Add a Moonshine (RoPE + SwiGLU) automatic-speech-recognition model with encoder-decoder architecture and…
Add a Moonshine (RoPE + SwiGLU) automatic-speech-recognition model with encoder-decoder architecture and…: a task in HF ML Bench v0 (Harbor dataset). Add first-class support for the Moonshine family of automatic-speech-recognition (ASR) models to the transformers library. Moonshine is an…
The task
Add first-class support for the **Moonshine** family of automatic-speech-recognition (ASR) models to the `transformers` library. Moonshine is an encoder-decoder speech model that builds on the Whisper-style architecture but replaces two of its ingredients:
Part of AdithyaSK/HF_ML_Bench_v0.