Sentence Similarity
sentence-transformers
Safetensors
xlm-roberta
feature-extraction
dense
Generated from Trainer
dataset_size:41454
loss:MultipleNegativesRankingLoss
Eval Results (legacy)
text-embeddings-inference
Instructions to use soof/miswag-category-mapper with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- sentence-transformers
How to use soof/miswag-category-mapper with sentence-transformers:
from sentence_transformers import SentenceTransformer model = SentenceTransformer("soof/miswag-category-mapper") sentences = [ "Bleu Eau De Parfum Men's Perfume عطر بلو للرجال", "Niche Perfumes | عطور النيش | عطور النيش | Beauty > Fragrance > Niche Perfumes | الجمال والعناية > عطور > عطور النيش", "Men Blouses | بلوز رجالي | بلوز رجالي | Clothes, Shoes & Bags > Men Clothes > Men Blouses > Men Blouses | ملابس، أحذية وحقائب > ملابس رجالية > بلوز وتيشرت رجالي > بلوز رجالي", "Men Shoes | حذاء رجالي | حذاء رجالي | Men Fashion > Men Shoes | ملابس رجالية > حذاء رجالي", "Men Niche Perfumes | عطور رجالية | عطور رجالية | Beauty > Fragrance > Niche Perfumes > Men Niche Perfumes | الجمال والعناية > عطور > عطور النيش > عطور رجالية", "Arabian, Oud & Makhmaria | عطور عربية ومخمرية | عطور عربية ومخمرية | Beauty > Fragrance > Arabian, Oud & Makhmaria | الجمال والعناية > عطور > عطور عربية ومخمرية", "Men Fragrance | عطور رجالية | بۆنی پیاوانی | Beauty > Fragrance > Fragrance > Men Fragrance | الجمال والعناية > عطور > العطور > عطور رجالية" ] embeddings = model.encode(sentences) similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [7, 7] - Notebooks
- Google Colab
- Kaggle
Handler loads model from model/ subdir
Browse files- handler.py +6 -1
handler.py
CHANGED
|
@@ -82,7 +82,12 @@ class EndpointHandler:
|
|
| 82 |
def __init__(self, path: str = ""):
|
| 83 |
model_dir = Path(path)
|
| 84 |
|
| 85 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 86 |
|
| 87 |
# Load precomputed category embeddings (numpy, no faiss needed)
|
| 88 |
self.category_embeddings = np.load(
|
|
|
|
| 82 |
def __init__(self, path: str = ""):
|
| 83 |
model_dir = Path(path)
|
| 84 |
|
| 85 |
+
# Load from model/ subdir to avoid ST auto-detection at repo root
|
| 86 |
+
model_path = model_dir / "model"
|
| 87 |
+
if model_path.exists():
|
| 88 |
+
self.bi_encoder = SentenceTransformer(str(model_path))
|
| 89 |
+
else:
|
| 90 |
+
self.bi_encoder = SentenceTransformer(str(model_dir))
|
| 91 |
|
| 92 |
# Load precomputed category embeddings (numpy, no faiss needed)
|
| 93 |
self.category_embeddings = np.load(
|