Basic implementation of the algorithm of "Session-based Recommendations With Recurrent Neural Networks" (http://arxiv.org/abs/1511.06939). With the extensions introduced in "Recurrent Neural Networks with Top-k Gains for Session-based Recommendations" (http://arxiv.org/abs/1706.03847).

Please always use the latest version as baseline and cite both papers when you do so! Parameters for the Recsys Challenge 2015 (RSC15) dataset can be found in examples/rsc15/run_rsc15.py

The code was optimized for GPU execution, running it on CPU is not optimal.

LICENSING: proper licence text coming soon! Meanwhile these are the main guidelines: for research and education purposes the code is and always will be free to use. Use in commercial systems of the code or parts of it requires a licence. If you've been using the code or any of its derivates in a commercial system, contact me.

I advise to use latest Theano version (Theano-1.0.0 or newer). The code was tested with various numpy and pandas versions under python 3.4 and 3.6. Python 2 is not supported.

• Refactor and cleaning.
• Speeding up execution.
• Ease of life improvements.
• Code for evaluating on GPU.

• Upgraded to the v2.0 version
• Added BPR-max and TOP1-max losses for cutting edge performance (coupled with additional sampling +30% in recall & MRR over the base results)
• Sacrificed some speed on CPU for faster GPU execution

• Fixed cross-entropy unstability. Very small predicted scores were rounded to 0 and thus their logarithm became NaN. Added a small epsilon (1e-24) to all scores before computing the logarithm. I got better results with this stabilized cross-entropy than with the TOP1 loss on networks with 100 hidden units.
• Added the option of using additional negative samples (besides the default, which is the other examples in the minibatch). The number of additional samples is given by the n_sample parameter. The probability of an item choosen as a sample is supp^sample_alpha, i.e. setting sample_alpha to 1 results in popularity based sampling, setting it to 0 results in uniform sampling. Using additional samples can slow down training, but depending on your config, the slowdown might not be noticable on GPU, up to 1000-2000 additional samples.
• Added an option to training to precompute a large batch of negative samples in advance. The number of int values (IDs) to be stored is determined by the sample_store parameter of the train function (default: 10M). This option is for the additional negative samples only, so only takes effect when n_sample > 0. Computing negative samples in each step results in very inefficient GPU utilization as computations are often interrupted by sample generation (which runs on the CPU). Precomputing samples for several steps in advance makes the process more efficient. However one should avoid setting the sample store too big as generating too many samples takes a long time, resulting in the GPU waiting for its completion for a long time. It also increases the memory footprint.

• Optimized code for GPU execution. Training is ~2x faster now.
• Added retrain functionality.