Hugging Face Vision Transformer (ViT)
Optimum Habana Vision Transformer (ViT) for PyTorch
BERT-5B
BERT model with 5 Billion Parameters using DeepSpeed for NLP
BERT-1.5B
BERT model with 1.5 Billion Parameters using DeepSpeed for NLP
PyTorch
These are the available PyTorch Container images in the Habana Vault.
These images contain SynapseAI SW and PyTorch framework that allow a user to run models on Gaudi.
For more details on docker usage, please refer to our installation guide on instructions to setup Container Runtime and Installation of docker images.
Vision Transformer
Vision Transformer – Image classification using transformers
BERT Fine tuning
BERT Fine Tuning Training and Inference
BERT Pretraining
BERT Pretraining for PyTorch