423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-23 22:32:54 +08:00
Code Issues Packages Projects Releases Wiki Activity

tiny fix

Browse Source 
Former-commit-id: 6953096c9d8f85d56cc980a4bec3a052411fb4a0
This commit is contained in:
hoshi-hiyouga 2023-12-13 10:21:29 +08:00
parent 5b211cfbe9
commit 657dff438c
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
src/llmtuner/model/utils.py
View File

@ -181,8 +181,8 @@ def resize_embedding_layer(model: "PreTrainedModel", tokenizer: "PreTrainedToken
logger.warning("Current model does not support resizing token embeddings.")
return
old_vocab_size = model.get_input_embeddings().weight.size(0)
if len(tokenizer) > old_vocab_size:
current_embedding_size = model.get_input_embeddings().weight.size(0)
if len(tokenizer) > current_embedding_size:
model.resize_token_embeddings(len(tokenizer), pad_to_multiple_of=64)
new_vocab_size = model.get_input_embeddings().weight.size(0)
logger.info("Resized token embeddings from {} to {}.".format(old_vocab_size, new_vocab_size))
new_embedding_size = model.get_input_embeddings().weight.size(0)
logger.info("Resized token embeddings from {} to {}.".format(current_embedding_size, new_embedding_size))