summaryrefslogtreecommitdiff
path: root/research/flossing/multi4_eval_compare/trm_multi4_eval_full.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/multi4_eval_compare/trm_multi4_eval_full.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/multi4_eval_compare/trm_multi4_eval_full.csv')
-rw-r--r--research/flossing/multi4_eval_compare/trm_multi4_eval_full.csv11
1 files changed, 11 insertions, 0 deletions
diff --git a/research/flossing/multi4_eval_compare/trm_multi4_eval_full.csv b/research/flossing/multi4_eval_compare/trm_multi4_eval_full.csv
new file mode 100644
index 0000000..3411ff5
--- /dev/null
+++ b/research/flossing/multi4_eval_compare/trm_multi4_eval_full.csv
@@ -0,0 +1,11 @@
+run,step,all/exact_accuracy,all/accuracy,all/lm_loss,all/q_halt_accuracy,all/q_halt_loss,all/steps
+trm_multi4_loguniform_repro,26041,0.7394047975540161,0.9067130088806152,0.21417337656021118,0.9997232556343079,0.0024572687689214945,16
+trm_multi4_loguniform_repro,52082,0.8449901342391968,0.9424432516098022,0.1342233270406723,0.9998746514320374,0.0010290677892044187,16
+trm_multi4_loguniform_repro,78123,0.8417639136314392,0.9411031007766724,0.13769488036632538,0.9997493028640747,0.0022251552436500788,16
+trm_multi4_loguniform_repro,104164,0.8547161221504211,0.9456510543823242,0.12779666483402252,0.999834418296814,0.0019055778393521905,16
+trm_multi4_loguniform_repro,130205,0.8536233305931091,0.9453508853912354,0.1282763034105301,0.9998888373374939,0.0018422487191855907,16
+trm_multi4_loguniform_repro,156246,0.8489472270011902,0.9433866739273071,0.13273237645626068,0.999782383441925,0.003051365725696087,16
+trm_multi4_loguniform_repro,182287,0.8558868765830994,0.9459810256958008,0.12728126347064972,0.9997469186782837,0.0024399051908403635,16
+trm_multi4_loguniform_repro,208328,0.8404204249382019,0.9403222799301147,0.13971956074237823,0.9996499419212341,0.0029723909683525562,16
+trm_multi4_loguniform_repro,234369,0.845432460308075,0.9419097304344177,0.13668429851531982,0.9997681975364685,0.0016449446557089686,16
+trm_multi4_loguniform_repro,260410,0.826143741607666,0.9344042539596558,0.15468436479568481,0.999701976776123,0.001810370129533112,16