LM Evaluation Harness

---
schema: "https://insidethemachine.ai/schema/listing.schema.json"
slug: "lm-evaluation-harness"
name: "LM Evaluation Harness"
kind: "repo"
tier: "catalog"
category: "evals"
category_label: "Evals & benchmarks"
stack_layer: "tools"
homepage: "https://www.eleuther.ai"
repo: "https://github.com/EleutherAI/lm-evaluation-harness"
license: "MIT"
stars: 14234
last_push: "2026-09-14"
archived: false
tags: []
seen_in_videos: []
gap_score: null
gap_score_status: "pending test on Mac Studio"
verified_on: "2026-10-15"
---

# LM Evaluation Harness

**What:** The standard harness for academic LLM benchmarks.

**Why:** Measures how close you are to the frontier.

**Category:** Evals & benchmarks · **Stack layer:** tools

## Status

Catalog entry. Existence and repo metadata checked 2026-10-15 via the GitHub API. Install commands below are auto-extracted from the README and have NOT been executed or verified. Read the README before running anything; prefer a sandbox or venv.

## Install (unverified)

1. [any] source: README (auto-extracted); status: unverified, not executed

```bash
git clone --depth 1 https://github.com/EleutherAI/lm-evaluation-harness
```

2. [any] source: README (auto-extracted); status: unverified, not executed

```bash
pip install "lm_eval[hf]"
```

3. [any] source: README (auto-extracted); status: unverified, not executed

```bash
pip install "lm_eval[vllm]"
```

4. [any] source: README (auto-extracted); status: unverified, not executed

```bash
pip install "lm_eval[api]"
```

## Verify

No verified check yet. Minimum: the install command exits 0 and the tool's own help/version flag runs (exact flag: see README).

## Rollback (unverified)

1. source: derived inverse of install command

```bash
rm -rf https://github.com/EleutherAI/lm-evaluation-harness
```

2. source: derived inverse of install command

```bash
pip uninstall -y lm_eval
```

## Links

- Homepage: https://www.eleuther.ai
- Repo: https://github.com/EleutherAI/lm-evaluation-harness
- README: https://github.com/EleutherAI/lm-evaluation-harness#readme