norrisaftcc / norrisaftcc/algorithm-liza

M3: Eval runner and JSON report

Open
#12 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

area:eval milestone:M3 type:feat
Dominant language
Python
Stars
0
Forks
0
Avg merge
1m
Merged PRs (30d)
1

Description

liza-eval --model <m> --protocol <p> runs the full task suite unattended in isolated working directories and emits a JSON report: per task, pass/fail, iteration count, wall time, token counts, and the transcript path.

Unattended is the operative word. A local model will hang, loop, and produce output that crashes the harness; the runner has to survive all three and record them as results rather than aborting the suite. Per-task timeouts are mandatory.

Done when a full suite run completes against a local model without supervision and the report is diffable between runs.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start at the liza-eval --model <m> --protocol <p> CLI entry point and trace how the task suite currently runs, if implemented. Define the runner's handling of hangs, loops, harness crashes, timeouts, and per-task results; done means an unattended full-suite run completes and emits a diffable JSON report containing the required fields.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
ai, cli, testing-qa
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.