admin/lerobot
1
0
Fork 0
mirror of https://github.com/huggingface/lerobot.git synced 2026-05-17 01:30:14 +00:00
Code Issues Packages Projects Releases Wiki Activity

refactor(processor): improve processor pipeline typing with generic type (#1810)

Browse Source 
* refactor(processor): introduce generic type for to_output

- Always return `TOutput`
- Remove `_prepare_transition`, so `__call__` now always returns `TOutput`
- Update tests accordingly
- This refactor paves the way for adding settings for `to_transition` and `to_output` in `make_processor` and the post-processor

* refactor(processor): consolidate ProcessorKwargs usage across policies

- Removed the ProcessorTypes module and integrated ProcessorKwargs directly into the processor pipeline.
- Updated multiple policy files to utilize the new ProcessorKwargs structure for preprocessor and postprocessor arguments.
- Simplified the handling of processor kwargs by initializing them to empty dictionaries when not provided.
This commit is contained in:
Adil Zouitine
2025-09-02 12:57:14 +02:00
committed by GitHub
parent 08fb310eaa
commit d32b76cc66
26 changed files with 847 additions and 220 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/processor/test_smolvla_processor.py
+30 -5
View File
@@ -89,7 +89,12 @@ def test_make_smolvla_processor_basic():
stats = create_default_stats()
with patch("lerobot.policies.smolvla.processor_smolvla.TokenizerProcessor"):
preprocessor, postprocessor = make_smolvla_pre_post_processors(config, stats)
preprocessor, postprocessor = make_smolvla_pre_post_processors(
config,
stats,
preprocessor_kwargs={"to_transition": lambda x: x, "to_output": lambda x: x},
postprocessor_kwargs={"to_transition": lambda x: x, "to_output": lambda x: x},
)
# Check processor names
assert preprocessor.name == "robot_preprocessor"
@@ -188,7 +193,12 @@ def test_smolvla_processor_cuda():
return features
with patch("lerobot.policies.smolvla.processor_smolvla.TokenizerProcessor", MockTokenizerProcessor):
preprocessor, postprocessor = make_smolvla_pre_post_processors(config, stats)
preprocessor, postprocessor = make_smolvla_pre_post_processors(
config,
stats,
preprocessor_kwargs={"to_transition": lambda x: x, "to_output": lambda x: x},
postprocessor_kwargs={"to_transition": lambda x: x, "to_output": lambda x: x},
)
# Create CPU data
observation = {
@@ -238,7 +248,12 @@ def test_smolvla_processor_accelerate_scenario():
return features
with patch("lerobot.policies.smolvla.processor_smolvla.TokenizerProcessor", MockTokenizerProcessor):
preprocessor, postprocessor = make_smolvla_pre_post_processors(config, stats)
preprocessor, postprocessor = make_smolvla_pre_post_processors(
config,
stats,
preprocessor_kwargs={"to_transition": lambda x: x, "to_output": lambda x: x},
postprocessor_kwargs={"to_transition": lambda x: x, "to_output": lambda x: x},
)
# Simulate Accelerate: data already on GPU and batched
device = torch.device("cuda:0")
@@ -289,7 +304,12 @@ def test_smolvla_processor_multi_gpu():
return features
with patch("lerobot.policies.smolvla.processor_smolvla.TokenizerProcessor", MockTokenizerProcessor):
preprocessor, postprocessor = make_smolvla_pre_post_processors(config, stats)
preprocessor, postprocessor = make_smolvla_pre_post_processors(
config,
stats,
preprocessor_kwargs={"to_transition": lambda x: x, "to_output": lambda x: x},
postprocessor_kwargs={"to_transition": lambda x: x, "to_output": lambda x: x},
)
# Simulate data on different GPU
device = torch.device("cuda:1")
@@ -315,7 +335,12 @@ def test_smolvla_processor_without_stats():
# Mock the tokenizer processor
with patch("lerobot.policies.smolvla.processor_smolvla.TokenizerProcessor"):
preprocessor, postprocessor = make_smolvla_pre_post_processors(config, dataset_stats=None)
preprocessor, postprocessor = make_smolvla_pre_post_processors(
config,
dataset_stats=None,
preprocessor_kwargs={"to_transition": lambda x: x, "to_output": lambda x: x},
postprocessor_kwargs={"to_transition": lambda x: x, "to_output": lambda x: x},
)
# Should still create processors
assert preprocessor is not None