BAAI
BGE M3
現行
31.0Mダウンロード3.1KいいねJan 2024公開日8K トークンコンテキストMITライセンス84 優秀品質
BGE M3 (0.5680000185966492B parameters) requires approximately 4.1 GB of VRAM with F16 quantization. For the best balance of quality and speed, we recommend hardware with at least 5 GB of VRAM.
はじめに
— コピー&ペーストでローカル実行Copy-paste commands to run BGE M3 on your machine.
Run
docker run --rm -it ghcr.io/ggerganov/llama.cpp:full \
--hf-repo "BAAI/bge-m3" \
--hf-file "bge-m3-F16.gguf" \
-c 4096 -ngl 99Quick specs
Parameters0.57B
Architecturedense
Context8K tokens
Modalityembedding
Min RAM0.2 GB
Rec. RAM1.2 GB (F16)
LicenseMIT
FamilyBGE
✓ RAG
About this model
- •Multi-Functionality: It can simultaneously perform the three common retrieval functionalities of embedding model: dense retrieval, multi-vector...
- •Multi-Linguality: It can support more than 100 working languages
- •Multi-Granularity: It is able to process inputs of different granularities, spanning from short sentences to long documents of up to 8192 tokens
関連モデル
おすすめ
最適なハードウェア
BGE M3のおすすめ
このモデルを実行
量子化オプション
量子化レベル別VRAM推定値
No hardware detected — fit column shows raw VRAM estimates
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 0.2 GB | Low | — |
Q3_K_S | 3 | 0.3 GB | Low | — |
NVFP4 | 4 | 0.3 GB | Medium | — |
Q4_K_M | 4 | 0.3 GB | Medium | — |
Q5_K_M | 5 | 0.4 GB | High | — |
Q6_K | 6 | 0.5 GB | High | — |
Q8_0 | 8 | 0.6 GB | Very High | — |
F16 | 16 | 1.2 GB | Maximum | — |
ハードウェア互換性
全ハードウェアの適合度推定
Computing compatibility...
メモリ内訳
Reference: RTX 2060 6GB
Weights1.2 GB
KV Cache1.1 GB
Runtime1.2 GB
Headroom0.6 GB
よくある質問
FAQ — BGE M3
関連項目