[Bug] LiveMath default config expects 2-1-7 split, but downloader and run.sh use 2-2-6

Open Beginner friendly
#7 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Assessment

Difficulty
2/5
Estimated time
1-3 hours
Newbie friendliness
75/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Quiet
Tech stack
python, shell

Research direction

Compare configs/livemathematicianbench/default.yaml with the split path in data/download.py and copilot_example/livemath/run.sh. Start by following the documented download workflow, then run the LiveMath training example with the default config. Done means the default config points to the directory created by the downloader and used by the runner.

Written by the indexing model from the issue text.

Description

Description

The default LiveMathematicianBench training configuration points to a split directory that is inconsistent with both the official dataset downloader and the LiveMath example runner.

The config expects:

data/ablation_splits/livemathematicianbench/2-1-7_seed42

However, the official downloader creates, and the LiveMath runner uses:

data/ablation_splits/livemathematicianbench/2-2-6_seed42

As a result, following the documented data-download workflow and then running training with the default LiveMath config fails because the configured directory does not exist.

Relevant files

The default config uses 2-1-7_seed42:

env:
  split_mode: split_dir
  split_dir: data/ablation_splits/livemathematicianbench/2-1-7_seed42

The official downloader creates 2-2-6_seed42:

"livemath": (
    "ablation_splits/livemathematicianbench/2-2-6_seed42",
    {"train": "items.json", "val": "items.json", "test": "items.json"},
),

The LiveMath example runner also uses 2-2-6_seed42:

SPLIT_DIR="${PROJECT_ROOT}/data/ablation_splits/livemathematicianbench/2-2-6_seed42"
Dominant language
Python
Stars
187
Forks
13
PR merge metrics
No merged PRs in 30d

Contributor guide

No contributing guide indexed for this repository

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.

More from EvolvingLMMs-Lab/SkillOpt-Lite

All issues in EvolvingLMMs-Lab/SkillOpt-Lite

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.