chore: enable Dependabot weekly GitHub Actions bumps

.github/dependabot.yml

@@ -0,0 +1,11 @@
+version: 2
+updates:
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      interval: "weekly"
+    cooldown:
+      default-days: 7
+    groups:
+      actions:
+        patterns: ["*"]

docs/source/_toctree.yml

@@ -9,8 +9,6 @@
 - sections:
   - local: il_robots
     title: Imitation Learning for Robots
-  - local: lelab
-    title: LeLab - Lerobot GUI
   - local: bring_your_own_policies
     title: Adding a Policy
   - local: integrate_hardware

docs/source/lelab.mdx

@@ -1,42 +0,0 @@
-# LeLab - LeRobot Guide
-Graphical user interfaces are the easiest to use for beginners because it's easy to just click everything without remembering the proper commands. That's why we built LeLab which is a GUI built on top of the LeRobot library. With this app you will be able to add robots, collect datasets, train and deploy models.
-
-### Installation
-To install lerobot you can simply copy the following command and paste into your terminal. For it to work you need to have `uv` installed, [here is how to do it.](https://docs.astral.sh/uv/getting-started/installation/) 
-```
-uv tool install git+https://github.com/huggingface/leLab.git && lelab
-```
-
-Once installed you will be able to run lelab anytime you want with `lelab` command from your terminal (above command has it included at the end so it will run it right after installation).
-
-### Adding robots
-
-##### Calibration
-You will need to select the proper arm type (leader or follower) and calibrate each arm as shown in the video available inside LeLab. Make sure that all joints are in the middle position when starting the calibration.
-
-##### Adding cameras
-At the bottom of the add robot page you can also add the cameras and name them accordingly. 
-
-### Teleoperation
-Once the robots have been configured you can go back and click the teleoperation button. You will see the 3D visualization of the arm and will be able to control the follower with the leader. If something doesn't work there, remove and add your robot again following the steps described in LeLab.
-
-### Recording a dataset
-Type a new name for your dataset and press on the plus button. You will need to provide:
-- Task description, for example "put the cube in a container"
-- Number of episodes that you want to record, at least 30 recommended
-- Episode and reset durations. These are max durations and can be shortened while recording with a spacebar press.
-- If you configured your cameras earlier you don't need to do that again.
-
-Press start recording, wait for it to load, perform the task with confident movements but don't rush. Once the task is finished and you moved your robot to the initial position press the spacebar. You will have time to reset the environment for example grab the cube from the container and placing it on the desk again. Once ready press the spacebar and record the next episode. Repeat until all the episodes are recorded.
-
-### Training a model
-This is the most powerful function with LeLab! You can easily train models locally on your own computer but also with [HF Jobs](https://huggingface.co/docs/huggingface_hub/en/guides/jobs) which gives you easy access to very powerful GPUs with clear pricing.
-
-> [!TIP]
-> To use HF Jobs make sure that you are logged in to HF, you can do that by running `hf auth login` in the terminal.
-
-In the training tab select if you want to train locally or specific HF hardware you want to use. You will also need to provide the dataset that will be used for training. Your own datasets will be listed in a dropdown list, you can also use other datasets by providing its id. Set the policy you want to train, batch size and number of steps. For guide on choosing hardware and batch size check out our [Compute HW Guide for LeRobot Training.](hardware_guide.mdx) 
-
-Once you start training the progress will be visualized inside LeLab. Checkpoints will be saved as well.
-### Running the model on a robot
-In the main view of the LeLab under jobs you will see all the models that you trained. To run the policy on the robot just click the green run button and press start inference. After loading the policy the robot should start solving the task that it learned during training.

src/lerobot/configs/parser.py

@@ -255,6 +255,7 @@ def extract_path_fields_from_config(config_path: str, path_fields: list[str]) ->
             remaining = config_data[field]
             if remaining:
                 _config_yaml_overrides[field] = _flatten_to_cli_args(remaining)
+            else:
                 del config_data[field]
             modified = True
 
