Downloads · 30 days
12
8% of all-time downloads
minilm-bo-en-sim-onnx
khyentsevision/minilm-bo-en-sim-onnx
minilm-bo-en-sim-onnx is a sentence similarity model from khyentsevision. Use it when you need a score for how close two texts are. It is set up for transformers.js. The card lists the license as apache-2.0.
ONNX export of khyentsevision/minilm-bo-en-sim for browser inference via transformers.js.
All-time downloads
148
Public
Repo size
113 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.onnx113 MB · 99%
At a glance
- Task
- Sentence Similarity
- Library
- transformers.js
- License
- apache-2.0
- Model type
- bert
- Access
- Public
- Created
- Sep 14, 2026
- Updated
- Sep 14, 2026
- SHA
- a8770448
Base models
Model card
From the Hugging Face model README
minilm-bo-en-sim (ONNX)
ONNX export of khyentsevision/minilm-bo-en-sim for browser inference via transformers.js.
onnx/model.onnx— fp32, direct export via optimum.onnx/model_quantized.onnx— int8 dynamic quantization (~22MB), used by default in the tibetan-english-aligner Space. Validated against the original PyTorch model: per-sentence cosine similarity0.98 on a held-out sample, with the pairwise similarity structure (what alignment scoring actually depends on) well preserved.
This repo holds converted weights only — no code, no pipeline. Use the mean-pooled,
L2-normalized last hidden state over attention_mask (standard sentence-transformers
mean pooling) to reproduce the original model's sentence embeddings.
- Task
- Sentence Similarity
- Library
- transformers.js
- Type
- bert
- License
- apache-2.0
- Created
- Sep 14, 2026
- Updated
- Sep 14, 2026