draft_models:-name:Qwen3-8B-EAGLE3algorithm:eagle3source:huggingfacehuggingface_repo_id:Tengyunw/qwen3_8b_eagle3model_sets:-name:Deepseek R1 0528description:DeepSeek-R1-0528 is a minor version of the DeepSeek R1 model that features enhanced reasoning depth and inference capabilities. These improvements are achieved through increased computational resources and algorithmic optimizations applied during post-training. The model delivers strong performance across a range of benchmark evaluations, including mathematics, programming, and general logic, with overall capabilities approaching those of leading models such as O3 and Gemini 2.5 Pro.home:https://www.deepseek.comicon:/static/catalog_icons/deepseek.pngcategories:-llmcapabilities:-context/128Ksize:671licenses:-mitrelease_date:"2025-05-28"specs:-mode:throughputquantization:FP8gpu_filters:vendor:nvidiacompute_capability:">=9.0"# Hopper or latersource:huggingfacehuggingface_repo_id:deepseek-ai/DeepSeek-R1-0528backend:SGLangbackend_parameters:---enable-dp-attention---context-length=32768-mode:standardquantization:FP8source:huggingfacehuggingface_repo_id:deepseek-ai/DeepSeek-R1-0528backend:vLLMbackend_parameters:---max-model-len=32768
在隔离环境中使用模型目录
内置模型目录从 Hugging Face 或 ModelScope 获取模型。如果在无法访问互联网的隔离环境中使用 GPUStack,可以自定义模型目录,改用本地路径模型来源。以下是示例:
model_sets:-name:Deepseek R1 0528description:DeepSeek-R1-0528 is a minor version of the DeepSeek R1 model that features enhanced reasoning depth and inference capabilities. These improvements are achieved through increased computational resources and algorithmic optimizations applied during post-training. The model delivers strong performance across a range of benchmark evaluations, including mathematics, programming, and general logic, with overall capabilities approaching those of leading models such as O3 and Gemini 2.5 Pro.home:https://www.deepseek.comicon:/static/catalog_icons/deepseek.pngcategories:-llmcapabilities:-context/128Ksize:671licenses:-mitrelease_date:"2025-05-28"specs:-mode:throughputquantization:FP8gpu_filters:vendor:nvidiacompute_capability:">=9.0"# Hopper or latersource:local_pathlocal_path:/path/to/DeepSeek-R1-0528backend:SGLangbackend_parameters:---enable-dp-attention---context-length=32768-mode:standardquantization:FP8source:local_path# assuming you have /path/to/DeepSeek-R1-0528 directorylocal_path:/path/to/DeepSeek-R1-0528backend:vLLMbackend_parameters:---max-model-len=32768