whowins 0.1.0__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (372) hide show
  1. whowins-0.1.0/CITATION.cff +23 -0
  2. whowins-0.1.0/LICENSE +190 -0
  3. whowins-0.1.0/MANIFEST.in +9 -0
  4. whowins-0.1.0/PKG-INFO +119 -0
  5. whowins-0.1.0/README_PYPI.md +83 -0
  6. whowins-0.1.0/configs/chat-templates/qwen3.5-large.jinja +154 -0
  7. whowins-0.1.0/configs/chat-templates/qwen3.5-small.jinja +154 -0
  8. whowins-0.1.0/configs/chat-templates/qwen3.jinja +89 -0
  9. whowins-0.1.0/configs/decoding.json +165 -0
  10. whowins-0.1.0/configs/environments.json +110 -0
  11. whowins-0.1.0/configs/models.json +380 -0
  12. whowins-0.1.0/configs/runs/balanced-qwen3-1.7b.json +26 -0
  13. whowins-0.1.0/configs/runs/balanced-qwen3-14b.json +25 -0
  14. whowins-0.1.0/configs/runs/initial-qwen3-1.7b.json +22 -0
  15. whowins-0.1.0/configs/runs/ladder-qwen3-14b.json +30 -0
  16. whowins-0.1.0/configs/runs/main-free-gpt-oss-120b-think.json +22 -0
  17. whowins-0.1.0/configs/runs/main-free-gpt-oss-20b-think.json +22 -0
  18. whowins-0.1.0/configs/runs/main-free-qwen3-0.6b-nothink.json +22 -0
  19. whowins-0.1.0/configs/runs/main-free-qwen3-0.6b-think.json +22 -0
  20. whowins-0.1.0/configs/runs/main-free-qwen3-1.7b-nothink.json +22 -0
  21. whowins-0.1.0/configs/runs/main-free-qwen3-1.7b-think.json +22 -0
  22. whowins-0.1.0/configs/runs/main-free-qwen3-14b-nothink.json +22 -0
  23. whowins-0.1.0/configs/runs/main-free-qwen3-14b-think.json +22 -0
  24. whowins-0.1.0/configs/runs/main-free-qwen3-32b-nothink-sm80.json +22 -0
  25. whowins-0.1.0/configs/runs/main-free-qwen3-32b-nothink.json +22 -0
  26. whowins-0.1.0/configs/runs/main-free-qwen3-32b-think-sm80.json +22 -0
  27. whowins-0.1.0/configs/runs/main-free-qwen3-32b-think.json +22 -0
  28. whowins-0.1.0/configs/runs/main-free-qwen3-4b-nothink.json +22 -0
  29. whowins-0.1.0/configs/runs/main-free-qwen3-4b-think.json +22 -0
  30. whowins-0.1.0/configs/runs/main-free-qwen3-8b-nothink.json +22 -0
  31. whowins-0.1.0/configs/runs/main-free-qwen3-8b-think.json +22 -0
  32. whowins-0.1.0/configs/runs/main-free-qwen3.5-0.8b-nothink.json +22 -0
  33. whowins-0.1.0/configs/runs/main-free-qwen3.5-0.8b-think.json +22 -0
  34. whowins-0.1.0/configs/runs/main-free-qwen3.5-122b-a10b-nothink.json +22 -0
  35. whowins-0.1.0/configs/runs/main-free-qwen3.5-122b-a10b-think.json +22 -0
  36. whowins-0.1.0/configs/runs/main-free-qwen3.5-27b-nothink.json +22 -0
  37. whowins-0.1.0/configs/runs/main-free-qwen3.5-27b-think.json +22 -0
  38. whowins-0.1.0/configs/runs/main-free-qwen3.5-2b-nothink.json +22 -0
  39. whowins-0.1.0/configs/runs/main-free-qwen3.5-2b-think.json +22 -0
  40. whowins-0.1.0/configs/runs/main-free-qwen3.5-35b-a3b-nothink.json +22 -0
  41. whowins-0.1.0/configs/runs/main-free-qwen3.5-35b-a3b-think.json +22 -0
  42. whowins-0.1.0/configs/runs/main-free-qwen3.5-4b-nothink.json +22 -0
  43. whowins-0.1.0/configs/runs/main-free-qwen3.5-4b-think.json +22 -0
  44. whowins-0.1.0/configs/runs/main-free-qwen3.5-9b-nothink.json +22 -0
  45. whowins-0.1.0/configs/runs/main-free-qwen3.5-9b-think.json +22 -0
  46. whowins-0.1.0/configs/runs/main-ladder-gpt-oss-120b-think.json +22 -0
  47. whowins-0.1.0/configs/runs/main-ladder-gpt-oss-20b-think.json +22 -0
  48. whowins-0.1.0/configs/runs/main-ladder-qwen3-0.6b-nothink.json +22 -0
  49. whowins-0.1.0/configs/runs/main-ladder-qwen3-0.6b-think.json +22 -0
  50. whowins-0.1.0/configs/runs/main-ladder-qwen3-1.7b-nothink.json +22 -0
  51. whowins-0.1.0/configs/runs/main-ladder-qwen3-1.7b-think.json +22 -0
  52. whowins-0.1.0/configs/runs/main-ladder-qwen3-14b-nothink.json +22 -0
  53. whowins-0.1.0/configs/runs/main-ladder-qwen3-14b-think.json +22 -0
  54. whowins-0.1.0/configs/runs/main-ladder-qwen3-32b-nothink-sm80.json +22 -0
  55. whowins-0.1.0/configs/runs/main-ladder-qwen3-32b-nothink.json +22 -0
  56. whowins-0.1.0/configs/runs/main-ladder-qwen3-32b-think-sm80.json +22 -0
  57. whowins-0.1.0/configs/runs/main-ladder-qwen3-32b-think.json +22 -0
  58. whowins-0.1.0/configs/runs/main-ladder-qwen3-4b-nothink.json +22 -0
  59. whowins-0.1.0/configs/runs/main-ladder-qwen3-4b-think.json +22 -0
  60. whowins-0.1.0/configs/runs/main-ladder-qwen3-8b-nothink.json +22 -0
  61. whowins-0.1.0/configs/runs/main-ladder-qwen3-8b-think.json +22 -0
  62. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-0.8b-nothink.json +22 -0
  63. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-0.8b-think.json +22 -0
  64. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-122b-a10b-nothink.json +22 -0
  65. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-122b-a10b-think.json +22 -0
  66. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-27b-nothink.json +22 -0
  67. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-27b-think.json +22 -0
  68. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-2b-nothink.json +22 -0
  69. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-2b-think.json +22 -0
  70. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-35b-a3b-nothink.json +22 -0
  71. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-35b-a3b-think.json +22 -0
  72. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-4b-nothink.json +22 -0
  73. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-4b-think.json +22 -0
  74. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-9b-nothink.json +22 -0
  75. whowins-0.1.0/configs/runs/main-ladder-qwen3.5-9b-think.json +22 -0
  76. whowins-0.1.0/configs/runs/main-load-gpt-oss-120b-think.json +22 -0
  77. whowins-0.1.0/configs/runs/main-load-gpt-oss-20b-think.json +22 -0
  78. whowins-0.1.0/configs/runs/main-load-qwen3-0.6b-nothink.json +22 -0
  79. whowins-0.1.0/configs/runs/main-load-qwen3-0.6b-think.json +22 -0
