625, + "evaluation_wall_s": 0.006386034190654755, + "final_success": 0.98828125 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006408873945474625, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11856388673186302 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006265223026275635, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13579446077346802 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 10094, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25718.875, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.140625, + 0.265625, + 0.375, + 0.71875, + 0.8125, + 0.921875, + 0.984375, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.012890398502349854, + "final_success": 0.9921875, + "gain_over_fixed_vectorizer_final": 0.9921875, + "late_success": 0.96875, + "learning_gain": 0.96875, + "training_wall_s": 0.19716917350888252 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.140625, + 0.296875, + 0.453125, + 0.75, + 0.84375, + 0.921875, + 0.984375, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005914624780416489, + "final_success": 0.9921875, + "late_success": 0.96875, + "learning_gain": 0.96875, + "training_wall_s": 0.14532073959708214 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006630841642618179, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11889376118779182 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.0065752193331718445, + "final_success": 0.9921875 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 441.76171875, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 19734, + "causal_role_sign_inversion_index": -0.0013616858451298147, + "decoder_distance_residual_corr": 0.27630749191826237, + "early_residual_late_activity_change_corr": 0.8615257306939526, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.9921875, + "mean_abs_raw_soma_corr": 0.9982728359495686, + "mean_abs_residual_soma_corr": 0.128785890931579, + "raw_minus_residual_abs_soma_corr": 0.8694869450179896, + "residual_minus_soma_outcome_acc": 0.11220472440944884, + "residual_outcome_decoder_balanced_acc": 0.5984251968503937, + "role_aligned_error_cv_corr": 0.2886398292942647, + "role_aligned_velocity_cv_corr": 0.028017735898898083, + "soma_outcome_decoder_balanced_acc": 0.4862204724409449, + "surrounding_event_decoder_balanced_acc": 0.7406409719980934, + "velocity_minus_error_abs_cv_corr": -0.26062209339536657 + }, + "wall_s": 1.5115188881754875 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t2_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t2_m0.json new file mode 100644 index 0000000..a6a95cc --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p1_eta0p03_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.005887255072593689, + "final_success": 0.99609375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005963407456874847, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11890537664294243 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006410978734493256, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1378590576350689 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 9884, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25705.75, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.09375, + 0.3125, + 0.453125, + 0.71875, + 0.921875, + 0.953125, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.01342979446053505, + "final_success": 0.99609375, + "gain_over_fixed_vectorizer_final": 0.99609375, + "late_success": 0.984375, + "learning_gain": 0.984375, + "training_wall_s": 0.1972711645066738 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.09375, + 0.3125, + 0.46875, + 0.734375, + 0.921875, + 0.953125, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005903154611587524, + "final_success": 0.99609375, + "late_success": 0.984375, + "learning_gain": 0.984375, + "training_wall_s": 0.14897293224930763 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005950756371021271, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12048821151256561 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.005916975438594818, + "final_success": 0.99609375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 435.07421875, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 19315, + "causal_role_sign_inversion_index": -0.0017520959075886383, + "decoder_distance_residual_corr": 0.28826815037677467, + "early_residual_late_activity_change_corr": 0.8462440592593323, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.99609375, + "mean_abs_raw_soma_corr": 0.9983236944582827, + "mean_abs_residual_soma_corr": 0.1252722799663563, + "raw_minus_residual_abs_soma_corr": 0.8730514144919265, + "residual_minus_soma_outcome_acc": -0.07450980392156864, + "residual_outcome_decoder_balanced_acc": 0.42549019607843136, + "role_aligned_error_cv_corr": 0.3172469224536509, + "role_aligned_velocity_cv_corr": 0.01890155072120642, + "soma_outcome_decoder_balanced_acc": 0.5, + "surrounding_event_decoder_balanced_acc": 0.7470678833518141, + "velocity_minus_error_abs_cv_corr": -0.29834537173244446 + }, + "wall_s": 1.4246256463229656 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t0_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t0_m0.json new file mode 100644 index 0000000..62813ac --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p3_eta0p01_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.015625, + "evaluation_wall_s": 0.006328854709863663, + "final_success": 0.0859375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006274580955505371, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12642864882946014 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006715584546327591, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13471213355660439 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12506, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25869.625, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.0, + 0.03125, + 0.171875 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013494089245796204, + "final_success": 0.1015625, + "gain_over_fixed_vectorizer_final": 0.1015625, + "late_success": 0.06770833333333333, + "learning_gain": 0.06770833333333333, + "training_wall_s": 0.2000165916979313 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.0, + 0.078125, + 0.25 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006301164627075195, + "final_success": 0.15625, + "late_success": 0.109375, + "learning_gain": 0.109375, + "training_wall_s": 0.1461787223815918 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006269574165344238, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1260518804192543 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006272308528423309, + "final_success": 0.1015625 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 459.6484375, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24996, + "causal_role_sign_inversion_index": -0.0026150713256167524, + "decoder_distance_residual_corr": 0.09050405511644932, + "early_residual_late_activity_change_corr": 0.718295931089254, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.1015625, + "mean_abs_raw_soma_corr": 0.9982436631410534, + "mean_abs_residual_soma_corr": 0.10239964053834512, + "raw_minus_residual_abs_soma_corr": 0.8958440226027082, + "residual_minus_soma_outcome_acc": 0.05334448160535121, + "residual_outcome_decoder_balanced_acc": 0.8637123745819398, + "role_aligned_error_cv_corr": 0.07040575557195189, + "role_aligned_velocity_cv_corr": 0.02494323712339446, + "soma_outcome_decoder_balanced_acc": 0.8103678929765886, + "surrounding_event_decoder_balanced_acc": 0.8500883850295358, + "velocity_minus_error_abs_cv_corr": -0.04546251844855743 + }, + "wall_s": 1.8335933312773705 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t1_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t1_m0.json new file mode 100644 index 0000000..b820c5d --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p3_eta0p01_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.046875, + "evaluation_wall_s": 0.0057619065046310425, + "final_success": 0.140625 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005758307874202728, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11737091466784477 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006212051957845688, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13478930667042732 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12476, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25867.75, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.0, + 0.015625, + 0.046875, + 0.171875 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.012907076627016068, + "final_success": 0.1875, + "gain_over_fixed_vectorizer_final": 0.1875, + "late_success": 0.078125, + "learning_gain": 0.078125, + "training_wall_s": 0.19523229077458382 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.046875, + 0.03125, + 0.078125, + 0.234375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005783010274171829, + "final_success": 0.24609375, + "late_success": 0.11458333333333333, + "learning_gain": 0.11458333333333333, + "training_wall_s": 0.14686867222189903 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005775231868028641, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11763830855488777 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.005784999579191208, + "final_success": 0.1875 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 465.4453125, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24929, + "causal_role_sign_inversion_index": -0.0029546118827365186, + "decoder_distance_residual_corr": 0.09472044967561526, + "early_residual_late_activity_change_corr": 0.7484836622861659, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.1875, + "mean_abs_raw_soma_corr": 0.9982623683441751, + "mean_abs_residual_soma_corr": 0.10479825777836274, + "raw_minus_residual_abs_soma_corr": 0.8934641105658124, + "residual_minus_soma_outcome_acc": 0.02644230769230771, + "residual_outcome_decoder_balanced_acc": 0.8237179487179487, + "role_aligned_error_cv_corr": 0.09748128851164035, + "role_aligned_velocity_cv_corr": 0.028474830457139017, + "soma_outcome_decoder_balanced_acc": 0.797275641025641, + "surrounding_event_decoder_balanced_acc": 0.8464402232408762, + "velocity_minus_error_abs_cv_corr": -0.06900645805450134 + }, + "wall_s": 1.7975875921547413 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t2_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t2_m0.json new file mode 100644 index 0000000..a1382a3 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p3_eta0p01_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.05078125, + "evaluation_wall_s": 0.006505992263555527, + "final_success": 0.12109375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006319597363471985, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12864483520388603 