diff --git a/README.md b/README.md index 01f8388f5eeec043a75216d1d2858eb8100cbbab..f67e77f4d65fb49a2046cda7fa5bfd3e527e4668 100644 --- a/README.md +++ b/README.md @@ -5,6 +5,33 @@ DeepSparkHub甄选上百个应用算法和模型，覆盖AI和通用计算各领 ## 模型列表 +### LLM (Large Language Model) + +| Model | Framework | ToolBox | Dataset/Weight | +|-------------------------------------------------------------|-----------|--------------------|-----------------------| +| [Aquila2-34B](nlp/llm/aquila2-34b/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | Bookcorpus | +| [Baichuan2-7B](nlp/llm/baichuan2-7b/Baichuan2) | PyTorch | DeepSpeed | baichuan2-7b-base | +| [Bloom-7B1](nlp/llm/bloom-7b1/firefly) | PyTorch | Firefly | school_math_0.25M | +| [ChatGLM-6B](nlp/llm/chatglm-6b/deepspeed) | PyTorch | DeepSpeed | ADGEN & chatglm-6b | +| [ChatGLM2-6B SFT](nlp/llm/ChatGLM2-6b-sft) | PyTorch | DeepSpeed | ADGEN & chatglm2-6b | +| [ChatGLM3-6B](nlp/llm/chatglm3-6b/deepspeed/finetune_demo) | PyTorch | DeepSpeed | ADGEN & chatglm3-6b | +| [DeepSeekMoE 7B](nlp/llm/deepseek_moe_7b/colossalai) | PyTorch | ColossalAI | deepseek-moe-16b-base | +| [Llama-7B](nlp/llm/llama-7b/colossalai) | PyTorch | ColossalAI | llama-7b-hf | +| [Llama2-7B](nlp/llm/llama2-7b/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | Bookcorpus | +| [Llama2-7B RMF](nlp/llm/llama2-7b_reward_sft/deepspeed) | PyTorch | DeepSpeed | Dahoas/rm-static | +| [Llama2-7B RLHF](nlp/llm/llama2-7b_rlhf/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | llama2-7b&tiny-llama | +| [Llama2-7B SFT](nlp/llm/llama2-7b_sft/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | GPT Small-117M | +| [Llama2-13B](nlp/llm/llama2-13b/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | Bookcorpus | +| [Llama2-34B](nlp/llm/llama2-34b/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | Bookcorpus | +| [Llama3-8B](nlp/llm/llama3_8b/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | Bookcorpus | +| [Llama3-8B SFT](nlp/llm/llama3_8b/colossalai) | PyTorch | ColossalAI | school_math_0.25M | +| [Mamba-2](nlp/llm/mamba-2/megatron-lm) | PyTorch | Megatron-LM | GPT Small-117M | +| [Mixtral 8x7B](nlp/llm/mixtral/megatron-lm) | PyTorch | Megatron-LM | GPT Small-117M | +| [QWen-7B](nlp/llm/qwen-7b/firefly) | PyTorch | Firefly | qwen-7b | +| [QWen1.5-7B](nlp/llm/qwen1.5-7b/firefly) | PyTorch | Firefly | school_math | +| [QWen1.5-14B](nlp/llm/qwen1.5-14b/firefly) | PyTorch | Firefly | school_math | +| [Qwen2.5-7B SFT](nlp/llm/qwen2.5-7b/LLaMA-Factory) | PyTorch | LLaMA-Factory | qwen2.5-7b | + ### Computer Vision #### Classification @@ -308,33 +335,6 @@ DeepSparkHub甄选上百个应用算法和模型，覆盖AI和通用计算各领 | [DeepSORT](cv/tracking/deep_sort/pytorch) | PyTorch | Market-1501 | | [FairMOT](cv/tracking/fairmot/pytorch) | PyTorch | MOT17 | -### LLM (Large Language Model) - -| Model | Framework | ToolBox | Dataset/Weight | -|-------------------------------------------------------------|-----------|--------------------|-----------------------| -| [Aquila2-34B](nlp/llm/aquila2-34b/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | Bookcorpus | -| [Baichuan2-7B](nlp/llm/baichuan2-7b/Baichuan2) | PyTorch | DeepSpeed | baichuan2-7b-base | -| [Bloom-7B1](nlp/llm/bloom-7b1/firefly) | PyTorch | Firefly | school_math_0.25M | -| [ChatGLM-6B](nlp/llm/chatglm-6b/deepspeed) | PyTorch | DeepSpeed | ADGEN & chatglm-6b | -| [ChatGLM2-6B SFT](nlp/llm/ChatGLM2-6b-sft) | PyTorch | DeepSpeed | ADGEN & chatglm2-6b | -| [ChatGLM3-6B](nlp/llm/chatglm3-6b/deepspeed/finetune_demo) | PyTorch | DeepSpeed | ADGEN & chatglm3-6b | -| [DeepSeekMoE 7B](nlp/llm/deepseek_moe_7b/colossalai) | PyTorch | ColossalAI | deepseek-moe-16b-base | -| [Llama-7B](nlp/llm/llama-7b/colossalai) | PyTorch | ColossalAI | llama-7b-hf | -| [Llama2-7B](nlp/llm/llama2-7b/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | Bookcorpus | -| [Llama2-7B RMF](nlp/llm/llama2-7b_reward_sft/deepspeed) | PyTorch | DeepSpeed | Dahoas/rm-static | -| [Llama2-7B RLHF](nlp/llm/llama2-7b_rlhf/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | llama2-7b&tiny-llama | -| [Llama2-7B SFT](nlp/llm/llama2-7b_sft/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | GPT Small-117M | -| [Llama2-13B](nlp/llm/llama2-13b/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | Bookcorpus | -| [Llama2-34B](nlp/llm/llama2-34b/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | Bookcorpus | -| [Llama3-8B](nlp/llm/llama3_8b/megatron-deepspeed) | PyTorch | Megatron-DeepSpeed | Bookcorpus | -| [Llama3-8B SFT](nlp/llm/llama3_8b/colossalai) | PyTorch | ColossalAI | school_math_0.25M | -| [Mamba-2](nlp/llm/mamba-2/megatron-lm) | PyTorch | Megatron-LM | GPT Small-117M | -| [Mixtral 8x7B](nlp/llm/mixtral/megatron-lm) | PyTorch | Megatron-LM | GPT Small-117M | -| [QWen-7B](nlp/llm/qwen-7b/firefly) | PyTorch | Firefly | qwen-7b | -| [QWen1.5-7B](nlp/llm/qwen1.5-7b/firefly) | PyTorch | Firefly | school_math | -| [QWen1.5-14B](nlp/llm/qwen1.5-14b/firefly) | PyTorch | Firefly | school_math | -| [Qwen2.5-7B SFT](nlp/llm/qwen2.5-7b/LLaMA-Factory) | PyTorch | LLaMA-Factory | qwen2.5-7b | - ### Multimodal | Model | Framework | Dataset | diff --git a/cv/face/retinaface/pytorch/LICENSE.MIT b/cv/face_detection/retinaface/pytorch/LICENSE.MIT similarity index 100% rename from cv/face/retinaface/pytorch/LICENSE.MIT rename to cv/face_detection/retinaface/pytorch/LICENSE.MIT diff --git a/cv/face/retinaface/pytorch/README.md b/cv/face_detection/retinaface/pytorch/README.md similarity index 36% rename from cv/face/retinaface/pytorch/README.md rename to cv/face_detection/retinaface/pytorch/README.md index 850c3591263fdb088724dcd76e791f1ba407f5b4..36267d7929ddd3f41a867e3fc1a6cd72b0d9756e 100644 --- a/cv/face/retinaface/pytorch/README.md +++ b/cv/face_detection/retinaface/pytorch/README.md @@ -2,18 +2,29 @@ ## Model description -Though tremendous strides have been made in uncontrolled face detection, accurate and efficient face localisation in the wild remains an open challenge. This paper presents a robust single-stage face detector, named RetinaFace, which performs pixel-wise face localisation on various scales of faces by taking advantages of joint extra-supervised and self-supervised multi-task learning. Specifically, We make contributions in the following five aspects: (1) We manually annotate five facial landmarks on the WIDER FACE dataset and observe significant improvement in hard face detection with the assistance of this extra supervision signal. (2) We further add a self-supervised mesh decoder branch for predicting a pixel-wise 3D shape face information in parallel with the existing supervised branches. (3) On the WIDER FACE hard test set, RetinaFace outperforms the state of the art average precision (AP) by 1.1% (achieving AP equal to 91.4%). (4) On the IJB-C test set, RetinaFace enables state of the art methods (ArcFace) to improve their results in face verification (TAR=89.59% for FAR=1e-6). (5) By employing light-weight backbone networks, RetinaFace can run real-time on a single CPU core for a VGA-resolution image. +Though tremendous strides have been made in uncontrolled face detection, accurate and efficient face localisation in the +wild remains an open challenge. This paper presents a robust single-stage face detector, named RetinaFace, which +performs pixel-wise face localisation on various scales of faces by taking advantages of joint extra-supervised and +self-supervised multi-task learning. Specifically, We make contributions in the following five aspects: (1) We manually +annotate five facial landmarks on the WIDER FACE dataset and observe significant improvement in hard face detection with +the assistance of this extra supervision signal. (2) We further add a self-supervised mesh decoder branch for predicting +a pixel-wise 3D shape face information in parallel with the existing supervised branches. (3) On the WIDER FACE hard +test set, RetinaFace outperforms the state of the art average precision (AP) by 1.1% (achieving AP equal to 91.4%). (4) +On the IJB-C test set, RetinaFace enables state of the art methods (ArcFace) to improve their results in face +verification (TAR=89.59% for FAR=1e-6). (5) By employing light-weight backbone networks, RetinaFace can run real-time on +a single CPU core for a VGA-resolution image. ## Prepare ### Install packages - ### Download dataset -1. Download the [WIDERFACE](http://shuoyang1213.me/WIDERFACE/WiderFace_Results.html) dataset. +1. Download the [WIDER FACE](http://shuoyang1213.me/WIDERFACE/WiderFace_Results.html) dataset. -2. Download annotations (face bounding boxes & five facial landmarks) from [baidu cloud](https://pan.baidu.com/s/1Laby0EctfuJGgGMgRRgykA) or [dropbox](https://www.dropbox.com/s/7j70r3eeepe4r2g/retinaface_gt_v1.1.zip?dl=0) +2. Download annotations (face bounding boxes & five facial landmarks) from [baidu + cloud](https://pan.baidu.com/s/1Laby0EctfuJGgGMgRRgykA) or + [dropbox](https://www.dropbox.com/s/7j70r3eeepe4r2g/retinaface_gt_v1.1.zip?dl=0) 3. Organise the dataset directory as follows: @@ -29,7 +40,9 @@ Though tremendous strides have been made in uncontrolled face detection, accurat ``` -4. Download pretrained models from [google cloud](https://drive.google.com/open?id=1oZRSG0ZegbVkVwUd8wUIQx8W7yfZ_ki1) and [baidu cloud](https://pan.baidu.com/s/12h97Fy1RYuqMMIV-RpzdPg) Password: fstq . The model could be put as follows: +4. Download pretrained models from [google cloud](https://drive.google.com/open?id=1oZRSG0ZegbVkVwUd8wUIQx8W7yfZ_ki1) + and [baidu cloud](https://pan.baidu.com/s/12h97Fy1RYuqMMIV-RpzdPg) Password: fstq . The model could be put as + follows: ```Shell @@ -58,7 +71,7 @@ python3 train.py --network resnet50 ``` -## Evaluate +## Evaluate 1. Generate txt file @@ -67,6 +80,7 @@ python3 train.py --network resnet50 python3 test_widerface.py --trained_model ${weight_file} --network mobile0.25 or resnet50 ``` + 2. Evaluate txt results. Demo come from [Here](https://github.com/wondervictor/WiderFace-Evaluation) ```Shell @@ -77,11 +91,13 @@ python3 evaluation.py ``` -3. You can also use widerface official Matlab evaluate demo in [Here](http://mmlab.ie.cuhk.edu.hk/projects/WIDERFace/WiderFace_Results.html) +3. You can also use widerface official Matlab evaluate demo in + [Here](http://mmlab.ie.cuhk.edu.hk/projects/WIDERFace/WiderFace_Results.html) -## Results on BI-V100 +## Results -| Easy Val | AP: 0.940523547724767 | -|------------|------------------------| -| Medium Val | AP: 0.9315975713948582 | -| Hard Val | AP: 0.8303531408987751 | \ No newline at end of file +| Model | GPU | Type | AP | +|------------|---------|------------|--------------------| +| RetinaFace | BI-V100 | Easy Val | 0.940523547724767 | +| RetinaFace | BI-V100 | Medium Val | 0.9315975713948582 | +| RetinaFace | BI-V100 | Hard Val | 0.8303531408987751 | diff --git a/cv/face/retinaface/pytorch/convert_to_onnx.py b/cv/face_detection/retinaface/pytorch/convert_to_onnx.py similarity index 100% rename from cv/face/retinaface/pytorch/convert_to_onnx.py rename to cv/face_detection/retinaface/pytorch/convert_to_onnx.py diff --git a/cv/face/retinaface/pytorch/curve/1.jpg b/cv/face_detection/retinaface/pytorch/curve/1.jpg similarity index 100% rename from cv/face/retinaface/pytorch/curve/1.jpg rename to cv/face_detection/retinaface/pytorch/curve/1.jpg diff --git a/cv/face/retinaface/pytorch/curve/FDDB.png b/cv/face_detection/retinaface/pytorch/curve/FDDB.png similarity index 100% rename from cv/face/retinaface/pytorch/curve/FDDB.png rename to cv/face_detection/retinaface/pytorch/curve/FDDB.png diff --git a/cv/face/retinaface/pytorch/curve/Widerface.jpg b/cv/face_detection/retinaface/pytorch/curve/Widerface.jpg similarity index 100% rename from cv/face/retinaface/pytorch/curve/Widerface.jpg rename to cv/face_detection/retinaface/pytorch/curve/Widerface.jpg diff --git a/cv/face/retinaface/pytorch/curve/test.jpg b/cv/face_detection/retinaface/pytorch/curve/test.jpg similarity index 100% rename from cv/face/retinaface/pytorch/curve/test.jpg rename to cv/face_detection/retinaface/pytorch/curve/test.jpg diff --git a/cv/face/retinaface/pytorch/data/FDDB/img_list.txt b/cv/face_detection/retinaface/pytorch/data/FDDB/img_list.txt similarity index 100% rename from cv/face/retinaface/pytorch/data/FDDB/img_list.txt rename to cv/face_detection/retinaface/pytorch/data/FDDB/img_list.txt diff --git a/cv/face/retinaface/pytorch/data/__init__.py b/cv/face_detection/retinaface/pytorch/data/__init__.py similarity index 100% rename from cv/face/retinaface/pytorch/data/__init__.py rename to cv/face_detection/retinaface/pytorch/data/__init__.py diff --git a/cv/face/retinaface/pytorch/data/config.py b/cv/face_detection/retinaface/pytorch/data/config.py similarity index 100% rename from cv/face/retinaface/pytorch/data/config.py rename to cv/face_detection/retinaface/pytorch/data/config.py diff --git a/cv/face/retinaface/pytorch/data/data_augment.py b/cv/face_detection/retinaface/pytorch/data/data_augment.py similarity index 100% rename from cv/face/retinaface/pytorch/data/data_augment.py rename to cv/face_detection/retinaface/pytorch/data/data_augment.py diff --git a/cv/face/retinaface/pytorch/data/wider_face.py b/cv/face_detection/retinaface/pytorch/data/wider_face.py similarity index 100% rename from cv/face/retinaface/pytorch/data/wider_face.py rename to cv/face_detection/retinaface/pytorch/data/wider_face.py diff --git a/cv/face/retinaface/pytorch/detect.py b/cv/face_detection/retinaface/pytorch/detect.py similarity index 100% rename from cv/face/retinaface/pytorch/detect.py rename to cv/face_detection/retinaface/pytorch/detect.py diff --git a/cv/face/retinaface/pytorch/layers/__init__.py b/cv/face_detection/retinaface/pytorch/layers/__init__.py similarity index 100% rename from cv/face/retinaface/pytorch/layers/__init__.py rename to cv/face_detection/retinaface/pytorch/layers/__init__.py diff --git a/cv/face/retinaface/pytorch/layers/functions/prior_box.py b/cv/face_detection/retinaface/pytorch/layers/functions/prior_box.py similarity index 100% rename from cv/face/retinaface/pytorch/layers/functions/prior_box.py rename to cv/face_detection/retinaface/pytorch/layers/functions/prior_box.py diff --git a/cv/face/retinaface/pytorch/layers/modules/__init__.py b/cv/face_detection/retinaface/pytorch/layers/modules/__init__.py similarity index 100% rename from cv/face/retinaface/pytorch/layers/modules/__init__.py rename to cv/face_detection/retinaface/pytorch/layers/modules/__init__.py diff --git a/cv/face/retinaface/pytorch/layers/modules/multibox_loss.py b/cv/face_detection/retinaface/pytorch/layers/modules/multibox_loss.py similarity index 100% rename from cv/face/retinaface/pytorch/layers/modules/multibox_loss.py rename to cv/face_detection/retinaface/pytorch/layers/modules/multibox_loss.py diff --git a/cv/face/retinaface/pytorch/models/__init__.py b/cv/face_detection/retinaface/pytorch/models/__init__.py similarity index 100% rename from cv/face/retinaface/pytorch/models/__init__.py rename to cv/face_detection/retinaface/pytorch/models/__init__.py diff --git a/cv/face/retinaface/pytorch/models/net.py b/cv/face_detection/retinaface/pytorch/models/net.py similarity index 100% rename from cv/face/retinaface/pytorch/models/net.py rename to cv/face_detection/retinaface/pytorch/models/net.py diff --git a/cv/face/retinaface/pytorch/models/retinaface.py b/cv/face_detection/retinaface/pytorch/models/retinaface.py similarity index 100% rename from cv/face/retinaface/pytorch/models/retinaface.py rename to cv/face_detection/retinaface/pytorch/models/retinaface.py diff --git a/cv/face/retinaface/pytorch/test_fddb.py b/cv/face_detection/retinaface/pytorch/test_fddb.py similarity index 100% rename from cv/face/retinaface/pytorch/test_fddb.py rename to cv/face_detection/retinaface/pytorch/test_fddb.py diff --git a/cv/face/retinaface/pytorch/test_widerface.py b/cv/face_detection/retinaface/pytorch/test_widerface.py similarity index 100% rename from cv/face/retinaface/pytorch/test_widerface.py rename to cv/face_detection/retinaface/pytorch/test_widerface.py diff --git a/cv/face/retinaface/pytorch/train.py b/cv/face_detection/retinaface/pytorch/train.py similarity index 100% rename from cv/face/retinaface/pytorch/train.py rename to cv/face_detection/retinaface/pytorch/train.py diff --git a/cv/face/retinaface/pytorch/utils/__init__.py b/cv/face_detection/retinaface/pytorch/utils/__init__.py similarity index 100% rename from cv/face/retinaface/pytorch/utils/__init__.py rename to cv/face_detection/retinaface/pytorch/utils/__init__.py diff --git a/cv/face/retinaface/pytorch/utils/box_utils.py b/cv/face_detection/retinaface/pytorch/utils/box_utils.py similarity index 100% rename from cv/face/retinaface/pytorch/utils/box_utils.py rename to cv/face_detection/retinaface/pytorch/utils/box_utils.py diff --git a/cv/face/retinaface/pytorch/utils/nms/__init__.py b/cv/face_detection/retinaface/pytorch/utils/nms/__init__.py similarity index 100% rename from cv/face/retinaface/pytorch/utils/nms/__init__.py rename to cv/face_detection/retinaface/pytorch/utils/nms/__init__.py diff --git a/cv/face/retinaface/pytorch/utils/nms/py_cpu_nms.py b/cv/face_detection/retinaface/pytorch/utils/nms/py_cpu_nms.py similarity index 100% rename from cv/face/retinaface/pytorch/utils/nms/py_cpu_nms.py rename to cv/face_detection/retinaface/pytorch/utils/nms/py_cpu_nms.py diff --git a/cv/face/retinaface/pytorch/utils/timer.py b/cv/face_detection/retinaface/pytorch/utils/timer.py similarity index 100% rename from cv/face/retinaface/pytorch/utils/timer.py rename to cv/face_detection/retinaface/pytorch/utils/timer.py diff --git a/cv/face/retinaface/pytorch/widerface_evaluate/README.md b/cv/face_detection/retinaface/pytorch/widerface_evaluate/README.md similarity index 100% rename from cv/face/retinaface/pytorch/widerface_evaluate/README.md rename to cv/face_detection/retinaface/pytorch/widerface_evaluate/README.md diff --git a/cv/face/retinaface/pytorch/widerface_evaluate/box_overlaps.pyx b/cv/face_detection/retinaface/pytorch/widerface_evaluate/box_overlaps.pyx similarity index 100% rename from cv/face/retinaface/pytorch/widerface_evaluate/box_overlaps.pyx rename to cv/face_detection/retinaface/pytorch/widerface_evaluate/box_overlaps.pyx diff --git a/cv/face/retinaface/pytorch/widerface_evaluate/evaluation.py b/cv/face_detection/retinaface/pytorch/widerface_evaluate/evaluation.py similarity index 100% rename from cv/face/retinaface/pytorch/widerface_evaluate/evaluation.py rename to cv/face_detection/retinaface/pytorch/widerface_evaluate/evaluation.py diff --git a/cv/face/retinaface/pytorch/widerface_evaluate/ground_truth/wider_easy_val.mat b/cv/face_detection/retinaface/pytorch/widerface_evaluate/ground_truth/wider_easy_val.mat similarity index 100% rename from cv/face/retinaface/pytorch/widerface_evaluate/ground_truth/wider_easy_val.mat rename to cv/face_detection/retinaface/pytorch/widerface_evaluate/ground_truth/wider_easy_val.mat diff --git a/cv/face/retinaface/pytorch/widerface_evaluate/ground_truth/wider_face_val.mat b/cv/face_detection/retinaface/pytorch/widerface_evaluate/ground_truth/wider_face_val.mat similarity index 100% rename from cv/face/retinaface/pytorch/widerface_evaluate/ground_truth/wider_face_val.mat rename to cv/face_detection/retinaface/pytorch/widerface_evaluate/ground_truth/wider_face_val.mat diff --git a/cv/face/retinaface/pytorch/widerface_evaluate/ground_truth/wider_hard_val.mat b/cv/face_detection/retinaface/pytorch/widerface_evaluate/ground_truth/wider_hard_val.mat similarity index 100% rename from cv/face/retinaface/pytorch/widerface_evaluate/ground_truth/wider_hard_val.mat rename to cv/face_detection/retinaface/pytorch/widerface_evaluate/ground_truth/wider_hard_val.mat diff --git a/cv/face/retinaface/pytorch/widerface_evaluate/ground_truth/wider_medium_val.mat b/cv/face_detection/retinaface/pytorch/widerface_evaluate/ground_truth/wider_medium_val.mat similarity index 100% rename from cv/face/retinaface/pytorch/widerface_evaluate/ground_truth/wider_medium_val.mat rename to cv/face_detection/retinaface/pytorch/widerface_evaluate/ground_truth/wider_medium_val.mat diff --git a/cv/face/retinaface/pytorch/widerface_evaluate/setup.py b/cv/face_detection/retinaface/pytorch/widerface_evaluate/setup.py similarity index 100% rename from cv/face/retinaface/pytorch/widerface_evaluate/setup.py rename to cv/face_detection/retinaface/pytorch/widerface_evaluate/setup.py