Skip to content

Parse agent traces#8113

Open
lhoestq wants to merge 7 commits intomainfrom
parse_agent_traces
Open

Parse agent traces#8113
lhoestq wants to merge 7 commits intomainfrom
parse_agent_traces

Conversation

@lhoestq
Copy link
Copy Markdown
Member

@lhoestq lhoestq commented Apr 1, 2026

Supports claude_code, codex and pi

JSON Lines files are parsed so one file = one row

The resulting row has three fields: harness, session_id, traces, and file_name with features:

AGENT_TRACES_FEATURES = datasets.Features(
    {
        "harness": datasets.Value("string"),
        "session_id": datasets.Value("string"),
        "traces": datasets.List(datasets.Json()),
        "file_name": datasets.Value("string"),
    }
)

cc @cfahlgren1

@HuggingFaceDocBuilderDev
Copy link
Copy Markdown

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

Copy link
Copy Markdown

@cfahlgren1 cfahlgren1 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

parsing seems to match

lhoestq and others added 3 commits April 8, 2026 21:01
…#8133)

* use full repo-relative file path instead of basename for agent traces file_name

* set base_path to hf://... instead of https://huggingface.co/...

* use relpath

---------

Co-authored-by: Quentin Lhoest <lhoest.q@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants