A large Wav2Vec2 Conformer model with rotary position embeddings fine-tuned on 960 hours of LibriSpeech for English ASR.