Update README.md

README.md

@@ -39,3 +39,16 @@ from optimum.exporters.onnx import main_export
 main_export('sentence-transformers/all-distilroberta-v1', "./output", cache_dir='./cache', optimize='O1') 
 ```
 
+Please note, this ONNX model does not contain the mean pooling layer, it needs to be done in code afterwards or the embeddings won't work.
+
+Code like this:
+
+```python
+#Mean Pooling - Take attention mask into account for correct averaging
+def mean_pooling(model_output, attention_mask):
+    token_embeddings = model_output[0] #First element of model_output contains all token embeddings
+    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
+    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)
+```
+
+See the example code from the original model in the "Usage (HuggingFace Transformers)" section.