423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-04 12:42:51 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix #1819

Browse Source 
Former-commit-id: 1fcd545c3dd78bd2113cde8ef788c5395de11c34
This commit is contained in:
hoshi-hiyouga 2023-12-13 10:14:01 +08:00
parent 15b321da8e
commit 5b211cfbe9
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/llmtuner/model/utils.py
View File

@ -182,7 +182,7 @@ def resize_embedding_layer(model: "PreTrainedModel", tokenizer: "PreTrainedToken
return return
old_vocab_size = model.get_input_embeddings().weight.size(0) old_vocab_size = model.get_input_embeddings().weight.size(0)
if len(tokenizer) != old_vocab_size: if len(tokenizer) > old_vocab_size:
model.resize_token_embeddings(len(tokenizer), pad_to_multiple_of=64) model.resize_token_embeddings(len(tokenizer), pad_to_multiple_of=64)
new_vocab_size = model.get_input_embeddings().weight.size(0) new_vocab_size = model.get_input_embeddings().weight.size(0)
logger.info("Resized token embeddings from {} to {}.".format(old_vocab_size, new_vocab_size)) logger.info("Resized token embeddings from {} to {}.".format(old_vocab_size, new_vocab_size))