Update README.md

README.md

@@ -16,6 +16,13 @@ license: apache-2.0
 
 # Wav2Vec2-Large-Robust - Finetuned on Librispeech (960 hours)
 
+## Note : Model has not been initialized. If you want to use it without further finetuning, do a forward pass first to recalculate the normalized weights of the positional convolutional layer :
+
+```ipython
+ with torch.no_grad():
+    model_hf(torch.randn((1,300_000)))
+```
+
 [Facebook's Wav2Vec2](https://ai.facebook.com/blog/wav2vec-20-learning-the-structure-of-speech-from-raw-audio/)
 
 The base model pretrained on 16kHz sampled speech audio.