[Improvement] Accept chat template string in vLLM models #768

VincentYCYao · 2025-07-21T15:01:00Z

Minor improvement following the previous PR (#652 (comment)).

Add comments and doc
Accept both chat template file path and chat template string

Summary by CodeRabbit

Documentation
- Enhanced the documentation for the VLLM class with detailed descriptions, supported models and formats, argument explanations, and usage examples.
Bug Fixes
- Improved chat template handling to support both file paths and direct template strings, ensuring better flexibility and error handling.

Minor improvement following the previous PR (EvolvingLMMs-Lab#652 (comment)). - Add comments and doc - Accept both chat template file path and chat template string

coderabbitai · 2025-07-21T15:01:07Z

Walkthrough

A comprehensive docstring was added to the VLLM class in lmms_eval/models/vllm.py, detailing its functionality, supported models, media formats, arguments, and usage examples. The __init__ method was enhanced to interpret the chat_template parameter as either a file path or a direct template string, raising an error if the file does not exist. The generate_until method was updated to use the preloaded chat template string, with added clarifying comments.

Changes

File(s)	Change Summary
lmms_eval/models/vllm.py	Added detailed class docstring; enhanced `__init__` to load chat templates from file or string with validation; updated `generate_until` to use cached chat template string; added clarifying comments.

Estimated code review effort

1 (~4 minutes)

Poem

In the warren where code and models dwell,
A docstring now weaves its magic spell.
Chat templates flex, file or string—
The VLLM class can handle anything!
With comments clear and usage bright,
Reviewers hop with sheer delight. 🐇✨

📜 Recent review details

Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 62b9500 and 7b87d91.

📒 Files selected for processing (1)

lmms_eval/models/vllm.py (3 hunks)

🚧 Files skipped from review as they are similar to previous changes (1)

lmms_eval/models/vllm.py

⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)

GitHub Check: Cursor Bugbot

✨ Finishing Touches

📝 Generate Docstrings

🧪 Generate unit tests

Create PR with unit tests
Post copyable unit tests in a comment

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

🪧 Tips

Chat

There are 3 ways to chat with CodeRabbit:

Review comments: Directly reply to a review comment made by CodeRabbit. Example:
- I pushed a fix in commit <commit_id>, please review it.
- Explain this complex logic.
- Open a follow-up GitHub issue for this discussion.
Files and specific lines of code (under the "Files changed" tab): Tag @coderabbitai in a new review comment at the desired location with your query. Examples:
- @coderabbitai explain this code block.
- @coderabbitai modularize this function.
PR comments: Tag @coderabbitai in a new PR comment to ask questions about the PR branch. For the best results, please provide a very specific query, as very limited context is provided in this mode. Examples:
- @coderabbitai gather interesting stats about this repository and render them as a table. Additionally, render a pie chart showing the language distribution in the codebase.
- @coderabbitai read src/utils.ts and explain its main purpose.
- @coderabbitai read the files in the src/scheduler package and generate a class diagram using mermaid and a README in the markdown format.
- @coderabbitai help me debug CodeRabbit configuration file.

Support

Need help? Create a ticket on our support page for assistance with any issues or questions.

Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments.

CodeRabbit Commands (Invoked using PR comments)

@coderabbitai pause to pause the reviews on a PR.
@coderabbitai resume to resume the paused reviews.
@coderabbitai review to trigger an incremental review. This is useful when automatic reviews are disabled for the repository.
@coderabbitai full review to do a full review from scratch and review all the files again.
@coderabbitai summary to regenerate the summary of the PR.
@coderabbitai generate docstrings to generate docstrings for this PR.
@coderabbitai generate sequence diagram to generate a sequence diagram of the changes in this PR.
@coderabbitai generate unit tests to generate unit tests for this PR.
@coderabbitai resolve resolve all the CodeRabbit review comments.
@coderabbitai configuration to show the current CodeRabbit configuration for the repository.
@coderabbitai help to get help.

Other keywords and placeholders

Add @coderabbitai ignore anywhere in the PR description to prevent this PR from being reviewed.
Add @coderabbitai summary to generate the high-level summary at a specific location in the PR description.
Add @coderabbitai anywhere in the PR title to generate the title automatically.

CodeRabbit Configuration File (`.coderabbit.yaml`)

You can programmatically configure CodeRabbit by adding a .coderabbit.yaml file to the root of your repository.
Please see the configuration documentation for more information.
If your editor has YAML language server enabled, you can add the path at the top of this file to enable auto-completion and validation: # yaml-language-server: $schema=https://coderabbit.ai/integrations/schema.v2.json

Documentation and Community

Visit our Documentation for detailed information on how to use CodeRabbit.
Join our Discord Community to get help, request features, and share feedback.
Follow us on X/Twitter for updates and announcements.

coderabbitai

Actionable comments posted: 1

🧹 Nitpick comments (1)

lmms_eval/models/vllm.py (1)

33-144: Excellent comprehensive documentation with minor formatting improvements needed.

The docstring provides thorough documentation covering all essential aspects of the VLLM class. However, the Python examples could be more Pythonic and better formatted.

