Set `tie_word_embeddings` to False
#7
by
aismlv
- opened
Fixes the warning below when using multiple GPUs. The parameter is not utilised by Llava (and set to false for Llama)
2023-12-14 19:03:27,491 - accelerate.utils.modeling - WARNING - The model weights are not tied. Please use the `tie_weights` method before using the `infer_auto_device` function.
Makes sense, thanks! would you be happy to apply the same changes for llava 13b as well?
ybelkada
changed pull request status to
merged
nice! 😊