  80. whowins-0.1.0/configs/runs/main-load-qwen3-1.7b-nothink.json +22 -0
  81. whowins-0.1.0/configs/runs/main-load-qwen3-1.7b-think.json +22 -0
  82. whowins-0.1.0/configs/runs/main-load-qwen3-14b-nothink.json +22 -0
  83. whowins-0.1.0/configs/runs/main-load-qwen3-14b-think.json +22 -0
  84. whowins-0.1.0/configs/runs/main-load-qwen3-32b-nothink-sm80.json +22 -0
  85. whowins-0.1.0/configs/runs/main-load-qwen3-32b-nothink.json +22 -0
  86. whowins-0.1.0/configs/runs/main-load-qwen3-32b-think-sm80.json +22 -0
  87. whowins-0.1.0/configs/runs/main-load-qwen3-32b-think.json +22 -0
  88. whowins-0.1.0/configs/runs/main-load-qwen3-4b-nothink.json +22 -0
  89. whowins-0.1.0/configs/runs/main-load-qwen3-4b-think.json +22 -0
  90. whowins-0.1.0/configs/runs/main-load-qwen3-8b-nothink.json +22 -0
  91. whowins-0.1.0/configs/runs/main-load-qwen3-8b-think.json +22 -0
  92. whowins-0.1.0/configs/runs/main-load-qwen3.5-0.8b-nothink.json +22 -0
  93. whowins-0.1.0/configs/runs/main-load-qwen3.5-0.8b-think.json +22 -0
  94. whowins-0.1.0/configs/runs/main-load-qwen3.5-122b-a10b-nothink.json +22 -0
  95. whowins-0.1.0/configs/runs/main-load-qwen3.5-122b-a10b-think.json +22 -0
  96. whowins-0.1.0/configs/runs/main-load-qwen3.5-27b-nothink.json +22 -0
  97. whowins-0.1.0/configs/runs/main-load-qwen3.5-27b-think.json +22 -0
  98. whowins-0.1.0/configs/runs/main-load-qwen3.5-2b-nothink.json +22 -0
  99. whowins-0.1.0/configs/runs/main-load-qwen3.5-2b-think.json +22 -0
  100. whowins-0.1.0/configs/runs/main-load-qwen3.5-35b-a3b-nothink.json +22 -0
  101. whowins-0.1.0/configs/runs/main-load-qwen3.5-35b-a3b-think.json +22 -0
  102. whowins-0.1.0/configs/runs/main-load-qwen3.5-4b-nothink.json +22 -0
  103. whowins-0.1.0/configs/runs/main-load-qwen3.5-4b-think.json +22 -0
  104. whowins-0.1.0/configs/runs/main-load-qwen3.5-9b-nothink.json +22 -0
  105. whowins-0.1.0/configs/runs/main-load-qwen3.5-9b-think.json +22 -0
  106. whowins-0.1.0/configs/runs/main-message-gpt-oss-120b-think.json +22 -0
  107. whowins-0.1.0/configs/runs/main-message-gpt-oss-20b-think.json +22 -0
  108. whowins-0.1.0/configs/runs/main-message-qwen3-0.6b-nothink.json +22 -0
  109. whowins-0.1.0/configs/runs/main-message-qwen3-0.6b-think.json +22 -0
  110. whowins-0.1.0/configs/runs/main-message-qwen3-1.7b-nothink.json +22 -0
  111. whowins-0.1.0/configs/runs/main-message-qwen3-1.7b-think.json +22 -0
  112. whowins-0.1.0/configs/runs/main-message-qwen3-14b-nothink.json +22 -0
  113. whowins-0.1.0/configs/runs/main-message-qwen3-14b-think.json +22 -0
  114. whowins-0.1.0/configs/runs/main-message-qwen3-32b-nothink-sm80.json +22 -0
  115. whowins-0.1.0/configs/runs/main-message-qwen3-32b-nothink.json +22 -0
  116. whowins-0.1.0/configs/runs/main-message-qwen3-32b-think-sm80.json +22 -0
  117. whowins-0.1.0/configs/runs/main-message-qwen3-32b-think.json +22 -0
  118. whowins-0.1.0/configs/runs/main-message-qwen3-4b-nothink.json +22 -0
  119. whowins-0.1.0/configs/runs/main-message-qwen3-4b-think.json +22 -0
  120. whowins-0.1.0/configs/runs/main-message-qwen3-8b-nothink.json +22 -0
  121. whowins-0.1.0/configs/runs/main-message-qwen3-8b-think.json +22 -0
  122. whowins-0.1.0/configs/runs/main-message-qwen3.5-0.8b-nothink.json +22 -0
  123. whowins-0.1.0/configs/runs/main-message-qwen3.5-0.8b-think.json +22 -0
  124. whowins-0.1.0/configs/runs/main-message-qwen3.5-122b-a10b-nothink.json +22 -0
  125. whowins-0.1.0/configs/runs/main-message-qwen3.5-122b-a10b-think.json +22 -0
  126. whowins-0.1.0/configs/runs/main-message-qwen3.5-27b-nothink.json +22 -0
  127. whowins-0.1.0/configs/runs/main-message-qwen3.5-27b-think.json +22 -0
  128. whowins-0.1.0/configs/runs/main-message-qwen3.5-2b-nothink.json +22 -0
  129. whowins-0.1.0/configs/runs/main-message-qwen3.5-2b-think.json +22 -0
  130. whowins-0.1.0/configs/runs/main-message-qwen3.5-35b-a3b-nothink.json +22 -0
  131. whowins-0.1.0/configs/runs/main-message-qwen3.5-35b-a3b-think.json +22 -0
  132. whowins-0.1.0/configs/runs/main-message-qwen3.5-4b-nothink.json +22 -0
  133. whowins-0.1.0/configs/runs/main-message-qwen3.5-4b-think.json +22 -0
  134. whowins-0.1.0/configs/runs/main-message-qwen3.5-9b-nothink.json +22 -0
  135. whowins-0.1.0/configs/runs/main-message-qwen3.5-9b-think.json +22 -0
  136. whowins-0.1.0/configs/runs/main-operator-gpt-oss-120b-think.json +22 -0
  137. whowins-0.1.0/configs/runs/main-operator-gpt-oss-20b-think.json +22 -0
  138. whowins-0.1.0/configs/runs/main-operator-qwen3-0.6b-nothink.json +22 -0
  139. whowins-0.1.0/configs/runs/main-operator-qwen3-0.6b-think.json +22 -0
  140. whowins-0.1.0/configs/runs/main-operator-qwen3-1.7b-nothink.json +22 -0
  141. whowins-0.1.0/configs/runs/main-operator-qwen3-1.7b-think.json +22 -0
  142. whowins-0.1.0/configs/runs/main-operator-qwen3-14b-nothink.json +22 -0
  143. whowins-0.1.0/configs/runs/main-operator-qwen3-14b-think.json +22 -0
  144. whowins-0.1.0/configs/runs/main-operator-qwen3-32b-nothink-sm80.json +22 -0
  145. whowins-0.1.0/configs/runs/main-operator-qwen3-32b-nothink.json +22 -0
  146. whowins-0.1.0/configs/runs/main-operator-qwen3-32b-think-sm80.json +22 -0
  147. whowins-0.1.0/configs/runs/main-operator-qwen3-32b-think.json +22 -0
  148. whowins-0.1.0/configs/runs/main-operator-qwen3-4b-nothink.json +22 -0
  149. whowins-0.1.0/configs/runs/main-operator-qwen3-4b-think.json +22 -0
  150. whowins-0.1.0/configs/runs/main-operator-qwen3-8b-nothink.json +22 -0
  151. whowins-0.1.0/configs/runs/main-operator-qwen3-8b-think.json +22 -0
  152. whowins-0.1.0/configs/runs/main-operator-qwen3.5-0.8b-nothink.json +22 -0
  153. whowins-0.1.0/configs/runs/main-operator-qwen3.5-0.8b-think.json +22 -0
  154. whowins-0.1.0/configs/runs/main-operator-qwen3.5-122b-a10b-nothink.json +22 -0
