Update README.md

README.md

@@ -62,7 +62,7 @@ This model is a fine-tuned version of intfloat/multilingual-e5-large for documen
     'patents': 4,
     'scientific_articles': 5
 }
-
+```
 ## Training procedure
 
 Trained on single gpu for 2 epochs for apx. 20 minutes.
@@ -82,3 +82,38 @@ hyperparameters:
 
 ## Evaluation results
 Test Loss:  0.5192, Test Acc: 0.9719
+
+
+## Usage:
+
+```python
+
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+
+# Load model and tokenizer
+tokenizer = AutoTokenizer.from_pretrained("kaixkhazaki/multilingual-e5-doclaynet")
+model = AutoModelForSequenceClassification.from_pretrained("kaixkhazaki/multilingual-e5-doclaynet")
+
+# Prepare text (note the "passage: " prefix required for E5 models)
+text = "passage: " + your_document_text
+
+# Tokenize and predict
+inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=512)
+outputs = model(**inputs)
+predictions = outputs.logits.softmax(dim=-1)
+
+# Get predicted class
+predicted_class = predictions.argmax().item()
+
+# Map to label (assuming you've loaded the label mapping)
+label_mapping = {
+    0: 'financial_reports',
+    1: 'government_tenders',
+    2: 'laws_and_regulations',
+    3: 'manuals',
+    4: 'patents',
+    5: 'scientific_articles'
+}
+predicted_label = label_mapping[predicted_class]
+
+```