VastModelZOO是瀚博半导体维护的AI模型平台,提供了人工智能多个领域(CV、AUDIO、NLP、LLM、MLLM等)的开源模型在瀚博训推芯片上的部署、训练示例。
VastModelZOO旨在基于瀚博半导体的硬件产品和软件SDK,展示最佳的编程实践,以达成模型的快速移植和最优性能。
为方便大家使用VastModelZOO,我们将持续增加典型模型和基础插件。
基于瀚博半导体的硬件产品使用VastModelZOO前,需联系销售代表获取部署软件包。
- 版本号:VVI-25.11
CV Models
- classification
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| CSPNet | timm | model name
| classification | Build_In |
| CSPNet | mmcls | model name
| classification | Build_In |
| CSPNet | ppcls | model name
| classification | Build_In |
| DenseNet | timm | model name
| classification | Build_In |
| DenseNet | torchvision | model name
| classification | Build_In |
| DenseNet | mmcls | model name
| classification | Build_In |
| DenseNet | keras | model name
| classification | Build_In |
| DenseNet | ppcls | model name
| classification | Build_In |
| DLANet | timm | model name
| classification | Build_In |
| DLANet | ppcls | model name
| classification | Build_In |
| DLANet | ucbdrive | model name
| classification | Build_In |
| EfficientNet | timm | model name
| classification | Build_In |
| EfficientNet_v2 | timm | model name
| classification | Build_In |
| ESNet | ppcls | model name
| classification | Build_In |
| GhostNet | timm | model name
| classification | Build_In |
| GhostNet | ppcls | model name
| classification | Build_In |
| HRNet | official | model name
| classification | Build_In |
| HRNet | mmcls | model name
| classification | Build_In |
| HRNet | timm | model name
| classification | Build_In |
| HRNet | ppcls | model name
| classification | Build_In |
| Inception_v3 | timm | model name
| classification | Build_In |
| Inception_v3 | torchvision | model name
| classification | Build_In |
| Inception_v3 | ppcls | model name
| classification | Build_In |
| Inception_v4 | timm | model name
| classification | Build_In |
| Inception_v4 | ppcls | model name
| classification | Build_In |
| IResNet | official | model name
| classification | Build_In |
| MNasNet | torchvision | model name
| classification | Build_In |
| MobileNet_v1 | keras | model name
| classification | Build_In |
| MobileNet_v1 | ppcls | model name
| classification | Build_In |
| MobileNet_v2 | torchvision | model name
| classification | Build_In |
| MobileNet_v2 | timm | model name
| classification | Build_In |
| MobileNet_v2 | mmcls | model name
| classification | Build_In |
| MobileNet_v2 | keras | model name
| classification | Build_In |
| MobileNet_v2 | ppcls | model name
| classification | Build_In |
| MobileNet_v3 | torchvision | model name
| classification | Build_In |
| MobileNet_v3 | timm | model name
| classification | Build_In |
| MobileNet_v3 | ppcls | model name
| classification | Build_In |
| MobileNet_v3 | showlo | model name
| classification | Build_In |
| MobileNet_v3 | sqlai | model name
| classification | Build_In |
| MobileOne | official | model name
| classification | Build_In |
| MobileViT | apple | model name
| classification | Build_In |
| PPLCNet | ppcls | model name
| classification | Build_In |
| PPLCNet_v2 | ppcls | model name
| classification | Build_In |
| RepOPT | official | model name
| classification | Build_In |
| RepVGG | official | model name
| classification | Build_In |
| Res2Net | ppcls | model name
| classification | Build_In |
| ResNeSt | official | model name
| classification | Build_In |
| ResNet | timm | model name
| classification | Build_In |
| ResNet | torchvision | model name
| classification | Build_In |
| ResNet | mmcls | model name
| classification | Build_In |
| ResNet | ppcls | model name
| classification | Build_In |
| ResNet | keras | model name
| classification | Build_In |
| ResNet | oneflow | model name
| classification | Build_In |
| ResNeXt | timm | model name
| classification | Build_In |
| ResNeXt | torchvision | model name
| classification | Build_In |
| ResNeXt | mmcls | model name
| classification | Build_In |
| ResNeXt | ppcls | model name
| classification | Build_In |
| SENet | timm | model name
| classification | Build_In |
| ShuffleNet_v1 | mmcls | model name
| classification | Build_In |
| ShuffleNet_v1 | megvii | model name
| classification | Build_In |
| ShuffleNet_v2 | torchvision | model name
| classification | Build_In |
| ShuffleNet_v2 | mmcls | model name
| classification | Build_In |
| ShuffleNet_v2 | ppcls | model name
| classification | Build_In |
| ShuffleNet_v2 | megvii | model name
| classification | Build_In |
| Swin | microsoft | model name
| classification | Build_In |
| VGG | timm | model name
| classification | Build_In |
| VGG | torchvision | model name
| classification | Build_In |
| VGG | mmcls | model name
| classification | Build_In |
| VGG | ppcls | model name
| classification | Build_In |
| ViT | huggingface | model name
| classification | Build_In |
| VOVNet | timm | model name
| classification | Build_In |
| Wide ResNet | timm | model name
| classification | Build_In |
| Wide ResNet | torchvision | model name
| classification | Build_In |
| Xception | ppcls | model name