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006793726235628128, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14155425503849983 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12436, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25865.25, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.0625, + 0.0625, + 0.171875 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013466555625200272, + "final_success": 0.171875, + "gain_over_fixed_vectorizer_final": 0.171875, + "late_success": 0.09895833333333333, + "learning_gain": 0.09895833333333333, + "training_wall_s": 0.19755515828728676 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.09375, + 0.109375, + 0.203125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006670564413070679, + "final_success": 0.2578125, + "late_success": 0.13541666666666666, + "learning_gain": 0.13541666666666666, + "training_wall_s": 0.1600082963705063 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006350424140691757, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12801989167928696 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006341654807329178, + "final_success": 0.171875 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 453.03125, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24842, + "causal_role_sign_inversion_index": -0.0032190460550888048, + "decoder_distance_residual_corr": 0.08774776896870987, + "early_residual_late_activity_change_corr": 0.728712002609679, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.171875, + "mean_abs_raw_soma_corr": 0.9982599764103632, + "mean_abs_residual_soma_corr": 0.10134740562971903, + "raw_minus_residual_abs_soma_corr": 0.8969125707806442, + "residual_minus_soma_outcome_acc": 0.04309605488850776, + "residual_outcome_decoder_balanced_acc": 0.8048885077186965, + "role_aligned_error_cv_corr": 0.13175212583392132, + "role_aligned_velocity_cv_corr": 0.03732782174561859, + "soma_outcome_decoder_balanced_acc": 0.7617924528301887, + "surrounding_event_decoder_balanced_acc": 0.8336863264565851, + "velocity_minus_error_abs_cv_corr": -0.09442430408830273 + }, + "wall_s": 1.7469113022089005 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t0_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t0_m0.json new file mode 100644 index 0000000..e351960 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p3_eta0p03_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0078125, + "evaluation_wall_s": 0.006482135504484177, + "final_success": 0.98046875 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006426282227039337, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13199982419610023 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006878208369016647, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13582409918308258 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 10556, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25747.75, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.109375, + 0.296875, + 0.515625, + 0.734375, + 0.8125, + 0.953125, + 0.96875 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013821639120578766, + "final_success": 0.98828125, + "gain_over_fixed_vectorizer_final": 0.98828125, + "late_success": 0.9114583333333334, + "learning_gain": 0.9114583333333334, + "training_wall_s": 0.20037396252155304 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0625, + 0.15625, + 0.34375, + 0.609375, + 0.78125, + 0.84375, + 0.984375, + 0.96875 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006515678018331528, + "final_success": 0.98828125, + "late_success": 0.9322916666666666, + "learning_gain": 0.9322916666666666, + "training_wall_s": 0.15468092262744904 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0064478591084480286, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13180949166417122 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006555285304784775, + "final_success": 0.98828125 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 439.9453125, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 20700, + "causal_role_sign_inversion_index": -0.0010676039791819303, + "decoder_distance_residual_corr": 0.21366016956107425, + "early_residual_late_activity_change_corr": 0.8394936443932649, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.98828125, + "mean_abs_raw_soma_corr": 0.998301094138981, + "mean_abs_residual_soma_corr": 0.120541613114923, + "raw_minus_residual_abs_soma_corr": 0.8777594810240581, + "residual_minus_soma_outcome_acc": 0.18906455862977595, + "residual_outcome_decoder_balanced_acc": 0.6831357048748352, + "role_aligned_error_cv_corr": 0.24999453979685807, + "role_aligned_velocity_cv_corr": 0.05367577875805521, + "soma_outcome_decoder_balanced_acc": 0.49407114624505927, + "surrounding_event_decoder_balanced_acc": 0.7271238691140229, + "velocity_minus_error_abs_cv_corr": -0.19631876103880286 + }, + "wall_s": 1.576721228659153 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t1_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t1_m0.json new file mode 100644 index 0000000..07cecd0 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p3_eta0p03_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.00390625, + "evaluation_wall_s": 0.006443891674280167, + "final_success": 0.98046875 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006427440792322159, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13042357936501503 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006965789943933487, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.15329063683748245 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 10212, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25726.25, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.125, + 0.234375, + 0.34375, + 0.625, + 0.8125, + 0.921875, + 0.984375, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.014198534190654755, + "final_success": 0.984375, + "gain_over_fixed_vectorizer_final": 0.984375, + "late_success": 0.96875, + "learning_gain": 0.96875, + "training_wall_s": 0.20616186782717705 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.140625, + 0.296875, + 0.453125, + 0.75, + 0.84375, + 0.921875, + 0.984375, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0065779611468315125, + "final_success": 0.9921875, + "late_success": 0.96875, + "learning_gain": 0.96875, + "training_wall_s": 0.16229357197880745 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006568562239408493, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1308167241513729 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006421893835067749, + "final_success": 0.984375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 447.04296875, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 19982, + "causal_role_sign_inversion_index": -0.0014496980554480898, + "decoder_distance_residual_corr": 0.22578275252639934, + "early_residual_late_activity_change_corr": 0.8499729285934847, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.984375, + "mean_abs_raw_soma_corr": 0.9982742968563556, + "mean_abs_residual_soma_corr": 0.1278523791165752, + "raw_minus_residual_abs_soma_corr": 0.8704219177397804, + "residual_minus_soma_outcome_acc": -0.14285714285714285, + "residual_outcome_decoder_balanced_acc": 0.3392857142857143, + "role_aligned_error_cv_corr": 0.23866171725442586, + "role_aligned_velocity_cv_corr": 0.048681608286546954, + "soma_outcome_decoder_balanced_acc": 0.48214285714285715, + "surrounding_event_decoder_balanced_acc": 0.7194165220295211, + "velocity_minus_error_abs_cv_corr": -0.18998010896787892 + }, + "wall_s": 1.6434127055108547 +} diff --git a/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t2_m0.json b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t2_m0.json new file mode 100644 index 0000000..fb9d390 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberr_k0p3_eta0p03_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.005801308900117874, + "final_success": 0.99609375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005768779665231705, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11594508960843086 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006162688136100769, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1331852450966835 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 10038, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25715.375, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.078125, + 0.28125, + 0.375, + 0.671875, + 0.84375, + 0.9375, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.012813683599233627, + "final_success": 0.99609375, + "gain_over_fixed_vectorizer_final": 0.99609375, + "late_success": 0.9791666666666666, + "learning_gain": 0.9791666666666666, + "training_wall_s": 0.19336237013339996 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.09375, + 0.3125, + 0.46875, + 0.734375, + 0.921875, + 0.953125, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005798365920782089, + "final_success": 0.99609375, + "late_success": 0.984375, + "learning_gain": 0.984375, + "training_wall_s": 0.14407265186309814 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.005801517516374588, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.11680164560675621 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.005825255066156387, + "final_success": 0.99609375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 441.2578125, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 19645, + "causal_role_sign_inversion_index": -0.0016557727802265117, + "decoder_distance_residual_corr": 0.24136044888097166, + "early_residual_late_activity_change_corr": 0.8343171402992363, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.99609375, + "mean_abs_raw_soma_corr": 0.9983149075059433, + "mean_abs_residual_soma_corr": 0.12421084235979667, + "raw_minus_residual_abs_soma_corr": 0.8741040651461467, + "residual_minus_soma_outcome_acc": -0.06274509803921569, + "residual_outcome_decoder_balanced_acc": 0.4372549019607843, + "role_aligned_error_cv_corr": 0.2701120930705655, + "role_aligned_velocity_cv_corr": 0.04138275262352257, + "soma_outcome_decoder_balanced_acc": 0.5, + "surrounding_event_decoder_balanced_acc": 0.7246404749670241, + "velocity_minus_error_abs_cv_corr": -0.22872934044704293 + }, + "wall_s": 1.4912168271839619 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t0_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t0_m0.json new file mode 100644 index 0000000..6449e18 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error_velocity", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p0_eta0p01_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.006250098347663879, + "final_success": 0.2578125 