mirror of https://github.com/explosion/spaCy.git
f57bea8ab6
The output weights often return negative scores for classes, especially via the bias terms. This means that when we add a new class, we can't rely on just zeroing the weights, or we'll end up with positive predictions for those labels. To solve this, we use nan values as the initial weights for new labels. This prevents them from ever coming out on top. During backprop, we replace the nan values with the minimum assigned score, so that we're still able to learn these classes. |
||
|---|---|---|
| .. | ||
| __init__.pxd | ||
| __init__.py | ||
| _beam_utils.pxd | ||
| _beam_utils.pyx | ||
| _parser_model.pxd | ||
| _parser_model.pyx | ||
| _state.pxd | ||
| _state.pyx | ||
| arc_eager.pxd | ||
| arc_eager.pyx | ||
| ner.pxd | ||
| ner.pyx | ||
| nn_parser.pxd | ||
| nn_parser.pyx | ||
| nonproj.pxd | ||
| nonproj.pyx | ||
| stateclass.pxd | ||
| stateclass.pyx | ||
| transition_system.pxd | ||
| transition_system.pyx | ||