sonoisa / feature-extraction updated 2 years ago

sentence-bert-base-ja-mean-tokens-v2

バージョン1よりも良いロス関数であるMultipleNegativesRankingLossを用いて学習した改良版です。 手元の非公開データセットでは、バージョン1よりも1.5〜2ポイントほど精度が高い結果が得られました。

Params
110 M
Context
512
Downloads 30d
575 K
Likes
51
Licence: cc-by-sa-4.0 Not gated SAFETENSORS 1 languages View on Hugging Face ↗

Download history

daily snapshots · 10 days
575 K554 K
Jul 28Jul 31Aug 3Aug 6

Can you run it?

Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.

FileQuantSizeEst. VRAMVerdict on RTX 4090 · 24 GB
model.safetensors f32 0.4 GB 1.0 GB ✅ Runs comfortably
Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.

Specifications

Architecture
BertModel
Parameters
110 M
Tensor type
F32
Context length
512
Vocabulary
32,000
Layers / heads
12 / 12
Licence
cc-by-sa-4.0
First seen on the Hub
2022-03-02
Training datasets
undisclosed
Added to our catalog
2026-07-28