in-progress eval log viewer (part 1)#22

Merged

dphuang2 merged 55 commits intomainfrom

in-progress-eval-viewer

Aug 6, 2025

Commits on Aug 5, 2025

initial server file watching works
Dylan Huang
committed
save
Dylan Huang
committed
fix error
Dylan Huang
committed
include dist/ as part of production packaging
Dylan Huang
committed
add favicon
Dylan Huang
committed
implement logger and use in agent rollout processor
Dylan Huang
committed
log rows after evals too
Dylan Huang
committed
add data
Dylan Huang
committed
Merge branch 'main' into in-progress-eval-viewer
Dylan Huang
committed
feedback -> reason
Dylan Huang
committed
add zod types for eval protocol
Dylan Huang
committed
save. fix reloading
Dylan Huang
committed
Update FastAPI dependency to version 0.116.1 and refactor LogsServer to use async context manager for file watching lifecycle. Enhance WebSocket connection handling in the frontend with exponential…
Dylan Huang
committed
Refactor InputMetadata to make row_id optional and update type annotations in Agent class for tool results and tool call execution.
Dylan Huang
committed
save
Dylan Huang
committed
Add Python debug configuration for Uvicorn logs server and refactor LocalFSDatasetLoggerAdapter to use current JSONL file path. Enhance read method to aggregate rows from all JSONL files in the dat…
Dylan Huang
committed
fix
Dylan Huang
committed
fix
Dylan Huang
committed
save
Dylan Huang
committed
remove unnecessary test
Dylan Huang
committed
remove classnames
Dylan Huang
committed
make threshold of succes lower
Dylan Huang
committed
remove old links
Dylan Huang
committed
revert
Dylan Huang
committed
add mobx
Dylan Huang
committed
save
Dylan Huang
committed
set threshold lower
Dylan Huang
committed
data is being set properly
Dylan Huang
committed

Commits on Aug 6, 2025

refactor App component layout and styling for improved UI
Dylan Huang
committed
looks good
Dylan Huang
committed
save
Dylan Huang
committed
refactor state management by moving GlobalState to a separate file
Dylan Huang
committed
more updates
Dylan Huang
committed
use table
Dylan Huang
committed
refactor Dashboard component to simplify dataset summary display by replacing table with div elements
Dylan Huang
committed
log for debugging
Dylan Huang
committed
refactor Row component to replace MessageBubble with ChatInterface and simplify message display
Dylan Huang
committed
Add logo image and update App component header
Dylan Huang
committed
Update chat width calculation to set maximum width as 80% of container width
Dylan Huang
committed
Refactor Row component to use MetadataSection for displaying Evaluation Result and Ground Truth, improving code readability and maintainability.
Dylan Huang
committed
first pass eval metadata
Dylan Huang
committed
add more fields
Dylan Huang
committed
save
Dylan Huang
committed
Refactor ChatInterface component to improve resize handle functionality and positioning, enhancing user experience for chat window adjustments.
Dylan Huang
committed
Add 'passed' field to EvalMetadata and update evaluation logic to determine if the evaluation passed based on the threshold of success.
Dylan Huang
committed
Update Row component to change created date text color from gray-900 to gray-600 for improved visual consistency.
Dylan Huang
committed
Refactor StatusIndicator component to accept status prop, enhancing flexibility for displaying connection and evaluation statuses. Update App and Row components to utilize the new StatusIndicator i…
Dylan Huang
committed
Refactor Agent class to accept EvaluationRow directly in the constructor, simplifying initialization. Update _call_model method to return a Message object with structured response data. Modify defa…
Dylan Huang
committed
fix rollout
Dylan Huang
committed
Refactor evaluation_test to improve error handling and metadata initialization. Ensure eval_metadata is set for each row before rollouts, and enhance exception management to log errors appropriatel…
Dylan Huang
committed
Add .eval_protocol to .gitignore and remove outdated JSONL dataset file. This update ensures that unnecessary files are ignored during version control, streamlining project management.
Dylan Huang
committed
Update test_pytest_mcp_config.py docstring to clarify test purpose for Stdio MCP Config usage in decorator
Dylan Huang
committed
Remove redundant Metadata header from Row component for cleaner UI.
Dylan Huang
committed
Update ChatInterface component to adjust maximum width calculation for chat window resizing from 80% to 66% of the available width, improving layout responsiveness.
Dylan Huang
committed
make messages more dense
Dylan Huang
committed