Merge pull request #2069 from RainFrost1/dynamic_benchmark

add r50 dynamic fp16 train for benchmark

ppcls/data/dataloader/dali.py

@@ -23,7 +23,6 @@ import nvidia.dali.types as types
 import paddle
 from nvidia.dali import fn
 from nvidia.dali.pipeline import Pipeline
-from nvidia.dali.plugin.base_iterator import LastBatchPolicy
 from nvidia.dali.plugin.paddle import DALIGenericIterator
 
 

test_tipc/config/ResNet/ResNet50_train_ampfp16_infer_python.txt

@@ -13,7 +13,7 @@ train_infer_img_dir:./dataset/ILSVRC2012/val
 null:null
 ##
 trainer:amp_train
-amp_train:tools/train.py -c ppcls/configs/ImageNet/ResNet/ResNet50.yaml -o Global.seed=1234 -o DataLoader.Train.sampler.shuffle=False -o DataLoader.Train.loader.num_workers=0 -o DataLoader.Train.loader.use_shared_memory=False -o AMP.scale_loss=128 -o AMP.use_dynamic_loss_scaling=True -o AMP.level=O2
+amp_train:tools/train.py -c ppcls/configs/ImageNet/ResNet/ResNet50_amp_O1.yaml -o Global.seed=1234 -o DataLoader.Train.sampler.shuffle=False -o DataLoader.Train.loader.num_workers=0 -o DataLoader.Train.loader.use_shared_memory=False -o AMP.scale_loss=128 -o AMP.use_dynamic_loss_scaling=True -o AMP.level=O1
 pact_train:null
 fpgm_train:null
 distill_train:null
@@ -50,3 +50,7 @@ inference:python/predict_cls.py -c configs/inference_cls.yaml
 -o Global.benchmark:True
 null:null
 null:null
+===========================train_benchmark_params==========================
+batch_size:128
+fp_items:amp_fp16
+epoch:1

test_tipc/config/ResNet/ResNet50_train_purefp16_infer_python.txt

@@ -0,0 +1,56 @@
+===========================train_params===========================
+model_name:ResNet50
+python:python3.7
+gpu_list:0|0,1
+-o Global.device:gpu
+-o Global.auto_cast:null
+-o Global.epochs:lite_train_lite_infer=2|whole_train_whole_infer=120
+-o Global.output_dir:./output/
+-o DataLoader.Train.sampler.batch_size:8
+-o Global.pretrained_model:null
+train_model_name:latest
+train_infer_img_dir:./dataset/ILSVRC2012/val
+null:null
+##
+trainer:amp_train
+amp_train:tools/train.py -c ppcls/configs/ImageNet/ResNet/ResNet50_amp_O1.yaml -o Global.seed=1234 -o DataLoader.Train.sampler.shuffle=False -o DataLoader.Train.loader.num_workers=0 -o DataLoader.Train.loader.use_shared_memory=False -o AMP.scale_loss=128 -o AMP.use_dynamic_loss_scaling=True -o AMP.level=O2
+pact_train:null
+fpgm_train:null
+distill_train:null
+to_static_train:-o Global.to_static=True
+null:null
+##
+===========================eval_params=========================== 
+eval:tools/eval.py -c ppcls/configs/ImageNet/ResNet/ResNet50.yaml
+null:null
+##
+===========================infer_params==========================
+-o Global.save_inference_dir:./inference
+-o Global.pretrained_model:
+norm_export:tools/export_model.py -c ppcls/configs/ImageNet/ResNet/ResNet50.yaml
+quant_export:null
+fpgm_export:null
+distill_export:null
+kl_quant:null
+export2:null
+pretrained_model_url:https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet50_pretrained.pdparams
+infer_model:../inference/
+infer_export:True
+infer_quant:Fasle
+inference:python/predict_cls.py -c configs/inference_cls.yaml
+-o Global.use_gpu:True|False
+-o Global.enable_mkldnn:True|False
+-o Global.cpu_num_threads:1|6
+-o Global.batch_size:1|16
+-o Global.use_tensorrt:True|False
+-o Global.use_fp16:True|False
+-o Global.inference_model_dir:../inference
+-o Global.infer_imgs:../dataset/ILSVRC2012/val
+-o Global.save_log_path:null
+-o Global.benchmark:True
+null:null
+null:null
+===========================train_benchmark_params==========================
+batch_size:128
+fp_items:pure_fp16
+epoch:1

test_tipc/docs/benchmark_train.md

@@ -9,7 +9,7 @@
 
 ```shell
 # 运行格式：bash test_tipc/prepare.sh  train_benchmark.txt  mode
-bash test_tipc/prepare.sh test_tipc/configs/MobileNetV2/MobileNetV2_train_infer_python.txt benchmark_train
+bash test_tipc/prepare.sh test_tipc/config/MobileNetV2/MobileNetV2_train_infer_python.txt benchmark_train
 ```
 
 ## 1.2 功能测试
@@ -24,7 +24,7 @@ bash test_tipc/benchmark_train.sh test_tipc/config/MobileNetV2/MobileNetV2_train
 `test_tipc/benchmark_train.sh`支持根据传入的第三个参数实现只运行某一个训练配置，如下：
 ```shell
 # 运行格式：bash test_tipc/benchmark_train.sh train_benchmark.txt mode params
-bash test_tipc/benchmark_train.sh test_tipc/configs/MobileNetV2/MobileNetV2_train_infer_python.txt benchmark_train  dynamic_bs8_fp32_DP_N1C1
+bash test_tipc/benchmark_train.sh test_tipc/config/MobileNetV2/MobileNetV2_train_infer_python.txt benchmark_train  dynamic_bs8_fp32_DP_N1C1
 ```
 dynamic_bs8_fp32_DP_N1C1为test_tipc/benchmark_train.sh传入的参数，格式如下：
 `${modeltype}_${batch_size}_${fp_item}_${run_mode}_${device_num}`
@@ -33,7 +33,7 @@ dynamic_bs8_fp32_DP_N1C1为test_tipc/benchmark_train.sh传入的参数，格式
 
 ## 2. 日志输出
 
-运行后将保存模型的训练日志和解析日志，使用 `test_tipc/configs/MobileNetV2/MobileNetV2_train_infer_python.txt` 参数文件的训练日志解析结果是：
+运行后将保存模型的训练日志和解析日志，使用 `test_tipc/config/MobileNetV2/MobileNetV2_train_infer_python.txt` 参数文件的训练日志解析结果是：
 
 ```
 {"model_branch": "dygaph", "model_commit": "7c39a1996b19087737c05d883fd346d2f39dbcc0", "model_name": "cls_MobileNetV2_bs8_fp32_SingleP_DP", "batch_size": 8, "fp_item": "fp32", "run_process_type": "SingleP", "run_mode": "DP", "convergence_value": "5.413110", "convergence_key": "loss:", "ips": 19.333, "speed_unit": "samples/s", "device_num": "N1C1", "model_run_time": "0", "frame_commit": "8cc09552473b842c651ead3b9848d41827a3dbab", "frame_version": "0.0.0"}