linear regression model with MeanSquaredError optimized by mini-batch SGD feature_size: the input size, usually the feature size output_size: the output is a continuous value f32 punalty: penalty the weights, default is none weight_init_fn: it is strongly recommended to init the weights, e.g., randomly init: |item| item.for_each(|i| *i = rng.gen_f32())
(feature_size: usize, penalty: Option<Penalty>, weight_init_fn: F)
source not stored for this graph (policy: none)
nothing calls this directly
no test coverage detected