{"id":15779519,"url":"https://github.com/jim60105/docker-moe-llava","last_synced_at":"2026-02-24T09:32:32.774Z","repository":{"id":241584840,"uuid":"807000543","full_name":"jim60105/docker-MoE-LLaVA","owner":"jim60105","description":"This is the docker image for gesen2egee/MoE-LLaVA-hf, a script that uses MoE-LLaVA to describe images. It is designed to prepare the training set caption for stable diffusion model training. (Dockerfile, CI image build)","archived":false,"fork":false,"pushed_at":"2024-05-31T04:49:47.000Z","size":111,"stargazers_count":2,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"master","last_synced_at":"2025-05-29T09:57:07.549Z","etag":null,"topics":["docker-image","dockerfile"],"latest_commit_sha":null,"homepage":"","language":"Dockerfile","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"gpl-3.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/jim60105.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,"publiccode":null,"codemeta":null}},"created_at":"2024-05-28T09:51:02.000Z","updated_at":"2024-08-07T12:36:37.000Z","dependencies_parsed_at":"2024-05-29T06:19:14.644Z","dependency_job_id":"6461a579-3db7-4165-95bf-d8d041c39b4c","html_url":"https://github.com/jim60105/docker-MoE-LLaVA","commit_stats":null,"previous_names":["jim60105/docker-moe-llava"],"tags_count":0,"template":false,"template_full_name":"jim60105/Containerfile-template","purl":"pkg:github/jim60105/docker-MoE-LLaVA","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jim60105%2Fdocker-MoE-LLaVA","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jim60105%2Fdocker-MoE-LLaVA/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jim60105%2Fdocker-MoE-LLaVA/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jim60105%2Fdocker-MoE-LLaVA/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/jim60105","download_url":"https://codeload.github.com/jim60105/docker-MoE-LLaVA/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jim60105%2Fdocker-MoE-LLaVA/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":29777881,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-02-24T04:54:30.205Z","status":"ssl_error","status_checked_at":"2026-02-24T04:53:58.628Z","response_time":75,"last_error":"SSL_read: 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":["docker-image","dockerfile"],"created_at":"2024-10-04T18:20:23.148Z","updated_at":"2026-02-24T09:32:32.743Z","avatar_url":"https://github.com/jim60105.png","language":"Dockerfile","funding_links":[],"categories":[],"sub_categories":[],"readme":"# docker-MoE-LLaVA\n\nThis is the docker image for [gesen2egee/MoE-LLaVA-hf](https://github.com/gesen2egee/MoE-LLaVA-hf), a script that uses [MoE-LLaVA](https://github.com/PKU-YuanGroup/MoE-LLaVA) to describe images. It is designed to prepare the training set caption for stable diffusion model training.\n\nGet the Dockerfile at [GitHub](https://github.com/jim60105/docker-MoE-LLaVA), or pull the image from [ghcr.io](https://ghcr.io/jim60105/moe-llava).\n\n## 🚀 Get your Docker ready for GPU support\n\n### Windows\n\nOnce you have installed [**Docker Desktop**](https://www.docker.com/products/docker-desktop/), [**CUDA Toolkit**](https://developer.nvidia.com/cuda-downloads), [**NVIDIA Windows Driver**](https://www.nvidia.com.tw/Download/index.aspx), and ensured that your Docker is running with [**WSL2**](https://docs.docker.com/desktop/wsl/#turn-on-docker-desktop-wsl-2), you are ready to go.\n\nHere is the official documentation for further reference.  \n\u003chttps://docs.nvidia.com/cuda/wsl-user-guide/index.html#nvidia-compute-software-support-on-wsl-2\u003e\n\u003chttps://docs.docker.com/desktop/wsl/use-wsl/#gpu-support\u003e\n\n### Linux, OSX\n\nInstall an NVIDIA GPU Driver if you do not already have one installed.  \n\u003chttps://docs.nvidia.com/datacenter/tesla/tesla-installation-notes/index.html\u003e\n\nInstall the NVIDIA Container Toolkit with this guide.  \n\u003chttps://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html\u003e\n\n## 📦 Available Pre-built Image\n\nYou can pull the pre-build image which **does not include the models** from the GitHub Container Registry.  \nThese images will download the models at runtime.