  155. whowins-0.1.0/configs/runs/main-operator-qwen3.5-122b-a10b-think.json +22 -0
  156. whowins-0.1.0/configs/runs/main-operator-qwen3.5-27b-nothink.json +22 -0
  157. whowins-0.1.0/configs/runs/main-operator-qwen3.5-27b-think.json +22 -0
  158. whowins-0.1.0/configs/runs/main-operator-qwen3.5-2b-nothink.json +22 -0
  159. whowins-0.1.0/configs/runs/main-operator-qwen3.5-2b-think.json +22 -0
  160. whowins-0.1.0/configs/runs/main-operator-qwen3.5-35b-a3b-nothink.json +22 -0
  161. whowins-0.1.0/configs/runs/main-operator-qwen3.5-35b-a3b-think.json +22 -0
  162. whowins-0.1.0/configs/runs/main-operator-qwen3.5-4b-nothink.json +22 -0
  163. whowins-0.1.0/configs/runs/main-operator-qwen3.5-4b-think.json +22 -0
  164. whowins-0.1.0/configs/runs/main-operator-qwen3.5-9b-nothink.json +22 -0
  165. whowins-0.1.0/configs/runs/main-operator-qwen3.5-9b-think.json +22 -0
  166. whowins-0.1.0/configs/runs/main-realism-gpt-oss-120b-think.json +22 -0
  167. whowins-0.1.0/configs/runs/main-realism-gpt-oss-20b-think.json +22 -0
  168. whowins-0.1.0/configs/runs/main-realism-qwen3-0.6b-nothink.json +22 -0
  169. whowins-0.1.0/configs/runs/main-realism-qwen3-0.6b-think.json +22 -0
  170. whowins-0.1.0/configs/runs/main-realism-qwen3-1.7b-nothink.json +22 -0
  171. whowins-0.1.0/configs/runs/main-realism-qwen3-1.7b-think.json +22 -0
  172. whowins-0.1.0/configs/runs/main-realism-qwen3-14b-nothink.json +22 -0
  173. whowins-0.1.0/configs/runs/main-realism-qwen3-14b-think.json +22 -0
  174. whowins-0.1.0/configs/runs/main-realism-qwen3-32b-nothink-sm80.json +22 -0
  175. whowins-0.1.0/configs/runs/main-realism-qwen3-32b-nothink.json +22 -0
  176. whowins-0.1.0/configs/runs/main-realism-qwen3-32b-think-sm80.json +22 -0
  177. whowins-0.1.0/configs/runs/main-realism-qwen3-32b-think.json +22 -0
  178. whowins-0.1.0/configs/runs/main-realism-qwen3-4b-nothink.json +22 -0
  179. whowins-0.1.0/configs/runs/main-realism-qwen3-4b-think.json +22 -0
  180. whowins-0.1.0/configs/runs/main-realism-qwen3-8b-nothink.json +22 -0
  181. whowins-0.1.0/configs/runs/main-realism-qwen3-8b-think.json +22 -0
  182. whowins-0.1.0/configs/runs/main-realism-qwen3.5-0.8b-nothink.json +22 -0
  183. whowins-0.1.0/configs/runs/main-realism-qwen3.5-0.8b-think.json +22 -0
  184. whowins-0.1.0/configs/runs/main-realism-qwen3.5-122b-a10b-nothink.json +22 -0
  185. whowins-0.1.0/configs/runs/main-realism-qwen3.5-122b-a10b-think.json +22 -0
  186. whowins-0.1.0/configs/runs/main-realism-qwen3.5-27b-nothink.json +22 -0
  187. whowins-0.1.0/configs/runs/main-realism-qwen3.5-27b-think.json +22 -0
  188. whowins-0.1.0/configs/runs/main-realism-qwen3.5-2b-nothink.json +22 -0
  189. whowins-0.1.0/configs/runs/main-realism-qwen3.5-2b-think.json +22 -0
  190. whowins-0.1.0/configs/runs/main-realism-qwen3.5-35b-a3b-nothink.json +22 -0
  191. whowins-0.1.0/configs/runs/main-realism-qwen3.5-35b-a3b-think.json +22 -0
  192. whowins-0.1.0/configs/runs/main-realism-qwen3.5-4b-nothink.json +22 -0
  193. whowins-0.1.0/configs/runs/main-realism-qwen3.5-4b-think.json +22 -0
  194. whowins-0.1.0/configs/runs/main-realism-qwen3.5-9b-nothink.json +22 -0
  195. whowins-0.1.0/configs/runs/main-realism-qwen3.5-9b-think.json +22 -0
  196. whowins-0.1.0/configs/runs/main-tier-gpt-oss-120b-think.json +22 -0
  197. whowins-0.1.0/configs/runs/main-tier-gpt-oss-20b-think.json +22 -0
  198. whowins-0.1.0/configs/runs/main-tier-qwen3-0.6b-nothink.json +22 -0
  199. whowins-0.1.0/configs/runs/main-tier-qwen3-0.6b-think.json +22 -0
  200. whowins-0.1.0/configs/runs/main-tier-qwen3-1.7b-nothink.json +22 -0
  201. whowins-0.1.0/configs/runs/main-tier-qwen3-1.7b-think.json +22 -0
  202. whowins-0.1.0/configs/runs/main-tier-qwen3-14b-nothink.json +22 -0
  203. whowins-0.1.0/configs/runs/main-tier-qwen3-14b-think.json +22 -0
  204. whowins-0.1.0/configs/runs/main-tier-qwen3-32b-nothink-sm80.json +22 -0
  205. whowins-0.1.0/configs/runs/main-tier-qwen3-32b-nothink.json +22 -0
  206. whowins-0.1.0/configs/runs/main-tier-qwen3-32b-think-sm80.json +22 -0
  207. whowins-0.1.0/configs/runs/main-tier-qwen3-32b-think.json +22 -0
  208. whowins-0.1.0/configs/runs/main-tier-qwen3-4b-nothink.json +22 -0
  209. whowins-0.1.0/configs/runs/main-tier-qwen3-4b-think.json +22 -0
  210. whowins-0.1.0/configs/runs/main-tier-qwen3-8b-nothink.json +22 -0
  211. whowins-0.1.0/configs/runs/main-tier-qwen3-8b-think.json +22 -0
  212. whowins-0.1.0/configs/runs/main-tier-qwen3.5-0.8b-nothink.json +22 -0
  213. whowins-0.1.0/configs/runs/main-tier-qwen3.5-0.8b-think.json +22 -0
  214. whowins-0.1.0/configs/runs/main-tier-qwen3.5-122b-a10b-nothink.json +22 -0
  215. whowins-0.1.0/configs/runs/main-tier-qwen3.5-122b-a10b-think.json +22 -0
  216. whowins-0.1.0/configs/runs/main-tier-qwen3.5-27b-nothink.json +22 -0
  217. whowins-0.1.0/configs/runs/main-tier-qwen3.5-27b-think.json +22 -0
  218. whowins-0.1.0/configs/runs/main-tier-qwen3.5-2b-nothink.json +22 -0
  219. whowins-0.1.0/configs/runs/main-tier-qwen3.5-2b-think.json +22 -0
  220. whowins-0.1.0/configs/runs/main-tier-qwen3.5-35b-a3b-nothink.json +22 -0
  221. whowins-0.1.0/configs/runs/main-tier-qwen3.5-35b-a3b-think.json +22 -0
  222. whowins-0.1.0/configs/runs/main-tier-qwen3.5-4b-nothink.json +22 -0
  223. whowins-0.1.0/configs/runs/main-tier-qwen3.5-4b-think.json +22 -0
  224. whowins-0.1.0/configs/runs/main-tier-qwen3.5-9b-nothink.json +22 -0
  225. whowins-0.1.0/configs/runs/main-tier-qwen3.5-9b-think.json +22 -0
  226. whowins-0.1.0/configs/vllm/gpt-oss-120b.yaml +18 -0
  227. whowins-0.1.0/configs/vllm/gpt-oss-20b.yaml +18 -0
  228. whowins-0.1.0/configs/vllm/qwen3-0.6b.yaml +20 -0
  229. whowins-0.1.0/configs/vllm/qwen3-1.7b.yaml +20 -0
  230. whowins-0.1.0/configs/vllm/qwen3-14b.yaml +20 -0
  231. whowins-0.1.0/configs/vllm/qwen3-32b-sm80.yaml +22 -0
