admin/lerobot
1
0
Fork 0
mirror of https://github.com/huggingface/lerobot.git synced 2026-07-24 02:06:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

groot: fix stale Eagle VLM docstring in processor (N1.7 uses Qwen3-VL backbone)

Browse Source 
Addresses checker nit: processor_groot.py docstring still described the N1.5
Eagle VLM path with eagle_content/eagle_* keys that no longer exist in the code.
This commit is contained in:
nv-sachdevkartik
2026-06-11 18:10:46 +00:00
parent bba996ef8d
commit 885f55ef04
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/lerobot/policies/groot/processor_groot.py
+2 -2
View File
@@ -579,8 +579,8 @@ def make_groot_pre_post_processors(
1. Optional key renaming (dataset-specific key mapping) 1. Optional key renaming (dataset-specific key mapping)
2. Add batch dimension to unbatched data 2. Add batch dimension to unbatched data
3. Pack video/state/action/language/embodiment and apply optional min-max normalization before padding 3. Pack video/state/action/language/embodiment and apply optional min-max normalization before padding
4. Encode video+language with Eagle VLM into intermediate eagle_content 4. Encode video+language with the GR00T N1.7 VLM backbone (Qwen3-VL) into intermediate VLM content
5. Collate eagle_content into batched eagle_* tensors 5. Collate the VLM content into batched backbone input tensors
6. Move tensors to device (GPU) 6. Move tensors to device (GPU)
NOTE: We optionally apply min-max normalization to STATE and ACTION using NOTE: We optionally apply min-max normalization to STATE and ACTION using