Add Qwen3 LLM

[sarapapi]

Description

• Model Name: Qwen3
• Model Type: LLM
• Checkpoint: https://huggingface.co/Qwen/Qwen3-30B-A3B-Instruct-2507
• Paper/Technical Report: https://arxiv.org/abs/2505.09388
• License: Apache 2.0
• Motivation: Qwen3-Omni was added, and it's interesting to see the comparison with a cascaded version with the same LLM (Qwen3)

Outputs

List of benchmarks for which outputs are provided:

• FLEURS
• CoVoST2 (Canary done, Seamless Patricia running, OWSM Patricia and Javi, Whisper Sara and Javi running)
• Europarl-ST
• WMT
• WinoST (Patricia running)
• CommonAccent
• ManDi
• CS-Dialogue
• CS-FLEURS
• LibriStutter
• NoisyFLEURS (ambient) (Patricia in queue)
• NoisyFLEURS (babble) (Patricia in queue)
• EmotionTalk
• mExpresso (Patricia running)
• ACL 60/60 (long)
• ACL 60/60 (short)
• MCIF (long)
• MCIF (short)