neuronxcc-2.16.345.0+69131dd3/0_REGISTRY/0.0.28.dev0/inference/diffusion-transformer/hf-internal-testing/tiny-pixart-alpha-pipe/44b00dfe26235af3373e.json

{"model_type": "diffusion-transformer", "text_encoder": {"_attn_implementation_autoset": true, "architectures": ["T5EncoderModel"], "classifier_dropout": 0.0, "d_ff": 37, "d_kv": 8, "d_model": 32, "decoder_start_token_id": 0, "dense_act_fn": "relu", "dropout_rate": 0.1, "feed_forward_proj": "relu", "gradient_checkpointing": false, "initializer_factor": 0.002, "is_encoder_decoder": true, "is_gated_act": false, "layer_norm_epsilon": 1e-06, "model_type": "t5", "neuron": {"auto_cast": null, "auto_cast_type": null, "compiler_type": "neuronx-cc", "compiler_version": "2.16.345.0+69131dd3", "dynamic_batch_size": false, "inline_weights_to_neff": false, "optlevel": "2", "output_attentions": false, "output_hidden_states": false, "static_batch_size": 1, "static_sequence_length": 16, "tensor_parallel_size": 1}, "num_decoder_layers": 5, "num_heads": 4, "num_layers": 5, "relative_attention_max_distance": 128, "relative_attention_num_buckets": 8, "task": "feature-extraction", "use_cache": true, "vocab_size": 1103}, "transformer": {"_class_name": "PixArtTransformer2DModel", "activation_fn": "gelu-approximate", "attention_bias": true, "attention_head_dim": 8, "attention_type": "default", "caption_channels": 32, "cross_attention_dim": 24, "dropout": 0.0, "in_channels": 4, "interpolation_scale": null, "neuron": {"auto_cast": null, "auto_cast_type": null, "compiler_type": "neuronx-cc", "compiler_version": "2.16.345.0+69131dd3", "dynamic_batch_size": false, "inline_weights_to_neff": false, "optlevel": "2", "output_attentions": false, "output_hidden_states": false, "static_batch_size": 1, "static_encoder_hidden_size": 32, "static_height": 8, "static_num_channels": 4, "static_patch_size": 2, "static_sequence_length": 16, "static_vae_scale_factor": 1, "static_width": 8, "tensor_parallel_size": 1}, "norm_elementwise_affine": false, "norm_eps": 1e-06, "norm_num_groups": 32, "norm_type": "ada_norm_single", "num_attention_heads": 3, "num_embeds_ada_norm": 1000, "num_layers": 2, "out_channels": 8, "patch_size": 2, "task": "semantic-segmentation", "upcast_attention": false, "use_additional_conditions": null}}