Merge pull request #6877 from LDOUBLEV/dyg_pts

[tipc]fix fp16 benchmark train

test_tipc/benchmark_train.sh

@@ -139,8 +139,8 @@ else
     device_num=${params_list[4]}
     IFS=";"
 
-    if [ ${precision} = "null" ];then
-        precision="fp32"
+    if [ ${precision} = "fp16" ];then
+        precision="amp"
     fi
 
     fp_items_list=($precision)
@@ -150,10 +150,16 @@ fi
 
 IFS="|"
 for batch_size in ${batch_size_list[*]}; do 
-    for precision in ${fp_items_list[*]}; do
+    for train_precision in ${fp_items_list[*]}; do
         for device_num in ${device_num_list[*]}; do
             # sed batchsize and precision
-            func_sed_params "$FILENAME" "${line_precision}" "$precision"
+            if [ ${train_precision} = "amp" ];then
+                precision="fp16"
+            else
+                precision="fp32"
+            fi
+
+            func_sed_params "$FILENAME" "${line_precision}" "$train_precision"
             func_sed_params "$FILENAME" "${line_batchsize}" "$MODE=$batch_size"
             func_sed_params "$FILENAME" "${line_epoch}" "$MODE=$epoch"
             gpu_id=$(set_gpu_id $device_num)

test_tipc/test_train_inference_python.sh

@@ -265,7 +265,7 @@ else
                 if [ ${run_train} = "null" ]; then
                     continue
                 fi
-                set_autocast=$(func_set_params "${autocast_key}" "${autocast}")
+
                 set_epoch=$(func_set_params "${epoch_key}" "${epoch_num}")
                 set_pretrain=$(func_set_params "${pretrain_model_key}" "${pretrain_model_value}")
                 set_batchsize=$(func_set_params "${train_batch_key}" "${train_batch_value}")
@@ -287,11 +287,11 @@ else
 
                 set_save_model=$(func_set_params "${save_model_key}" "${save_log}")
                 if [ ${#gpu} -le 2 ];then  # train with cpu or single gpu
-                    cmd="${python} ${run_train} ${set_use_gpu}  ${set_save_model} ${set_epoch} ${set_pretrain} ${set_autocast} ${set_batchsize} ${set_train_params1} ${set_amp_config} "
+                    cmd="${python} ${run_train} ${set_use_gpu}  ${set_save_model} ${set_epoch} ${set_pretrain}  ${set_batchsize} ${set_train_params1} ${set_amp_config} "
                 elif [ ${#ips} -le 15 ];then  # train with multi-gpu
-                    cmd="${python} -m paddle.distributed.launch --gpus=${gpu} ${run_train} ${set_use_gpu} ${set_save_model} ${set_epoch} ${set_pretrain} ${set_autocast} ${set_batchsize} ${set_train_params1} ${set_amp_config}"
+                    cmd="${python} -m paddle.distributed.launch --gpus=${gpu} ${run_train} ${set_use_gpu} ${set_save_model} ${set_epoch} ${set_pretrain}  ${set_batchsize} ${set_train_params1} ${set_amp_config}"
                 else     # train with multi-machine
-                    cmd="${python} -m paddle.distributed.launch --ips=${ips} --gpus=${gpu} ${run_train} ${set_use_gpu} ${set_save_model} ${set_pretrain} ${set_epoch} ${set_autocast} ${set_batchsize} ${set_train_params1} ${set_amp_config}"
+                    cmd="${python} -m paddle.distributed.launch --ips=${ips} --gpus=${gpu} ${run_train} ${set_use_gpu} ${set_save_model} ${set_pretrain} ${set_epoch}  ${set_batchsize} ${set_train_params1} ${set_amp_config}"
                 fi
                 # run train
                 eval $cmd