  232. whowins-0.1.0/configs/vllm/qwen3-32b.yaml +20 -0
  233. whowins-0.1.0/configs/vllm/qwen3-4b.yaml +20 -0
  234. whowins-0.1.0/configs/vllm/qwen3-8b.yaml +20 -0
  235. whowins-0.1.0/configs/vllm/qwen3.5-0.8b.yaml +19 -0
  236. whowins-0.1.0/configs/vllm/qwen3.5-122b-a10b.yaml +19 -0
  237. whowins-0.1.0/configs/vllm/qwen3.5-27b.yaml +19 -0
  238. whowins-0.1.0/configs/vllm/qwen3.5-2b.yaml +19 -0
  239. whowins-0.1.0/configs/vllm/qwen3.5-35b-a3b.yaml +19 -0
  240. whowins-0.1.0/configs/vllm/qwen3.5-4b.yaml +19 -0
  241. whowins-0.1.0/configs/vllm/qwen3.5-9b.yaml +19 -0
  242. whowins-0.1.0/data/load/lb-02.json +81 -0
  243. whowins-0.1.0/data/load/lb-03.json +80 -0
  244. whowins-0.1.0/data/load/lb-04.json +80 -0
  245. whowins-0.1.0/data/load/lb-05.json +82 -0
  246. whowins-0.1.0/data/load/lb-06.json +82 -0
  247. whowins-0.1.0/data/load/lb-07.json +82 -0
  248. whowins-0.1.0/data/load/lb-08.json +81 -0
  249. whowins-0.1.0/data/load/lb-09.json +81 -0
  250. whowins-0.1.0/data/load/lb-10.json +81 -0
  251. whowins-0.1.0/data/realism/batches.json +65 -0
  252. whowins-0.1.0/data/realism/rs-01.json +564 -0
  253. whowins-0.1.0/data/realism/rs-02.json +586 -0
  254. whowins-0.1.0/data/realism/rs-03.json +594 -0
  255. whowins-0.1.0/data/realism/rs-04.json +608 -0
  256. whowins-0.1.0/data/realism/rs-05.json +594 -0
  257. whowins-0.1.0/data/scenarios/format/fmt-0001-0010.json +202 -0
  258. whowins-0.1.0/data/scenarios/format/fmt-0011-0060.json +1010 -0
  259. whowins-0.1.0/data/scenarios/format/fmt-0061-0110.json +1016 -0
  260. whowins-0.1.0/data/scenarios/format/fmt-0111-0160.json +1011 -0
  261. whowins-0.1.0/data/scenarios/format/fmt-0161-0210.json +1010 -0
  262. whowins-0.1.0/data/scenarios/format/fmt-0211-0240.json +604 -0
  263. whowins-0.1.0/data/scenarios/ordering/ord-0001-0010.json +218 -0
  264. whowins-0.1.0/data/scenarios/ordering/ord-0011-0060.json +1056 -0
  265. whowins-0.1.0/data/scenarios/ordering/ord-0061-0110.json +1055 -0
  266. whowins-0.1.0/data/scenarios/ordering/ord-0111-0160.json +1052 -0
  267. whowins-0.1.0/data/scenarios/ordering/ord-0161-0210.json +1052 -0
  268. whowins-0.1.0/data/scenarios/ordering/ord-0211-0240.json +632 -0
  269. whowins-0.1.0/data/scenarios/path/path-0001-0010.json +219 -0
  270. whowins-0.1.0/data/scenarios/path/path-0011-0060.json +1104 -0
  271. whowins-0.1.0/data/scenarios/path/path-0061-0110.json +1103 -0
  272. whowins-0.1.0/data/scenarios/path/path-0111-0160.json +1103 -0
  273. whowins-0.1.0/data/scenarios/path/path-0161-0210.json +1102 -0
  274. whowins-0.1.0/data/scenarios/path/path-0211-0240.json +662 -0
  275. whowins-0.1.0/data/scenarios/refusal/ref-0001-0010.json +196 -0
  276. whowins-0.1.0/data/scenarios/refusal/ref-0011-0060.json +954 -0
  277. whowins-0.1.0/data/scenarios/refusal/ref-0061-0110.json +952 -0
  278. whowins-0.1.0/data/scenarios/refusal/ref-0111-0120.json +192 -0
  279. whowins-0.1.0/data/scenarios/tool/tool-0001-0010.json +219 -0
  280. whowins-0.1.0/data/scenarios/tool/tool-0011-0060.json +1102 -0
  281. whowins-0.1.0/data/scenarios/tool/tool-0061-0110.json +1102 -0
  282. whowins-0.1.0/data/scenarios/tool/tool-0111-0160.json +1102 -0
  283. whowins-0.1.0/data/scenarios/tool/tool-0161-0210.json +1102 -0
  284. whowins-0.1.0/data/scenarios/tool/tool-0211-0240.json +662 -0
  285. whowins-0.1.0/data/splits/axis.json +1226 -0
  286. whowins-0.1.0/data/splits/domain.json +1234 -0
  287. whowins-0.1.0/data/splits/item.json +4783 -0
  288. whowins-0.1.0/data/splits/set.json +2286 -0
  289. whowins-0.1.0/data/splits/source_pair.json +656 -0
  290. whowins-0.1.0/data/trials/01-decomposition.json +214 -0
  291. whowins-0.1.0/data/trials/02-composition.json +258 -0
  292. whowins-0.1.0/data/trials/03-load.json +158 -0
  293. whowins-0.1.0/data/trials/04-declared-tier.json +197 -0
  294. whowins-0.1.0/data/trials/05-binding.json +196 -0
  295. whowins-0.1.0/data/trials/06-authorship.json +158 -0
  296. whowins-0.1.0/data/trials/07-harness.json +171 -0
  297. whowins-0.1.0/pyproject.toml +52 -0
  298. whowins-0.1.0/setup.cfg +4 -0
  299. whowins-0.1.0/setup.py +30 -0
  300. whowins-0.1.0/src/whowins/__init__.py +9 -0
  301. whowins-0.1.0/src/whowins/checkers/__init__.py +59 -0
  302. whowins-0.1.0/src/whowins/checkers/core.py +557 -0
  303. whowins-0.1.0/src/whowins/checkers/coverage.py +59 -0
  304. whowins-0.1.0/src/whowins/checkers/fmt/__init__.py +13 -0
  305. whowins-0.1.0/src/whowins/checkers/fmt/_lib.py +284 -0
  306. whowins-0.1.0/src/whowins/checkers/fmt/domain.py +404 -0
  307. whowins-0.1.0/src/whowins/checkers/fmt/editorial.py +250 -0
  308. whowins-0.1.0/src/whowins/checkers/fmt/prose.py +1122 -0
  309. whowins-0.1.0/src/whowins/checkers/fmt/structural.py +2427 -0
  310. whowins-0.1.0/src/whowins/checkers/fmt/surface.py +520 -0
  311. whowins-0.1.0/src/whowins/checkers/fmt/syntax.py +386 -0
  312. whowins-0.1.0/src/whowins/checkers/fmt/tooling.py +1063 -0
  313. whowins-0.1.0/src/whowins/checkers/fmt/values.py +358 -0
  314. whowins-0.1.0/src/whowins/checkers/fmt/wire.py +243 -0
  315. whowins-0.1.0/src/whowins/checkers/trace.py +339 -0
  316. whowins-0.1.0/src/whowins/data.py +75 -0
  317. whowins-0.1.0/src/whowins/logs.py +135 -0
  318. whowins-0.1.0/src/whowins/progress.py +311 -0
  319. whowins-0.1.0/src/whowins/runner/__init__.py +31 -0
  320. whowins-0.1.0/src/whowins/runner/analyse.py +866 -0
  321. whowins-0.1.0/src/whowins/runner/batch.py +743 -0
  322. whowins-0.1.0/src/whowins/runner/bundles.py +109 -0
  323. whowins-0.1.0/src/whowins/runner/compare.py +270 -0
  324. whowins-0.1.0/src/whowins/runner/design.py +1115 -0
  325. whowins-0.1.0/src/whowins/runner/fillers.py +291 -0
  326. whowins-0.1.0/src/whowins/runner/layouts.py +225 -0
  327. whowins-0.1.0/src/whowins/runner/preflight.py +192 -0
