ITA: An Energy-Efficient Attention and Softmax Accelerator for Quantized Transformers

Open in new window