--- title: "instruct-eval vs lm-evaluation-harness" type: "comparison" canonical_url: "https://www.graphcanon.com/compare/declare-lab-instruct-eval-vs-eleutherai-lm-evaluation-harness" tools: ["declare-lab-instruct-eval", "eleutherai-lm-evaluation-harness"] --- # instruct-eval vs lm-evaluation-harness *GraphCanon updated Aug 7, 2026* ## Verdict Pick instruct-eval if key facts about instruct-eval; pick lm-evaluation-harness if lm-evaluation-harness is a Python framework for evaluating language models in various parallelism modes using different checkpoint formats, compatible with the Megatron-LM backend. [instruct-eval](https://declare-lab.github.io/instruct-eval/) reports 552 GitHub stars, 45 forks, and 24 open issues, last pushed Mar 10, 2024. [lm-evaluation-harness](https://www.eleuther.ai) has 14k stars, 3.5k forks, and 938 open issues, last pushed Jul 13, 2026. Figures are from public GitHub metadata via [instruct-eval's repository](https://github.com/declare-lab/instruct-eval) and [lm-evaluation-harness's repository](https://github.com/EleutherAI/lm-evaluation-harness). | | [instruct-eval](/tools/declare-lab-instruct-eval.md) | [lm-evaluation-harness](/tools/eleutherai-lm-evaluation-harness.md) | | --- | --- | --- | | Tagline | Quantitative evaluation for instruction-tuned language models | A framework for few-shot evaluation of language models. | | Stars | 552 | 13,560 | | Forks | 45 | 3,467 | | Open issues | 24 | 938 | | Language | Python | Python | | Adopt for | Key facts about instruct-eval | lm-evaluation-harness is a Python framework for evaluating language models in various parallelism modes using different checkpoint formats, compatible with the Megatron-LM backend. | | Persona | - | - | | Runtime | - | - | | License | The tool is distributed under Apache-2.0 license | MIT | | Categories | Evaluation & Observability | Evaluation & Observability | ## Trust and health _Sourced signals - not a safety guarantee. No winner column._ | | [instruct-eval](/tools/declare-lab-instruct-eval.md) | [lm-evaluation-harness](/tools/eleutherai-lm-evaluation-harness.md) | | --- | --- | --- | | Maintenance | Dormant (18%) | Active (82%) | | Days since push | 879d | 24d | | Open issues (now) | 24 | 938 | | Full report | [trust report](/tools/declare-lab-instruct-eval/trust.md) | [trust report](/tools/eleutherai-lm-evaluation-harness/trust.md) | ## Decision facts: instruct-eval - **Requirements:** Min 8 GB RAM; Requires Python environment setup and specific dependencies as outlined in the repository's documentation. - **Adopt for:** Key facts about instruct-eval - **License detail:** The tool is distributed under Apache-2.0 license ## Decision facts: lm-evaluation-harness - **Adopt for:** lm-evaluation-harness is a Python framework for evaluating language models in various parallelism modes using different checkpoint formats, compatible with the Megatron-LM backend. ## Choose when ### Choose instruct-eval if… - License: instruct-eval is Apache-2.0, lm-evaluation-harness is MIT. - Requirements: Min 8 GB RAM; Requires Python environment setup and specific dependencies as outlined in the repository's documentation.. - Tags unique to instruct-eval: benchmarking, evaluation, instruct-tuning, llm. - When you need to quantitatively evaluate the performance of instruction-tuned large language models such as Alpaca and Flan-T5 on held-out tasks. ### Choose lm-evaluation-harness if… - License: lm-evaluation-harness is MIT, instruct-eval is Apache-2.0. - Tags unique to lm-evaluation-harness: data-parallelism, evaluation-framework, expert-parallelism, language-model. - - When you need to evaluate large language models across multiple GPUs in data or tensor parallel configurations. ## When NOT to use instruct-eval - When primarily interested in general model evaluation without a focus on instruction-tuned LMs. - If your primary interest lies in qualitative assessment rather than quantitative metrics. - If you need support for non-HuggingFace Transformer models, as instruct-eval mainly supports models from