| classification | Build_In |
- object detection
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| CenterNet | official | model name
| object detection | Build_In |
| DETR | model name
| object detection | Build_In | |
| FastestDet | official | model name
| object detection | Build_In |
| Grounding-DINO | official | model name
| object detection | Build_In |
| NanoDet | official | model name
| object detection | Build_In |
| RetinaNet | pytorch | model name
| object detection | Build_In |
| RetinaNet | mlcommons | model name
| object detection | Build_In |
| RetinaNet | mmdet | model name
| object detection | Build_In |
| RT-DETR | official | model name
| object detection | Build_In |
| YOLO-World | official | model name
| object detection | Build_In |
| YOLOF | mmdet | model name
| object detection | Build_In |
| YoloFastV2 | official | model name
| object detection | Build_In |
| YOLOR | official | model name
| object detection | Build_In |
| Yolov3 | pytorch(u) | model name
| object detection | Buid_In |
| Yolov4 | darknet | model name
| object detection | Buid_In |
| Yolov4 | bubbliiiing | model name
| object detection | Buid_In |
| Yolov4 | tianxiaomo | model name
| object detection | Buid_In |
| Yolov5 | pytorch(u) | model name
| object detection | Build_In |
| Yolov5 | mmyolo | model name
| object detection | Build_In |
| yolov6 | yolov6 | model name
| object detection | Build_In |
| yolov7 | yolov7 | model name
| object detection | Build_In |
| yolov8 | yolov8 | model name
| object detection | Build_In |
| yolov10 | yolov10 | model name
| object detection | Build_In |
| Yolov11 | official | model name
| object detection | Build_In |
| Yolov12 | official | model name
| object detection | Build_In |
| Yolox | official | model name
| object detection | Build_In |
| Yolox | mmyolo | model name
| object detection | Build_In |
- segmentation
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| BiSeNet | pytorch | model name
| segmentation | Build_In |
| BiSeNet | pytorch | model name
| face segmentation | Build_In |
| Deeplab_v3 | pytorch | model name
| segmentation | Build_In |
| Deeplab_v3_plus | pytorch | model name
| segmentation | Build_In |
| FCN | pytorch | model name
| segmentation | Build_In |
| FCN | mmseg | model name
| segmentation | Build_In |
| HRNet_Seg | pytorch | model name
| segmentation | Build_In |
| Human_Seg | pytorch | model name
| human segmentation | Build_In |
| Mask2Former | official | model name
| instance segmentation | Build_In |
| MODNet | official | model name
| matting | Build_In |
| ParseNet | pytorch | model name
| face segmentation | Build_In |
| R2U_Net | pytorch | model name
| segmentation | Build_In |
| RefineNet | pytorch | model name
| segmentation | Build_In |
| STDC | official | model name
| segmentation | Build_In |
| Unet | bubbliiiing | model name
| segmentation | Build_In |
| Unet | milesial | model name
| segmentation | Build_In |
| Unet | keras | model name
| segmentation | Build_In |
| Unet3P | pytorch | model name
| segmentation | Build_In |
| UnetPP | pytorch | model name
| segmentation | Build_In |
| Yolov8-seg | ultralytics | model name
| instance segmentation | Build_In |
- face alignment
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| HIH | pytorch | model name
| face alignment | Build_In |
| HRNet_Face_Alignment | pytorch | model name
| face alignment | Build_In |
| PFLD | pytorch | model name
| face alignment | Build_In |
| PIPNet | official | model name
| face alignment | Build_In |
- face detection
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| CenterFace | onnx | model name
| face detection | Build_In |
| RetinaFace | pytorch | model name
| face detection | Build_In |
| SCRFD | insightface | model name
| face detection | Build_In |
| YOLO5Face | pytorch | model name
| face detection | Build_In |
- facial quality
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| SDD-FIQA | sdd_fiqa | model name
| face quality | Build_In |
- face recognize
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| FaceNet | tensorflow | model name
| face recognize | Build_In |
| FaceNet | pytorch | model name
| face recognize | Build_In |
- facial attribute
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| FairFace | official | model name
| face attribute | Build_In |
- image colorization
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| GPEN | official | model name
| image colorization | Build_In |
- image compress
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| Elic | official | model name
| image compress | Build_In |
| Mlicpp | official | model name
| image compress | Build_In |
- image retrieval
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| DINOv2 | official | model name
| image_retrieval | Build_In |
- low light image enhancement
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| PairLIE | official | model name
| low light image enhancement | Build_In |
