Model Adaptation List (Continuously Updated)
Note
If you have adaptation requirements for other software/models, please contact mailto:hpc@hkust-gz.edu.cn
Training
| Full Model Name | Adaptation Status | Guidance Documentation |
|---|---|---|
| qwen3-30b-a3b | Adapted | Train/fine-tune using MindSpeed-LLM framework, refer to guide: Reference Document, LlamaFactory training reference: Reference Document, training script: Reference Document |
| Qwen2.5-Coder-32B/14B/7B-Instruct; Qwen3-Coder-30B-A3B-Instruct | Adapted | Fine-tuning using MindSpeed-LLM framework reference guide: Reference Document, Reference Document |
| Qwen3-8B-Instruct | Adapted | Full-parameter SFT fine-tuning training using MindSpeed-LLM framework: Reference Document, LlamaFactory training reference: Reference Document, training script: Reference Document |
| Qwen/Qwen3-VL-8B-Instruct | Adapted | Fine-tuning using MindSpeed-MM framework: Reference Document, LlamaFactory training reference: Reference Document, training script: Reference Document |
| Qwen/Qwen3-8B | Adapted | Full-parameter SFT fine-tuning training using MindSpeed-LLM framework: Reference Document, LlamaFactory training reference: Reference Document, training script: Reference Document |
| Qwen/Qwen3-14B | Adapted | Full-parameter SFT fine-tuning training using MindSpeed-LLM framework: Reference Document, LlamaFactory training reference: Reference Document, training script: Reference Document |
| Qwen/Qwen3-235B-A22B-Thinking-2507 | Adapted | MindSpeed LLM installation guide: Reference Document, Large model instruction fine-tuning: Reference Document, fine-tuning script: Reference Document |
| Qwen3-8B | Adapted | Full-parameter SFT fine-tuning training using MindSpeed-LLM framework: Reference Document |
| Qwen3-8B | Adapted | Reference Document |
| Qwen3-8B | Adapted | Reference Document |
| Qwen3-8B | Adapted | Reference links: Reference Document, Reference Document |
| Qwen3-8B | Adapted | Full-parameter SFT fine-tuning training using MindSpeed-LLM framework: Reference Document, LlamaFactory training reference: Reference Document, training script: Reference Document |
| Wan 2.2 | Adapted | Fine-tuning using MindSpeed-MM framework: Reference Document, MindSpeed-MM fine-tuning practice for Wan2.2-T2V-A14B model: Reference Document |
| Qwen2.5-72B-Instruct | Adapted | MindSpeed LLM installation guide: Reference Document, Pre-training: Reference Document, LoRA fine-tuning: Reference Document, Model LoRA fine-tuning script: Reference Document, Pre-training: Reference Document |
| Qwen3-8b | Adapted | MindSpeed-LLM preset dense large models: Reference Document, Installation guide: Reference Document, Fine-tuning guide: Reference Document |
| Qwen3-32b | Adapted | MindSpeed-LLM preset dense large models: Reference Document, Installation guide: Reference Document, Fine-tuning guide: Reference Document |
| Qwen3-VL-32b | Adapted | Fine-tuning using MindSpeed-MM framework: Reference Document, LlamaFactory training reference: Reference Document, training script: Reference Document |
| Qwen3(VL)-4B/8B | Adapted | MindSpeed-MM reinforcement learning: Reference Document |
| Qwen2-7B / LLaMA2-7B | Adapted | MindSpeed-LLM preset dense large models: Reference Document, Installation guide: Reference Document, LoRA fine-tuning: Reference Document, LlaMA2-7B fine-tuning script: Reference Document |
| Qwen-3 series, such as 14B, 32B; LLaMA-3.2 8B/14B | MindSpeed-LLM official website supports LLaMA3.2-1B/3B; VeRL does not currently support LLaMA-3.2 8B/14B and qwen3-14b | DAPO operation instructions: Reference Document, Installation guide: Reference Document, Qwen3-32B model mindspeed-rl reinforcement learning script: Reference Document |
| LLaMA3-8B-Instruct / LLaMA3.1-8B-Instruct | Adapted | MindSpeed-LLM preset dense large models: Reference Document, LoRA fine-tuning: Reference Document, Installation guide: Reference Document, LlamaFactory training reference: Reference Document, LlamaFactory framework training script: Reference Document |
| DeepSeek-R1-Distill-Llama-70B or Llama-3-70B | Adapted | MindSpeed LLM installation guide: Reference Document, Distributed pre-training: Reference Document, llama3-70B pre-training script: Reference Document |
| Deepseek V3.2 | Adapted | MindSpeed LLM installation guide: Reference Document, Fine-tuning script: Reference Document, Model fine-tuning script: Reference Document |
| openai/gpt-oss-120b | gpt-oss-20b supported, 120B not yet | Installation guide: Reference Document, Operation instructions: Reference Document, gpt-oss-20b model fine-tuning script: Reference Document |
Inference
Note
Some reference documents are deployment guidance documents for the same framework and can be used as references.
