Fix modeling code (typos/bugs)

#5
by Xenova HF staff - opened

Same as https://huggingface.co/microsoft/Florence-2-large/discussions/12

Two changes:

  • Fix Florence2Seq2SeqLMOutput
  • Only set attention_mask device if inputs_embeds is not None
leoxiaobin changed pull request status to merged

Sign up or log in to comment