Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -1,3 +1,43 @@
|
|
| 1 |
---
|
| 2 |
license: mit
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: mit
|
| 3 |
+
base_model: intfloat/multilingual-e5-base
|
| 4 |
+
library_name: gguf
|
| 5 |
+
tags:
|
| 6 |
+
- feature-extraction
|
| 7 |
+
- sentence-similarity
|
| 8 |
+
- embeddings
|
| 9 |
+
- gguf
|
| 10 |
+
pipeline_tag: feature-extraction
|
| 11 |
---
|
| 12 |
+
|
| 13 |
+
# multilingual-e5-base — GGUF (Q8_0)
|
| 14 |
+
|
| 15 |
+
GGUF build of [intfloat/multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) — a multilingual XLM-RoBERTa embedding model (278M params, 768 dimensions, 100 languages, `passage:`/`query:` asymmetric prefixes).
|
| 16 |
+
|
| 17 |
+
Built **from first-party intfloat safetensors**, because no reputable quantizer publishes this model — this repo exists so downstream projects (vectorlink, TerminusDB) have an auditable supply chain rather than pulling a random community upload.
|
| 18 |
+
|
| 19 |
+
## Provenance
|
| 20 |
+
|
| 21 |
+
| Step | Tool | Pin |
|
| 22 |
+
|------|------|-----|
|
| 23 |
+
| Source weights | `intfloat/multilingual-e5-base` | commit `d128750597153bb5987e10b1c3493a34e5a4502a` |
|
| 24 |
+
| Conversion | llama.cpp `convert_hf_to_gguf.py` | release `b6987` |
|
| 25 |
+
| Quantization | `llama-quantize` → Q8_0 | llama.cpp `0.6.0` |
|
| 26 |
+
|
| 27 |
+
Build script: [`make-multilingual-e5-base-gguf.sh`](https://github.com/terminusdb-org/vectorlink/blob/main/scripts/make-multilingual-e5-base-gguf.sh) — downloads pinned first-party weights, verifies the safetensors SHA-256, converts, asserts mean pooling + 768-dim in the GGUF metadata, then quantizes.
|
| 28 |
+
|
| 29 |
+
## Verification
|
| 30 |
+
|
| 31 |
+
Same probe text, cosine similarity vs the original safetensors weights (f16 conversion): `0.9998`. Mean pooling, `passage:`/`query:` prefixes applied by the caller per E5 convention.
|
| 32 |
+
|
| 33 |
+
## Usage (Ollama)
|
| 34 |
+
|
| 35 |
+
```bash
|
| 36 |
+
ollama pull hf.co/vectorlink/multilingual-e5-base-GGUF:Q8_0
|
| 37 |
+
ollama cp hf.co/vectorlink/multilingual-e5-base-GGUF:Q8_0 multilingual-e5-base
|
| 38 |
+
curl localhost:11434/api/embed -d '{"model":"multilingual-e5-base","input":"query: test"}' # → 768-dim
|
| 39 |
+
```
|
| 40 |
+
|
| 41 |
+
## License
|
| 42 |
+
|
| 43 |
+
MIT — inherited from `intfloat/multilingual-e5-base`. This repo performs format conversion and quantization only; no training, fine-tuning, or modification of weights. See `LICENSE`.
|