zh-segmenter.master_spec 3.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188
  1. component {
  2. name: "lookahead"
  3. transition_system {
  4. registered_name: "shift-only"
  5. parameters {
  6. key: "left_to_right"
  7. value: "false"
  8. }
  9. }
  10. resource {
  11. name: "word-map"
  12. part {
  13. file_pattern: "word-map"
  14. }
  15. }
  16. resource {
  17. name: "tag-map"
  18. part {
  19. file_pattern: "tag-map"
  20. }
  21. }
  22. resource {
  23. name: "tag-to-category"
  24. part {
  25. file_pattern: "tag-to-category"
  26. }
  27. }
  28. resource {
  29. name: "lcword-map"
  30. part {
  31. file_pattern: "lcword-map"
  32. }
  33. }
  34. resource {
  35. name: "category-map"
  36. part {
  37. file_pattern: "category-map"
  38. }
  39. }
  40. resource {
  41. name: "char-map"
  42. part {
  43. file_pattern: "char-map"
  44. }
  45. }
  46. resource {
  47. name: "char-ngram-map"
  48. part {
  49. file_pattern: "char-ngram-map"
  50. }
  51. }
  52. resource {
  53. name: "label-map"
  54. part {
  55. file_pattern: "label-map"
  56. }
  57. }
  58. resource {
  59. name: "prefix-table"
  60. part {
  61. file_pattern: "prefix-table"
  62. }
  63. }
  64. resource {
  65. name: "suffix-table"
  66. part {
  67. file_pattern: "suffix-table"
  68. }
  69. }
  70. fixed_feature {
  71. name: "char"
  72. fml: "input(-1).char input.char input(1).char"
  73. embedding_dim: 32
  74. vocabulary_size: 3521
  75. size: 3
  76. }
  77. fixed_feature {
  78. name: "char-bigram"
  79. fml: "input.char-bigram"
  80. embedding_dim: 32
  81. vocabulary_size: 6579
  82. size: 1
  83. }
  84. network_unit {
  85. registered_name: "wrapped_units.LayerNormBasicLSTMNetwork"
  86. parameters {
  87. key: "hidden_layer_sizes"
  88. value: "256"
  89. }
  90. }
  91. backend {
  92. registered_name: "SyntaxNetComponent"
  93. }
  94. num_actions: 1
  95. component_builder {
  96. registered_name: "DynamicComponentBuilder"
  97. }
  98. }
  99. component {
  100. name: "segmenter"
  101. transition_system {
  102. registered_name: "binary-segment-transitions"
  103. }
  104. resource {
  105. name: "word-map"
  106. part {
  107. file_pattern: "word-map"
  108. }
  109. }
  110. resource {
  111. name: "tag-map"
  112. part {
  113. file_pattern: "tag-map"
  114. }
  115. }
  116. resource {
  117. name: "tag-to-category"
  118. part {
  119. file_pattern: "tag-to-category"
  120. }
  121. }
  122. resource {
  123. name: "lcword-map"
  124. part {
  125. file_pattern: "lcword-map"
  126. }
  127. }
  128. resource {
  129. name: "category-map"
  130. part {
  131. file_pattern: "category-map"
  132. }
  133. }
  134. resource {
  135. name: "char-map"
  136. part {
  137. file_pattern: "char-map"
  138. }
  139. }
  140. resource {
  141. name: "char-ngram-map"
  142. part {
  143. file_pattern: "char-ngram-map"
  144. }
  145. }
  146. resource {
  147. name: "label-map"
  148. part {
  149. file_pattern: "label-map"
  150. }
  151. }
  152. resource {
  153. name: "prefix-table"
  154. part {
  155. file_pattern: "prefix-table"
  156. }
  157. }
  158. resource {
  159. name: "suffix-table"
  160. part {
  161. file_pattern: "suffix-table"
  162. }
  163. }
  164. linked_feature {
  165. name: "lookahead"
  166. fml: "input.focus stack.focus"
  167. embedding_dim: 32
  168. size: 2
  169. source_component: "lookahead"
  170. source_translator: "reverse-token"
  171. source_layer: "state_h_0"
  172. }
  173. network_unit {
  174. registered_name: "wrapped_units.LayerNormBasicLSTMNetwork"
  175. parameters {
  176. key: "hidden_layer_sizes"
  177. value: "128"
  178. }
  179. }
  180. backend {
  181. registered_name: "SyntaxNetComponent"
  182. }
  183. num_actions: 2
  184. component_builder {
  185. registered_name: "DynamicComponentBuilder"
  186. }
  187. }