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006242528557777405, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12353946641087532 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006771989166736603, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14909863099455833 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12418, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25864.125, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.109375, + 0.34375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013225801289081573, + "final_success": 0.2578125, + "gain_over_fixed_vectorizer_final": 0.2578125, + "late_success": 0.15625, + "learning_gain": 0.15625, + "training_wall_s": 0.2055460885167122 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.109375, + 0.34375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006329037249088287, + "final_success": 0.2578125, + "late_success": 0.15625, + "learning_gain": 0.15625, + "training_wall_s": 0.15284299105405807 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006254822015762329, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12366009503602982 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006259225308895111, + "final_success": 0.2578125 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 464.91015625, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24797, + "causal_role_sign_inversion_index": -0.007845451341592362, + "decoder_distance_residual_corr": -0.029846957245025556, + "early_residual_late_activity_change_corr": 0.8940712703645787, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.2578125, + "mean_abs_raw_soma_corr": 0.9990703097635842, + "mean_abs_residual_soma_corr": 0.1186711728250458, + "raw_minus_residual_abs_soma_corr": 0.8803991369385384, + "residual_minus_soma_outcome_acc": 0.09449760765550241, + "residual_outcome_decoder_balanced_acc": 0.9134768740031898, + "role_aligned_error_cv_corr": 0.5812394823554998, + "role_aligned_velocity_cv_corr": 0.22121793281007843, + "soma_outcome_decoder_balanced_acc": 0.8189792663476874, + "surrounding_event_decoder_balanced_acc": 0.7248198655806746, + "velocity_minus_error_abs_cv_corr": -0.3600215495454214 + }, + "wall_s": 1.9226362220942974 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t1_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t1_m0.json new file mode 100644 index 0000000..40d4769 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error_velocity", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p0_eta0p01_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.006943900138139725, + "final_success": 0.34375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.00688939169049263, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13791662454605103 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007425196468830109, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.15716198831796646 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12364, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25860.75, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.03125, + 0.046875, + 0.078125, + 0.140625, + 0.328125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.014243204146623611, + "final_success": 0.34375, + "gain_over_fixed_vectorizer_final": 0.34375, + "late_success": 0.18229166666666666, + "learning_gain": 0.18229166666666666, + "training_wall_s": 0.21286562457680702 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.03125, + 0.046875, + 0.078125, + 0.140625, + 0.328125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007102441042661667, + "final_success": 0.34375, + "late_success": 0.18229166666666666, + "learning_gain": 0.18229166666666666, + "training_wall_s": 0.17122868448495865 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007043372839689255, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1386127658188343 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006931040436029434, + "final_success": 0.34375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 458.875, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24672, + "causal_role_sign_inversion_index": -0.008684546650026242, + "decoder_distance_residual_corr": -0.017016422287269704, + "early_residual_late_activity_change_corr": 0.9042082337444448, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.34375, + "mean_abs_raw_soma_corr": 0.9989957342987432, + "mean_abs_residual_soma_corr": 0.12021397246480989, + "raw_minus_residual_abs_soma_corr": 0.8787817618339333, + "residual_minus_soma_outcome_acc": 0.14123376623376627, + "residual_outcome_decoder_balanced_acc": 0.9047619047619048, + "role_aligned_error_cv_corr": 0.5897636157812169, + "role_aligned_velocity_cv_corr": 0.2201562608474383, + "soma_outcome_decoder_balanced_acc": 0.7635281385281385, + "surrounding_event_decoder_balanced_acc": 0.6878809640522314, + "velocity_minus_error_abs_cv_corr": -0.3696073549337786 + }, + "wall_s": 1.942064419388771 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t2_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t2_m0.json new file mode 100644 index 0000000..c0d2ee2 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error_velocity", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p0_eta0p01_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.006492838263511658, + "final_success": 0.40234375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006357703357934952, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12479226663708687 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0068312399089336395, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14262231439352036 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12292, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25856.25, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.03125, + 0.125, + 0.140625, + 0.34375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013507824391126633, + "final_success": 0.40234375, + "gain_over_fixed_vectorizer_final": 0.40234375, + "late_success": 0.203125, + "learning_gain": 0.203125, + "training_wall_s": 0.20231231674551964 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.03125, + 0.125, + 0.140625, + 0.34375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006460081785917282, + "final_success": 0.40234375, + "late_success": 0.203125, + "learning_gain": 0.203125, + "training_wall_s": 0.15404890477657318 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006399795413017273, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12508311122655869 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.0064074136316776276, + "final_success": 0.40234375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 458.2578125, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24529, + "causal_role_sign_inversion_index": -0.008862245646814898, + "decoder_distance_residual_corr": 0.008903072549980118, + "early_residual_late_activity_change_corr": 0.92069164829789, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.40234375, + "mean_abs_raw_soma_corr": 0.9990173004741096, + "mean_abs_residual_soma_corr": 0.11991301371524118, + "raw_minus_residual_abs_soma_corr": 0.8791042867588684, + "residual_minus_soma_outcome_acc": 0.1296719334983183, + "residual_outcome_decoder_balanced_acc": 0.8775937559489815, + "role_aligned_error_cv_corr": 0.6110992488387482, + "role_aligned_velocity_cv_corr": 0.22427046673738615, + "soma_outcome_decoder_balanced_acc": 0.7479218224506632, + "surrounding_event_decoder_balanced_acc": 0.6910911029239888, + "velocity_minus_error_abs_cv_corr": -0.38682878210136207 + }, + "wall_s": 1.8243730068206787 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t0_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t0_m0.json new file mode 100644 index 0000000..82adb0b --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error_velocity", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p0_eta0p03_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.006400451064109802, + "final_success": 1.0 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0063620880246162415, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12568441405892372 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006796989589929581, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14286509528756142 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 9502, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25681.875, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.125, + 0.359375, + 0.640625, + 0.9375, + 0.984375, + 0.96875, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.01346665620803833, + "final_success": 1.0, + "gain_over_fixed_vectorizer_final": 1.0, + "late_success": 0.9895833333333334, + "learning_gain": 0.9895833333333334, + "training_wall_s": 0.19994772970676422 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.125, + 0.359375, + 0.640625, + 0.9375, + 0.984375, + 0.96875, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006394829601049423, + "final_success": 1.0, + "late_success": 0.9895833333333334, + "learning_gain": 0.9895833333333334, + "training_wall_s": 0.15124044194817543 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0064504072070121765, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1257457099854946 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006362009793519974, + "final_success": 1.0 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 432.68359375, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 18458, + "causal_role_sign_inversion_index": -0.006142993077369978, + "decoder_distance_residual_corr": 0.2402017439967244, + "early_residual_late_activity_change_corr": 0.9334692289466362, + "evaluation_episodes": 256, + "evaluation_success_fraction": 1.0, + "mean_abs_raw_soma_corr": 0.9992589307734457, + "mean_abs_residual_soma_corr": 0.12446825336865082, + "raw_minus_residual_abs_soma_corr": 0.8747906774047949, + "residual_minus_soma_outcome_acc": 0.0, + "residual_outcome_decoder_balanced_acc": 0.5, + "role_aligned_error_cv_corr": 0.7133651088999409, + "role_aligned_velocity_cv_corr": 0.13450332588570119, + "soma_outcome_decoder_balanced_acc": 0.5, + "surrounding_event_decoder_balanced_acc": 0.6296902348864057, + "velocity_minus_error_abs_cv_corr": -0.5788617830142397 + }, + "wall_s": 1.5230983793735504 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t1_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t1_m0.json new file mode 100644 index 0000000..5ae8417 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error_velocity", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p0_eta0p03_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.0063506439328193665, + "final_success": 0.99609375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006313655525445938, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12519824132323265 