INNER CODE UNIT · Python

input_ids

FoundationVision/Liquid · evaluation/app.py:256

            input_ids = torch.cat([input_ids, next_token], dim=-1)
            model_kwargs = vqllm._update_model_kwargs_for_generation(
                outputs,
                model_kwargs,
                is_encoder_decoder=vqllm.config.is_encoder_decoder,
            )

        del sampling_kwargs
        del model_inputs
        del outputs
        image_vq_id = torch.cat(pred_tokens,dim=1)-ori_vocabe_size
        image_vq_id = torch.clamp(image_vq_id, min=0, max=8191)
        
        generated_image_list = []
        for index, generate_id in enumerate(image_vq_id):
            rec_img = image_tokenizer.pil_from_img_toks(generate_id)
            generated_image_list.append(rec_img)
            # rec_img.save('{}/{}.jpg'.format(image_save_pth,str(idx)))

View source record →