train finetune

This commit is contained in:
游雁 2024-02-20 18:01:15 +08:00
parent 45d9ccafef
commit 2e8dc0933f
5 changed files with 12 additions and 8 deletions

View File

@ -105,7 +105,8 @@ if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
echo "stage 4: ASR Training"
mkdir -p ${exp_dir}/exp/${model_dir}
log_file="${exp_dir}/exp/${model_dir}/train.log.txt"
current_time=$(date "+%Y-%m-%d_%H-%M")
log_file="${exp_dir}/exp/${model_dir}/train.log.txt.${current_time}"
echo "log_file: ${log_file}"
gpu_num=$(echo $CUDA_VISIBLE_DEVICES | awk -F "," '{print NF}')

View File

@ -105,7 +105,8 @@ if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
echo "stage 4: ASR Training"
mkdir -p ${exp_dir}/exp/${model_dir}
log_file="${exp_dir}/exp/${model_dir}/train.log.txt"
current_time=$(date "+%Y-%m-%d_%H-%M")
log_file="${exp_dir}/exp/${model_dir}/train.log.txt.${current_time}"
echo "log_file: ${log_file}"
gpu_num=$(echo $CUDA_VISIBLE_DEVICES | awk -F "," '{print NF}')

View File

@ -105,7 +105,8 @@ if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
echo "stage 4: ASR Training"
mkdir -p ${exp_dir}/exp/${model_dir}
log_file="${exp_dir}/exp/${model_dir}/train.log.txt"
current_time=$(date "+%Y-%m-%d_%H-%M")
log_file="${exp_dir}/exp/${model_dir}/train.log.txt.${current_time}"
echo "log_file: ${log_file}"
gpu_num=$(echo $CUDA_VISIBLE_DEVICES | awk -F "," '{print NF}')

View File

@ -105,7 +105,8 @@ if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
echo "stage 4: ASR Training"
mkdir -p ${exp_dir}/exp/${model_dir}
log_file="${exp_dir}/exp/${model_dir}/train.log.txt"
current_time=$(date "+%Y-%m-%d_%H-%M")
log_file="${exp_dir}/exp/${model_dir}/train.log.txt.${current_time}"
echo "log_file: ${log_file}"
gpu_num=$(echo $CUDA_VISIBLE_DEVICES | awk -F "," '{print NF}')

View File

@ -188,7 +188,7 @@ class Trainer:
epoch (int): The current epoch number.
"""
self.model.train()
pbar = tqdm(colour="blue", desc=f"Training Epoch: {epoch + 1}", total=len(self.dataloader_train),
pbar = tqdm(colour="blue", desc=f"rank: {self.local_rank}, Training Epoch: {epoch + 1}", total=len(self.dataloader_train),
dynamic_ncols=True)
# Set the number of steps for gradient accumulation
@ -278,7 +278,7 @@ class Trainer:
f"epoch: {epoch}/{self.max_epoch}, "
f"step: {batch_idx}/{len(self.dataloader_train)}, total: {self.batch_total}, "
f"(loss: {loss.detach().cpu().item():.3f}), "
f"{[(k, round(v.cpu().item(), 3)) for k, v in stats.items()]}"
f"{[(k, round(v.cpu().item(), 3)) for k, v in stats.items()]}, "
f"{speed_stats}, "
f"{gpu_info}"
)
@ -307,7 +307,7 @@ class Trainer:
"""
self.model.eval()
with torch.no_grad():
pbar = tqdm(colour="red", desc=f"Training Epoch: {epoch + 1}", total=len(self.dataloader_val),
pbar = tqdm(colour="red", desc=f"rank: {self.local_rank}, Validation Epoch: {epoch + 1}", total=len(self.dataloader_val),
dynamic_ncols=True)
speed_stats = {}
time5 = time.perf_counter()
@ -343,7 +343,7 @@ class Trainer:
f"validation epoch: {epoch}/{self.max_epoch}, "
f"step: {batch_idx}/{len(self.dataloader_val)}, "
f"(loss: {loss.detach().cpu().item():.3f}), "
f"{[(k, round(v.cpu().item(), 3)) for k, v in stats.items()]}"
f"{[(k, round(v.cpu().item(), 3)) for k, v in stats.items()]}, "
f"{speed_stats}, "
)
pbar.set_description(description)