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006946448236703873, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14218781888484955 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 9226, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25664.625, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.265625, + 0.59375, + 0.703125, + 0.953125, + 0.96875, + 0.984375, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013822562992572784, + "final_success": 0.99609375, + "gain_over_fixed_vectorizer_final": 0.99609375, + "late_success": 0.9947916666666666, + "learning_gain": 0.9947916666666666, + "training_wall_s": 0.20022496581077576 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.265625, + 0.59375, + 0.703125, + 0.953125, + 0.96875, + 0.984375, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006692767143249512, + "final_success": 0.99609375, + "late_success": 0.9947916666666666, + "learning_gain": 0.9947916666666666, + "training_wall_s": 0.14937333390116692 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006394337862730026, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12529880180954933 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.0063323527574539185, + "final_success": 0.99609375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 430.640625, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 17805, + "causal_role_sign_inversion_index": -0.006959803702230309, + "decoder_distance_residual_corr": 0.2701095063149012, + "early_residual_late_activity_change_corr": 0.9355533920872674, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.99609375, + "mean_abs_raw_soma_corr": 0.9992073302924226, + "mean_abs_residual_soma_corr": 0.12656676277229884, + "raw_minus_residual_abs_soma_corr": 0.8726405675201238, + "residual_minus_soma_outcome_acc": -0.0019607843137254832, + "residual_outcome_decoder_balanced_acc": 0.4980392156862745, + "role_aligned_error_cv_corr": 0.6964950406882578, + "role_aligned_velocity_cv_corr": 0.1456255045993032, + "soma_outcome_decoder_balanced_acc": 0.5, + "surrounding_event_decoder_balanced_acc": 0.6228071558714278, + "velocity_minus_error_abs_cv_corr": -0.5508695360889546 + }, + "wall_s": 1.409028884023428 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t2_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t2_m0.json new file mode 100644 index 0000000..e8caabf --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error_velocity", + "kappa": 0.0, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p0_eta0p03_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.007011532783508301, + "final_success": 0.99609375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.00702957808971405, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.136286448687315 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007391933351755142, + "final_success": 0.0, + "late_success": 0.005208333333333333, + "learning_gain": 0.005208333333333333, + "training_wall_s": 0.15550277009606361 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 9036, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25652.75, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.109375, + 0.28125, + 0.5625, + 0.765625, + 0.953125, + 0.984375, + 1.0, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.014254752546548843, + "final_success": 0.99609375, + "gain_over_fixed_vectorizer_final": 0.99609375, + "late_success": 1.0, + "learning_gain": 1.0, + "training_wall_s": 0.20594780892133713 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.109375, + 0.28125, + 0.5625, + 0.765625, + 0.953125, + 0.984375, + 1.0, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006980877369642258, + "final_success": 0.99609375, + "late_success": 1.0, + "learning_gain": 1.0, + "training_wall_s": 0.1640157848596573 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006980404257774353, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13694464787840843 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006960630416870117, + "final_success": 0.99609375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.0, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 429.59765625, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 17475, + "causal_role_sign_inversion_index": -0.007109588457054589, + "decoder_distance_residual_corr": 0.2831837713315748, + "early_residual_late_activity_change_corr": 0.9410941696893125, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.99609375, + "mean_abs_raw_soma_corr": 0.9992374187316144, + "mean_abs_residual_soma_corr": 0.12597111914673131, + "raw_minus_residual_abs_soma_corr": 0.8732662995848831, + "residual_minus_soma_outcome_acc": -0.009803921568627472, + "residual_outcome_decoder_balanced_acc": 0.49019607843137253, + "role_aligned_error_cv_corr": 0.7116792761740364, + "role_aligned_velocity_cv_corr": 0.15040963400276214, + "soma_outcome_decoder_balanced_acc": 0.5, + "surrounding_event_decoder_balanced_acc": 0.6339148191728725, + "velocity_minus_error_abs_cv_corr": -0.5612696421712742 + }, + "wall_s": 1.4930438101291656 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t0_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t0_m0.json new file mode 100644 index 0000000..12975e3 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error_velocity", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p1_eta0p01_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.01953125, + "evaluation_wall_s": 0.007043611258268356, + "final_success": 0.19921875 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006851181387901306, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1369473822414875 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007549561560153961, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14320910722017288 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12422, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25864.375, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.109375, + 0.3125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.014395080506801605, + "final_success": 0.21875, + "gain_over_fixed_vectorizer_final": 0.21875, + "late_success": 0.14583333333333334, + "learning_gain": 0.14583333333333334, + "training_wall_s": 0.2060721330344677 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.109375, + 0.34375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006937149912118912, + "final_success": 0.2578125, + "late_success": 0.15625, + "learning_gain": 0.15625, + "training_wall_s": 0.1539219170808792 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007046587765216827, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13443706184625626 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006838671863079071, + "final_success": 0.21875 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 465.16796875, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24810, + "causal_role_sign_inversion_index": -0.007758845209598214, + "decoder_distance_residual_corr": -0.046757830564177316, + "early_residual_late_activity_change_corr": 0.8877927198874914, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.21875, + "mean_abs_raw_soma_corr": 0.9991023849286567, + "mean_abs_residual_soma_corr": 0.12025916308067822, + "raw_minus_residual_abs_soma_corr": 0.8788432218479785, + "residual_minus_soma_outcome_acc": 0.08250000000000002, + "residual_outcome_decoder_balanced_acc": 0.9110714285714285, + "role_aligned_error_cv_corr": 0.5594523770594855, + "role_aligned_velocity_cv_corr": 0.21503550650920192, + "soma_outcome_decoder_balanced_acc": 0.8285714285714285, + "surrounding_event_decoder_balanced_acc": 0.7227741619206323, + "velocity_minus_error_abs_cv_corr": -0.3444168705502836 + }, + "wall_s": 1.9601136557757854 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t1_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t1_m0.json new file mode 100644 index 0000000..895bf50 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error_velocity", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p1_eta0p01_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0234375, + "evaluation_wall_s": 0.0063621290028095245, + "final_success": 0.2734375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006364382803440094, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13385027647018433 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006696052849292755, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14331360906362534 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12382, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25861.875, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.03125, + 0.03125, + 0.078125, + 0.125, + 0.3125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013725094497203827, + "final_success": 0.296875, + "gain_over_fixed_vectorizer_final": 0.296875, + "late_success": 0.171875, + "learning_gain": 0.171875, + "training_wall_s": 0.20390445366501808 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.03125, + 0.046875, + 0.078125, + 0.140625, + 0.328125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006363272666931152, + "final_success": 0.34375, + "late_success": 0.18229166666666666, + "learning_gain": 0.18229166666666666, + "training_wall_s": 0.16501661390066147 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006387878209352493, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1342468038201332 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.0063500069081783295, + "final_success": 0.296875 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 458.6484375, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24713, + "causal_role_sign_inversion_index": -0.008605818715404931, + "decoder_distance_residual_corr": -0.04225435280696725, + "early_residual_late_activity_change_corr": 0.8992166684313505, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.296875, + "mean_abs_raw_soma_corr": 0.9990344410752721, + "mean_abs_residual_soma_corr": 0.12127518791283871, + "raw_minus_residual_abs_soma_corr": 0.8777592531624333, + "residual_minus_soma_outcome_acc": 0.10979532163742689, + "residual_outcome_decoder_balanced_acc": 0.9138888888888889, + "role_aligned_error_cv_corr": 0.5678919306613218, + "role_aligned_velocity_cv_corr": 0.21437071089141166, + "soma_outcome_decoder_balanced_acc": 0.804093567251462, + "surrounding_event_decoder_balanced_acc": 0.6881166732460898, + "velocity_minus_error_abs_cv_corr": -0.35352121976991013 + }, + "wall_s": 1.8743064366281033 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t2_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t2_m0.json new file mode 100644 index 0000000..8cef366 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error_velocity", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p1_eta0p01_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0234375, + "evaluation_wall_s": 