hmellor
changed the title [Transformers backend] Support LoRA on lm_head and aux hidden state layer 0 [Transformers backend] Support LoRA on lm_head13 days ago
Derive the input embedding name for LoRA from the Transformers model
8b6486c5
Compare the embed_in LoRA test against HF in fp16
9597f6be
hmellor
changed the title [Transformers backend] Support LoRA on lm_head [Transformers backend] Support LoRA on lm_head and on any input embedding name13 days ago
Allow embedding_modules to be set per instance
a36e6c6f
Rename input_embedding_modules to embedding_modules
bb3c58fe
Trim the embed_in LoRA test
8bda795a
hmellormarked this pull request as ready for review 13 days ago
Login to write a write a comment.
Login via GitHub