Open subhalingamd opened 1 month ago
You're right, with flan-t5, there is position_bias
where I never handle this before in tensor parallel.
Not sure why there is an error when stopping the program in v4.3.0 but I'll do a fix in the next version.
Hi @minhthuc2502, thanks for the response.
You're right, with flan-t5, there is
position_bias
where I never handle this before in tensor parallel.
could you please share if there are any plans to fix this?
Hi,
I was running Flan-t5 XXL with ctranslate2 and observed completely different results when run with tensor parallelism.
To convert from HF to CT2:
Code:
Outputs: Case 1: No TP When run as
python run.py
ormpirun -n 1 python run.py
Case 2: With TP When run as
mpirun -n 2 python run.py
I hope this is not an expected behaviour.
Further, with v4.3.0, I get an extra error at the end (after the output) which I didn't use to get with v4.1.0 (with the same code). The error goes like this:
Your help would be greatly appreciated.