Value Error: Cannot create a tensor proto whose content is larger than 2GB.

[ByUnal]

Hi there,

I was just trying re-run the code by given default paramaters and dataset, but word2vec model. I'm using word2vec-google-news-300 model therefore the embedding-dim in my case is 300. So the final command I run :

python3 train_harnn.py --epochs 5 --batch-size 2 --embedding-dim 300 --embedding-type 0

But It throws me the error below.

...
2024-04-19 11:49:46,589 - INFO - Loading data...
2024-04-19 11:49:46,589 - INFO - Data processing...
2024-04-19 11:49:46.734756: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:995] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2024-04-19 11:49:46.756399: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1960] 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...
Traceback (most recent call last):
  File "train_harnn.py", line 275, in <module>
    train_harnn()
  File "train_harnn.py", line 55, in train_harnn
    harnn = TextHARNN(
  File "/home/ybkaratas/Desktop/HieararchMC/Hierarchical-Multi-Label-Text-Classification/HARNN/text_harnn.py", line 165, in __init__
    self.embedding = tf.constant(pretrained_embedding, dtype=tf.float32, name="embedding")
  File "/home/ybkaratas/miniconda3/envs/cht3.8/lib/python3.8/site-packages/tensorflow/python/framework/constant_op.py", line 162, in constant_v1
    return _constant_impl(value, dtype, shape, name, verify_shape=verify_shape,
  File "/home/ybkaratas/miniconda3/envs/cht3.8/lib/python3.8/site-packages/tensorflow/python/framework/constant_op.py", line 277, in _constant_impl
    const_tensor = ops._create_graph_constant(  # pylint: disable=protected-access
  File "/home/ybkaratas/miniconda3/envs/cht3.8/lib/python3.8/site-packages/tensorflow/python/framework/ops.py", line 1008, in _create_graph_constant
    tensor_util.make_tensor_proto(
  File "/home/ybkaratas/miniconda3/envs/cht3.8/lib/python3.8/site-packages/tensorflow/python/framework/tensor_util.py", line 585, in make_tensor_proto
    raise ValueError(
ValueError: Cannot create a tensor proto whose content is larger than 2GB.

Initially I thougt it was due to insufficient memory, but I got the same error even when I reduced the both batch size and training data size. What could be the reason ? Can you help me ?

My hardware features: Nvidia GeForce 3060 12GB