Issues

• [Bug] Batch-size probing swallows non-OOM RuntimeError and retries indefinitely at batch size zero

  FlagOpen/FlagEmbedding#1607 · Filter by author dongyikuan919 opened on Sep 30, 2026

• [Bug] With negatives_cross_device and kd_loss_type=kl_div, the distillation weight grows with the number of GPUs

  FlagOpen/FlagEmbedding#1603 · Filter by author gss10282025 opened on Sep 26, 2026

• Tutorials/2_Metrics /2.1_Similarity_Metrics.ipynb 示例中的最终结论写反

  FlagOpen/FlagEmbedding#1601 · Filter by author 1104056452 opened on Sep 16, 2026
  1

• Consider using pre-built Flash Attention kernels via kernels

  FlagOpen/FlagEmbedding#1582 · Filter by author sayakpaul opened on May 28, 2026

• bge-reranker-v2-minicpm-layerwise微调报错

  FlagOpen/FlagEmbedding#1581 · Filter by author yvway opened on May 19, 2026

• Silent trust_remote_code=True enable in FlagAutoModel.from_finetuned via basename-only registry lookup

  FlagOpen/FlagEmbedding#1579 · Filter by author beanduan22 opened on May 7, 2026
  6

• mac上使用omlx部署报错

  FlagOpen/FlagEmbedding#1578 · Filter by author liumingliang0305-sudo opened on May 7, 2026

• Embedder for m3-memory

  FlagOpen/FlagEmbedding#1577 · Filter by author skynetcmd opened on May 6, 2026

• rerank微调数据格式

  FlagOpen/FlagEmbedding#1576 · Filter by author yvway opened on Apr 24, 2026
  1

• unable to use with transformers==5.2.0

  FlagOpen/FlagEmbedding#1569 · Filter by author liu-yx17 opened on Mar 30, 2026
  1

• Activation Beacon，关于Loss定义的疑问：平均 token loss 是否会稀释真正依赖 beacon memory 的位置？

  FlagOpen/FlagEmbedding#1565 · Filter by author Strandding opened on Mar 20, 2026

• 这行注释掉会导致微调模型时flashattention2报错(fla2只支持fp16、bf16，关还关不掉)，赶紧修复了吧。

  FlagOpen/FlagEmbedding#1564 · Filter by author yy9996 opened on Mar 11, 2026

• 📋 Documentation Enhancement Suggestion

  FlagOpen/FlagEmbedding#1562 · Filter by author croviatrust opened on Mar 4, 2026

• question: calculate input_tokens about bge-reranker-v2-gemma

  FlagOpen/FlagEmbedding#1559 · Filter by author llyycchhee opened on Dec 29, 2025

• [Question] Seeking Spring Boot Starter or Best Practice for BGE Models in a RAG Project

  FlagOpen/FlagEmbedding#1555 · Filter by author wangyz351 opened on Dec 25, 2025

• Reproducing Llara on MSMARCO

  FlagOpen/FlagEmbedding#1554 · Filter by author LuisPB7 opened on Dec 21, 2025

• 多模态/知识图谱等技能什么时候可以融入进去

  FlagOpen/FlagEmbedding#1552 · Filter by author jinec opened on Dec 15, 2025

• 啥时候开源bge-vl的微调方法

  FlagOpen/FlagEmbedding#1550 · Filter by author lixiaofen474-beep opened on Nov 27, 2025

• torch.distributed.elastic.multiprocessing.errors.ChildFailedError

  FlagOpen/FlagEmbedding#1549 · Filter by author liwei6677 opened on Nov 19, 2025

• Activation beacon输出乱码

  FlagOpen/FlagEmbedding#1548 · Filter by author bestfleer opened on Nov 18, 2025
  6

• Why do the embeddings from my trained BGE-M3 model match those from the original model?

  FlagOpen/FlagEmbedding#1547 · Filter by author shyzzz521 opened on Nov 15, 2025

• peft 0.18 is incompatible

  FlagOpen/FlagEmbedding#1546 · Filter by author ripineros opened on Nov 13, 2025

• Error __init__. py position in visual_bge

  FlagOpen/FlagEmbedding#1545 · Filter by author yuzaoyah opened on Nov 13, 2025

• finetune是否支持50系显卡

  FlagOpen/FlagEmbedding#1543 · Filter by author Havedream6 opened on Nov 4, 2025

• 微调后bge-m3每次测试结果都不相同

  FlagOpen/FlagEmbedding#1542 · Filter by author May-Sunshine opened on Nov 3, 2025