{"id":16270788,"url":"https://github.com/fboulnois/llama-cpp-docker","last_synced_at":"2026-03-07T01:34:57.625Z","repository":{"id":213444621,"uuid":"734138537","full_name":"fboulnois/llama-cpp-docker","owner":"fboulnois","description":"Run llama.cpp in a GPU accelerated Docker container","archived":false,"fork":false,"pushed_at":"2025-12-16T22:58:28.000Z","size":43,"stargazers_count":62,"open_issues_count":0,"forks_count":15,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-02-25T21:09:35.948Z","etag":null,"topics":["chatgpt","docker","docker-compose","llama","llama-cpp","llama2","llama3","llm","mistral"],"latest_commit_sha":null,"homepage":"","language":"Dockerfile","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/fboulnois.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","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,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2023-12-21T00:58:33.000Z","updated_at":"2026-02-22T21:12:16.000Z","dependencies_parsed_at":"2025-12-26T03:07:34.759Z","dependency_job_id":null,"html_url":"https://github.com/fboulnois/llama-cpp-docker","commit_stats":{"total_commits":57,"total_committers":1,"mean_commits":57.0,"dds":0.0,"last_synced_commit":"9a35ebc6bcee56dc5029506f1b019c7f305ba702"},"previous_names":["fboulnois/llama-cpp-docker"],"tags_count":25,"template":false,"template_full_name":null,"purl":"pkg:github/fboulnois/llama-cpp-docker","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/fboulnois%2Fllama-cpp-docker","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/fboulnois%2Fllama-cpp-docker/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/fboulnois%2Fllama-cpp-docker/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/fboulnois%2Fllama-cpp-docker/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/fboulnois","download_url":"https://codeload.github.com/fboulnois/llama-cpp-docker/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/fboulnois%2Fllama-cpp-docker/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30205196,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-06T19:07:06.838Z","status":"ssl_error","status_checked_at":"2026-03-06T18:57:34.882Z","response_time":250,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.6:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"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":["chatgpt","docker","docker-compose","llama","llama-cpp","llama2","llama3","llm","mistral"],"created_at":"2024-10-10T18:11:36.493Z","updated_at":"2026-03-07T01:34:57.586Z","avatar_url":"https://github.com/fboulnois.png","language":"Dockerfile","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Llama.cpp in Docker\n\nRun [llama.cpp](https://github.com/ggerganov/llama.cpp) in a GPU accelerated\nDocker container.\n\n## Minimum requirements\n\nBy default, the service requires a CUDA capable GPU with at least 8GB+ of VRAM. \nIf you don't have an Nvidia GPU with CUDA then the CPU version will be built and\nused instead.\n\n## Quickstart\n\n```bash\nmake build\nmake llama-3-8b\nmake up\n```\n\nAfter starting up the chat server will be available at `http://localhost:8080`.\n\n## Options\n\nOptions are specified as environment variables in the `docker-compose.yml` file.\nBy default, the following options are set:\n\n* `GGML_CUDA_NO_PINNED`: Disable pinned memory for compatability (default is 1)\n* `LLAMA_ARG_CTX_SIZE`: The context size to use (default is 2048)\n* `LLAMA_ARG_MODEL`: The name of the model to use (default is `/models/Meta-Llama-3.1-8B-Instruct-Q5_K_M.gguf`)\n* `LLAMA_ARG_N_GPU_LAYERS`: The number of layers to run on the GPU (default is 99)\n\nSee the [llama.cpp documentation](https://github.com/ggerganov/llama.cpp/tree/master/examples/server)\nfor the complete list of server options.\n\n## Models\n\nThe [`docker-entrypoint.sh`](docker-entrypoint.sh) has targets for downloading\npopular models. Run `./docker-entrypoint.sh --help` to list available models.\nDownload models by running `./docker-entrypoint.sh \u003cmodel\u003e` where `\u003cmodel\u003e` is\nthe name of the model. By default, these will download the `_Q5_K_M.gguf`\nversions of the models. These models are quantized to 5 bits which provide a\ngood balance between speed and accuracy.\n\nConfused about which model to use? Below is a list of popular models, ranked by\n[ELO rating](https://en.wikipedia.org/wiki/Elo_rating_system). Generally, the\nhigher the ELO rating the better the model.\n\n| Target | Model | Parameters | Size | [~ELO](https://chat.lmsys.org/?leaderboard) | Notes |\n| --- | --- | --- | --- | --- | --- |\n| deepseek-r1-qwen-14b | [`deepseek-r1-distill-qwen-14b`](https://huggingface.co/bartowski/DeepSeek-R1-Distill-Qwen-14B-GGUF) | 14B | 10.5 GB | 1375 | The best small thinking model |\n| gemma-3-27b | [`gemma-3-27b-it`](https://huggingface.co/bartowski/google_gemma-3-27b-it-GGUF) | 27B | 19.27 GB | 1361 | Google's best medium model |\n| mistral-small-3 | [`mistral-small-3.2-24b-instruct`](https://huggingface.co/bartowski/mistralai_Mistral-Small-3.2-24B-Instruct-2506-GGUF) | 24B | 16.76 GB | 1273 | Mistral AI's best small model |\n| llama-3-8b | [`meta-llama-3.1-8b-instruct`](https://huggingface.co/bartowski/Meta-Llama-3.1-8B-Instruct-GGUF) | 8B | 5.73 GB | 1193 | Meta's best small model |\n| phi-4-mini | [`phi-4-mini-instruct`](https://huggingface.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF) | 4B | 2.85 GB | 1088++ | Microsoft's best tiny model |\n\n\u003e [!NOTE]\n\u003e Values with `+` are minimum estimates from previous versions of the model due\n\u003e to missing data.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ffboulnois%2Fllama-cpp-docker","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Ffboulnois%2Fllama-cpp-docker","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ffboulnois%2Fllama-cpp-docker/lists"}