summaryrefslogtreecommitdiff
path: root/research/flossing/problem_tracks/hrm_learning_groups.csv
diff options
context:
space:
mode:
authorYurenHao0426 <blackhao0426@gmail.com>2026-06-13 12:35:36 -0500
committerYurenHao0426 <blackhao0426@gmail.com>2026-06-13 12:35:36 -0500
commit66e0d8b9fd4d0f7a2231d689c055e26fdf1cf04a (patch)
treec29cba61124018755a19b02c9d33e3ad5f2e05cc /research/flossing/problem_tracks/hrm_learning_groups.csv
rrm workspace: TRM/HRM/SRM code, Maze dataset, dynamical-analysis pipelineHEADmain
Curated export for clone-and-run Maze training (2x A6000) + diagnostics. trm/hrm pretrain.py carry trajectory-augmentation code (backward-compatible). Heavy artifacts (checkpoints/wandb/npz) gitignored; see PROVENANCE.md. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Diffstat (limited to 'research/flossing/problem_tracks/hrm_learning_groups.csv')
-rw-r--r--research/flossing/problem_tracks/hrm_learning_groups.csv6
1 files changed, 6 insertions, 0 deletions
diff --git a/research/flossing/problem_tracks/hrm_learning_groups.csv b/research/flossing/problem_tracks/hrm_learning_groups.csv
new file mode 100644
index 0000000..862a54a
--- /dev/null
+++ b/research/flossing/problem_tracks/hrm_learning_groups.csv
@@ -0,0 +1,6 @@
+kind,group,n,fraction,givens_mean,final_success_rate,final_token_acc_mean,initial_lambda_max,final_lambda_max,initial_mean8,final_mean8,initial_tail_mean_5_8,final_tail_mean_5_8,initial_positive_count,final_positive_count,initial_positive_sum,final_positive_sum
+HRM,always,7,0.013671875,33.857142857142854,1.0,1.0,-0.11943858010428292,-0.14200945092099054,-0.1954429488895195,-0.1990994273551873,-0.23143203716192925,-0.2316632675273078,0.0,0.0,0.0,0.0
+HRM,flaky_then_stable,136,0.265625,25.264705882352942,1.0,1.0,-0.09145989122471827,-0.14467768998891936,-0.11213886570562061,-0.18874008171084877,-0.12149259587749839,-0.20950002090402825,0.007352941176470588,0.029411764705882353,0.00026379939278259,0.0007114800186279942
+HRM,never,145,0.283203125,24.613793103448277,0.0,0.6184759497642517,-0.09740061664889599,0.030114806785874453,-0.11696181520819664,-0.029440750938995515,-0.12532571852207183,-0.05453352611592232,0.0,1.7310344827586206,0.0,0.06679639958993307
+HRM,stable_learned,113,0.220703125,25.911504424778762,1.0,1.0,-0.09373763412962445,-0.14458168358995324,-0.1166969095294949,-0.18998062464952534,-0.126381192636569,-0.2119278870522976,0.0,0.008849557522123894,0.0,0.00017139836486461944
+HRM,transient_or_regressed,111,0.216796875,24.26126126126126,0.0,0.6783450135776589,-0.09775018336268158,0.03993409140552562,-0.11691337225639874,-0.019361971456525788,-0.1249423275055649,-0.04386983102941067,0.0,2.2162162162162162,0.0,0.08383428943679072