| USUIR | official | model name
| low light image enhancement | Build_In |
- mot
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| ByteTrack | official | model name
| mot | Build_In |
| DeepSort | pytorch | model name
| mot | Build_In |
| FairMOT | official | model name
| mot | Build_In |
| JDE | pytorch | model name
| mot | Build_In |
| OCSORT | pytorch | model name
| mot | Build_In |
- pose
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| OpenPose | pytorch | model name
| pose | Build_In |
| HRNet_Pose | mmpose | model name
| pose | Build_In |
| Yolov8_Pose | ultralytics | model name
| pose | Build_In |
- reid
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| Fast_Reid | official | model name
| reid | Build_In |
- salient object detection
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| BSANet | official | model name
| salient object detection | Build_In |
| BSANet | salod | model name
| salient object detection | Build_In |
| DHSNet | salod | model name
| salient object detection | Build_In |
| EGNet | salod | model name
| salient object detection | Build_In |
| F3Net | official | model name
| salient object detection | Build_In |
| GCPANet | salod | model name
| salient object detection | Build_In |
| ISNet | official | model name
| salient object detection | Build_In |
| SCRN | salod | model name
| salient object detection | Build_In |
| U2Net | official | model name
| salient object detection | Build_In |
- super_resolution
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| Bicubic_PlusPlus | official | model name
| super resolution | Build_In |
| BSRGAN | official | model name
| super resolution | Build_In |
| ECBSR | official | model name
| super resolution | Build_In |
| ECBSR | basicsr | model name
| super resolution | Build_In |
| EDSR | basicsr | model name
| super resolution | Build_In |
| EDSR | official | model name
| super resolution | Build_In |
| ESRGAN | official | model name
| super resolution | Build_In |
| FSRNet* | pytorch | model name
| super resolution | Build_In |
| GPEN | official | model name
| super resolution | Build_In |
| LESRCNN | official | model name
| super resolution | Build_In |
| MSRN | official | model name
| super resolution | Build_In |
| NCNet | official | model name
| super resolution | Build_In |
| PFSRNet | official | model name
| face super resolution | Build_In |
| RCAN | official | model name
| super resolution | Build_In |
| RCAN | basicsr | model name
| super resolution | Build_In |
| Real_ESRGAN | official | model name
| super resolution | Build_In |
| RealSR | official | model name
| super resolution | Build_In |
| SPARNet | official | model name
| super resolution | Build_In |
| VDSR | pytorch | model name
| super resolution | Build_In |
- text detection
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| CT | ppocr | model name
| text detection | Build_In |
| DBNet | ppocr | model name
| text detection | Build_In |
| DBNet | official | model name
| text detection | Build_In |
| DBNet | mmocr | model name
| text detection | Build_In |
| EAST | ppocr | model name
| text detection | Build_In |
| FAST | pytorch | model name
| text detection | Build_In |
| FCENet | mmocr | model name
| text detection | Build_In |
| PAN | pan_pp.pytorch | model name
| text detection | Build_In |
| PAN++ detection only | pan_pp.pytorch | model name
| text detection | Build_In |
| PSENet | pytorch | model name
| text detection | Build_In |
| Text Snake | mmocr | model name
| text detection | Build_In |
- text recognition
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| CNN_CTC | pytorch | model name
| ocr recognize | Build_In |
| CRNN | ppocr | model name
| text recognition | Build_In |
| PPOCR_V4_REC | ppocr | model name
| ocr recognize | Build_In |
NLP Models
- information extraction
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| uie | uie_pytorch | model name
| Information extraction | Build_In |
- named entity recognition
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| BERT | huggingface | model name
| named entity recognition | Build_In |
| RoBERTa | huggingface | model name
| named entity recognition | Build_In |
| ERNIE | huggingface | model name
| named entity recognition | Build_In |
| RoBERTa | huggingface | model name
| named entity recognition | Build_In |
| XLMRoBERTa | huggingface | model name
| named entity recognition | Build_In |
- question answering
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| BERT | tensorflow | model name
| question answering | Build_In |
| BERT | huggingface | model name
| question answering | Build_In |
| ERNIE | huggingface | model name
| question answering | Build_In |
- sentence classification
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| BERT | tensorflow | model name
| sentence classification | Build_In |
| BERT | huggingface | model name
| sentence classification | Build_In |
| BERT | modelscope | model name
| sentence classification | Build_In |
