Skip to content

Commit f6f8202

Browse files
fix emb script(#5345)
1 parent d9924cc commit f6f8202

File tree

1 file changed

+3
-0
lines changed

1 file changed

+3
-0
lines changed

examples/train/embedding/train_emb.sh

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -5,7 +5,10 @@ nproc_per_node=2
55
# --use_chat_template must be false to use generation template
66
# --dataloader_drop_last must be true or eval gather will throw error
77
# --model iic/gte-modernbert-base iic/gte_Qwen2-7B-instruct also supported
8+
# INFONCE_TEMPERATURE default value is 0.01, here we use 0.1 because it makes
9+
# the `sentence-transformers/stsb:positive` dataset result to a zero loss
810
CUDA_VISIBLE_DEVICES=0,1 \
11+
INFONCE_TEMPERATURE=0.1 \
912
NPROC_PER_NODE=$nproc_per_node \
1013
swift sft \
1114
--model Qwen/Qwen3-Embedding-0.6B \

0 commit comments

Comments
 (0)