| 12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152 |
- # -*- coding: utf-8 -*-
- """Viterbi 解码算法。
- 按 ARCHITECTURE.md Phase 3 拆分建议，从 ``common/Utils.py`` 迁出。
- 类型：CORE（模型推理后处理，人工主导）。
- 原位置：``common/Utils.py`` 第 129-157 行的 ``viterbi_decode``。
- ``common/Utils.py`` 仍 re-export ``viterbi_decode``，老 import 不受影响。
- 说明：
- ARCHITECTURE.md 提到 ``decode``、``viterbi_decode``。``common/Utils.py``
- 中只有 ``viterbi_decode``；名为 ``decode`` 的函数分散在 ``foolnltk/``、
- ``product/``、``complaint/`` 等模块中，不属于 Utils.py 拆分范围。
- 本文件只迁 ``viterbi_decode``。
- """
- from __future__ import absolute_import
- import numpy as np
- __all__ = ["viterbi_decode"]
- def viterbi_decode(score, transition_params):
- """Decode the highest scoring sequence of tags outside of TensorFlow.
- This should only be used at test time.
- Args:
- score: A [seq_len, num_tags] matrix of unary potentials.
- transition_params: A [num_tags, num_tags] matrix of binary potentials.
- Returns:
- viterbi: A [seq_len] list of integers containing the highest scoring tag
- indices.
- viterbi_score: A float containing the score for the Viterbi sequence.
- """
- trellis = np.zeros_like(score)
- backpointers = np.zeros_like(score, dtype=np.int32)
- trellis[0] = score[0]
- for t in range(1, score.shape[0]):
- v = np.expand_dims(trellis[t - 1], 1) + transition_params
- trellis[t] = score[t] + np.max(v, 0)
- backpointers[t] = np.argmax(v, 0)
- viterbi = [np.argmax(trellis[-1])]
- for bp in reversed(backpointers[1:]):
- viterbi.append(bp[viterbi[-1]])
- viterbi.reverse()
- viterbi_score = np.max(trellis[-1])
- return viterbi, viterbi_score
|
