admin/lerobot
1
0
Fork 0
mirror of https://github.com/huggingface/lerobot.git synced 2026-05-21 11:39:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

docs(rollout): edit rename_map instructions

Browse Source
This commit is contained in:
Steven Palma
2026-04-22 13:59:55 +02:00
parent 8d1401abe3
commit bff2d50dc1
2 changed files with 9 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/source/rename_map.mdx
+7 -18
View File
@@ -61,17 +61,6 @@ lerobot-eval \
--rename_map='{"observation.images.image": "observation.images.base_0_rgb", "observation.images.image2": "observation.images.left_wrist_0_rgb"}' --rename_map='{"observation.images.image": "observation.images.base_0_rgb", "observation.images.image2": "observation.images.left_wrist_0_rgb"}'
``` ```
### Recording
`lerobot-record` also supports rename maps, nested under the dataset config:
```bash
lerobot-record \ # When running inference
--policy.path="<user>/smolVLA_finetuned" \
... \
--dataset.rename_map='{"observation.images.glove2": "observation.images.image"}'
```
## Alternative: edit the policy config directly ## Alternative: edit the policy config directly
If you always use the same dataset or environment, you can **edit the policy's `config.json`** so its observation keys match your data source. Then no rename map is needed. If you always use the same dataset or environment, you can **edit the policy's `config.json`** so its observation keys match your data source. Then no rename map is needed.
@@ -105,10 +94,10 @@ XVLA-base has three visual inputs and `empty_cameras=0` by default. Your dataset
## Quick reference ## Quick reference
| Goal | What to do | | Goal | What to do |
| ----------------------------------------- | --------------------------------------------------------------------------- | | --------------------------------------- | --------------------------------------------------------------------------- |
| Dataset keys ≠ policy keys | `--rename_map='{"dataset_key": "policy_key", ...}'` | | Dataset keys ≠ policy keys | `--rename_map='{"dataset_key": "policy_key", ...}'` |
| Env keys ≠ policy keys (eval) | `--rename_map='{"env_key": "policy_key", ...}'` | | Env keys ≠ policy keys (eval) | `--rename_map='{"env_key": "policy_key", ...}'` |
| Recording with different keys (inference) | `--dataset.rename_map='{"source_key": "policy_key", ...}'`. | | Rollout with different keys (inference) | `--rename_map='{"source_key": "policy_key", ...}'`. |
| Fewer cameras than policy expects | `--policy.empty_cameras=N` (supported by PI0, PI05, PI0Fast, SmolVLA, XVLA) | | Fewer cameras than policy expects | `--policy.empty_cameras=N` (supported by PI0, PI05, PI0Fast, SmolVLA, XVLA) |
| Avoid passing a rename map | Edit the policy's `config.json` so its keys match your data source | | Avoid passing a rename map | Edit the policy's `config.json` so its keys match your data source |
src/lerobot/rollout/configs.py
+2 -3
View File
@@ -289,9 +289,8 @@ class RolloutConfig:
raise ValueError("--policy.path is required for rollout") raise ValueError("--policy.path is required for rollout")
# --- Task resolution --- # --- Task resolution ---
# When --dataset.rename_map (or any --dataset.* flag) is passed, draccus # When any --dataset.* flag is passed, draccus creates a DatasetRecordConfig with single_task="".
# creates a DatasetRecordConfig with single_task="". If the user set # If the user set the task via the top-level --task flag, propagate it so that all
# the task via the top-level --task flag, propagate it so that all
# downstream consumers (inference engine, dataset frame builders) see it. # downstream consumers (inference engine, dataset frame builders) see it.
if self.dataset is not None and not self.dataset.single_task and self.task: if self.dataset is not None and not self.dataset.single_task and self.task:
logger.info("Propagating top-level task '%s' to dataset config", self.task) logger.info("Propagating top-level task '%s' to dataset config", self.task)