| Full Model Name | Inference Engine | Adaptation Status | Reference Document |
|---|---|---|---|
| Qwen3-VL-30B-A3B-Instruct | vLLM | Adapted | Reference Document, Reference Document |
| qwen3-30b-a3b | vLLM | Adapted | Reference Document |
| Qwen3-VL 235B-A22B | vLLM | Adapted | Reference Document |
| Qwen3-VL-32B-Thinking | vLLM | Adapted | Reference Document |
| Qwen2.5-Coder-32B/14B/7B-Instruct; Qwen3-Coder-30B-A3B-Instruct | vLLM, sglang | Adapted | Reference Document, Reference Document |
| Qwen3-8B | MindIE / vLLM | Adapted | Reference Document, Reference Document |
| Qwen2-7B | vLLM / MindIE | Adapted | Reference Document, Reference Document |
| Qwen/Qwen3-8B | vLLM | Adapted | Reference Document |
| Qwen/Qwen3-14B | vLLM | Adapted | Reference Document |
| Qwen/Qwen2.5-7B-Instruct | vLLM | Adapted | Reference Document |
| Qwen/Qwen2.5-VL-7B-Instruct | vLLM | Adapted | Reference Document |
| Qwen/Qwen2.5-14B-Instruct | vLLM | Adapted | Reference Document |
| Qwen/Qwen3-VL-Embedding-2B | vLLM | Adapted | Reference Document |
| Qwen/Qwen3-VL-Embedding-8B | vLLM | Adapted | Reference Document |
| Qwen/Qwen3-Embedding-8B | vLLM | Adapted | Reference Document |
| Qwen/Qwen3-Embedding-4B | vLLM | Adapted | Reference Document |
| Qwen/Qwen3-Embedding-0.6B | vLLM | Adapted | Reference Document |
| Qwen/Qwen3-235B-A22B-Thinking-2507 | liteLLM, vLLM | Adapted | Reference Document, Reference Document, Reference Document |
| Qwen3-8B | vLLM | Adapted | Reference Document |
| Qwen3-8B | sglang | Adapted | Reference Document, Reference Document |
| Qwen3-8B | vLLM | Adapted | Reference Document |
| Qwen3-32B | vLLM | Adapted | Reference Document |
| Qwen3-235B | vLLM | Adapted | Reference Document |
| Qwen3-VL-235B | vLLM | Adapted | Reference Document |
| Qwen3-Omini | vLLM | Adapted | Reference Document |
| Qwen3(VL)-4B/8B/32B | vLLM | Adapted | Reference Document |
| Qwen3-235B-A22B/Qwen3-235B-A22B-W8A8 | vllm/omni_infer | Adapted | Reference Document |
| Wan 2.2 | Adapted | Reference Document | |
| DeepSeek-R1-Distill-70B (Int8/W8A8 Quantized Version) | vLLM or MindIE | Adapted | Reference Document |
| Deepseek V3.2 | liteLLM, vLLM | Adapted | Reference Document, Reference Document, Reference Document |
| DeepSeek-V3 | vLLM | Adapted | Reference Document |
| Deepseekocr | vLLM | Adapted | Reference Document |
| Kimi-K2-Thinking | liteLLM, vLLM | Adapted | Reference Document, Reference Document, Reference Document |
| Kimi-Audio | pytorch | Adapted | None available |
| LLaMA3-8B-Instruct | MindIE | Adapted | Reference Document |
| openai/gpt-oss-120b | liteLLM, vLLM | Adapted | Reference Document, Reference Document, Reference Document |
| Whisper-Large-V3 | pytorch | Adapted | Reference Document |
| BAAI/bge-base-en-v1.5 | vLLM | TEI adapted, vllm not yet adapted | Reference Document |
| BAAI/bge-large-en-v1.5 | vLLM | TEI adapted, vllm not yet adapted | Reference Document |
| LLaDA2.0-flash | SgLang | Submitted to model adaptation team, migration adaptation in progress; Successfully served and called based on SGLang -- version 1.20 | Reference Document |
| HunyuanVideo-1.5 | Adapted | Reference Document | |
| speaker-diarization-3.1 | pytorch | Adapted | Reference Document |