  328. whowins-0.1.0/src/whowins/runner/rendering.py +283 -0
  329. whowins-0.1.0/src/whowins/runner/rescore.py +230 -0
  330. whowins-0.1.0/src/whowins/runner/response.py +208 -0
  331. whowins-0.1.0/src/whowins/runner/run.py +475 -0
  332. whowins-0.1.0/src/whowins/runner/serving.py +1249 -0
  333. whowins-0.1.0/src/whowins/runner/spec.py +363 -0
  334. whowins-0.1.0/src/whowins/runner/tools.py +226 -0
  335. whowins-0.1.0/src/whowins/runner/trials.py +342 -0
  336. whowins-0.1.0/src/whowins/runner/vocabulary.py +232 -0
  337. whowins-0.1.0/src/whowins/splits/__init__.py +15 -0
  338. whowins-0.1.0/src/whowins/splits/axes.py +78 -0
  339. whowins-0.1.0/src/whowins/splits/build.py +575 -0
  340. whowins-0.1.0/src/whowins/splits/check.py +491 -0
  341. whowins-0.1.0/src/whowins/splits/grouping.py +210 -0
  342. whowins-0.1.0/src/whowins/splits/sources.py +127 -0
  343. whowins-0.1.0/src/whowins/stats/__init__.py +11 -0
  344. whowins-0.1.0/src/whowins/stats/capacity.py +287 -0
  345. whowins-0.1.0/src/whowins/stats/choice.py +312 -0
  346. whowins-0.1.0/src/whowins/stats/families.py +386 -0
  347. whowins-0.1.0/src/whowins/stats/ladder.py +394 -0
  348. whowins-0.1.0/src/whowins/stats/layouts.py +253 -0
  349. whowins-0.1.0/src/whowins/stats/load_cell.py +298 -0
  350. whowins-0.1.0/src/whowins/stats/mixed.py +403 -0
  351. whowins-0.1.0/src/whowins/stats/models.py +540 -0
  352. whowins-0.1.0/src/whowins/stats/normal.py +185 -0
  353. whowins-0.1.0/src/whowins/stats/paired.py +89 -0
  354. whowins-0.1.0/src/whowins/stats/phase1.py +514 -0
  355. whowins-0.1.0/src/whowins/stats/pooled.py +128 -0
  356. whowins-0.1.0/src/whowins/stats/power.py +175 -0
  357. whowins-0.1.0/src/whowins/stats/rankings.py +173 -0
  358. whowins-0.1.0/src/whowins/stats/regularity.py +90 -0
  359. whowins-0.1.0/src/whowins/stats/score.py +197 -0
  360. whowins-0.1.0/src/whowins/stats/simulate.py +355 -0
  361. whowins-0.1.0/src/whowins/stats/strengths.py +371 -0
  362. whowins-0.1.0/src/whowins/validation/__init__.py +4 -0
  363. whowins-0.1.0/src/whowins/validation/build_registry.py +88 -0
  364. whowins-0.1.0/src/whowins/validation/check_batch.py +364 -0
  365. whowins-0.1.0/src/whowins/validation/check_load.py +247 -0
  366. whowins-0.1.0/src/whowins/validation/check_realism.py +221 -0
  367. whowins-0.1.0/src/whowins/validation/validate_collection.py +237 -0
  368. whowins-0.1.0/src/whowins.egg-info/PKG-INFO +119 -0
  369. whowins-0.1.0/src/whowins.egg-info/SOURCES.txt +370 -0
  370. whowins-0.1.0/src/whowins.egg-info/dependency_links.txt +1 -0
  371. whowins-0.1.0/src/whowins.egg-info/requires.txt +11 -0
  372. whowins-0.1.0/src/whowins.egg-info/top_level.txt +1 -0
@@ -0,0 +1,23 @@
1
+ cff-version: 1.2.0
2
+ message: "If you use WhoWins, please cite it as below."
3
+ title: "Position, Role or Label? A Controlled Study of Instruction Conflicts in LLM Agents"
4
+ type: software
5
+ authors:
6
+ - given-names: Anish
7
+ family-names: Laddha
8
+ affiliation: "LNMIIT, India"
9
+ - given-names: Nitesh
10
+ family-names: Pradhan
11
+ affiliation: "LNMIIT, India"
12
+ - given-names: Gaurav
13
+ family-names: Srivastava
14
+ affiliation: "Virginia Tech, USA"
15
+ version: 0.1.0
16
+ license: Apache-2.0
17
+ repository-code: "https://github.com/anishh15/WhoWins"
18
+ url: "https://anishh15.github.io/WhoWins/"
19
+ keywords:
20
+ - instruction hierarchy
21
+ - instruction conflicts
22
+ - LLM agents
23
+ - benchmark
whowins-0.1.0/LICENSE ADDED
@@ -0,0 +1,190 @@
1
+ Apache License
2
+ Version 2.0, January 2004
3
+ http://www.apache.org/licenses/
4
+
5
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
6
+
7
+ 1. Definitions.
8
+
9
+ "License" shall mean the terms and conditions for use, reproduction,
10
+ and distribution as defined by Sections 1 through 9 of this document.
11
+
12
+ "Licensor" shall mean the copyright owner or entity authorized by
13
+ the copyright owner that is granting the License.
14
+
15
+ "Legal Entity" shall mean the union of the acting entity and all
16
+ other entities that control, are controlled by, or are under common
17
+ control with that entity. For the purposes of this definition,
18
+ "control" means (i) the power, direct or indirect, to cause the
19
+ direction or management of such entity, whether by contract or
20
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
21
+ outstanding shares, or (iii) beneficial ownership of such entity.
22
+
23
+ "You" (or "Your") shall mean an individual or Legal Entity
24
+ exercising permissions granted by this License.
25
+
26
+ "Source" form shall mean the preferred form for making modifications,
27
+ including but not limited to software source code, documentation
28
+ source, and configuration files.
29
+
30
+ "Object" form shall mean any form resulting from mechanical
31
+ transformation or translation of a Source form, including but
32
+ not limited to compiled object code, generated documentation,
33
+ and conversions to other media types.
34
+
35
+ "Work" shall mean the work of authorship, whether in Source or
36
+ Object form, made available under the License, as indicated by a
37
+ copyright notice that is included in or attached to the work
38
+ (an example is provided in the Appendix below).
39
+
40
+ "Derivative Works" shall mean any work, whether in Source or Object
41
+ form, that is based on (or derived from) the Work and for which the
42
+ editorial revisions, annotations, elaborations, or other modifications
43
+ represent, as a whole, an original work of authorship. For the purposes
44
+ of this License, Derivative Works shall not include works that remain
45
+ separable from, or merely link (or bind by name) to the interfaces of,
46
+ the Work and Derivative Works thereof.
