{"id":13652321,"url":"https://github.com/arthur-ai/bench","last_synced_at":"2025-12-14T07:36:42.711Z","repository":{"id":188792832,"uuid":"663544319","full_name":"arthur-ai/bench","owner":"arthur-ai","description":"A tool for evaluating LLMs","archived":false,"fork":false,"pushed_at":"2024-05-10T14:20:46.000Z","size":10932,"stargazers_count":424,"open_issues_count":1,"forks_count":43,"subscribers_count":11,"default_branch":"main","last_synced_at":"2025-09-02T13:54:06.580Z","etag":null,"topics":["llm","mlops"],"latest_commit_sha":null,"homepage":"https://bench.readthedocs.io","language":"TypeScript","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/arthur-ai.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null}},"created_at":"2023-07-07T14:40:39.000Z","updated_at":"2025-08-04T06:33:18.000Z","dependencies_parsed_at":"2023-08-16T22:27:37.677Z","dependency_job_id":"c34d57a7-ce5c-43d0-a029-8b9cd40cc8cd","html_url":"https://github.com/arthur-ai/bench","commit_stats":null,"previous_names":["arthur-ai/bench"],"tags_count":12,"template":false,"template_full_name":null,"purl":"pkg:github/arthur-ai/bench","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/arthur-ai%2Fbench","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/arthur-ai%2Fbench/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/arthur-ai%2Fbench/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/arthur-ai%2Fbench/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/arthur-ai","download_url":"https://codeload.github.com/arthur-ai/bench/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/arthur-ai%2Fbench/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":27721609,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-12-14T02:00:11.348Z","response_time":56,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["llm","mlops"],"created_at":"2024-08-02T02:00:58.306Z","updated_at":"2025-12-14T07:36:42.694Z","avatar_url":"https://github.com/arthur-ai.png","language":"TypeScript","funding_links":[],"categories":["Tools","A01_文本生成_文本对话"],"sub_categories":["大语言对话模型及数据"],"readme":"# Bench\n\nBench is a tool for evaluating LLMs for production use cases. Whether you are comparing different LLMs, considering different prompts, or testing generation hyperparameters like temperature and # tokens, Bench provides one touch point for all your LLM performance evaluation.\n\nIf you have encountered a need for any of the following in your LLM work, then Bench can help with your evaluation:\n\n- to standardize the workflow of LLM evaluation with a common interface across tasks and use cases\n- to test whether open source LLMs can do as well as the top closed-source LLM API providers on your specific data\n- to translate the rankings on LLM leaderboards and benchmarks into scores that you care about for your actual use case\n\nJoin the bench community on [Discord](https://discord.gg/tdfUAtaVHz).\n\nFor bug fixes and feature requests, please file a Github issue.\n\n## Package installation\n\nInstall Bench to your python environment with optional dependencies for serving results locally (recommended):\n`pip install 'arthur-bench[server]'`\n\nAlternatively, install Bench to your python environment with minimum dependencies:\n`pip install arthur-bench`\n\nFor further setup instructions visit our [installation guide](https://bench.readthedocs.io/en/latest/setup.html)\n\n## Using Bench\n\nFor a more in-depth walkthrough of using bench, visit our [quickstart walkthrough](https://bench.readthedocs.io/en/latest/quickstart.html) and our [test suite creation guide](https://bench.readthedocs.io/en/latest/creating_test_suites.html) on our docs.\n\nTo make sure you can run test suites in bench, you can run the following code snippets to create a test suite and run it to give a score to candidate outputs.\n\n```python\nfrom arthur_bench.run.testsuite import TestSuite\nsuite = TestSuite(\n    \"bench_quickstart\",\n    \"exact_match\",\n    input_text_list=[\"What year was FDR elected?\", \"What is the opposite of down?\"],\n    reference_output_list=[\"1932\", \"up\"]\n)\nsuite.run(\"quickstart_run\", candidate_output_list=[\"1932\", \"up is the opposite of down\"])\n```\n\nSaved test suites can be loaded later on to benchmark test performance over time, without needing to re-prepare reference data:\n\n```python\nexisting_suite = TestSuite(\"bench_quickstart\", \"exact_match\")\nexisting_suite.run(\"quickstart_new_run\", candidate_output_list=[\"1936\", \"up\"])\n```\n\nTo view the results for these runs in the local UI that comes with the `bench` package, run `bench` from the command line (this requires the bench optional server dependencies to be installed):\n\n```\nbench\n```\n\nViewing examples in the bench UI will look something like this:\n\u003cp align=\"center\"\u003e\n\u003cimg src=\"docs/source/_static/img/Bench_UI_Screenshot.png\" alt=\"Examples UI\" width=\"1100\"/\u003e\n\n## Running Bench from source\n\nTo launch Bench from source:\n\n1. Install the dependencies\n    * `pip install -e '.[server]'`\n2. Build the Front End\n    * `cd arthur_bench/server/js`\n    * `npm i`\n    * `npm run build`\n3. Launch the server\n    * `bench`\n\nBecause the server was installed with `pip -e`, local changes will be picked up. However, the server will need to be restarted between\nchanges in order for those changes to be picked up.\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Farthur-ai%2Fbench","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Farthur-ai%2Fbench","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Farthur-ai%2Fbench/lists"}