0.006839442998170853, + "final_success": 0.32421875 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006811194121837616, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13399991765618324 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.00724228098988533, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1417798548936844 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12322, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25858.125, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.03125, + 0.109375, + 0.125, + 0.3125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013575725257396698, + "final_success": 0.34765625, + "gain_over_fixed_vectorizer_final": 0.34765625, + "late_success": 0.18229166666666666, + "learning_gain": 0.18229166666666666, + "training_wall_s": 0.20463567599654198 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.03125, + 0.125, + 0.140625, + 0.34375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006874799728393555, + "final_success": 0.40234375, + "late_success": 0.203125, + "learning_gain": 0.203125, + "training_wall_s": 0.15967321395874023 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0067849308252334595, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1330891214311123 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.0068716853857040405, + "final_success": 0.34765625 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 452.53125, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24588, + "causal_role_sign_inversion_index": -0.008691079596347698, + "decoder_distance_residual_corr": -0.008483104455547088, + "early_residual_late_activity_change_corr": 0.9157601840342005, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.34765625, + "mean_abs_raw_soma_corr": 0.9990531130456036, + "mean_abs_residual_soma_corr": 0.12084188226660464, + "raw_minus_residual_abs_soma_corr": 0.878211230778999, + "residual_minus_soma_outcome_acc": 0.11084572428177353, + "residual_outcome_decoder_balanced_acc": 0.8839736257821436, + "role_aligned_error_cv_corr": 0.5904202744967263, + "role_aligned_velocity_cv_corr": 0.21817995999307133, + "soma_outcome_decoder_balanced_acc": 0.77312790150037, + "surrounding_event_decoder_balanced_acc": 0.6885768186618957, + "velocity_minus_error_abs_cv_corr": -0.37224031450365497 + }, + "wall_s": 1.8288945853710175 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t0_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t0_m0.json new file mode 100644 index 0000000..eda16f5 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error_velocity", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p1_eta0p03_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.006992604583501816, + "final_success": 1.0 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0069955140352249146, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13849567621946335 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0075439102947711945, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.15397954359650612 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 9564, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25685.75, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.109375, + 0.359375, + 0.640625, + 0.890625, + 0.984375, + 0.96875, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.014785513281822205, + "final_success": 1.0, + "gain_over_fixed_vectorizer_final": 1.0, + "late_success": 0.9895833333333334, + "learning_gain": 0.9895833333333334, + "training_wall_s": 0.20387889072299004 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.125, + 0.359375, + 0.640625, + 0.9375, + 0.984375, + 0.96875, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007059283554553986, + "final_success": 1.0, + "late_success": 0.9895833333333334, + "learning_gain": 0.9895833333333334, + "training_wall_s": 0.16868066415190697 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007101424038410187, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14773177728056908 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.00718211755156517, + "final_success": 1.0 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 433.55859375, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 18590, + "causal_role_sign_inversion_index": -0.006139542578201915, + "decoder_distance_residual_corr": 0.22450390475663165, + "early_residual_late_activity_change_corr": 0.9334353987614218, + "evaluation_episodes": 256, + "evaluation_success_fraction": 1.0, + "mean_abs_raw_soma_corr": 0.9992703157890134, + "mean_abs_residual_soma_corr": 0.12592115789131259, + "raw_minus_residual_abs_soma_corr": 0.8733491578977008, + "residual_minus_soma_outcome_acc": 0.0, + "residual_outcome_decoder_balanced_acc": 0.5, + "role_aligned_error_cv_corr": 0.7009350042728064, + "role_aligned_velocity_cv_corr": 0.1250699937600068, + "soma_outcome_decoder_balanced_acc": 0.5, + "surrounding_event_decoder_balanced_acc": 0.6225737005901573, + "velocity_minus_error_abs_cv_corr": -0.5758650105127996 + }, + "wall_s": 1.5047636926174164 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t1_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t1_m0.json new file mode 100644 index 0000000..e447198 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error_velocity", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p1_eta0p03_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.006329528987407684, + "final_success": 0.99609375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006439197808504105, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1299189180135727 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006652858108282089, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1456008143723011 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 9290, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25668.625, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.234375, + 0.5625, + 0.703125, + 0.953125, + 0.96875, + 0.984375, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013262268155813217, + "final_success": 0.99609375, + "gain_over_fixed_vectorizer_final": 0.99609375, + "late_success": 0.9947916666666666, + "learning_gain": 0.9947916666666666, + "training_wall_s": 0.19930467009544373 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.265625, + 0.59375, + 0.703125, + 0.953125, + 0.96875, + 0.984375, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006318964064121246, + "final_success": 0.99609375, + "late_success": 0.9947916666666666, + "learning_gain": 0.9947916666666666, + "training_wall_s": 0.16212405264377594 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006250225007534027, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13105998188257217 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006280824542045593, + "final_success": 0.99609375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 436.0625, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 17949, + "causal_role_sign_inversion_index": -0.006889803927151911, + "decoder_distance_residual_corr": 0.25750191688759305, + "early_residual_late_activity_change_corr": 0.9331268345500653, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.99609375, + "mean_abs_raw_soma_corr": 0.9992196242862024, + "mean_abs_residual_soma_corr": 0.12843288051454596, + "raw_minus_residual_abs_soma_corr": 0.8707867437716564, + "residual_minus_soma_outcome_acc": 0.0, + "residual_outcome_decoder_balanced_acc": 0.5, + "role_aligned_error_cv_corr": 0.6841129309153174, + "role_aligned_velocity_cv_corr": 0.1347889964992377, + "soma_outcome_decoder_balanced_acc": 0.5, + "surrounding_event_decoder_balanced_acc": 0.6192532946063222, + "velocity_minus_error_abs_cv_corr": -0.5493239344160797 + }, + "wall_s": 1.4199837744235992 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t2_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t2_m0.json new file mode 100644 index 0000000..cb67281 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error_velocity", + "kappa": 0.1, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p1_eta0p03_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.006869271397590637, + "final_success": 0.99609375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006893862038850784, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13531124964356422 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0075135380029678345, + "final_success": 0.0, + "late_success": 0.005208333333333333, + "learning_gain": 0.005208333333333333, + "training_wall_s": 0.14413100853562355 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 9116, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25657.75, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.109375, + 0.234375, + 0.5625, + 0.71875, + 0.9375, + 0.96875, + 1.0, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.01448969915509224, + "final_success": 0.99609375, + "gain_over_fixed_vectorizer_final": 0.99609375, + "late_success": 1.0, + "learning_gain": 1.0, + "training_wall_s": 0.20029856637120247 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.109375, + 0.28125, + 0.5625, + 0.765625, + 0.953125, + 0.984375, + 1.0, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.00699218362569809, + "final_success": 0.99609375, + "late_success": 1.0, + "learning_gain": 1.0, + "training_wall_s": 0.16103898733854294 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006982915103435516, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13519801944494247 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006851688027381897, + "final_success": 0.99609375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.1, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 430.03515625, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 17667, + "causal_role_sign_inversion_index": -0.006905458915154327, + "decoder_distance_residual_corr": 0.2744983632160322, + "early_residual_late_activity_change_corr": 0.9410733698033948, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.99609375, + "mean_abs_raw_soma_corr": 0.9992476170584551, + "mean_abs_residual_soma_corr": 0.12799383011952623, + "raw_minus_residual_abs_soma_corr": 0.8712537869389289, + "residual_minus_soma_outcome_acc": -0.011764705882352955, + "residual_outcome_decoder_balanced_acc": 0.48823529411764705, + "role_aligned_error_cv_corr": 0.6999000336489246, + "role_aligned_velocity_cv_corr": 0.1387715561751899, + "soma_outcome_decoder_balanced_acc": 0.5, + "surrounding_event_decoder_balanced_acc": 0.6228296906972537, + "velocity_minus_error_abs_cv_corr": -0.5611284774737347 + }, + "wall_s": 1.454554207623005 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t0_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t0_m0.json new file mode 100644 index 