47
+
48
+ "Contribution" shall mean any work of authorship, including
49
+ the original version of the Work and any modifications or additions
50
+ to that Work or Derivative Works thereof, that is intentionally
51
+ submitted to the Licensor for inclusion in the Work by the copyright owner
52
+ or by an individual or Legal Entity authorized to submit on behalf of
53
+ the copyright owner. For the purposes of this definition, "submitted"
54
+ means any form of electronic, verbal, or written communication sent
55
+ to the Licensor or its representatives, including but not limited to
56
+ communication on electronic mailing lists, source code control systems,
57
+ and issue tracking systems that are managed by, or on behalf of, the
58
+ Licensor for the purpose of discussing and improving the Work, but
59
+ excluding communication that is conspicuously marked or otherwise
60
+ designated in writing by the copyright owner as "Not a Contribution."
61
+
62
+ "Contributor" shall mean Licensor and any individual or Legal Entity
63
+ on behalf of whom a Contribution has been received by the Licensor and
64
+ subsequently incorporated within the Work.
65
+
66
+ 2. Grant of Copyright License. Subject to the terms and conditions of
67
+ this License, each Contributor hereby grants to You a perpetual,
68
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
69
+ copyright license to reproduce, prepare Derivative Works of,
70
+ publicly display, publicly perform, sublicense, and distribute the
71
+ Work and such Derivative Works in Source or Object form.
72
+
73
+ 3. Grant of Patent License. Subject to the terms and conditions of
74
+ this License, each Contributor hereby grants to You a perpetual,
75
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
76
+ (except as stated in this section) patent license to make, have made,
77
+ use, offer to sell, sell, import, and otherwise transfer the Work,
78
+ where such license applies only to those patent claims licensable
79
+ by such Contributor that are necessarily infringed by their
80
+ Contribution(s) alone or by combination of their Contribution(s)
81
+ with the Work to which such Contribution(s) was submitted. If You
82
+ institute patent litigation against any entity (including a
83
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
84
+ or a Contribution incorporated within the Work constitutes direct
85
+ or contributory patent infringement, then any patent licenses
86
+ granted to You under this License for that Work shall terminate
87
+ as of the date such litigation is filed.
88
+
89
+ 4. Redistribution. You may reproduce and distribute copies of the
90
+ Work or Derivative Works thereof in any medium, with or without
91
+ modifications, and in Source or Object form, provided that You
92
+ meet the following conditions:
93
+
94
+ (a) You must give any other recipients of the Work or
95
+ Derivative Works a copy of this License; and
96
+
97
+ (b) You must cause any modified files to carry prominent notices
98
+ stating that You changed the files; and
99
+
100
+ (c) You must retain, in the Source form of any Derivative Works
101
+ that You distribute, all copyright, patent, trademark, and
102
+ attribution notices from the Source form of the Work,
103
+ excluding those notices that do not pertain to any part of
104
+ the Derivative Works; and
105
+
106
+ (d) If the Work includes a "NOTICE" text file as part of its
107
+ distribution, then any Derivative Works that You distribute must
108
+ include a readable copy of the attribution notices contained
109
+ within such NOTICE file, excluding any notices that do not
110
+ pertain to any part of the Derivative Works, in at least one
111
+ of the following places: within a NOTICE text file distributed
112
+ as part of the Derivative Works; within the Source form or
113
+ documentation, if provided along with the Derivative Works; or,
114
+ within a display generated by the Derivative Works, if and
115
+ wherever such third-party notices normally appear. The contents
116
+ of the NOTICE file are for informational purposes only and
117
+ do not modify the License. You may add Your own attribution
118
+ notices within Derivative Works that You distribute, alongside
119
+ or as an addendum to the NOTICE text from the Work, provided
120
+ that such additional attribution notices cannot be construed
121
+ as modifying the License.
122
+
123
+ You may add Your own copyright statement to Your modifications and
124
+ may provide additional or different license terms and conditions
125
+ for use, reproduction, or distribution of Your modifications, or
126
+ for any such Derivative Works as a whole, provided Your use,
127
+ reproduction, and distribution of the Work otherwise complies with
128
+ the conditions stated in this License.
129
+
130
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
131
+ any Contribution intentionally submitted for inclusion in the Work
132
+ by You to the Licensor shall be under the terms and conditions of
133
+ this License, without any additional terms or conditions.
134
+ Notwithstanding the above, nothing herein shall supersede or modify
135
+ the terms of any separate license agreement you may have executed
136
+ with Licensor regarding such Contributions.
137
+
138
+ 6. Trademarks. This License does not grant permission to use the trade
139
+ names, trademarks, service marks, or product names of the Licensor,
140
+ except as required for reasonable and customary use in describing the
141
+ origin of the Work and reproducing the content of the NOTICE file.
142
+
143
+ 7. Disclaimer of Warranty. Unless required by applicable law or
144
+ agreed to in writing, Licensor provides the Work (and each
145
+ Contributor provides its Contributions) on an "AS IS" BASIS,
146
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
147
+ implied, including, without limitation, any warranties or conditions
148
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
149
+ PARTICULAR PURPOSE. You are solely responsible for determining the
150
+ appropriateness of using or redistributing the Work and assume any
151
+ risks associated with Your exercise of permissions under this License.
152
+
153
+ 8. Limitation of Liability. In no event and under no legal theory,
154
+ whether in tort (including negligence), contract, or otherwise,
155
+ unless required by applicable law (such as deliberate and grossly
156
+ negligent acts) or agreed to in writing, shall any Contributor be
157
+ liable to You for damages, including any direct, indirect, special,
158
+ incidental, or consequential damages of any character arising as a
159
+ result of this License or out of the use or inability to use the
160
+ Work (including but not limited to damages for loss of goodwill,
161
+ work stoppage, computer failure or malfunction, or any and all
162
+ other commercial damages or losses), even if such Contributor
163
+ has been advised of the possibility of such damages.
164
+
165
+ 9. Accepting Warranty or Additional Liability. While redistributing
166
+ the Work or Derivative Works thereof, You may choose to offer,
167
+ and charge a fee for, acceptance of support, warranty, indemnity,
168
+ or other liability obligations and/or rights consistent with this
169
+ License. However, in accepting such obligations, You may act only
170
+ on Your own behalf and on Your sole responsibility, not on behalf
171
+ of any other Contributor, and only if You agree to indemnify,
172
+ defend, and hold each Contributor harmless for any liability
173
+ incurred by, or claims asserted against, such Contributor by reason
174
+ of your accepting any such warranty or additional liability.
175
+
176
+ END OF TERMS AND CONDITIONS
177
+
178
+ Copyright 2026 Anish Laddha, Nitesh Pradhan, Gaurav Srivastava
179
+
180
+ Licensed under the Apache License, Version 2.0 (the "License");
181
+ you may not use this file except in compliance with the License.
182
+ You may obtain a copy of the License at
183
+
184
+ http://www.apache.org/licenses/LICENSE-2.0
185
+
186
+ Unless required by applicable law or agreed to in writing, software
187
+ distributed under the License is distributed on an "AS IS" BASIS,
188
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
189
+ See the License for the specific language governing permissions and
190
+ limitations under the License.
@@ -0,0 +1,9 @@
1
+ # The sdist carries the files setup.py copies into whowins/_bundle, so a wheel built from it has them.
2
+ recursive-include data/scenarios *.json
3
+ include data/trials/*.json data/load/*.json data/realism/*.json data/splits/*.json
4
+ include configs/*.json configs/runs/*.json configs/vllm/*.yaml configs/chat-templates/*.jinja
5
+ include CITATION.cff
6
+ # The tests stay in the repository, with the fixtures and tools they need.
