Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
Python agentaichatgptdeepseekfine-tuninggemmaimage-generationllamallmllmsopenaipythonqwenreinforcement-learningself-hostedstable-diffusiontext-to-speechttsuiunsloth
Docker Hub README: xformers is amd64 only, GB10 runs through PTX, canonical docs link (#10492)
Daniel Han7e931e984ea3
diff --git a/docker/DOCKERHUB.md b/docker/DOCKERHUB.mdindex aee967f1493..e2dadd3489e 100644--- a/docker/DOCKERHUB.md+++ b/docker/DOCKERHUB.md@@ -1,8 +1,8 @@# Unsloth Docker Image-Pre-built images for [Unsloth](https://github.com/unslothai/unsloth): fine-tune and run LLMs, vision, audio and diffusion models with no setup. Every image carries the full training stack (PyTorch 2.11 with CUDA 12.8, Unsloth, unsloth-zoo, bitsandbytes, xformers, TRL, PEFT), JupyterLab with the [Unsloth notebooks](https://github.com/unslothai/notebooks) pre-synced, and prebuilt llama.cpp and whisper.cpp for GGUF work.+Pre-built images for [Unsloth](https://github.com/unslothai/unsloth): fine-tune and run LLMs, vision, audio and diffusion models with no setup. Every image carries the full training stack (PyTorch 2.11 with CUDA 12.8, Unsloth, unsloth-zoo, bitsandbytes, TRL, PEFT, plus xformers on `linux/amd64`), JupyterLab with the [Unsloth notebooks](https://github.com/unslothai/notebooks) pre-synced, and prebuilt llama.cpp and whisper.cpp for GGUF work.-Source: [`docker/`](https://github.com/unslothai/unsloth/tree/main/docker) in the main repository. Guide: [docs.unsloth.ai](https://docs.unsloth.ai/get-started/install-and-update/docker).+Source: [`docker/`](https://github.com/unslothai/unsloth/tree/main/docker) in the main repository. Guide: [docs.unsloth.ai](https://docs.unsloth.ai/get-started/install/docker).## Tags@@ -75,7 +75,7 @@ docker run -d -e UNSLOTH_ALLOW_CPU=1 -p 8000:8000 -p 8888:8888 unsloth/unsloth## Supported GPUs-Compiled for `sm_75 sm_80 sm_86 sm_90 sm_100 sm_120`: Turing (T4, RTX 20), Ampere (A100, A10, RTX 30), Ada (L4, L40, RTX 40), Hopper (H100, H200, GH200), Blackwell (B200, GB200, RTX 50, RTX PRO 6000) and GB10 (DGX Spark). The container prints the detected GPU on start and explains what to do when the driver is too old.+Compiled for `sm_75 sm_80 sm_86 sm_90 sm_100 sm_120`: Turing (T4, RTX 20), Ampere (A100, A10, RTX 30), Ada (L4, L40, RTX 40), Hopper (H100, H200, GH200) and Blackwell (B200, GB200, RTX 50, RTX PRO 6000). GB10 (DGX Spark, `sm_121`) runs the `sm_120` binaries through Blackwell forward compatibility; only kernels compiled at run time, such as Triton, use the CUDA 13 compiler the container switches to on that GPU. The container prints the detected GPU on start and explains what to do when the driver is too old.Driver requirements: