diff --git a/deploy/slim/prune/README.md b/deploy/slim/prune/README.md index e4407455b162019938b10a791948048bc4d5d97b..7b8dd169c5fa9d01421070f1ccc2bd4e8ed543a2 100644 --- a/deploy/slim/prune/README.md +++ b/deploy/slim/prune/README.md @@ -23,13 +23,13 @@ ```bash git clone https://github.com/PaddlePaddle/PaddleSlim.git +cd PaddleSlim git checkout develop -cd Paddleslim python3 setup.py install ``` ### 2. 获取预训练模型 -模型裁剪需要加载事先训练好的模型，PaddleOCR也提供了一系列(模型)[../../../doc/doc_ch/models_list.md]，开发者可根据需要自行选择模型或使用自己的模型。 +模型裁剪需要加载事先训练好的模型，PaddleOCR也提供了一系列[模型](../../../doc/doc_ch/models_list.md)，开发者可根据需要自行选择模型或使用自己的模型。 ### 3. 敏感度分析训练 @@ -49,14 +49,14 @@ python3 setup.py install 进入PaddleOCR根目录，通过以下命令对模型进行敏感度分析训练： ```bash -python3.7 deploy/slim/prune/sensitivity_anal.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model="your trained model" +python3.7 deploy/slim/prune/sensitivity_anal.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model="your trained model" Global.save_model_dir=./output/prune_model/ ``` ### 4. 导出模型、预测部署 在得到裁剪训练保存的模型后，我们可以将其导出为inference_model： ```bash -pytho3.7 deploy/slim/prune/export_prune_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model=./output/det_db/best_accuracy Global.save_inference_dir=inference_model +pytho3.7 deploy/slim/prune/export_prune_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model=./output/det_db/best_accuracy Global.save_inference_dir=./prune/prune_inference_model ``` inference model的预测和部署参考： diff --git a/deploy/slim/prune/README_en.md b/deploy/slim/prune/README_en.md index c2bb9d537e808d132d35571407e1ed9081970f87..fe9c5dcd6660757d11a884d4dbe077e95fad8afe 100644 --- a/deploy/slim/prune/README_en.md +++ b/deploy/slim/prune/README_en.md @@ -22,15 +22,15 @@ Five steps for OCR model prune: ```bash git clone https://github.com/PaddlePaddle/PaddleSlim.git +cd PaddleSlim git checkout develop -cd Paddleslim python3 setup.py install ``` ### 2. Download Pretrain Model Model prune needs to load pre-trained models. -PaddleOCR also provides a series of (models)[../../../doc/doc_en/models_list_en.md]. Developers can choose their own models or use their own models according to their needs. +PaddleOCR also provides a series of [models](../../../doc/doc_en/models_list_en.md). Developers can choose their own models or use their own models according to their needs. ### 3. Pruning sensitivity analysis @@ -54,7 +54,7 @@ Enter the PaddleOCR root directory，perform sensitivity analysis on the model w ```bash -python3.7 deploy/slim/prune/sensitivity_anal.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model="your trained model" +python3.7 deploy/slim/prune/sensitivity_anal.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model="your trained model" Global.save_model_dir=./output/prune_model/ ``` @@ -63,7 +63,7 @@ python3.7 deploy/slim/prune/sensitivity_anal.py -c configs/det/ch_ppocr_v2.0/ch_ We can export the pruned model as inference_model for deployment: ```bash -python deploy/slim/prune/export_prune_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model=./output/det_db/best_accuracy Global.save_inference_dir=inference_model +python deploy/slim/prune/export_prune_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model=./output/det_db/best_accuracy Global.save_inference_dir=./prune/prune_inference_model ``` Reference for prediction and deployment of inference model: diff --git a/deploy/slim/quantization/README.md b/deploy/slim/quantization/README.md index 41ed3ef5d755c53898180f6fc24b6beb777811ed..62bc408f5eeda6d8366834200e8d8a20d1dc82cd 100644 --- a/deploy/slim/quantization/README.md +++ b/deploy/slim/quantization/README.md @@ -23,7 +23,7 @@ ```bash git clone https://github.com/PaddlePaddle/PaddleSlim.git -cd Paddleslim +cd PaddleSlim python setup.py install ``` @@ -37,12 +37,12 @@ PaddleOCR提供了一系列训练好的[模型](../../../doc/doc_ch/models_list. 量化训练的代码位于slim/quantization/quant.py 中，比如训练检测模型，训练指令如下： ```bash -python deploy/slim/quantization/quant.py -c configs/det/det_mv3_db.yml -o Global.pretrained_model='your trained model' Global.save_model_dir=./output/quant_model +python deploy/slim/quantization/quant.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model='your trained model' Global.save_model_dir=./output/quant_model # 比如下载提供的训练模型 wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_train.tar tar -xf ch_ppocr_mobile_v2.0_det_train.tar -python deploy/slim/quantization/quant.py -c configs/det/det_mv3_db.yml -o Global.pretrained_model=./ch_ppocr_mobile_v2.0_det_train/best_accuracy Global.save_inference_dir=./output/quant_inference_model +python deploy/slim/quantization/quant.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model=./ch_ppocr_mobile_v2.0_det_train/best_accuracy Global.save_model_dir=./output/quant_inference_model ``` 如果要训练识别模型的量化，修改配置文件和加载的模型参数即可。 @@ -52,7 +52,7 @@ python deploy/slim/quantization/quant.py -c configs/det/det_mv3_db.yml -o Global 在得到量化训练保存的模型后，我们可以将其导出为inference_model，用于预测部署： ```bash -python deploy/slim/quantization/export_model.py -c configs/det/det_mv3_db.yml -o Global.checkpoints=output/quant_model/best_accuracy Global.save_model_dir=./output/quant_inference_model +python deploy/slim/quantization/export_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.checkpoints=output/quant_model/best_accuracy Global.save_inference_dir=./output/quant_inference_model ``` ### 5. 量化模型部署 diff --git a/deploy/slim/quantization/README_en.md b/deploy/slim/quantization/README_en.md index 231cdb6488b758b0678573133de59f6f22b21cc6..bf3e91d69a298a7aaa8da21c22c45e62713e50cf 100644 --- a/deploy/slim/quantization/README_en.md +++ b/deploy/slim/quantization/README_en.md @@ -26,7 +26,7 @@ After training, if you want to further compress the model size and accelerate th ```bash git clone https://github.com/PaddlePaddle/PaddleSlim.git -cd Paddleslim +cd PaddlSlim python setup.py install ``` @@ -43,12 +43,12 @@ After the quantization strategy is defined, the model can be quantified. The code for quantization training is located in `slim/quantization/quant.py`. For example, to train a detection model, the training instructions are as follows: ```bash -python deploy/slim/quantization/quant.py -c configs/det/det_mv3_db.yml -o Global.pretrained_model='your trained model' Global.save_model_dir=./output/quant_model +python deploy/slim/quantization/quant.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model='your trained model' Global.save_model_dir=./output/quant_model # download provided model wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_train.tar tar -xf ch_ppocr_mobile_v2.0_det_train.tar -python deploy/slim/quantization/quant.py -c configs/det/det_mv3_db.yml -o Global.pretrained_model=./ch_ppocr_mobile_v2.0_det_train/best_accuracy Global.save_model_dir=./output/quant_model +python deploy/slim/quantization/quant.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model=./ch_ppocr_mobile_v2.0_det_train/best_accuracy Global.save_model_dir=./output/quant_model ``` @@ -57,7 +57,7 @@ python deploy/slim/quantization/quant.py -c configs/det/det_mv3_db.yml -o Global After getting the model after pruning and finetuning we, can export it as inference_model for predictive deployment: ```bash -python deploy/slim/quantization/export_model.py -c configs/det/det_mv3_db.yml -o Global.checkpoints=output/quant_model/best_accuracy Global.save_inference_dir=./output/quant_inference_model +python deploy/slim/quantization/export_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.checkpoints=output/quant_model/best_accuracy Global.save_inference_dir=./output/quant_inference_model ``` ### 5. Deploy diff --git a/deploy/slim/quantization/quant.py b/deploy/slim/quantization/quant.py index 7671e5f871ce6769fc51876d1fa2e5f0af63d904..315e3b4321a544e77795c43d493873fcf46e1930 100755 --- a/deploy/slim/quantization/quant.py +++ b/deploy/slim/quantization/quant.py @@ -112,10 +112,6 @@ def main(config, device, logger, vdl_writer): config['Architecture']["Head"]['out_channels'] = char_num model = build_model(config['Architecture']) - # prepare to quant - quanter = QAT(config=quant_config, act_preprocess=PACT) - quanter.quantize(model) - if config['Global']['distributed']: model = paddle.DataParallel(model) @@ -136,31 +132,15 @@ def main(config, device, logger, vdl_writer): logger.info('train dataloader has {} iters, valid dataloader has {} iters'. format(len(train_dataloader), len(valid_dataloader))) + quanter = QAT(config=quant_config, act_preprocess=PACT) + quanter.quantize(model) + # start train program.train(config, train_dataloader, valid_dataloader, device, model, loss_class, optimizer, lr_scheduler, post_process_class, eval_class, pre_best_model_dict, logger, vdl_writer) -def test_reader(config, device, logger): - loader = build_dataloader(config, 'Train', device, logger) - import time - starttime = time.time() - count = 0 - try: - for data in loader(): - count += 1 - if count % 1 == 0: - batch_time = time.time() - starttime - starttime = time.time() - logger.info("reader: {}, {}, {}".format( - count, len(data[0]), batch_time)) - except Exception as e: - logger.info(e) - logger.info("finish reader: {}, Success!".format(count)) - - if __name__ == '__main__': config, device, logger, vdl_writer = program.preprocess(is_train=True) main(config, device, logger, vdl_writer) - # test_reader(config, device, logger)