from axengine import InferenceSession import numpy as np session = InferenceSession("./compiled_slice_quant_onnx/cfg_00_timestep_to_model_t_embedder_mlp_mlp_2_Gemm_output_0_config.axmodel") input_feed = { "timestep": np.array([1.0], dtype=np.float32), } output = session.run(None, input_feed)[0] print(output.shape)