aced125

Results 9 comments of aced125

improving distributed logging story

I'm experiencing this issue! Of wandb spawning multiple runs when doing ddp training. Will do the rank zero thing to avoid for now.

📚 Inaccurate pre-trained model predictions master thread

Hi all - have been noticing lately the sentence boundary detection with the default parser on both small and medium models seems to be a bit more off than what...

💫 Incompatibilities & support on later versions of spaCy

Any update on this?

Allow passing in tol and max sweeps arguments?

I sometimes get a bug: ``` RuntimeError: Expected stat == success to be true, but got false. (Could this error message be improved? If so, please report an enhancement request...

deepspeed strategy can't save checkpoint, TypeError: cannot pickle `torch._C._distributed_c10d.ProcessGroup` object

same issue

GPU memory not cleaned properly when using multiple workers in DataLoader

I get this issue as well. Currently I solve it with `ray stop` and `ray start --head --port=6379 --object-manager-port=8076 --autoscaling-config=~/ray_bootstrap_config.yaml --metrics-export-port=8080` but this is harder when you have multiple nodes

DataLoader multiprocessing causes jobs to fail

I get this as well

Zero-Shot Learning in Modern NLP | Joe Davison Blog

Nice blog - I only had time to skim through the high level of each method. Which method does the transformers pipeline use?

NameError: name 'PAD_WORD' is not defined

Am facing the same issue as well - can't find pad_word anywhere else.