Document the GGUF alongside the ONNX
Browse files
README.md
CHANGED
|
@@ -1,6 +1,17 @@
|
|
| 1 |
A small English embedding model by BAAI (Beijing Academy of Artificial Intelligence).
|
| 2 |
|
| 3 |
-
This repository hosts the
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
|
| 5 |
---
|
| 6 |
license: mit
|
|
|
|
| 1 |
A small English embedding model by BAAI (Beijing Academy of Artificial Intelligence).
|
| 2 |
|
| 3 |
+
This repository hosts the versions of the model used by Understand to generate embeddings for semantic search, allowing users to search their codebase by meaning rather than exact keyword matches.
|
| 4 |
+
|
| 5 |
+
- `bge-small-en-v1.5-f16.gguf` — GGUF (F16), served by ullama with `--embeddings`.
|
| 6 |
+
Used by Understand 2026 and later. 384 dimensions, CLS pooling, 512-token context.
|
| 7 |
+
- `bge-small-en-v1.5.onnx` + `bge-small-en-v1.5-tokenizer.json` — ONNX, used by the
|
| 8 |
+
retired undaiserver. Kept for older releases.
|
| 9 |
+
|
| 10 |
+
The GGUF was converted from `BAAI/bge-small-en-v1.5` with llama.cpp's
|
| 11 |
+
`convert_hf_to_gguf.py`; its vectors match the reference implementation
|
| 12 |
+
(cosine similarity 1.00000). Note the ONNX path used mean pooling while the
|
| 13 |
+
GGUF uses BGE's specified CLS pooling, so indexes built with one are not
|
| 14 |
+
comparable with the other.
|
| 15 |
|
| 16 |
---
|
| 17 |
license: mit
|