search

google-research / simclr

SimCLRv2 - Big Self-Supervised Models are Strong Semi-Supervised Learners
https://arxiv.org/abs/2006.10029
Apache License 2.0
4.05k stars 623 forks source link

Command about fine-tune and semi-supervised learning on tf2 #161

Open lucasliunju opened 3 years ago

lucasliunju commented 3 years ago

Could you please provide the command about finetune and semi-supervised learning on tf2. I just find the command on tf1 and I find that is not appliable for tf2

Thank you!

Yong

wit543 commented 2 years ago

I was successful using this command

python run.py --mode=train_then_eval --train_mode=finetune \
  --fine_tune_after_block=4 --zero_init_logits_layer=True \
  --global_bn=False --optimizer=momentum --learning_rate=0.1 --weight_decay=0.0 \
  --train_epochs=100 --train_batch_size=63 --warmup_epochs=0 \
  --dataset=imagenet --image_size=224 --eval_split=test --resnet_depth=152 \
  --checkpoint=pretrianed_model/r152_1x_sk0/saved_model --model_dir=pretrained_ft --use_tpu=False
deepankarvarma commented 8 months ago

@wit543 can you please share your directory structure, I am a bit unclear about the --checkpoint and --model_dir what is the directory contents