yesl16

lora微调gte embedding, merge后推理结果跟微调的结果相差很大

1

lora微调gte embedding, 使用merge后的模型进行推理，结果跟微调的结果相差很大，甚至比初始模型效果还差 shell ``` swift sft \ --model 'iic/gte_Qwen2-1.5B-instruct' \ --train_type lora \ --dataset '/workspace/train_df.csv' \ --val_dataset '/workspace/test_df.csv' \ --torch_dtype bfloat16 \ --num_train_epochs 3 \ --per_device_train_batch_size 4 \ --per_device_eval_batch_size...