| --- |
| license: apache-2.0 |
| base_model: sentence-transformers/all-MiniLM-L6-v2 |
| base_model_relation: quantized |
| library_name: onnx |
| tags: |
| - onnx |
| - int8 |
| - feature-extraction |
| - color-picker |
| - embed |
| --- |
| |
| # all-MiniLM-L6-v2 · INT8 ONNX (colour search embedder) |
|
|
| Single-file INT8 ONNX export of [`sentence-transformers/all-MiniLM-L6-v2`](https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2) |
| (384-dim sentence embeddings). Built as the English/Latin-query semantic embedder for |
| [toolve](https://toolve.net) colour-search tools: 23 MB, runs fully client-side via |
| transformers.js or onnxruntime-web, no server round-trip. |
|
|
| ## Files |
|
|
| | File | Size | Notes | |
| |---|---|---| |
| | `onnx/model_int8.onnx` | 23 MB | INT8 dynamic-quantized BertModel | |
| | `tokenizer.json` | — | unchanged from upstream | |
| | `config.json` / `tokenizer_config.json` | — | unchanged from upstream | |
| | `manifest.json` | — | sha256 + size of every file, for integrity checks | |
|
|
| ## Usage (transformers.js) |
|
|
| ```js |
| import { pipeline } from '@xenova/transformers'; |
| const extractor = await pipeline('feature-extraction', 'androidli/colorpicker-models', { quantized: true }); |
| const emb = await extractor('deep teal', { pooling: 'mean', normalize: true }); |
| ``` |
|
|
| ## License |
|
|
| Inherited from upstream: **Apache-2.0** (`sentence-transformers/all-MiniLM-L6-v2`, |
| base `nreimers/MiniLM-L6-H384-uncased`). |
|
|