@@ -310,13 +311,7 @@ def wrap(config_path: Path | None = None) -> Callable[[F], F]:
                     cli_args = filter_arg("config_path", cli_args)
                     cfg = argtype.from_pretrained(config_path_cli, cli_args=cli_args)
                 else:
-                    if config_path_cli:
-                        cli_args = filter_arg("config_path", cli_args)
-                    cfg = draccus.parse(
-                        config_class=argtype,
-                        config_path=config_path_cli or config_path,
-                        args=cli_args,
-                    )
+                    cfg = draccus.parse(config_class=argtype, config_path=config_path, args=cli_args)
             response = fn(cfg, *args, **kwargs)
             return response
 

src/lerobot/policies/groot/eagle2_hg_model/modeling_eagle2_5_vl.py

@@ -60,7 +60,6 @@ class Eagle25VLPreTrainedModel(PreTrainedModel):
         "SiglipEncoderLayer",
     ]
     _skip_keys_device_placement = "past_key_values"
-    _supports_flash_attn = True
     _supports_flash_attn_2 = True
     _supports_cache_class = True
     _supports_static_cache = True

src/lerobot/policies/groot/eagle2_hg_model/processing_eagle2_5_vl.py

@@ -124,6 +124,7 @@ class Eagle25VLProcessor(ProcessorMixin):
         "videos_kwargs",
         "text_kwargs",
     ]
+    image_processor_class = "AutoImageProcessor"
     tokenizer_class = "AutoTokenizer"
 
     def __init__(

src/lerobot/policies/groot/processor_groot.py

@@ -206,11 +206,7 @@ def _build_eagle_processor(tokenizer_assets_repo: str = DEFAULT_TOKENIZER_ASSETS
             "Vendor files are copied during model creation. Create the policy/model first, "
             "or call ensure_eagle_cache_ready() before building processors."
         )
-    proc = AutoProcessor.from_pretrained(
-        str(cache_dir),
-        trust_remote_code=True,
-        fix_mistral_regex=False,
-    )
+    proc = AutoProcessor.from_pretrained(str(cache_dir), trust_remote_code=True, use_fast=True)
     proc.tokenizer.padding_side = "left"
     return proc
 

tests/test_yaml_policy_path.py

@@ -1,14 +1,10 @@
 """Tests for policy.path support in YAML config files (issue #2957)."""
 
 import json
-import sys
 import tempfile
-from dataclasses import dataclass, field
-from unittest.mock import patch
 
 import yaml
 
-from lerobot.configs import parser
 from lerobot.configs.parser import (
     _config_path_args,
     _config_yaml_overrides,
@@ -20,8 +16,7 @@ from lerobot.configs.parser import (
 
 
 def test_extract_path_fields_from_yaml():
-    """Test that policy.path is extracted from a YAML config and the policy block
-    is removed entirely (siblings are captured separately as cli_overrides)."""
+    """Test that policy.path is extracted from a YAML config and removed."""
     config = {
         "dataset": {"repo_id": "lerobot/pusht"},
         "policy": {"type": "smolvla", "path": "lerobot/smolvla_base", "push_to_hub": False},
@@ -31,33 +26,26 @@ def test_extract_path_fields_from_yaml():
         config_path = f.name
 
     _config_path_args.clear()
-    _config_yaml_overrides.clear()
     cleaned_path = extract_path_fields_from_config(config_path, ["policy"])
 
     # Path should be extracted and stored
     assert _config_path_args["policy"] == "lerobot/smolvla_base"
 
-    # Cleaned config should not have the policy block at all -- draccus must not
-    # try to decode it as PreTrainedConfig; the actual config comes from
-    # from_pretrained(path) with the captured overrides applied on top.
+    # Cleaned config should not have the path field
     with open(cleaned_path) as f:
         cleaned = yaml.safe_load(f)
-    assert "policy" not in cleaned
+    assert "path" not in cleaned["policy"]
+    assert cleaned["policy"]["type"] == "smolvla"
+    assert cleaned["policy"]["push_to_hub"] is False
 
     # Original dataset should be untouched
     assert cleaned["dataset"]["repo_id"] == "lerobot/pusht"
 
-    # Sibling overrides (excluding type/path) captured for from_pretrained.
-    overrides = get_yaml_overrides("policy")
-    assert any("push_to_hub=false" in o for o in overrides)
-
     _config_path_args.clear()
-    _config_yaml_overrides.clear()
 
 
 def test_extract_path_fields_from_json():
-    """Test that policy.path is extracted from a JSON config and the policy
-    block is removed entirely."""
+    """Test that policy.path is extracted from a JSON config."""
     config = {
         "policy": {"type": "act", "path": "some/local/path"},
     }
@@ -66,17 +54,15 @@ def test_extract_path_fields_from_json():
         config_path = f.name
 