the HuggingFace ecosystem. ## When NOT to use lm-evaluation-harness - - If your evaluation setup requires pipeline parallelism not currently supported by this framework. ## Common questions ### What is the difference between instruct-eval and lm-evaluation-harness? instruct-eval: Quantitative evaluation for instruction-tuned language models. lm-evaluation-harness: A framework for few-shot evaluation of language models.. See the comparison table for live GitHub stats and shared categories. ### When should I choose instruct-eval over lm-evaluation-harness? Choose instruct-eval over lm-evaluation-harness when License: instruct-eval is Apache-2.0, lm-evaluation-harness is MIT; Requirements: Min 8 GB RAM; Requires Python environment setup and specific dependencies as outlined in the repository's documentation.; Tags unique to instruct-eval: benchmarking, evaluation, instruct-tuning, llm; When you need to quantitatively evaluate the performance of instruction-tuned large language models such as Alpaca and Flan-T5 on held-out tasks. ### When should I choose lm-evaluation-harness over instruct-eval? Choose lm-evaluation-harness over instruct-eval when License: lm-evaluation-harness is MIT, instruct-eval is Apache-2.0; Tags unique to lm-evaluation-harness: data-parallelism, evaluation-framework, expert-parallelism, language-model; - When you need to evaluate large language models across multiple GPUs in data or tensor parallel configurations. ### When should I avoid instruct-eval? When primarily interested in general model evaluation without a focus on instruction-tuned LMs. If your primary interest lies in qualitative assessment rather than quantitative metrics. If you need support for non-HuggingFace Transformer models, as instruct-eval mainly supports models from the HuggingFace ecosystem. ### When should I avoid lm-evaluation-harness? - If your evaluation setup requires pipeline parallelism not currently supported by this framework. ### Is instruct-eval or lm-evaluation-harness more popular on GitHub? lm-evaluation-harness has more GitHub stars (13,560 vs 552). Stars measure visibility, not whether either tool fits your constraints. ### Are instruct-eval and lm-evaluation-harness open source? Yes - both are open-source projects on GitHub (instruct-eval: Apache-2.0, lm-evaluation-harness: MIT). ### Where can I find alternatives to instruct-eval or lm-evaluation-harness? GraphCanon lists graph-backed alternatives at [instruct-eval alternatives](/tools/declare-lab-instruct-eval/alternatives) and [lm-evaluation-harness alternatives](/tools/eleutherai-lm-evaluation-harness/alternatives) ([instruct-eval markdown twin](/tools/declare-lab-instruct-eval/alternatives.md), [lm-evaluation-harness markdown twin](/tools/eleutherai-lm-evaluation-harness/alternatives.md)), ranked by typed relationship edges rather than popularity votes. ### Is there a machine-readable version of this comparison? Yes. The markdown twin at [this comparison](/compare/declare-lab-instruct-eval-vs-eleutherai-lm-evaluation-harness.md) mirrors this page for agents and LLM crawlers, with the same stats table and FAQ answers. ### Which is better maintained, instruct-eval or lm-evaluation-harness? instruct-eval: Dormant. lm-evaluation-harness: Active. Compare maintenance labels, days since push, and release cadence in the trust section below - stars alone do not measure maintenance. ### Where are the full trust reports for instruct-eval and lm-evaluation-harness? GraphCanon publishes per-repo trust reports with dated maintenance, provenance, and scan summaries: [instruct-eval trust report](/tools/declare-lab-instruct-eval/trust); [lm-evaluation-harness trust report](/tools/eleutherai-lm-evaluation-harness/trust). --- **Machine-readable endpoints** - JSON: [`/api/graphcanon/graph?tool=declare-lab-instruct-eval`](/api/graphcanon/graph?tool=declare-lab-instruct-eval) - LLM index: [/llms.txt](/llms.txt) - Full corpus: [/llms-full.txt](/llms-full.txt) _GraphCanon - The knowledge graph for AI development. https://www.graphcanon.com/_