spotlight v0.1.2 Release Notes
Release Date: 2017-09-19 // over 6 years ago-
โ Added
spotlight.layers.BloomEmbedding
: bloom embedding layers that reduce the number of
๐ parameters required by hashing embedding indices into some fixed smaller dimensionality,
following Serrร , Joan, and Alexandros Karatzoglou. "Getting deep recommenders fit: Bloom
๐ embeddings for sparse binary input/output networks."sequence_mrr_score
now accepts an option that excludes previously seen items from scoring.
๐ Changed
- โก๏ธ
optimizer
arguments is nowoptimizer_func
. It accepts a function that takes a single argument (list of model parameters) and return a PyTorch optimizer (thanks to Ethan Rosenthal). fit
calls will resume from previous model state when called repeatedly (Ethan Rosenthal).- โก๏ธ Updated to work with PyTorch v0.2.0.
๐ Fixed
- ๐ Factorization predict APIs now work as advertised in the documentation.