FlagEmbedding
FlagEmbedding copied to clipboard
FlagOpen
Retrieval and Retrieval-augmented LLMs
Hi - thank you for the great work. I was wondering which data did you use to finetune gemma-9b for the lightweight model? Is it the same dataset (bge-m3, quora,...
https://huggingface.co/spaces/Someshfengde/Visualized_BGE_demo I've developed a demo for Visualized_base_en_v1.5 model. I thought it'll be good if anyone just want to try out the model via huggingface spaces.
Hi: I wonder what's the base/source dataset used to create the following dataset - bio_book - one_details_book - multi_details_book - multi_details_paper_long - one_detail_paper_long Thanks! Best, Mingyi
能够提供activation beacon在LLaMA2-7B-base（非chat）上训练的checkpoints吗，我们希望引用作为baseline进行比较
考虑多用户使用，加载了多个reranker , 速度却变慢了？这个正常吗 1. 只开一个reranker. 300个文本，3个request，排队访问，大概各1s uvicorn api_rerank:app --host 0.0.0.0 --workers 1 --port 8004 INFO: Started server process [2821886] INFO: Waiting for application startup. INFO: Application startup complete. INFO: Uvicorn...
embedding model前向推理时，不同batchsize，相同文本，输出embedding小数点后几位不同，请问这是什么原因导致的？理论上相同的输入应该会有完全相同的输出才对？ 使用的是示例代码： ``` sentences_1 = ["样例数据"] sentences_2 = ["样例数据", "样例数据", "样例数据", "样例数据"] model_path='./model_dir/bge-base-zh-v1.5' model = FlagModel(model_path, query_instruction_for_retrieval="为这个句子生成表示以用于检索相关文章：", # query_instruction_for_retrieval="", use_fp16=False) # Setting use_fp16 to True speeds up computation with...
 这里第1、2句是不相似的，1、3句是相似的   但这里只有第二个欧几里得距离是满足1、3句更加相似。然而最后的总结句也是错误的。 我运行了代码，第三个cos距离确实是产生了相反结果。
The returning type was only `np.ndarray`, but it is possible to change the return type to `torch.Tensor` if we set `convert_to_numpy` to `False`. The origin version can lead to wordy...
i am using mteb==1.1.0, for eval reranker model but its showing following errorlog Reranking - AskUbuntuDupQuestions, s2s - MindSmallReranking, s2s - SciDocsRR, s2s - StackOverflowDupQuestions, s2s test.jsonl.gz: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████| 135k/135k [00:00
