admin/lerobot
1
0
Fork 0
mirror of https://github.com/huggingface/lerobot.git synced 2026-08-08 17:39:44 +00:00
Code Issues Packages Projects Releases Wiki Activity

docs(g05): record joint CoT training smoke

Browse Source
This commit is contained in:
Pepijn
2026-07-29 11:29:32 +02:00
parent 5eeee589a9
commit d94fef7b40
1 changed files with 6 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/source/g05.mdx
+6
View File
@@ -205,6 +205,12 @@ completed a batch-size-one BF16 forward, backward, gradient clip, and AdamW
step. It produced finite loss `2.77356`, finite pre-clip gradient norm `54.38`,
all six author optimizer groups, and 23.59 GiB peak allocated CUDA memory.
The same RTX 5090 also loaded the private `g05_so101` checkpoint and completed
a real joint `BBox → Subtask → Action` forward and backward through the
recipe-driven path. It produced finite total loss `4.38332`, including non-zero
`ce_loss=4.04244` and `fm_loss=0.340881`, with 15.58 GiB peak allocated CUDA
memory.
A 50-episode LIBERO/RoboTwin success-rate comparison additionally requires the
matching simulator, task assets, reset seeds, and author evaluator; no task-level
benchmark number is claimed until that separate gate runs.