0000000..50c82a4 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error_velocity", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p3_eta0p01_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.02734375, + "evaluation_wall_s": 0.007088668644428253, + "final_success": 0.12890625 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007156647741794586, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1513182632625103 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007629722356796265, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14410367980599403 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12460, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25866.75, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.09375, + 0.234375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.014401573687791824, + "final_success": 0.15625, + "gain_over_fixed_vectorizer_final": 0.15625, + "late_success": 0.11458333333333333, + "learning_gain": 0.11458333333333333, + "training_wall_s": 0.21120928600430489 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.109375, + 0.34375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007242176681756973, + "final_success": 0.2578125, + "late_success": 0.15625, + "learning_gain": 0.15625, + "training_wall_s": 0.17544353753328323 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.00708344578742981, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1420268565416336 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.007108330726623535, + "final_success": 0.15625 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 453.34765625, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24898, + "causal_role_sign_inversion_index": -0.00762898483692713, + "decoder_distance_residual_corr": -0.06117547409151871, + "early_residual_late_activity_change_corr": 0.8756076080647975, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.15625, + "mean_abs_raw_soma_corr": 0.9991560399524143, + "mean_abs_residual_soma_corr": 0.1255784612045066, + "raw_minus_residual_abs_soma_corr": 0.8735775787479076, + "residual_minus_soma_outcome_acc": 0.09212962962962956, + "residual_outcome_decoder_balanced_acc": 0.9041666666666667, + "role_aligned_error_cv_corr": 0.5174903311386221, + "role_aligned_velocity_cv_corr": 0.19779440961234948, + "soma_outcome_decoder_balanced_acc": 0.8120370370370371, + "surrounding_event_decoder_balanced_acc": 0.72333746192052, + "velocity_minus_error_abs_cv_corr": -0.3196959215262726 + }, + "wall_s": 1.9328845329582691 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t1_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t1_m0.json new file mode 100644 index 0000000..ccc83ea --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error_velocity", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p3_eta0p01_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.04296875, + "evaluation_wall_s": 0.006271157413721085, + "final_success": 0.19921875 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006298404186964035, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12539008259773254 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006900351494550705, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14197174087166786 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12428, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25864.75, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.015625, + 0.046875, + 0.125, + 0.21875 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013385161757469177, + "final_success": 0.2421875, + "gain_over_fixed_vectorizer_final": 0.2421875, + "late_success": 0.13020833333333334, + "learning_gain": 0.13020833333333334, + "training_wall_s": 0.20461491122841835 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.03125, + 0.046875, + 0.078125, + 0.140625, + 0.328125 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006414361298084259, + "final_success": 0.34375, + "late_success": 0.18229166666666666, + "learning_gain": 0.18229166666666666, + "training_wall_s": 0.15508219227194786 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0063520073890686035, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12526404485106468 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006268881261348724, + "final_success": 0.2421875 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 465.5546875, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24819, + "causal_role_sign_inversion_index": -0.008518311615923924, + "decoder_distance_residual_corr": -0.0697537334176312, + "early_residual_late_activity_change_corr": 0.885794067958317, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.2421875, + "mean_abs_raw_soma_corr": 0.9990981332161842, + "mean_abs_residual_soma_corr": 0.1255231327428768, + "raw_minus_residual_abs_soma_corr": 0.8735750004733074, + "residual_minus_soma_outcome_acc": 0.1299467908214167, + "residual_outcome_decoder_balanced_acc": 0.9252577319587629, + "role_aligned_error_cv_corr": 0.5245667246214586, + "role_aligned_velocity_cv_corr": 0.19871863555138106, + "soma_outcome_decoder_balanced_acc": 0.7953109411373462, + "surrounding_event_decoder_balanced_acc": 0.6976142823070874, + "velocity_minus_error_abs_cv_corr": -0.32584808907007756 + }, + "wall_s": 1.800193302333355 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t2_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t2_m0.json new file mode 100644 index 0000000..78fb893 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.01, + "feedback": "error_velocity", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p3_eta0p01_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.05078125, + "evaluation_wall_s": 0.007010702043771744, + "final_success": 0.21875 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006950356066226959, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1381758265197277 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007407184690237045, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.15614957362413406 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 12410, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25863.625, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.0625, + 0.078125, + 0.234375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.014417063444852829, + "final_success": 0.26953125, + "gain_over_fixed_vectorizer_final": 0.26953125, + "late_success": 0.125, + "learning_gain": 0.125, + "training_wall_s": 0.21103275939822197 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.03125, + 0.125, + 0.140625, + 0.34375 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.007080644369125366, + "final_success": 0.40234375, + "late_success": 0.203125, + "learning_gain": 0.203125, + "training_wall_s": 0.17015830427408218 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006973445415496826, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13772671297192574 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006973259150981903, + "final_success": 0.26953125 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.01, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 453.21875, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 24780, + "causal_role_sign_inversion_index": -0.008526297052818842, + "decoder_distance_residual_corr": -0.036049771169280535, + "early_residual_late_activity_change_corr": 0.9040668552528915, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.26953125, + "mean_abs_raw_soma_corr": 0.9991130308901424, + "mean_abs_residual_soma_corr": 0.12486343765516342, + "raw_minus_residual_abs_soma_corr": 0.874249593234979, + "residual_minus_soma_outcome_acc": 0.13275982329690772, + "residual_outcome_decoder_balanced_acc": 0.8938231419049834, + "role_aligned_error_cv_corr": 0.5493085497262201, + "role_aligned_velocity_cv_corr": 0.201005240688992, + "soma_outcome_decoder_balanced_acc": 0.7610633186080756, + "surrounding_event_decoder_balanced_acc": 0.696906923439218, + "velocity_minus_error_abs_cv_corr": -0.3483033090372281 + }, + "wall_s": 1.8161988146603107 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t0_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t0_m0.json new file mode 100644 index 0000000..d751548 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t0_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error_velocity", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p3_eta0p03_t0_m0", + "task_seed": 0 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.00629371777176857, + "final_success": 0.9921875 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006284419447183609, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12433725222945213 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006661858409643173, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14106907695531845 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 9742, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25696.875, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.09375, + 0.265625, + 0.59375, + 0.8125, + 0.96875, + 0.953125, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.013221479952335358, + "final_success": 0.9921875, + "gain_over_fixed_vectorizer_final": 0.9921875, + "late_success": 0.984375, + "learning_gain": 0.984375, + "training_wall_s": 0.19931010529398918 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.015625, + 0.125, + 0.359375, + 0.640625, + 0.9375, + 0.984375, + 0.96875, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.0063039883971214294, + "final_success": 1.0, + "late_success": 0.9895833333333334, + "learning_gain": 0.9895833333333334, + "training_wall_s": 0.14945119991898537 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006364766508340836, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.12456006556749344 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006289340555667877, + "final_success": 0.9921875 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 438.76171875, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100000, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 0 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 18969, + "causal_role_sign_inversion_index": -0.006057431612911789, + "decoder_distance_residual_corr": 0.20291729614588686, + "early_residual_late_activity_change_corr": 0.9308717101459166, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.9921875, + "mean_abs_raw_soma_corr": 0.9992892779628525, + "mean_abs_residual_soma_corr": 0.13190591570992197, + "raw_minus_residual_abs_soma_corr": 0.8673833622529306, + "residual_minus_soma_outcome_acc": 0.4921259842519685, + "residual_outcome_decoder_balanced_acc": 0.9881889763779528, + "role_aligned_error_cv_corr": 0.6753823390505134, + "role_aligned_velocity_cv_corr": 0.1027264997822552, + "soma_outcome_decoder_balanced_acc": 