Cannot use training_args.logging_steps of Hugging Face Trainer
See original GitHub issue- Weights and Biases version: version 0.8.35
- Python version: 3.7.6
- Operating System: Linux
Description
I wanted to use Pytorch Trainer, this works
from transformers import TrainingArguments
training_args = TrainingArguments("/kaggle/working")
...
# training_args.logging_steps=2
...
from transformers import Trainer
trainer = Trainer(
model=model,
args=training_args,...)
This does not, I received You can only call wandb.watch once per model
from transformers import TrainingArguments
training_args = TrainingArguments("/kaggle/working")
...
training_args.logging_steps=2
from transformers import Trainer
trainer = Trainer(
model=model,
args=training_args,...)
Issue Analytics
- State:
- Created 3 years ago
- Comments:9 (7 by maintainers)
Top Results From Across the Web
Trainer - Hugging Face
The Trainer class provides an API for feature-complete training in PyTorch for most standard use cases. It's used in most of the example...
Read more >Cannot use training_args.logging_steps of Hugging Face ...
Description. I wanted to use Pytorch Trainer, this works. from transformers import TrainingArguments training_args = ...
Read more >Pretrain Transformers Models in PyTorch Using Hugging Face ...
This notebook is designed to use an already pretrained transformers model and fine-tune it on your custom dataset, and also train a ...
Read more >Huggingface Trainer only doing 3 epochs no matter the ...
training_args = TrainingArguments( output_dir='./results', # output directory num_train_epochs=10, # total number of training epochs ...
Read more >Hugging Face Transformers - Documentation - Weights & Biases
A Weights & Biases integration for Hugging Face's Transformers library: solving NLP, one logged run ... from transformers import TrainingArguments, Trainer.
Read more >Top Related Medium Post
No results found
Top Related StackOverflow Question
No results found
Troubleshoot Live Code
Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start FreeTop Related Reddit Thread
No results found
Top Related Hackernoon Post
No results found
Top Related Tweet
No results found
Top Related Dev.to Post
No results found
Top Related Hashnode Post
No results found
Top GitHub Comments
Since this is on Kaggle, I would share the notebook here so that you can fork it. In the Settings menu on the right panel, please (1) Enable Internet for the kernel and (2) Use the lastest enviroment as seen in screenshot below
Can you try to restart your kernel? Also maybe remove the wandb folder created. I think it’s due to the importing of wandb which tries to look into this unrelated folder sometimes.
Let me know if that solves it and we can look into a permanent fix.