Consider these improvements for the examples:

-    Python Example 1: (example of passing model specific arguments)
-    # ---------------------
-    import subprocess
-    cmd = [
+    Example 1: Passing model-specific arguments
+    ```python
+    import subprocess
+    
+    cmd = [
             "python3",
             "-m",
             "lmms_eval",
             "--model",
             "vllm",
             "--model_args",
             "model_version=meta-llama/Llama-4-Scout-17B-16E-Instruct,"
             "tensor_parallel_size=4,"
             "dtype=bfloat16,"
             "max_model_len=10240,"
             "gpu_memory_utilization=0.9,"
-            'override_generation_config={"attn_temperature_tuning": true},' # example of passing model specific arguments, JSON string will be parsed automatically
+            'override_generation_config={"attn_temperature_tuning": true},'  # JSON string will be parsed automatically
             "enforce_eager=True,"
             "kv_cache_dtype=fp8",
             "--tasks",
-            task, # change this to your task
+            "your_task_name",
             "--batch_size",
             "1",
             "--limit",
             "10",
             "--log_samples",
             "--output_path",
             "logs",
-        ]
-    cmd_result = subprocess.run(cmd, check=False)
-    # ---------------------
+    ]
+    result = subprocess.run(cmd, check=False)
+    ```

📜 Review details

Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 71edf36 and ad319e2.

📒 Files selected for processing (1)

lmms_eval/models/vllm.py (2 hunks)

lmms_eval/models/vllm.py

- Load chat template during initialization, supporting both file paths and template strings. - Added error handling for non-existent file paths. - Simplified chat template usage in the chat method.

cursor

Bug: Template String Misinterpretation as File Path

The VLLM class's chat_template argument uses a flawed heuristic to distinguish between file paths and template strings. The conditions os.path.sep in chat_template or chat_template.endswith((".jinja", ".jinja2", ".j2")) can cause valid template strings (e.g., those containing path separators or ending with common template extensions) to be incorrectly interpreted as file paths. This results in a FileNotFoundError when the code attempts to open the non-existent "file", making the auto-detection unreliable.

lmms_eval/models/vllm.py#L173-L182

lmms-eval/lmms_eval/models/vllm.py

Lines 173 to 182 in 7b87d91

    
           # Check if it looks like a file path (contains path separators or has common template extensions) 
        
           if os.path.sep in chat_template or chat_template.endswith((".jinja", ".jinja2", ".j2")): 
        
               # It appears to be a file path, so it must exist 
        
               if not os.path.isfile(chat_template): 
        
                   raise FileNotFoundError(f"Chat template file not found: {chat_template}") 
        
               with open(chat_template, "r") as f: 
        
                   self.chat_template = f.read() 
        
           else: 
        
               # Treat as a template string 
        
               self.chat_template = chat_template

Fix in Cursor • Fix in Web

Was this report helpful? Give feedback by reacting with 👍 or 👎

[Improvement] Accept chat template string in vLLM models

ad319e2

Minor improvement following the previous PR (EvolvingLMMs-Lab#652 (comment)). - Add comments and doc - Accept both chat template file path and chat template string

coderabbitai bot reviewed Jul 21, 2025

View reviewed changes

lmms_eval/models/vllm.py Outdated Show resolved Hide resolved

Luodian approved these changes Jul 22, 2025

View reviewed changes

lint

395ae93

This comment was marked as outdated.

Sign in to view

[Enhancement] Improve chat template handling in VLLM initialization

62b9500

- Load chat template during initialization, supporting both file paths and template strings. - Added error handling for non-existent file paths. - Simplified chat template usage in the chat method.

This comment was marked as outdated.

Sign in to view

lint

7b87d91

cursor bot reviewed Jul 23, 2025

View reviewed changes

Luodian merged commit 8895505 into EvolvingLMMs-Lab:main Jul 23, 2025
3 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[Improvement] Accept chat template string in vLLM models #768

[Improvement] Accept chat template string in vLLM models #768

Uh oh!

VincentYCYao commented Jul 21, 2025 •

edited by coderabbitai bot

Loading

Uh oh!

coderabbitai bot commented Jul 21, 2025 •

edited

Loading

Chat

Support

CodeRabbit Commands (Invoked using PR comments)

Other keywords and placeholders

CodeRabbit Configuration File (`.coderabbit.yaml`)

Documentation and Community

Uh oh!

coderabbitai bot left a comment

Uh oh!

Uh oh!

This comment was marked as outdated.

Uh oh!

This comment was marked as outdated.

Uh oh!

cursor bot left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

	# Check if it looks like a file path (contains path separators or has common template extensions)
	if os.path.sep in chat_template or chat_template.endswith((".jinja", ".jinja2", ".j2")):
	# It appears to be a file path, so it must exist
	if not os.path.isfile(chat_template):
	raise FileNotFoundError(f"Chat template file not found: {chat_template}")
	with open(chat_template, "r") as f:
	self.chat_template = f.read()
	else:
	# Treat as a template string
	self.chat_template = chat_template

[Improvement] Accept chat template string in vLLM models #768

[Improvement] Accept chat template string in vLLM models #768

Uh oh!

Conversation

VincentYCYao commented Jul 21, 2025 • edited by coderabbitai bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary by CodeRabbit

Uh oh!

coderabbitai bot commented Jul 21, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Estimated code review effort

Poem

Chat

Support

CodeRabbit Commands (Invoked using PR comments)

Other keywords and placeholders

CodeRabbit Configuration File (.coderabbit.yaml)

Documentation and Community

Uh oh!

coderabbitai bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

This comment was marked as outdated.

Uh oh!

This comment was marked as outdated.

Uh oh!

cursor bot left a comment

Choose a reason for hiding this comment

Bug: Template String Misinterpretation as File Path

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

VincentYCYao commented Jul 21, 2025 •

edited by coderabbitai bot

Loading

coderabbitai bot commented Jul 21, 2025 •

edited

Loading

CodeRabbit Configuration File (`.coderabbit.yaml`)