Pretrained models

Pretrained models

The follow pre-trained language models are considered parts of the training data and freely usable to build the SLT systems:

• Wav2vec 2.0
• Hubert
• WavLM
• SpeechLM
• data2vec
• MBART
• MBART50
• M2M100
• Delta LM
• T5
• BLOOM(only the small 560m paramter version)