0.49606299212598426, + "surrounding_event_decoder_balanced_acc": 0.6084959675453069, + "velocity_minus_error_abs_cv_corr": -0.5726558392682581 + }, + "wall_s": 1.4734766706824303 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t1_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t1_m0.json new file mode 100644 index 0000000..f7c8a07 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t1_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error_velocity", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p3_eta0p03_t1_m0", + "task_seed": 1 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.0, + "evaluation_wall_s": 0.006848052144050598, + "final_success": 0.99609375 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.00685211643576622, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13389190658926964 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.00729263573884964, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.14569858834147453 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 9406, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25675.875, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.1875, + 0.5, + 0.671875, + 0.890625, + 0.953125, + 0.984375, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.014725316315889359, + "final_success": 0.99609375, + "gain_over_fixed_vectorizer_final": 0.99609375, + "late_success": 0.9947916666666666, + "learning_gain": 0.9947916666666666, + "training_wall_s": 0.20651720091700554 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.265625, + 0.59375, + 0.703125, + 0.953125, + 0.96875, + 0.984375, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006857927888631821, + "final_success": 0.99609375, + "late_success": 0.9947916666666666, + "learning_gain": 0.9947916666666666, + "training_wall_s": 0.1626637913286686 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006808057427406311, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1332736611366272 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006842322647571564, + "final_success": 0.99609375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 431.8828125, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100001, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 1 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 18233, + "causal_role_sign_inversion_index": -0.006836529479807937, + "decoder_distance_residual_corr": 0.22623503412120596, + "early_residual_late_activity_change_corr": 0.9266998543555044, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.99609375, + "mean_abs_raw_soma_corr": 0.9992435771416621, + "mean_abs_residual_soma_corr": 0.13359308599715133, + "raw_minus_residual_abs_soma_corr": 0.8656504911445108, + "residual_minus_soma_outcome_acc": 0.0, + "residual_outcome_decoder_balanced_acc": 0.5, + "role_aligned_error_cv_corr": 0.6553994699440479, + "role_aligned_velocity_cv_corr": 0.11491661003778803, + "soma_outcome_decoder_balanced_acc": 0.5, + "surrounding_event_decoder_balanced_acc": 0.6082382886952071, + "velocity_minus_error_abs_cv_corr": -0.5404828599062599 + }, + "wall_s": 1.5314385145902634 +} diff --git a/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t2_m0.json b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t2_m0.json new file mode 100644 index 0000000..b6d35a7 --- /dev/null +++ b/results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t2_m0.json @@ -0,0 +1,245 @@ +{ + "args": { + "eta": 0.03, + "feedback": "error_velocity", + "kappa": 0.3, + "model_seed": 0, + "outdir": "results/bci_dev", + "tag": "bci_dev_v1_fberrvel_k0p3_eta0p03_t2_m0", + "task_seed": 2 + }, + "conditions": { + "acute_online_lesion_after_intact": { + "drop_from_intact": 0.00390625, + "evaluation_wall_s": 0.006332863122224808, + "final_success": 0.9921875 + }, + "both_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006333433091640472, + "final_success": 0.0, + "fraction_of_intact_learning_gain": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.1346125267446041 + }, + "fixed_vectorizer": { + "cost": { + "causal_scalar_reward_observations": 0, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25088.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 0 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.015625, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006835956126451492, + "final_success": 0.0, + "late_success": 0.005208333333333333, + "learning_gain": 0.005208333333333333, + "training_wall_s": 0.15525603666901588 + }, + "intact": { + "cost": { + "causal_scalar_reward_observations": 9298, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25669.125, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.09375, + 0.1875, + 0.53125, + 0.578125, + 0.90625, + 0.96875, + 1.0, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.014085140079259872, + "final_success": 0.99609375, + "gain_over_fixed_vectorizer_final": 0.99609375, + "late_success": 1.0, + "learning_gain": 1.0, + "training_wall_s": 0.20123415812849998 + }, + "online_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.03125, + 0.109375, + 0.28125, + 0.5625, + 0.765625, + 0.953125, + 0.984375, + 1.0, + 1.0, + 1.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006440259516239166, + "final_success": 0.99609375, + "late_success": 1.0, + "learning_gain": 1.0, + "training_wall_s": 0.15970947965979576 + }, + "plasticity_lesion": { + "cost": { + "causal_scalar_reward_observations": 12544, + "forward_equivalent_definition": "ordinary full state step plus each perturbed cursor/loss readout at 1/context_dim of a context-to-soma state step", + "forward_equivalent_episode_steps": 25872.0, + "ordinary_state_episode_steps": 25088, + "perturbation_events": 98 + }, + "daily_success": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "early_success": 0.0, + "evaluation_wall_s": 0.006396383047103882, + "final_success": 0.0, + "late_success": 0.0, + "learning_gain": 0.0, + "training_wall_s": 0.13402202725410461 + }, + "sham_after_intact": { + "change_from_intact": 0.0, + "evaluation_wall_s": 0.006328493356704712, + "final_success": 0.99609375 + } + }, + "config": { + "context_ar": 0.8, + "context_dim": 16, + "coupling_scale": 1.0, + "days": 14, + "episodes_per_day": 64, + "feedback": "error_velocity", + "forward_eta": 0.03, + "inertia": 0.65, + "kappa": 0.3, + "n_background": 30, + "n_minus": 5, + "n_plus": 5, + "perturb_every": 4, + "perturb_sigma": 0.03, + "predictor_eta": 0.2, + "process_noise": 0.12, + "steps_per_episode": 28, + "target": 0.8, + "vectorizer_eta": 0.03 + }, + "finite": true, + "hardware": { + "device": "cpu", + "platform": "Linux-5.15.0-161-generic-x86_64-with-glibc2.35", + "processor": "x86_64", + "threads": 1, + "torch_version": "2.3.1+cu118" + }, + "peak_rss_mib": 431.28515625, + "protocol": { + "confirmation_seeds_touched": false, + "evaluation_plasticity": false, + "evaluation_task_seed": 100002, + "name": "oral_b_continuous_bci_development_v1", + "paired_environment_across_conditions": true, + "selection_split": "development", + "training_task_seed": 2 + }, + "provenance": { + "git_commit": "b4458cdb3533f7b342f3ff3c8e3582962181d09f", + "git_dirty": false, + "oral_b_protocol_sha256": "ce58956da6de01a174f845525c8f80a4ef7ad8b94453c2f8e76a8001c56d1b89", + "runner_and_protocol_tracked": true + }, + "schema_version": 1, + "signatures": { + "active_training_events": 18036, + "causal_role_sign_inversion_index": -0.006697710635998395, + "decoder_distance_residual_corr": 0.2526024649682273, + "early_residual_late_activity_change_corr": 0.9373464733073118, + "evaluation_episodes": 256, + "evaluation_success_fraction": 0.99609375, + "mean_abs_raw_soma_corr": 0.9992697520050593, + "mean_abs_residual_soma_corr": 0.13451463566484606, + "raw_minus_residual_abs_soma_corr": 0.8647551163402132, + "residual_minus_soma_outcome_acc": -0.009803921568627472, + "residual_outcome_decoder_balanced_acc": 0.49019607843137253, + "role_aligned_error_cv_corr": 0.6715398218142011, + "role_aligned_velocity_cv_corr": 0.11638738203997051, + "soma_outcome_decoder_balanced_acc": 0.5, + "surrounding_event_decoder_balanced_acc": 0.6149042391680266, + "velocity_minus_error_abs_cv_corr": -0.5551524397742307 + }, + "wall_s": 1.4669511802494526 +} diff --git a/results/bci_dev_selection.json b/results/bci_dev_selection.json new file mode 100644 index 0000000..f78d16f --- /dev/null +++ b/results/bci_dev_selection.json @@ -0,0 +1,816 @@ +{ + "candidates": [ + { + "eligible": false, + "eta": 0.01, + "feedback": "error", + "final_success": [ + 0.15625, + 0.24609375, + 0.2578125 + ], + "fixed_final_gap": [ + 0.15625, + 0.24609375, + 0.2578125 + ], + "kappa": 0.0, + "learning_gain": [ + 0.109375, + 0.11458333333333333, + 0.13541666666666666 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": false, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": true + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": true + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": true + } + ], + "sign_inversion": [ + -0.002889420693915609, + -0.0034216490242629308, + -0.0037756110210441854 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.15625, + "worst_sign_inversion": -0.0037756110210441854 + }, + { + "eligible": false, + "eta": 0.03, + "feedback": "error", + "final_success": [ + 0.98828125, + 0.9921875, + 0.99609375 + ], + "fixed_final_gap": [ + 0.98828125, + 0.9921875, + 0.99609375 + ], + "kappa": 0.0, + "learning_gain": [ + 0.9322916666666666, + 0.96875, + 0.984375 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + } + ], + "sign_inversion": [ + -0.0011943816763679907, + -0.00148484176075913, + -0.0018247418377719364 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.98828125, + "worst_sign_inversion": -0.0018247418377719364 + }, + { + "eligible": false, + "eta": 0.01, + "feedback": "error", + "final_success": [ + 0.140625, + 0.21484375, + 0.21875 + ], + "fixed_final_gap": [ + 0.140625, + 0.21484375, + 0.21875 + ], + "kappa": 0.1, + "learning_gain": [ + 0.08333333333333333, + 0.08854166666666667, + 0.11979166666666667 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": false, + "learning_gain_10pt": false, + "positive_sign_inversion": false, + "residual_soma_corr": true + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": false, + "positive_sign_inversion": false, + "residual_soma_corr": true + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": true + } + ], + "sign_inversion": [ + -0.0027520933435467784, + -0.0031658220936501905, + -0.003522021519430374 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.140625, + "worst_sign_inversion": -0.003522021519430374 + }, + { + "eligible": false, + "eta": 0.03, + "feedback": "error", + "final_success": [ + 0.98828125, + 0.9921875, + 0.99609375 + ], + "fixed_final_gap": [ + 0.98828125, + 0.9921875, + 0.99609375 + ], + "kappa": 0.1, + "learning_gain": [ + 0.921875, + 0.96875, + 0.984375 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + } + ], + "sign_inversion": [ + -0.0011604788136277272, + -0.0013616858451298147, + -0.0017520959075886383 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.98828125, + "worst_sign_inversion": -0.0017520959075886383 + }, + { + "eligible": false, + "eta": 0.01, + "feedback": "error", + "final_success": [ + 0.1015625, + 0.1875, + 0.171875 + ], + "fixed_final_gap": [ + 0.1015625, + 0.1875, + 0.171875 + ], + "kappa": 0.3, + "learning_gain": [ + 0.06770833333333333, + 0.078125, + 0.09895833333333333 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": false, + "learning_gain_10pt": false, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": false, + "learning_gain_10pt": false, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": false, + "learning_gain_10pt": false, + "positive_sign_inversion": false, + "residual_soma_corr": false + } + ], + "sign_inversion": [ + -0.0026150713256167524, + -0.0029546118827365186, + -0.0032190460550888048 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.1015625, + "worst_sign_inversion": -0.0032190460550888048 + }, + { + "eligible": false, + "eta": 0.03, + "feedback": "error", + "final_success": [ + 0.98828125, + 0.984375, + 0.99609375 + ], + "fixed_final_gap": [ + 0.98828125, + 0.984375, + 0.99609375 + ], + "kappa": 0.3, + "learning_gain": [ + 0.9114583333333334, + 0.96875, + 0.9791666666666666 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + } + ], + "sign_inversion": [ + -0.0010676039791819303, + -0.0014496980554480898, + -0.0016557727802265117 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.984375, + "worst_sign_inversion": -0.0016557727802265117 + }, + { + "eligible": false, + "eta": 0.01, + "feedback": "error_velocity", + "final_success": [ + 0.2578125, + 0.34375, + 0.40234375 + ], + "fixed_final_gap": [ + 0.2578125, + 0.34375, + 0.40234375 + ], + "kappa": 0.0, + "learning_gain": [ + 0.15625, + 0.18229166666666666, + 0.203125 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + } + ], + "sign_inversion": [ + -0.007845451341592362, + -0.008684546650026242, + -0.008862245646814898 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.2578125, + "worst_sign_inversion": -0.008862245646814898 + }, + { + "eligible": false, + "eta": 0.03, + "feedback": "error_velocity", + "final_success": [ + 1.0, + 0.99609375, + 0.99609375 + ], + "fixed_final_gap": [ + 1.0, + 0.99609375, + 0.99609375 + ], + "kappa": 0.0, + "learning_gain": [ + 0.9895833333333334, + 0.9947916666666666, + 1.0 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + } + ], + "sign_inversion": [ + -0.006142993077369978, + -0.006959803702230309, + -0.007109588457054589 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.99609375, + "worst_sign_inversion": -0.007109588457054589 + }, + { + "eligible": false, + "eta": 0.01, + "feedback": "error_velocity", + "final_success": [ + 0.21875, + 0.296875, + 0.34765625 + ], + "fixed_final_gap": [ + 0.21875, + 0.296875, + 0.34765625 + ], + "kappa": 0.1, + "learning_gain": [ + 0.14583333333333334, + 0.171875, + 0.18229166666666666 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + } + ], + "sign_inversion": [ + -0.007758845209598214, + -0.008605818715404931, + -0.008691079596347698 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.21875, + "worst_sign_inversion": -0.008691079596347698 + }, + { + "eligible": false, + "eta": 0.03, + "feedback": "error_velocity", + "final_success": [ + 1.0, + 0.99609375, + 0.99609375 + ], + "fixed_final_gap": [ + 1.0, + 0.99609375, + 0.99609375 + ], + "kappa": 0.1, + "learning_gain": [ + 0.9895833333333334, + 0.9947916666666666, + 1.0 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + } + ], + "sign_inversion": [ + -0.006139542578201915, + -0.006889803927151911, + -0.006905458915154327 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.99609375, + "worst_sign_inversion": -0.006905458915154327 + }, + { + "eligible": false, + "eta": 0.01, + "feedback": "error_velocity", + "final_success": [ + 0.15625, + 0.2421875, + 0.26953125 + ], + "fixed_final_gap": [ + 0.15625, + 0.2421875, + 0.26953125 + ], + "kappa": 0.3, + "learning_gain": [ + 0.11458333333333333, + 0.13020833333333334, + 0.125 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": false, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + } + ], + "sign_inversion": [ + -0.00762898483692713, + -0.008518311615923924, + -0.008526297052818842 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.15625, + "worst_sign_inversion": -0.008526297052818842 + }, + { + "eligible": false, + "eta": 0.03, + "feedback": "error_velocity", + "final_success": [ + 0.9921875, + 0.99609375, + 0.99609375 + ], + "fixed_final_gap": [ + 0.9921875, + 0.99609375, + 0.99609375 + ], + "kappa": 0.3, + "learning_gain": [ + 0.984375, + 0.9947916666666666, + 1.0 + ], + "seed_checks": [ + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + }, + { + "both_lesion_removes_half_gain": true, + "fixed_vectorizer_gap_20pt": true, + "learning_gain_10pt": true, + "positive_sign_inversion": false, + "residual_soma_corr": false + } + ], + "sign_inversion": [ + -0.006057431612911789, + -0.006836529479807937, + -0.006697710635998395 + ], + "source_commits": [ + "b4458cdb3533f7b342f3ff3c8e3582962181d09f" + ], + "source_paths": [ + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t0_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t1_m0.json", + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t2_m0.json" + ], + "task_seeds": [ + 0, + 1, + 2 + ], + "worst_final_success": 0.9921875, + "worst_sign_inversion": -0.006836529479807937 + } + ], + "complete_grid": true, + "confirmation_seeds_touched": false, + "protocol": "oral_b_continuous_bci_development_v1", + "schema_version": 1, + "selected": null, + "source_sha256": { + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t0_m0.json": "08f6fca72eb065a0438449256afcc1686c5e9ec019ae5388d1cb70939f5c8188", + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t1_m0.json": "6bd21f683f230a8836c143aaf3d0ce1107a9fcb186afb701a61b69666a5d8240", + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p01_t2_m0.json": "27aeb61a86639ef7f9f8878f884e95704fcc5289c3703b03a80860ad385983e4", + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t0_m0.json": "be8d0bf7590e849e53088fb61bf1842d969c2e5c52a79dfee422da0c44d5f4ba", + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t1_m0.json": "5eda652e90de8d69cdff31f107ce80cd07dce3a57d45b666120c22e25c7eae97", + "results/bci_dev/bci_dev_v1_fberr_k0p0_eta0p03_t2_m0.json": "d0d1a4e6ab72d88db761194a9332ba90f808e2b1cd9c363c117f81f717c41281", + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t0_m0.json": "e0ae1fa90723349f7dfb722f32ecd46898ab7dacfd45b045446650790f7dc66b", + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t1_m0.json": "2e07584cb4f35e496f5bec2402e02258c43d779e8fe1eadcd133763a053a434e", + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p01_t2_m0.json": "e5c09a5a5885a3857c4b4368f15d08fa4ee53641f6072562889afbdc95f36e18", + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t0_m0.json": "8630aa8bd1f526346a0db8d91eaa9a43f624578bfc871f2340f87f6310084b6c", + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t1_m0.json": "f4da9454e94c3f5ff9f3ba7e87103a6b26718500e2887bd949d89df40ece51e6", + "results/bci_dev/bci_dev_v1_fberr_k0p1_eta0p03_t2_m0.json": "d99cb3260d9e1c1739ae2e25c8ff48bccd3756a033350ce6ce2c4e4967fd1709", + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t0_m0.json": "5e8b6f7a286905910aa42527e0d5c78838338977d92ed472bee4e623f9671522", + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t1_m0.json": "7f25af029e930231fb26c4bb0437e4eb56d6c67a60481538863f3c3524fb77c3", + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p01_t2_m0.json": "dfa2bdbc82b1b148c1178f4cd5be73a24d55c609a2a08dcee4f0c83247d16e19", + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t0_m0.json": "c07038486bfc6bad7aec39e00097b0eb032a66da6dd4c21529e5fc7a13e662fa", + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t1_m0.json": "d0c7e7721fa0c8546f12f44e99e9863872493d1d79560cd079cd60f255687590", + "results/bci_dev/bci_dev_v1_fberr_k0p3_eta0p03_t2_m0.json": "2726cde0c3f61a61db89857cb32920e2d06e5790b03dec800c1ad9febf5daf52", + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t0_m0.json": "03d02926b58fdd4b357f0fe357a6dce21ae85a882ef29a08b4cca4bcfaa31d15", + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t1_m0.json": "8d112aa00520a5ece733707415d6da62a17eb179e7276abf23098cdf2e0389b8", + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p01_t2_m0.json": "cb152fc537f8a8c1fc21725a7bdc71044b4d277178bff2dbb07d4860440f79d8", + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t0_m0.json": "2e3db4eb3597cd51247e0bc07de88eeb7766a4c51737b40d26126547fb813335", + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t1_m0.json": "381e8245a036f5c0daa7854d46469763235f959cb5cde317df07c02089ccc9aa", + "results/bci_dev/bci_dev_v1_fberrvel_k0p0_eta0p03_t2_m0.json": "65f3e4c19818b35872ffec35bdd463e9499628aee2a2ca84bfe81f59d0a08663", + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t0_m0.json": "ce48e4540a1824140c8aed7e528a30e0fbea3433fec23795bc0265eb87c1b5f8", + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t1_m0.json": "13a4cc7d8af635a6d3e8c9cc8a3c7f2e676dcaabc3565362d80276eb76394131", + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p01_t2_m0.json": "ffa588b2c4e35577828d88ccaf306a82bd47cf2ee14eb7c1d2143980ffc2a7b3", + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t0_m0.json": "8b48edaa2697445578228be8f6437c6410d18820edce813299c2564a87cd3601", + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t1_m0.json": "b3697adbebf512ab8e7d3c6aafc326a951b6f03ac67999303f8ffffc574ac8cc", + "results/bci_dev/bci_dev_v1_fberrvel_k0p1_eta0p03_t2_m0.json": "cbf3dea4a6c929dda8d68b7b8db895bb3873a5a2a5571055141231ce4575b049", + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t0_m0.json": "1b32db2f45e6b9c01c8ceb7d0a1b9a5a4a70672e893db4769c34e472fad372c7", + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t1_m0.json": "7f890b690f9955f8c0c46bd9257c8ba4f7855f057c7e664015ccceb0d0957602", + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p01_t2_m0.json": "80066dfe3a6cc2a7efad3c5c2afaf04324f1550f8bf9bdffeba85b885bb38036", + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t0_m0.json": "525ce4dcf0605c434ab263d7f8a5eacfd7716a2b4194ef5ee417a837bfde23f5", + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t1_m0.json": "8c9e25a2d78d874fb2bcc2ad6cc76f53fc66901245433fe713be7d9aa866a621", + "results/bci_dev/bci_dev_v1_fberrvel_k0p3_eta0p03_t2_m0.json": "51da31a96dbac9d2249f8bf3c2dd144b3dab0f801b494a99396531e48f43dffa" + }, + "status": "failed_no_eligible_candidate" +} |