Safetensors
llama

Spire is a 7B parameter decoder-only model with strong abilities in machine translation, automatic speech recognition, and speech translation. SpireBase was created by applying speech-centric continued pretraining to TowerBase-7B-v0.1, which was itself created by applying continued pretraining to Llama 2.

For further details, check the repository and the our arXiv preprint.

Downloads last month
10
Safetensors
Model size
6.78B params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API: The model has no library tag.

Model tree for utter-project/SpireBase

Quantizations
1 model

Collection including utter-project/SpireBase