phi-1_5 / modeling_mixformer_sequential.py

Commit History

fix(phi-1_5): Checks length of `attention_mask`if it is passed as direct tensor.
f9f2ac7

gugarosa commited on

Support for `attention_mask` in forward pass.
3128bb6

gugarosa commited on

Upload MixFormerSequentialForCausalLM
d655135

suriyagunasekar commited on

Upload MixFormerSequentialForCausalLM
1698206

suriyagunasekar commited on