Platform Built-in Images/Models
Continuously Updated
Container Images
| Image Name | Version | Description |
|---|---|---|
| langchain-ascend | A3-ubuntu22.03-py311 | LangChain for NPU |
| lmsysorg/sglang | main-cann8.3.rc2-a3 | SGLang for NPU |
| pyscf | 2.11.0 | PySCF |
| ascend/vllm-ascend | v0.13.0rc1-a3 | vLLM for NPU |
| ascendhub/mis-tei | 7.2.RC1-800I-A3-aarch64 | TEI images for NPU |
| llama-factory | 910c-8.3.rc2-py3.11 | LlamaFactory fine-tuning |
| llama-factory | 910c-8.2.rc1-py3.11 | LlamaFactory |
| ascend-pytorch | 2.8.0-A3-ubuntu22.03-py311 | Ascend PyTorch |
| mindie | 2.1.RC2-800I-A3-py311-openeuler24.03-lts | MindIE |
| mindie | 2.2.RC1-800I-A3-py311-openeuler24.03-lts | MindIE |
| ascend-mindspore | 2.7.1-A3-ubuntu22.04-py39 | MindSpore |
| apharbor/ubuntu-arm64 | 18.04 | Ubuntu 18.04 (ARM64) |
| apharbor/ubuntu-arm64 | 22.04 | Ubuntu 22.04 (ARM64) |
| apharbor/ubuntu-arm64 | 20.04 | Ubuntu 20.04 (ARM64) |
Pre-configured Models
info
- Code Development: When creating a project under [Code Development], check the [Model Secondary Development] option. The model files will be automatically mounted to the ~/models directory in the container development environment
- Model Training: Pre-configured models can be selected for training under [Model Training]
- Central Deployment: Pre-configured models can be selected for inference deployment under [Central Deployment] - [Online Services]
| Model Name | Framework | Platform Address | Function |
|---|---|---|---|
| Qwen3-VL-235B-A22B | vllm-ascend | AI Studio (Login to HPC4 Platform required) | Inference |
| DeepSeek-V3.2-Exp-w8a8 | vllm-ascend | AI Studio (Login to HPC4 Platform required) | Inference |
| DeepSeek-V3-0324-w8a8-mindie | mindie | AI Studio (Login to HPC4 Platform required) | Inference |
| Qwen3-235B-A22B-Thinking-2507 | sglang | AI Studio (Login to HPC4 Platform required) | Inference |
| Qwen3-Embedding | vllm-ascend | AI Studio (Login to HPC4 Platform required) | Inference |
| LLamafactory-llama3-8b | llamafactory | AI Studio (Login to HPC4 Platform required) | Training, Evaluation, Fine-tuning |
| DeepSeek-V3.2-Exp-PD | vllm-ascend | AI Studio (Login to HPC4 Platform required) | PD Disaggregation Large Model Inference |
| Qwen-Lite-Model-Union | vllm-ascend | AI Studio (Login to HPC4 Platform required) | Qwen Lightweight Model Collection |
| Qwen-Coder-Union | vllm-ascend | AI Studio (Login to HPC4 Platform required) | Qwen-Coder Model Collection |
| DeepSeek-R1-Distill-Llama-70B | vllm-ascend | AI Studio (Login to HPC4 Platform required) | Distilled Model |