TensorFlow 1.0.0-alpha

[hunkim]

TensorFlow 1.0.0-alpha

[normanheckscher]

• I've pushed in some changes to a new branch.
• Need to look further into the tf.contrib.seq2seq
• I can't compare speed with older versions yet because TF1.0 from the pip install isn't optimally compiled for my computer yet.
• Need to investigate the deprecation of concat

WARNING:tensorflow:From /Users/norman/Documents/workspace/word-rnn-tensorflow/model.py:66: concat (from tensorflow.python.ops.array_ops) is deprecated and will be removed after 2016-12-14.
Instructions for updating:
This op will be removed after the deprecation date. Please switch to tf.concat_v2().

W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.

[normanheckscher]

I slipped in a new --gpu_mem switch with a default use of 66% of GPU memory. I've found that this helps load the graph onto my limited 1GB share notebook GPU (rMBP mid-2012). It might not be needed, or it may inhibit, when using GPUs with larger memory specs.

Can change or remove this.