FlagEmbedding
FlagEmbedding copied to clipboard
Retrieval and Retrieval-augmented LLMs
bge-rerank-base用onnx部署 显存持续增长 不会释放,直到溢出
请问这种loss产生偶尔上升的情况是否正常,又该如何判断预训练合适结束? [bge-m3-patent-retromae_batch56_max350.log](https://github.com/FlagOpen/FlagEmbedding/files/15360497/bge-m3-patent-retromae_batch56_max350.log)
pretrain
Is there away to pretrain the M_3 models?
稀疏向量的格式
你好,我正在用bge-m3生成的稀疏向量想对接qdrant数据库(我了解官方已有两个其他数据库的对接教程) 生成的稀疏向量格式类似这样 ``` { '\u8fd9': 0.1641845703125, '\u4ef6': 0.246826171875, '\u7684\u65f6\u5019': 0.0863037109375, '\u90a3\u4e48': 0.08343505859375, '\u4e5f\u5c31': 0.099853515625, '\u2581\u5f53': 0.084716796875, '\u60a8': 0.158203125, '\u770b\u5230': 0.1983642578125, '\u7684': 0.0579833984375, '\u800c': 0.033172607421875, '\u4e5f': 0.108154296875, '\u8131': 0.255126953125, '\u4e16\u754c': 0.1624755859375,...
报错如下 Traceback (most recent call last): File "/opt/tiger/ecom_Rag/FlagEmbedding/baai_general_embedding/finetune/run.py", line 222, in main() File "/opt/tiger/ecom_Rag/FlagEmbedding/baai_general_embedding/finetune/run.py", line 211, in main trainer.train() File "/usr/local/lib/python3.9/dist-packages/transformers/trainer.py", line 1885, in train return inner_training_loop( File "/usr/local/lib/python3.9/dist-packages/transformers/trainer.py", line...
在A5000的显卡上,对100条数据进行重排序,需要5s才能完成
添加噪声微调模型
I encountered the following error when trying to add noise parameters during fine-tuning of embeddings models. How should I resolve this? The command I ran is: CUDA_VISIBLE_DEVICES=7 torchrun --standalone --nnodes=1...
不懂这行代码有什么作用,引入时报错 
四卡40gH卡,batch开32,开了gradient_checkpointing,query_max_len 512 ,passage_max_len 8192 训着训着会突然OOM,求问稳定的解决方案,(尝试了batch开4,16,32都会炸)
Hello, I am using the officially provided method of loading the reranker to perform similarity calculations. During the calculation process, I found that after the cache stabilizes for a period...