Spaces:
Running
Running
Upload folder using huggingface_hub (part 2)
Browse files
src/lib/embeddings/client_fastembed.py
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from typing import List
|
| 2 |
+
from src.adapters.interfaces.embeddings import EmbeddingsClient
|
| 3 |
+
|
| 4 |
+
class FastEmbedEmbeddingsClient(EmbeddingsClient):
|
| 5 |
+
def __init__(self, model_name: str = "BAAI/bge-small-en-v1.5"):
|
| 6 |
+
self.model_name = model_name
|
| 7 |
+
self._model = None
|
| 8 |
+
|
| 9 |
+
@property
|
| 10 |
+
def model(self):
|
| 11 |
+
if self._model is None:
|
| 12 |
+
from fastembed import TextEmbedding
|
| 13 |
+
self._model = TextEmbedding(model_name=self.model_name)
|
| 14 |
+
return self._model
|
| 15 |
+
|
| 16 |
+
def embed_text(self, text: str) -> List[float]:
|
| 17 |
+
embeddings = list(self.model.embed([text]))
|
| 18 |
+
return embeddings[0].tolist()
|
| 19 |
+
|
| 20 |
+
def embed_texts(self, texts: List[str]) -> List[List[float]]:
|
| 21 |
+
embeddings = list(self.model.embed(texts))
|
| 22 |
+
return [e.tolist() for e in embeddings]
|
| 23 |
+
|
| 24 |
+
_fastembed_client = FastEmbedEmbeddingsClient()
|