ROZBEH comments

Results 9 comments of


                                            ROZBEH

Testing contextnet is extremely slow in the refactored main branch

@HesNobi @nglehuy Any suggestions on how to fix this? I am facing the same issue, the test.py is extremely slow. It takes 1 hour to run inference on 8 data...

training config used for training stt_en_quartznet15x5

I see - thanks @nithinraok Yeah knowing the actual config can be really beneficial.

Conversation Summary Buffer Memory does not accept a system message

Timeout DDP training using BioNeMo default configs

I am facing the same issue on a multi node multi GPU and without docker. I am utilizing slurm to run the job.

test.py not run with cpu=false

Hi there, Thanks for putting together this repo. Great resource. Any progress on this? Thanks.

Memory is fully eaten and training quit with errors for 40k hours ASR training

Hi there, Just checking here and wondering whether this is resolved? I am facing same issue. Thank you.

Memory is fully eaten and training quit with errors for 40k hours ASR training

Thanks @haihua I'm indeed 5 nodes with 5 GPU each. Is that what you mean?

Memory is fully eaten and training quit with errors for 40k hours ASR training

I see but the above issue is persistent with multi node and I'd like to get it working.

Animation Options

Any updates here?