fix dependence and redirect url

[DelinQu]

This pull request introduces several improvements and fixes across the codebase, focusing on enhanced configuration flexibility, improved sampling logic, dependency updates, and documentation corrections. The most significant changes include new configuration options for model freezing, refined action sampling logic for chunked actions, dependency and documentation updates to reflect repository renaming, and improved usability in testing and troubleshooting.

Model configuration and training improvements:

• Added a freeze_lm_head option to TrainPipelineConfig and updated configure_llm to respect this setting, allowing users to freeze the language model head independently from the rest of the LLM. [1] [2]
• Fixed logic so that vision_lora is only enabled if lora_enable is true, preventing configuration mismatches.

Action sampling and model logic:

• Refactored sample_actions in modeling_eo1.py to use a local chunk_size variable for clarity, improved cache handling, and ensured consistent use of chunk size in tensor slicing and action projection. [1] [2] [3] [4]
• Passed states to the model's forward call in modeling_eo1.py to support additional stateful operations.

Dependency and repository updates:

• Updated repository URLs and documentation links to use the new EO1 naming instead of EO-1, and added/updated dependencies in pyproject.toml (e.g., version pinning for lerobot, new dependencies like qwen_vl_utils, and ujson). [1] [2] [3] [4]
• Changed the installation order for flash-attn in the setup instructions and clarified recommended installation methods.

Testing and troubleshooting enhancements:

• Improved the VLM test script (now in tests/test_vlm.py) to handle image and grid inputs correctly across multiple turns, allowing for more robust interactive testing.
• Added a troubleshooting section to the README.md for common issues such as missing FFmpeg installations.

Other notable changes:

• Made root in LerobotConfig optional for better flexibility in dataset configuration.
• Cleaned up imports and removed unused code in modeling_qwen2_5_vl.py, and fixed logic in prepare_inputs_for_generation for state handling. [1] [2] [3]
• Removed unnecessary sourcing of env.sh in various experiment scripts, streamlining environment setup. [1] [2] [3] [4] [5]# What does this PR do?

Fixes # (issue)

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Was this discussed/approved via a GitHub issue? Please add a link to it if that's the case.
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

[Copilot]

Pull Request Overview

This pull request introduces several improvements to the EO1 codebase, focusing on configuration flexibility, model sampling logic improvements, dependency updates, and repository renaming. The changes enhance the model's training capabilities by adding granular control over component freezing, improve action sampling consistency for chunked actions, and update the project to reflect the new EO1 repository naming convention.

• Added freeze_lm_head configuration option for independent control of language model head freezing
• Refactored action sampling logic in modeling_eo1.py to use consistent chunk size variables and improve cache handling
• Updated repository URLs, dependencies, and documentation to reflect EO-1 to EO1 renaming

Reviewed Changes

Copilot reviewed 14 out of 15 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
tools/test_hf_model.py	Removed obsolete HuggingFace model test script
tests/test_vlm.py	Enhanced VLM test script with improved multi-turn conversation handling and image processing
pyproject.toml	Updated repository URLs and added new dependencies with version constraints
experiments/*.sh	Removed unnecessary environment script sourcing from training scripts
eo/train/train_utils.py	Updated LLM configuration to use new freeze_lm_head parameter
eo/train/pipeline_config.py	Added freeze_lm_head configuration option and improved LoRA logic
eo/model/modeling_qwen2_5_vl.py	Cleaned up imports, improved state handling, and removed unused generation methods
eo/model/modeling_eo1.py	Refactored action sampling with consistent chunk size usage and improved cache management
eo/data/schema.py	Made root parameter optional in LerobotConfig
README.md	Updated repository URLs and added troubleshooting section

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}