Table 3 Hyper-parameter values for word representation models. Parameters specific to individual models are set to their defaults
From: Bio-SimVerb and Bio-SimLex: wide-coverage evaluation sets of word similarity in biomedicine
Parameters | Values |
|---|---|
Context window size | 5 |
Vector dimension | 200 |
Learning rate | 0.05 |
Negative sampling | 5 |
Min-count | 5 |
Sampling rate | 1e-5 |