Supported Models
Common Models
The following models are commonly supported in LazyLLM. You can directly use the names provided here as arguments for the TrainableModule. LazyLLM will automatically download the model for you. If you are unable to access the network, you may also specify the absolute path to a local model directly.
Large Language Models
- Baichuan-13B-Chat
- Baichuan-7B
- Baichuan2-13B-Chat
- Baichuan2-13B-Chat-4bits
- Baichuan2-7B-Chat
- Baichuan2-7B-Chat-4bits
- Baichuan2-7B-Intermediate-Checkpoints
- chatglm3-6b
- chatglm3-6b-128k
- chatglm3-6b-32k
- CodeLlama-13b-hf
- CodeLlama-34b-hf
- CodeLlama-70b-hf
- CodeLlama-7b-hf
- glm-4-9b-chat
- internlm-20b
- internlm-7b
- internlm-chat-20b
- internlm-chat-20b-4bit
- internlm-chat-7b
- internlm2-1_8b
- internlm2-20b
- internlm2-7b
- internlm2-chat-1_8b
- internlm2-chat-1_8b-sft
- internlm2-chat-20b
- internlm2-chat-20b-4bits
- internlm2-chat-20b-sft
- internlm2-chat-7b
- internlm2-chat-7b-4bits
- internlm2-chat-7b-sft
- internlm2_5-7b-chat
- internlm2-math-20b
- internlm2-math-7b
- Llama-2-13b
- Llama-2-70b
- Llama-2-7b
- Meta-Llama-3-70B
- Meta-Llama-3-8B
- Qwen-14B
- Qwen-1_8B
- Qwen-72B
- Qwen-7B
- Qwen1.5-0.5B-Chat
- Qwen1.5-1.8B
- Qwen1.5-14B
- Qwen1.5-14B-Chat
- Qwen1.5-4B
- Qwen1.5-72B
- Qwen1.5-7B
- Qwen2-72B-Instruct
- Qwen2-72B-Instruct-AWQ
- Qwen2-7B-Instruct
Multimodal Models
- InternVL-Chat-V1-5
- llava-1.5-13b-hf
- llava-1.5-7b-hf
- Mini-InternVL-Chat-2B-V1-5
- bge-large-zh-v1.5
- bge-reranker-large
- bark
- ChatTTS
- musicgen-medium
- musicgen-stereo-small
- stable-diffusion-3-medium
- SenseVoiceSmall
Other Models
For LLM models not listed in the above list, you can also try using the full secondary name of models provided by huggingface or modelscope, such as: Shanghai_AI_Laboratory/internlm2-chat-1_8b. Pass this name as an argument for the TrainableModule, and LazyLLM will automatically download the model for you.