New discussion

Sequence length?

#5 opened 8 months ago by deleted

Handle model parallelism

#4 opened 10 months ago by sgugger

adds _no_split_block

1
#3 opened 11 months ago by staturecrane

Results are extremely poor

2
#2 opened 12 months ago by Steve72

Quantization support.

3
#1 opened 12 months ago by AV99