Error on config_from_keras_model: [IndexError: list index out of range]

franhervas99 commented 6 days ago

Prerequisites

Please make sure to check off these prerequisites before submitting a bug report.

[x] Checked on last version 0.8.1.
[x] Similar issue checked, not the same problem #172 .
[x] There are some steps down to reproduce.

Quick summary

I am trying to use hls4ml.utils.config_from_keras_model() to get the config from my keras model, and I get an error IndexError: list index out of range.

Details

I created a simple perceptron in keras, with pretrained weights and biases. Then, when trying to get the config for HLS4ML an error is produced.

Steps to Reproduce

This is the python script to generate the keras model:

import numpy as np
import tensorflow as tf
from tensorflow.keras import Model, Input
from tensorflow.keras.layers import Dense

# Neural network constants
gain1 = 0.000455892409391384
ymin1 = -1

b1 = -0.15387008833216031523
IW1 = np.array([0.1125322203431925322, -0.34492392384017417362, 0.7052591999722942484,
                -1.2719945400779701927, 2.089129767777872182, -1.2416911002851198642,
                0.53266405909136382846, -0.2257313111056203081, 0.076208495617127072763])

b2 = -0.34191592641085533089
LW2 = 1.2490161057030868541

ymin2 = -1
gain2 = 2165.98876

# Build the model
def build_model():
    # Input layer
    input_layer = Input(shape=(len(IW1),), name="Input")

    # Layer 1 (Dense with custom weights and bias, using tanh activation)
    dense1 = Dense(
        units=1,
        activation="tanh",
        use_bias=True,
        kernel_initializer=tf.constant_initializer(IW1.reshape(-1, 1)),
        bias_initializer=tf.constant_initializer(b1),
        name="Hidden",
    )(input_layer)

    # Layer 2 (Dense with single weight and bias)
    output_layer = Dense(
        units=1,
        activation="linear",
        use_bias=True,
        kernel_initializer=tf.constant_initializer(LW2),
        bias_initializer=tf.constant_initializer(b2),
        name="Output",
    )(dense1)

    # Create the model
    model = Model(inputs=input_layer, outputs=output_layer, name="Perceptron")
    return model

# Instantiate and compile the model
model = build_model()
model.compile(optimizer="adam", loss="mse")

# Summary of the model
model.summary()

# Save the model to a .keras file
model.save("./perceptron.keras")

This is the python script for using HLS4ML:

import keras
import hls4ml
import json

model_path = 'perceptron.keras'

# Load the model from the .keras file
keras_model = keras.models.load_model(model_path)

# Generate a simple configuration from keras model
config = hls4ml.utils.config_from_keras_model(keras_model, granularity='name')

# Convert to an hls model
hls_model = hls4ml.converters.convert_from_keras_model(keras_model, hls_config=config, output_dir='.')

Expected behavior

The HLS4ML script should read the config and convert the keras model to an HLS model.

Actual behavior

The HLS4ML script can't pass from the config stage.

Error

2024-11-20 09:20:22.246408: I tensorflow/core/util/port.cc:153] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.
2024-11-20 09:20:22.246749: I external/local_xla/xla/tsl/cuda/cudart_stub.cc:32] Could not find cuda drivers on your machine, GPU will not be used.
2024-11-20 09:20:22.248817: I external/local_xla/xla/tsl/cuda/cudart_stub.cc:32] Could not find cuda drivers on your machine, GPU will not be used.
2024-11-20 09:20:22.254286: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:477] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1732090822.263536   88589 cuda_dnn.cc:8310] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
E0000 00:00:1732090822.266125   88589 cuda_blas.cc:1418] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
2024-11-20 09:20:22.275680: I tensorflow/core/platform/cpu_feature_guard.cc:210] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 AVX_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
WARN: Unable to import optimizer(s) from expr_templates.py: No module named 'sympy'
/home/fran/.local/lib/python3.10/site-packages/hls4ml/converters/__init__.py:27: UserWarning: WARNING: Pytorch converter is not enabled!
  warnings.warn("WARNING: Pytorch converter is not enabled!", stacklevel=1)
WARNING: Failed to import handlers from core.py: No module named 'torch'.
WARNING: Failed to import handlers from convolution.py: No module named 'torch'.
WARNING: Failed to import handlers from merge.py: No module named 'torch'.
WARNING: Failed to import handlers from pooling.py: No module named 'torch'.
WARNING: Failed to import handlers from reshape.py: No module named 'torch'.
W0000 00:00:1732090823.476630   88589 gpu_device.cc:2344] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...
/home/fran/.local/lib/python3.10/site-packages/keras/src/saving/saving_lib.py:713: UserWarning: Skipping variable loading for optimizer 'adam', because it has 10 variables whereas the saved optimizer has 2 variables. 
  saveable.load_own_variables(weights_store.get(inner_path))
Interpreting Model
Topology:
Traceback (most recent call last):
  File "/home/fran/work_repository/machine_learning_fpga/python/issue_github/hls4ml_dev.py", line 11, in <module>
    config = hls4ml.utils.config_from_keras_model(keras_model, granularity='name')
  File "/home/fran/.local/lib/python3.10/site-packages/hls4ml/utils/config.py", line 159, in config_from_keras_model
    layer_list, _, _, _ = hls4ml.converters.p
arse_keras_model(model_arch, reader)
  File "/home/fran/.local/lib/python3.10/site-packages/hls4ml/converters/keras_to_hls.py", line 260, in parse_keras_model
    input_shapes = [output_shapes[inbound_node[0]] for inbound_node in keras_layer['inbound_nodes'][0]]
IndexError: list index out of range

Scripts and keras model files

issue_github.zip

vloncar commented 4 days ago

The serialized format of Keras models changed in recent versions, leading to this error. You have two options, use older TF and Keras, or use main branch of hls4ml (soon to be 1.0.0 release)

franhervas99 commented 4 days ago

Hi @vloncar , Updating to the last version (main branch), gives me the same error as before.

fastmachinelearning / hls4ml