7
+ prune tests
8
+ exclude README.md
9
+ global-exclude __pycache__ *.py[cod] .DS_Store
whowins-0.1.0/PKG-INFO ADDED
@@ -0,0 +1,119 @@
1
+ Metadata-Version: 2.4
2
+ Name: whowins
3
+ Version: 0.1.0
4
+ Summary: Position, Role or Label? A Controlled Study of Instruction Conflicts in LLM Agents
5
+ Author-email: Anish Laddha <anshladdha15@gmail.com>, Nitesh Pradhan <nitesh.pradhan@lnmiit.ac.in>, Gaurav Srivastava <gks@vt.edu>
6
+ Maintainer-email: Anish Laddha <anshladdha15@gmail.com>
7
+ License-Expression: Apache-2.0
8
+ Project-URL: Homepage, https://anishh15.github.io/WhoWins/
9
+ Project-URL: Repository, https://github.com/anishh15/WhoWins
10
+ Project-URL: Issues, https://github.com/anishh15/WhoWins/issues
11
+ Keywords: llm,agents,instruction hierarchy,instruction conflicts,benchmark
12
+ Classifier: Development Status :: 3 - Alpha
13
+ Classifier: Intended Audience :: Science/Research
14
+ Classifier: Intended Audience :: Developers
15
+ Classifier: Operating System :: OS Independent
16
+ Classifier: Programming Language :: Python :: 3
17
+ Classifier: Programming Language :: Python :: 3 :: Only
18
+ Classifier: Programming Language :: Python :: 3.10
19
+ Classifier: Programming Language :: Python :: 3.11
20
+ Classifier: Programming Language :: Python :: 3.12
21
+ Classifier: Programming Language :: Python :: 3.13
22
+ Classifier: Programming Language :: Python :: 3.14
23
+ Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
24
+ Requires-Python: >=3.10
25
+ Description-Content-Type: text/markdown
26
+ License-File: LICENSE
27
+ Provides-Extra: dev
28
+ Requires-Dist: pytest; extra == "dev"
29
+ Requires-Dist: PyYAML; extra == "dev"
30
+ Requires-Dist: Jinja2; extra == "dev"
31
+ Provides-Extra: download
32
+ Requires-Dist: huggingface_hub; extra == "download"
33
+ Provides-Extra: tokenizers
34
+ Requires-Dist: tokenizers; extra == "tokenizers"
35
+ Dynamic: license-file
36
+
37
+ # WhoWins
38
+
39
+ [![WhoWins: position, role or label? Five seats send five casings of one key, and the tool result wins](https://anishh15.github.io/WhoWins/og.png)](https://anishh15.github.io/WhoWins/)
40
+
41
+ [![PyPI](https://img.shields.io/pypi/v/whowins?style=flat-square&color=C9786E)](https://pypi.org/project/whowins/)
42
+ [![Python 3.10+](https://img.shields.io/badge/python-3.10%2B-6F87A6?style=flat-square)](https://www.python.org/downloads/)
43
+ [![License: Apache-2.0](https://img.shields.io/badge/license-Apache_2.0-7A3B34?style=flat-square)](https://github.com/anishh15/WhoWins/blob/main/LICENSE)
44
+ [![Website](https://img.shields.io/badge/website-WhoWins-C9786E?style=flat-square)](https://anishh15.github.io/WhoWins/)
45
+
46
+ *Position, Role or Label? A Controlled Study of Instruction Conflicts in LLM Agents*
47
+
48
+ An LLM agent takes instructions from the operator's system prompt, the repository's configuration,
49
+ the user, tool results and its own saved notes, and these often disagree. A source sets three things
50
+ at once: the instruction's position, the chat role that carries it and the label that names it.
51
+ WhoWins keeps each instruction's text fixed and changes only how it is delivered. It has
52
+ 1,080 scenarios in five task types, each with up to
53
+ five instructions that no reply can satisfy together, sent from
54
+ five seats, and seven experiments that each change one thing.
55
+ Code scores which instruction a reply followed; no LLM judge is used. The paper runs it on
56
+ 14 open-weight models, 1,887,356 scored replies in all.
57
+
58
+ ## Install
59
+
60
+ ```bash
61
+ pip install whowins
62
+ ```
63
+
64
+ Python 3.10 or newer; nothing outside the standard library. The package carries the scenarios and
65
+ the run configs, so it works without a checkout. The tests, the GPU scripts and the docs are in the
66
+ [repository](https://github.com/anishh15/WhoWins).
67
+
68
+ ## Quick start
69
+
70
+ ```bash
71
+ python -m whowins.validation.validate_collection # counts, family rules, the worked trials
72
+ python -m whowins.checkers.coverage # every scenario has a working checker
73
+ python -m whowins.splits.check # the frozen held-out lists still match the data
74
+ python -m whowins.runner.design main-ladder-qwen3-8b-nothink # what a Ladder run sends
75
+ ```
76
+
77
+ Build the paper's Box 1, a trial in which every seat asks for a different casing, and score a reply:
78
+
79
+ ```python
80
+ from whowins.checkers import Response, ToolCall, check
81
+ from whowins.runner.design import trials_for_run
82
+ from whowins.runner.rendering import render_dicts
83
+ from whowins.runner.serving import load_run
84
+ from whowins.runner.trials import load_scenarios
85
+
86
+ scenarios = load_scenarios() # every base scenario, by id
87
+ run = load_run("main-ladder-qwen3-8b-nothink") # a run config the package carries
88
+ trials = trials_for_run(run, scenarios) # every trial this run sends, in order
89
+ box1 = next(t for t in trials if t.trial_id == "fmt-0002@S1-S2-S3-S4-S5")
90
+ for m in render_dicts(box1, scenarios): # the messages the model receives
91
+ print(m["role"], repr(m["content"]), *(m[k] for k in ("tool_calls", "tool_call_id") if k in m))
92
+
93
+ reply = Response(text='{"ACCOUNT_ID": "acc_1042", "OWNER_NAME": "Ada"}')
94
+ print(check(scenarios["fmt-0002"], reply).winning_option) # o5: the tool result's option
95
+ print(check(scenarios["tool-0001"], Response(tool_calls=[ToolCall("check_types")])).winning_option) # o2
96
+ ```
97
+
98
+ Logs and run records go to `logs/` and `results/` under the working directory.
99
+
100
+ ## Links
101
+
102
+ - Website, with the findings and every result: <https://anishh15.github.io/WhoWins/>
103
+ - Repository, with the tests, the GPU scripts and the runbook: <https://github.com/anishh15/WhoWins>
104
+ - Issues: <https://github.com/anishh15/WhoWins/issues>
105
+ - Paper: coming soon
106
+
107
+ ## Citation
108
+
109
+ ```bibtex
110
+ @misc{laddha2026whowins,
111
+ title = {Position, Role or Label? A Controlled Study of Instruction Conflicts in LLM Agents},
112
+ author = {Anish Laddha and Nitesh Pradhan and Gaurav Srivastava},
113
+ year = {2026},
114
+ url = {https://github.com/anishh15/WhoWins}
115
+ }
116
+ ```
117
+
118
+ Apache License 2.0. Contact: [anshladdha15@gmail.com](mailto:anshladdha15@gmail.com),
119
+ [nitesh.pradhan@lnmiit.ac.in](mailto:nitesh.pradhan@lnmiit.ac.in), [gks@vt.edu](mailto:gks@vt.edu).
