Oct 7 9 days ago 15:41 Repair preserved evaluation references and label unpublished training sources Codex
15:39 Repair preserved evaluation references and label unpublished training sources Codex
15:32 Prune retired task and training artifacts; retain evaluated checkpoints and repair release references Codex
12:12 Import factor-sweep assets 401-527 12:11 Import factor-sweep assets 1-400 11:06 Add task-level fine-tuning-scope catalogue and preserve historical comparison limits 11:02 Complete fine-tuning-scope asset ownership transfer before removing old repositories 10:32 Add concise experiment navigation and task-level reproducibility manifests 10:21 Add concise experiment navigation and task-level reproducibility manifests 10:20 Store identical historical inference checkpoints once and retain per-run metadata 10:05 Add concise experiment navigation and task-level reproducibility manifests 09:37 Consolidate paper experiment assets by task and run (1-2) 09:18 Consolidate paper experiment assets by task and run (451-556) 09:16 Consolidate paper experiment assets by task and run (301-450) 09:15 Consolidate paper experiment assets by task and run (151-300) 09:13 Consolidate paper experiment assets by task and run (1-150) Oct 1 15 days ago 07:21 Archive four mean-trained QwenOFT profile-simulation evaluations, 100 episodes each 05:49 Record manifest-derived robot type in mean-profile portable inference configs Sep 30 16 days ago 12:59 Publish completed Flappy QwenOFT H1 mean-profile 5000-update Blackwell run Sep 28 18 days ago 15:16 Add Flappy fixed-mean RTX3090 teacher; evaluation pending Sep 24 22 days ago 03:32 Index 18 hist8 VLA models and publish portable configs with 27 evaluations 03:18 Publish hist8 profile flappy qwengr00t step5000 03:12 Publish hist8 profile flappy qwenpi_v3 step5000 03:09 Publish hist8 profile flappy qwenoft step5000 Sep 23 23 days ago 05:25 Remove nine superseded profiled small-policy bundles Remove old flappy, demon-attack, and deadly-corridor Sample Factory RTX 3090 v1 bundles from main. Preserve the temporal-hist8-capacity1 directories. Exact deleted paths and file hashes are recorded at the parent revision. 03:33 Publish five temporal H8 profiled small checkpoints Sep 22 24 days ago 14:58 Record Flappy temporal policy readback verification 14:57 Replace Flappy QwenOFT and QwenGR00T temporal 10 FPS policies Atomic replacement with seed-42 temporal HiddenRegimeSampler 25M training bundles and corrected 100-episode best-checkpoint evaluations. QwenPI v3 is intentionally unchanged. 12:13 Mark historical small-policy bundles published after readback 08:23 Publish historical Atari small-policy bundles and formal evaluations 
