Repository navigation
feat(web): add saved models and evaluation setup - #104
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Add a Models page for caller-owned Python policies and OpenAI-compatible HTTP models. Users can save bounded browser-local snapshot configurations, export runner-compatible agent.json files, and prepare balanced evaluations with native opponents, reproducible seeds, and optional JSONL/Parquet training trajectories. A downloadable episode script provides a small first run.
The page validates factory options, endpoint roots, environment variable names and runner limits. Credential values stay in the runner environment. Clear local-run instructions, mobile layouts, keyboard controls, storage recovery and clipboard fallback support the workflow without presenting saved configurations as running agents.
Validation: five browser tests passed for both model types, saved configuration lifecycle, validation, downloads, mobile layout and unavailable storage/clipboard. Browser-exported files were loaded by the real Python runner; a test Python policy and local HTTP model stub each completed balanced native games and wrote JSONL datasets. The downloaded episode script also completed a native game. TypeScript, lint and dependency-boundary checks passed; the complete browser suite and GitHub checks run against the integrated replay frontend.