jikechao opened a new issue, #15084: URL: https://github.com/apache/tvm/issues/15084
For the UpSampling2D operator in keras. If we set the value of attr `interpolation` to **bilinear** or **bicubic**, TVM will give different predicted results with Keras. ### Expected behavior TVM and Keras have the same inference results. ### Actual behavior TVM gave different inference results with Keras.  ### Environment TVM: latest Keras: 2.12.0 ### Steps to reproduce ``` import tvm import tvm.relay as relay import numpy as np from tensorflow import keras from tensorflow.keras import layers, models input_shape = (1, 1, 2, 3) input_data = np.random.random(size=input_shape) x = layers.Input(shape=input_shape[1:], dtype='float32') kwargs={'size':[2, 2],'data_format':"channels_last",'interpolation':"bilinear"} # bilinear, bicubic lead to the wrong result layer = keras.layers.UpSampling2D(**kwargs) layer.set_weights(layer.get_weights()) y = layer(x) model = models.Model(x, y) model.summary() res_keras = model(input_data) print(res_keras.shape) shape_dict = {'input_1': input_shape} mod, params = relay.frontend.from_keras(model, shape_dict,layout='NHWC') with tvm.transform.PassContext(opt_level=3): model = relay.build_module.create_executor("graph", mod, tvm.cpu(0), 'llvm', params).evaluate() test_x_tvm = input_data res_tvm = model(tvm.nd.array(test_x_tvm.astype('float32'))).numpy() np.testing.assert_allclose(res_keras, res_tvm, atol=1e-3, rtol=1e-3) ``` ### Triage Please refer to the list of label tags [here](https://github.com/apache/tvm/wiki/Issue-Triage-Labels) to find the relevant tags and add them below in a bullet format (example below). * needs-triage * frontend:keras * topi ### Analysis 1. if we use interpolation="nearest", the inference results are the same between keras and tvm. 2. the bug seems in the file [topi/image/resize](https://github.com/apache/tvm/blob/main/python/tvm/topi/image/resize.py) I tried to locate/fix this bug but failed. Wish someone could help me review and fix this bug. Thanks! -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
