Commit History
21 days ago
-
Invalidate WebShop SPrPO training seeds 1 and 2 affected by request-seed bug
Verified signatureWithdraw experiment selection and paper eligibility. Preserve checkpoints, original run configs, and logs. No retraining.37541c4
26 days ago
-
Use explicit download and catalog URLs in model card
Verified signatured550f14 -
Standardize retained training artifacts at step200 and organize ablations
Verified signaturePublish checksum-verified step200 adapters, remove superseded step300/400 paths, preserve historical provenance and existing remote-only artifacts.136f147
27 days ago
-
Add reproducible training overhead table for main experiments
Verified signaturefe9e3c9 -
Reduce edge margins while preserving the original motivation canvas size
Verified signatureaa65eff -
Center clean-rate glyphs and tighten motivation whitespace
Verified signature104e478 -
Align motivation condition names exactly with bar centers
Verified signatured61caac -
Equalize motivation panel widths and reduce condition typography
Verified signature03dcf6c -
Fit relative-drop labels between bars and clean reference lines
Verified signaturee404e3e -
Place ALFWorld perturbation labels to the right of bars
Verified signature898f770 -
Move motivation condition labels closer to the plots
Verified signaturea2cc857 -
Tighten motivation layout and remove explanatory footer
Verified signature58ab98b
28 days ago
-
Restyle introduction figure with relative-drop labels and method-inspired pastel colors
Verified signature8b708a8 -
Redesign half-width motivation figure with matched Gaussian 0.20 and dropout 0.05
Verified signature7cfe5a4 -
Update ALFWorld SAM three-seed main results
Verified signature42ab6dc -
Add ALFWorld 1.5B single-layer FFN Channel-SAM step100/step200 (rho=0.005, cap=0.05) (#6)
- Add ALFWorld 1.5B single-layer FFN Channel-SAM step100/step200 (rho=0.005, cap=0.05) (5a588e781f42aa37acc26ec822745a7a53619778) Co-authored-by: Daivd Li <Davidk1@users.noreply.huggingface.co>f667567 -
Add qwen2_5_1_5_stable_gaussian_noise_v11_step_200_seed0 step200 LoRA adapter and run config (#5)
- Add qwen2_5_1_5_stable_gaussian_noise_v11_step_200_seed0 step200 LoRA adapter and run config (ff5c4919d2bc99c5ab7be56bd42c94e6b37e1bb9) Co-authored-by: Daivd Li <Davidk1@users.noreply.huggingface.co>6a3865b -
Add qwen2_5_1_5_stable_gaussian_noise_v9_step_200_seed0 step200 LoRA adapter and run config (#4)
- Add qwen2_5_1_5_stable_gaussian_noise_v9_step_200_seed0 step200 LoRA adapter and run config (2486653e11f22a0b12c61aa38bae58bb581ba58c) Co-authored-by: Daivd Li <Davidk1@users.noreply.huggingface.co>3844509
29 days ago
-
Tighten main-result panel spacing in LaTeX and preview
Verified signaturede9a061 -
Publish step-200 main-result figures, plotting data, and LaTeX layout
Verified signaturef631c23
30 days ago
-
Rename shared analysis group to perturbation_effect_across_stage
Verified signaturea606550 -
Publish WebShop perturbation-geometry analysis figure and data
Verified signaturef389664 -
Normalize step100 LoRA adapter configs and add step100 to manifests
Verified signatureSame publication changes as step200: base_model_name_or_path -> Qwen/Qwen2.5-7B-Instruct, revision pinned to a09a354. Weights unchanged.90e03a6 -
Add results/alfworld/robust_training/qwen2_5_7_gaussian_noise_step_200_seed0/checkpoints/global_step_100/actor/lora_adapter
Verified signatureb5d10c9 -
Add results/alfworld/training/qwen2_5_7_step_200_seed0/checkpoints/global_step_100/actor/lora_adapter
Verified signatureb571fa8 -
Add results/webshop/training/qwen2_5_7_step_200_seed0/checkpoints/global_step_100/actor
Verified signatureb5a08cc -
Add WebShop 7B single-layer FFN Channel-SAM step200 (rho=0.005, cap=0.05) (#2)
- Add WebShop 7B single-layer FFN Channel-SAM step200 (rho=0.005, cap=0.05) (9bba64e8481e5f012b380f31c51a801bfcbc13e5) Co-authored-by: Daivd Li <Davidk1@users.noreply.huggingface.co>56fdbbc
-
Publish ALFWorld Qwen2.5-7B step200 clean-vs-gaussian evaluation results
Verified signature15aff1e -
Publish qwen2_5_7_stable_gaussian_noise_step_200_seed0 step200 LoRA adapter for collaborator ALFWorld 7B handoff
Verified signature230907a -
Publish qwen2_5_7_stable_gaussian_noise_step_200_seed0 step200 LoRA adapter for collaborator ALFWorld 7B handoff
Verified signatureb30a182 -
Publish qwen2_5_7_stable_gaussian_noise_step_200_seed0 step200 LoRA adapter for collaborator ALFWorld 7B handoff
Verified signaturee87bc1e -
Publish qwen2_5_7_stable_gaussian_noise_step_200_seed0 step200 LoRA adapter for collaborator ALFWorld 7B handoff
Verified signatureda23783
-
Remove redundant motivation README
Verified signature707f7f1 -
Remove redundant reward surface README
Verified signature56454c7 -
Organize reward surface and motivation figures with reproducible CSV inputs
Verified signaturebbaca61
-
Publish complete WebShop step200 three-seed evaluation results
Verified signature230ae6e -
Remove superseded SAM-named evaluation figure
Verified signaturef322c84 -
Publish complete WebShop step200 three-seed evaluation results
Verified signature5ad165d
-
Use fresh cap 0.05 SAM as official step200 baseline
Verified signatureb756470 -
Add WebShop single-layer FFN Channel-SAM step200 (rho=0.005, cap=0.05) (#1)
- Add WebShop single-layer FFN Channel-SAM step200 (rho=0.005, cap=0.05) (2be24292d326dab2fee8258dd1149b4dc5f8226a) Co-authored-by: Daivd Li <Davidk1@users.noreply.huggingface.co>aa7d451 -
Remove redundant evaluation package readme and manifest
Verified signaturec8c5c49 -
Update curated evaluation figures and reproducible data
Verified signatureadbb800 -
Use Vanilla, Plain Gaussian and Stable Gaussian algorithm labels
Verified signature057000e -
Publish WebShop 1.5B Stable Gaussian step200 main result
Verified signature9d5f63d -
Simplify model card around step200 baseline and model download table
Verified signature3cdd8a6
-
Document unified 200-step training default for 1.5B and 7B
Verified signatureSet the documented default to step200 while preserving historical step300 artifact paths and accurately describing published weights. Documentation only; no checkpoint changes.b562238
-
Publish four completed Qwen2.5-7B step-200 baseline adapters
Verified signatureALFWorld/WebShop clean and Gaussian baselines, seed 0; terminal LoRA adapters, original run configs, SHA256 manifests and loading instructions. Existing 1.5B artifacts retained.2f9d4a3
-
Document four finalized step-300 evaluation baselines
Verified signature7903637 -
Publish evaluation-only step-300 LoRA baselines; remove training archive
Verified signature903ca99 -
Archive baseline: alfworld/training/qwen2_5_1_5_step_300_seed0
Verified signature5dfb70c