diff options
| author | YurenHao0426 <blackhao0426@gmail.com> | 2026-06-29 12:15:51 -0500 |
|---|---|---|
| committer | YurenHao0426 <blackhao0426@gmail.com> | 2026-06-29 12:15:51 -0500 |
| commit | a6ec4288a2232988b130b2f00bb2565f81706966 (patch) | |
| tree | 1bb86e7f0b899b823b9e7fdf383e832d30a181e0 /initial_perturb_robustness/smoke_plots | |
Recursive reasoning dynamics: analysis pipeline, paper drafts, toy models
Failure=more-chaotic (task-general under validity labeling) reduces to convergence/completeness
detection; mechanism (transient chaos vs multistability vs input-induced) under investigation.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Diffstat (limited to 'initial_perturb_robustness/smoke_plots')
| -rw-r--r-- | initial_perturb_robustness/smoke_plots/initial_perturb_robustness_combined.csv | 4 |
1 files changed, 4 insertions, 0 deletions
diff --git a/initial_perturb_robustness/smoke_plots/initial_perturb_robustness_combined.csv b/initial_perturb_robustness/smoke_plots/initial_perturb_robustness_combined.csv new file mode 100644 index 0000000..f7c3ccc --- /dev/null +++ b/initial_perturb_robustness/smoke_plots/initial_perturb_robustness_combined.csv @@ -0,0 +1,4 @@ +label,sigma,n_samples,rollouts,ckpt_root,ckpt_name,perturb,noise_distribution,mean_rollout_exact,mean_rollout_token_acc,pass_at_k,all_k,correct_count_mean,correct_count_std,correct_count_q10,correct_count_q50,correct_count_q90,zero_frac,full_frac
+trm_baseline_smoke,0.0,16,2,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_repro,step_58590,both,gaussian,0.875,0.9537037014961243,0.875,0.875,1.75,0.6614378094673157,1.0,2.0,2.0,0.125,0.875
+trm_baseline_smoke,0.001,16,2,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_repro,step_58590,both,gaussian,0.8125,0.9309413433074951,0.875,0.75,1.625,0.6959705352783203,0.5,2.0,2.0,0.125,0.75
+trm_baseline_b32k8_smoke,0.0,32,8,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_repro,step_58590,both,gaussian,0.90625,0.9637345671653748,0.90625,0.90625,7.25,2.3318448066711426,8.0,8.0,8.0,0.09375,0.90625
|