     _config_path_args.clear()
-    _config_yaml_overrides.clear()
     cleaned_path = extract_path_fields_from_config(config_path, ["policy"])
 
     assert _config_path_args["policy"] == "some/local/path"
 
     with open(cleaned_path) as f:
         cleaned = json.load(f)
-    assert "policy" not in cleaned
+    assert "path" not in cleaned["policy"]
 
     _config_path_args.clear()
-    _config_yaml_overrides.clear()
 
 
 def test_extract_no_path_returns_original():
@@ -230,91 +216,3 @@ def test_flatten_nested_with_bools():
     args = _flatten_to_cli_args(d)
     assert "--optimizer.use_warmup=true" in args
     assert "--optimizer.lr=0.01" in args
-
-
-def test_extract_removes_field_with_siblings_and_no_type():
-    """Regression: when policy.path has siblings but no type:, the entire policy
-    block must still be removed from the cleaned config. Otherwise draccus tries
-    to decode the leftover dict as PreTrainedConfig and crashes on the missing
-    type discriminator.
-    """
-    config = {
-        "dataset": {"repo_id": "lerobot/pusht"},
-        "policy": {
-            "path": "lerobot/smolvla_base",
-            "n_action_steps": 10,
-            "dtype": "bfloat16",
-        },
-    }
-    with tempfile.NamedTemporaryFile(mode="w", suffix=".yaml", delete=False) as f:
-        yaml.dump(config, f)
-        config_path = f.name
-
-    _config_path_args.clear()
-    _config_yaml_overrides.clear()
-    cleaned_path = extract_path_fields_from_config(config_path, ["policy"])
-
-    with open(cleaned_path) as f:
-        cleaned = yaml.safe_load(f) or {}
-    assert "policy" not in cleaned, "policy block should be fully removed when path is present"
-    assert cleaned["dataset"]["repo_id"] == "lerobot/pusht"
-    assert _config_path_args["policy"] == "lerobot/smolvla_base"
-    overrides = get_yaml_overrides("policy")
-    assert any("n_action_steps=10" in o for o in overrides)
-    assert any("dtype=bfloat16" in o for o in overrides)
-
-    _config_path_args.clear()
-    _config_yaml_overrides.clear()
-
-
-@dataclass
-class _DummyNested:
-    foo: int = 0
-
-
-@dataclass
-class _DummyConfig:
-    nested: _DummyNested = field(default_factory=_DummyNested)
-    other: str = "default"
-
-    @classmethod
-    def __get_path_fields__(cls):
-        return ["nested"]
-
-
-def test_wrap_uses_cleaned_config_for_draccus_parse():
-    """Regression: wrap() updates config_path_cli to point at the cleaned temp
-    file but must propagate that to the draccus.parse fallback branch. Without
-    the fix, cli_args still contains --config_path=<original> and draccus reads
-    the original YAML with `path:` still in it, crashing on the unknown field.
-    """
-    config = {
-        "nested": {"path": "some/checkpoint", "foo": 42},
-        "other": "set-via-yaml",
-    }
-    with tempfile.NamedTemporaryFile(mode="w", suffix=".yaml", delete=False) as f:
-        yaml.dump(config, f)
-        config_path = f.name
-
-    _config_path_args.clear()
-    _config_yaml_overrides.clear()
-
-    captured: dict = {}
-
-    @parser.wrap()
-    def main(cfg: _DummyConfig) -> _DummyConfig:
-        captured["cfg"] = cfg
-        return cfg
-
-    with patch.object(sys, "argv", ["prog", f"--config_path={config_path}"]):
-        main()
-
-    assert captured["cfg"].other == "set-via-yaml"
-    assert _config_path_args["nested"] == "some/checkpoint"
-    # Cleaned config dropped `nested:` entirely; defaults stand for this wrapper
-    # class (a real PreTrainedConfig would now load the checkpoint and apply
-    # the captured yaml_overrides via from_pretrained()).
-    assert captured["cfg"].nested.foo == 0
-
-    _config_path_args.clear()
-    _config_yaml_overrides.clear()