Open mgoulao opened 8 months ago
Update: Increasing the micro_batch_size
to 8 seems to do the trick, however, I'm now wondering if it should be possible to use smaller batch sizes?
I have same issue when i'm using sample_packing: false
, micro_batch_size: 1
.
This looks like a dupe of https://github.com/OpenAccess-AI-Collective/axolotl/issues/1092
what GPU are you using?
I'm using 4x A100 80GB
Please check that this issue hasn't been reported before.
Expected Behavior
I'm currently testing if I can fit the
sequence_len
so at the very least it should give an OOM. I have tried changing thesample_packing
to false but it just returns a different error.My setup is not ideal since I'm using Pytorch with CUDA 11.7 and BitsAndBytes with CUDA 12.2 (the version of my GPUs driver).
Here is my pip list (filtered)
Current behaviour
I obtain the following errors:
sample_packing=true
sample_packing=false
Steps to reproduce
I have created a accelerate config file:
And I'm running the following command:
Possible solution
No response
Which Operating Systems are you using?
Python Version
3.10
axolotl branch-commit
main/d69ba2b0b76fad112acecd5a1fbb339e6244ff7b
Acknowledgements