\n\nMount the current directory as `/dataset` and run the script with additional input arguments.\n\n\u003e [!IMPORTANT]  \n\u003e Remember to prepend `--` before the arguments.\n\n```bash\ndocker run --gpus all -it -v \".:/dataset\" ghcr.io/jim60105/moe-llava:no_model -- [arguments]\n# Example\ndocker run --gpus all -it -v \".:/dataset\" ghcr.io/jim60105/moe-llava:no_model -- --moe --force --caption_style='mixed' --folder_name --modify_prompt --low_vram\n```\n\nThe `[arguments]` placeholder should be replaced with the [arguments for the script](https://github.com/gesen2egee/MoE-LLaVA-hf/blob/main/predict.py#L352-L360). Check the [original colab notebook](https://github.com/gesen2egee/MoE-LLaVA-hf/blob/main/MoE_LLaVA_jupyter.ipynb) for more information.\n\n## ⚡️ Preserve the download cache for the models\n\nYou can mount the `/.cache` to share model caches between containers.  \nIn this way, they will not be repeatedly downloaded every time when image start.\n\n```bash\ndocker run --gpus all -it -v \".:/dataset\" -v \"moe_cache:/.cache\" ghcr.io/jim60105/moe-llava:no_model -- --moe --force --caption_style='mixed' --folder_name --modify_prompt --low_vram\n```\n\n## 🛠️ Building the Image *include models*\n\n\u003e [!CAUTION]  \n\u003e These models are extremely big! They inflate the image size to a whopping 40GB 😕  \n\u003e It is too time-consuming to build and I suggest avoiding it.  \n\u003e Please use the `no_model` image and attaching the `/.cache` volume as instructed earlier.  \n\u003e ![image](https://github.com/jim60105/docker-MoE-LLaVA/assets/16995691/17a58c24-8e2f-4d73-aa77-9495f9a1ccfb)\n\n\u003e [!IMPORTANT]  \n\u003e Clone the Git repository recursively to include submodules:  \n\u003e `git clone --recursive https://github.com/jim60105/docker-MoE-LLaVA.git`\n\nYou can build the image which includes the models by targeting to the final stage.  \nUse the `LOW_VRAM` build argument and to choose the model to preload.\n\n- (No build-arg): Preload the `LanguageBind/MoE-LLaVA-Phi2-2.7B-4e` model.\n- `LOW_VRAM=1`: Preload the `LanguageBind/MoE-LLaVA-StableLM-1.6B-4e-384` model.\n\n```bash\ndocker build -t moe-llava --target final --build-arg LOW_VRAM=1 .\n```\n\n## 📝 LICENSE\n\n\u003e [!NOTE]  \n\u003e The main program, [PKU-YuanGroup/MoE-LLaVA](https://github.com/PKU-YuanGroup/MoE-LLaVA) and [the predict script](https://github.com/gesen2egee/MoE-LLaVA-hf/blob/main/LICENSE), is distributed under [Apache License 2.0](https://github.com/PKU-YuanGroup/MoE-LLaVA/blob/main/LICENSE).  \n\u003e Please consult their repository for access to the source code and licenses.  \n\u003e The following is the license for the Dockerfiles and CI workflows in this repository.\n\n\u003cimg src=\"https://github.com/jim60105/docker-MoE-LLaVA/assets/16995691/65f76d01-a00b-4a93-86b6-a06bc3667869\" alt=\"gplv3\" width=\"300\" /\u003e\n\n[GNU GENERAL PUBLIC LICENSE Version 3](LICENSE)\n\nThis program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.\n\nThis program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.\n\nYou should have received a copy of the GNU General Public License along with this program. If not, see \u003chttps://www.gnu.org/licenses/\u003e.\n\n\u003e [!CAUTION]\n\u003e A GPLv3 licensed Dockerfile means that you _**MUST**_ **distribute the source code with the same license**, if you\n\u003e\n\u003e - Re-distribute the image. (You can simply point to this GitHub repository if you doesn't made any code changes.)\n\u003e - Distribute a image that uses code from this repository.\n\u003e - Or **distribute a image based on this image**. (`FROM ghcr.io/jim60105/moe-llava` in your Dockerfile)\n\u003e\n\u003e \"Distribute\" means to make the image available for other people to download, usually by pushing it to a public registry. If you are solely using it for your personal purposes, this has no impact on you.\n\u003e\n\u003e Please consult the [LICENSE](LICENSE) for more details.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjim60105%2Fdocker-moe-llava","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fjim60105%2Fdocker-moe-llava","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjim60105%2Fdocker-moe-llava/lists"}