bge-small-zh-v1.5 · INT8 ONNX (colour search embedder, zh)

Single-file INT8 ONNX export of BAAI/bge-small-zh-v1.5 for Chinese / Chinese-character queries in toolve colour-search tools. 24 MB, lazy-loaded client-side (transformers.js / onnxruntime-web) on the first non-Latin query, so English-only users never download it — the Latin counterpart lives in androidli/colorpicker-models.

Files

File Size Notes
onnx/model_int8.onnx 24 MB INT8 dynamic-quantized BertModel (4 layers)
tokenizer.json / config.json / tokenizer_config.json — unchanged from upstream
manifest.json — sha256 + size of every file

Usage (transformers.js)

import { pipeline } from '@xenova/transformers';
const extractor = await pipeline('feature-extraction', 'androidli/colorpicker-models-zh', { quantized: true });
const emb = await extractor('黛蓝', { pooling: 'cls', normalize: true });

License

Inherited from upstream: MIT (BAAI bge-small-zh-v1.5).

Downloads last month
331
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for androidli/colorpicker-models-zh

Quantized
(11)
this model