Linslab commited on
Commit
a4c81a1
·
verified ·
1 Parent(s): 2d37ce9

Preserve HF-data campaign provenance alongside two newly published task peaks

Browse files
Files changed (1) hide show
  1. README.md +2 -1
README.md CHANGED
@@ -9,8 +9,9 @@ license: mit
9
  One PI0.5 LoRA checkpoint per ManiLadder task (92 tasks), each finetuned on 20 demos.
10
  Checkpoints are selected by the highest **50-episode** success rate; ties use the earliest step.
11
 
12
- - The 70 previously published tasks use fixed evaluation seeds 1000–1049.
13
  - The 22 completed retraining runs published through 2026-10-30 use seeds 800000–800049. All reached 40,000 training steps. 15 have all eight 5k–40k evaluations; 7 have seven evaluations because one intermediate checkpoint was not retained. Their selected weights are the best among the available evaluated checkpoints.
 
14
 
15
  See [retraining provenance and missing steps](single_task/pi05_retrain_eval50_20261022_manifest.json). These scores use different seed sets across the two groups and are not a controlled comparison.
16
 
 
9
  One PI0.5 LoRA checkpoint per ManiLadder task (92 tasks), each finetuned on 20 demos.
10
  Checkpoints are selected by the highest **50-episode** success rate; ties use the earliest step.
11
 
12
+ - The 60 previously published tasks use fixed evaluation seeds 1000–1049.
13
  - The 22 completed retraining runs published through 2026-10-30 use seeds 800000–800049. All reached 40,000 training steps. 15 have all eight 5k–40k evaluations; 7 have seven evaluations because one intermediate checkpoint was not retained. Their selected weights are the best among the available evaluated checkpoints.
14
+ - The 10 tasks from the HF-data retraining campaign `pi05_hf21_20261025` (added 2026-10-29; raw demos freshly downloaded from Linslab/ManiLadder_Dataset) use seeds 1000–1049. All reached 40,000 training steps and have all eight 5k–40k evaluations. See [its manifest](single_task/pi05_hf21_20261025_manifest.json).
15
 
16
  See [retraining provenance and missing steps](single_task/pi05_retrain_eval50_20261022_manifest.json). These scores use different seed sets across the two groups and are not a controlled comparison.
17