omniASR-LLM-300M components

This repository contains the pinned OmniASR-LLM-300M model as separately loadable BF16 SafeTensors components. encoder/ is the HF-native Wav2Vec2 encoder, projector/ is the acoustic projection, and decoder/ contains the LLM and its text/language projections.

The decoder source is ziywang50/omniASR-LLM-300M@5aa5b38d2ac81967b8166a8ba6cb8598e12954d5 and the encoder source is giangndm/omniASR-LLM-300M-encoder@acf827a7cd67cf579ff56e9064c72183a7956835. The root config.json retains the original OmniASR architecture metadata.

Downloads last month
10
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for giangndm/omniASR-LLM-300M

Finetuned
(1)
this model