| Electra | CIB | model name
| sentence classification | Build_In |
| Electra | huggingface | model name
| sentence classification | Build_In |
| ERNIE | huggingface | model name
| sentence classification | Build_In |
| RoBERTa | huggingface | model name
| sentence classification | Build_In |
| XLMRoBERTa | huggingface | model name
| sentence classification | Build_In |
-
Text2Vec
-
以下是Embedding模型
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| acge | huggingface | model name
| Embedding model | Build_In |
| bge | huggingface | model name
| Embedding model | Build_In |
| bce | huggingface | model name
| Embedding model | Build_In |
| e5 | huggingface | model name
| Embedding model | Build_In |
| gte | huggingface | model name
| Embedding model | Build_In |
| m3e | huggingface | model name
| Embedding model | Build_In |
| text2vec | huggingface | model name
| Embedding model | Build_In |
- 以下是Reranker模型
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| bge | huggingface | model name
| Reranker model | Build_In |
| bce | huggingface | model name
| Reranker model | Build_In |
LLM Models
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| Aquila2 | huggingface | model name
| large language model | Build_In |
| BaiChuan2 | huggingface | model name
| large language model | Build_In |
| ChatGLM2 | huggingface | model name
| large language model | Build_In |
| ChatGLM3 | huggingface | model name
| large language model | Build_In |
| ChatGLM4 | huggingface | model name
| large language model | Build_In |
| DeepSeek-R1-Distill | huggingface | model name
| large language model | Build_In |
| Falcon | huggingface | model name
| large language model | Build_In |
| Gemma | huggingface | model name
| large language model | Build_In |
| Gemma-1.1 | huggingface | model name
| large language model | Build_In |
| Internlm2 | huggingface | model name
| large language model | Build_In |
| Internlm2_5 | huggingface | model name
| large language model | Build_In |
| Internlm3 | huggingface | model name
| large language model | Build_In |
| LLaMA | huggingface | model name
| large language model | Build_In |
| LLaMA-2 | huggingface | model name
| large language model | Build_In |
| LLaMA-3 | huggingface | model name
| large language model | Build_In |
| LLaMA-3.1 | huggingface | model name
| large language model | Build_In |
| LLaMA-3.2 | huggingface | model name
| large language model | Build_In |
| LLaMA-3.3 | huggingface | model name
| large language model | Build_In |
| Mistral | huggingface | model name
| large language model | Build_In |
| Mobillama | huggingface | model name
| large language model | Build_In |
| OpenBuddy | huggingface | model name
| large language model | Build_In |
| Phi-3 | huggingface | model name
| large language model | Build_In |
| Qwen1.5 | huggingface | model name
| large language model | Build_In |
| Qwen2 | huggingface | model name
| large language model | Build_In |
| Qwen2.5 | huggingface | model name
| large language model | Build_In |
| Qwen3 | huggingface | model name
| large language model | Build_In |
| QWQ | huggingface | model name
| large language model | Build_In |
| Skywork-OR1 | huggingface | model name
| large language model | Build_In |
| Stablelm | huggingface | model name
| large language model | Build_In |
| Tigerbot | huggingface | model name
| large language model | Build_In |
| wizardlm | huggingface | model name
| large language model | Build_In |
| Yi | huggingface | model name
| large language model | Build_In |
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| Qwen2.5 | huggingface | model name
| large language model | vLLM |
| Qwen3 | huggingface | model name
| large language model | vLLM |
| DeepSeek-V3 | huggingface | model name
| large language model | vLLM |
| DeepSeek-R1 | huggingface | model name
| large language model | vLLM |
以下模型为Domain-LLM专业领域模型(医疗、金融、DeepResearch等),一般基于General-LLM模型为基座+专业领域数据集微调+强化学习而来
| model | codebase | model list | model type | runtime | priority |
|---|---|---|---|---|---|
| Tongyi-DeepResearch | huggingface | model name
| LLM | vLLM | p1 |
VLM Models
| model | codebase | model list | model type | runtime |
|---|---|---|---|---|
| CLIP | openai | model name
| vision-language model | Build_In |
| Logics_Parsing | huggingface | model name
| vision-language model | Build_In |
| OCRFlux | huggingface | model name
| vision-language model | Build_In |
| olmOCR2 | huggingface | model name
| vision-language model | Build_In |
| Qwen2-VL | huggingface | model name
| vision-language model | Build_In |
| Qwen2.5-VL | huggingface | model name
| vision-language model | Build_In |
| RolmOCR | huggingface | model name
| vision-language model | Build_In |
| SigLip | model name
| vision-language model | Build_In |
VastModelZOO提供的模型仅供您用于非商业目的,请参考原始模型来源许可证进行使用VastModelZOO描述的数据集均为开源数据集,如您使用这些数据集,请参考原始数据集来源许可证进行使用- 如您不希望您的数据集或模型公布在
VastModelZOO上,请您提交issue,我们将尽快处理
VastModelZOO提供的模型,如原始模型有许可证描述,请以该许可证为准VastModelZOO遵循Apache 2.0许可证许可