@@ -0,0 +1,83 @@
1
+ # WhoWins
2
+
3
+ [![WhoWins: position, role or label? Five seats send five casings of one key, and the tool result wins](https://anishh15.github.io/WhoWins/og.png)](https://anishh15.github.io/WhoWins/)
4
+
5
+ [![PyPI](https://img.shields.io/pypi/v/whowins?style=flat-square&color=C9786E)](https://pypi.org/project/whowins/)
6
+ [![Python 3.10+](https://img.shields.io/badge/python-3.10%2B-6F87A6?style=flat-square)](https://www.python.org/downloads/)
7
+ [![License: Apache-2.0](https://img.shields.io/badge/license-Apache_2.0-7A3B34?style=flat-square)](https://github.com/anishh15/WhoWins/blob/main/LICENSE)
8
+ [![Website](https://img.shields.io/badge/website-WhoWins-C9786E?style=flat-square)](https://anishh15.github.io/WhoWins/)
9
+
10
+ *Position, Role or Label? A Controlled Study of Instruction Conflicts in LLM Agents*
11
+
12
+ An LLM agent takes instructions from the operator's system prompt, the repository's configuration,
13
+ the user, tool results and its own saved notes, and these often disagree. A source sets three things
14
+ at once: the instruction's position, the chat role that carries it and the label that names it.
15
+ WhoWins keeps each instruction's text fixed and changes only how it is delivered. It has
16
+ 1,080 scenarios in five task types, each with up to
17
+ five instructions that no reply can satisfy together, sent from
18
+ five seats, and seven experiments that each change one thing.
19
+ Code scores which instruction a reply followed; no LLM judge is used. The paper runs it on
20
+ 14 open-weight models, 1,887,356 scored replies in all.
21
+
22
+ ## Install
23
+
24
+ ```bash
25
+ pip install whowins
26
+ ```
27
+
28
+ Python 3.10 or newer; nothing outside the standard library. The package carries the scenarios and
29
+ the run configs, so it works without a checkout. The tests, the GPU scripts and the docs are in the
30
+ [repository](https://github.com/anishh15/WhoWins).
31
+
32
+ ## Quick start
33
+
34
+ ```bash
35
+ python -m whowins.validation.validate_collection # counts, family rules, the worked trials
36
+ python -m whowins.checkers.coverage # every scenario has a working checker
37
+ python -m whowins.splits.check # the frozen held-out lists still match the data
38
+ python -m whowins.runner.design main-ladder-qwen3-8b-nothink # what a Ladder run sends
39
+ ```
40
+
41
+ Build the paper's Box 1, a trial in which every seat asks for a different casing, and score a reply:
42
+
43
+ ```python
44
+ from whowins.checkers import Response, ToolCall, check
45
+ from whowins.runner.design import trials_for_run
46
+ from whowins.runner.rendering import render_dicts
47
+ from whowins.runner.serving import load_run
48
+ from whowins.runner.trials import load_scenarios
49
+
50
+ scenarios = load_scenarios() # every base scenario, by id
51
+ run = load_run("main-ladder-qwen3-8b-nothink") # a run config the package carries
52
+ trials = trials_for_run(run, scenarios) # every trial this run sends, in order
53
+ box1 = next(t for t in trials if t.trial_id == "fmt-0002@S1-S2-S3-S4-S5")
54
+ for m in render_dicts(box1, scenarios): # the messages the model receives
55
+ print(m["role"], repr(m["content"]), *(m[k] for k in ("tool_calls", "tool_call_id") if k in m))
56
+
57
+ reply = Response(text='{"ACCOUNT_ID": "acc_1042", "OWNER_NAME": "Ada"}')
58
+ print(check(scenarios["fmt-0002"], reply).winning_option) # o5: the tool result's option
59
+ print(check(scenarios["tool-0001"], Response(tool_calls=[ToolCall("check_types")])).winning_option) # o2
60
+ ```
61
+
62
+ Logs and run records go to `logs/` and `results/` under the working directory.
63
+
64
+ ## Links
65
+
66
+ - Website, with the findings and every result: <https://anishh15.github.io/WhoWins/>
67
+ - Repository, with the tests, the GPU scripts and the runbook: <https://github.com/anishh15/WhoWins>
68
+ - Issues: <https://github.com/anishh15/WhoWins/issues>
69
+ - Paper: coming soon
70
+
71
+ ## Citation
72
+
73
+ ```bibtex
74
+ @misc{laddha2026whowins,
75
+ title = {Position, Role or Label? A Controlled Study of Instruction Conflicts in LLM Agents},
76
+ author = {Anish Laddha and Nitesh Pradhan and Gaurav Srivastava},
77
+ year = {2026},
78
+ url = {https://github.com/anishh15/WhoWins}
79
+ }
80
+ ```
81
+
82
+ Apache License 2.0. Contact: [anshladdha15@gmail.com](mailto:anshladdha15@gmail.com),
83
+ [nitesh.pradhan@lnmiit.ac.in](mailto:nitesh.pradhan@lnmiit.ac.in), [gks@vt.edu](mailto:gks@vt.edu).
@@ -0,0 +1,154 @@
1
+ {%- set image_count = namespace(value=0) %}
2
+ {%- set video_count = namespace(value=0) %}
3
+ {%- macro render_content(content, do_vision_count, is_system_content=false) %}
4
+ {%- if content is string %}
5
+ {{- content }}
6
+ {%- elif content is iterable and content is not mapping %}
7
+ {%- for item in content %}
8
+ {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
9
+ {%- if is_system_content %}
10
+ {{- raise_exception('System message cannot contain images.') }}
11
+ {%- endif %}
12
+ {%- if do_vision_count %}
13
+ {%- set image_count.value = image_count.value + 1 %}
14
+ {%- endif %}
15
+ {%- if add_vision_id %}
16
+ {{- 'Picture ' ~ image_count.value ~ ': ' }}
17
+ {%- endif %}
18
+ {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
19
+ {%- elif 'video' in item or item.type == 'video' %}
20
+ {%- if is_system_content %}
21
+ {{- raise_exception('System message cannot contain videos.') }}
22
+ {%- endif %}
23
+ {%- if do_vision_count %}
24
+ {%- set video_count.value = video_count.value + 1 %}
25
+ {%- endif %}
26
+ {%- if add_vision_id %}
27
+ {{- 'Video ' ~ video_count.value ~ ': ' }}
28
+ {%- endif %}
29
+ {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
30
+ {%- elif 'text' in item %}
31
+ {{- item.text }}
32
+ {%- else %}
33
+ {{- raise_exception('Unexpected item type in content.') }}
34
+ {%- endif %}
35
+ {%- endfor %}
36
+ {%- elif content is none or content is undefined %}
37
+ {{- '' }}
38
+ {%- else %}
39
+ {{- raise_exception('Unexpected content type.') }}
40
+ {%- endif %}
41
+ {%- endmacro %}
42
+ {%- if not messages %}
43
+ {{- raise_exception('No messages provided.') }}
44
+ {%- endif %}
45
+ {%- if tools and tools is iterable and tools is not mapping %}
46
+ {{- '<|im_start|>system\n' }}
47
+ {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
48
+ {%- for tool in tools %}
49
+ {{- "\n" }}
50
+ {{- tool | tojson }}
51
+ {%- endfor %}
52
+ {{- "\n</tools>" }}
53
+ {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
54
+ {%- if messages[0].role == 'system' %}
55
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
56
+ {%- if content %}
57
+ {{- '\n\n' + content }}
58
+ {%- endif %}
59
+ {%- endif %}
60
+ {{- '<|im_end|>\n' }}
61
+ {%- else %}
62
+ {%- if messages[0].role == 'system' %}
63
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
64
+ {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
65
+ {%- endif %}
66
+ {%- endif %}
67
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
68
+ {%- for message in messages[::-1] %}
69
+ {%- set index = (messages|length - 1) - loop.index0 %}
70
+ {%- if ns.multi_step_tool and message.role == "user" %}
71
+ {%- set content = render_content(message.content, false)|trim %}
72
+ {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
73
+ {%- set ns.multi_step_tool = false %}
74
+ {%- set ns.last_query_index = index %}
75
+ {%- endif %}
76
+ {%- endif %}
77
+ {%- endfor %}
78
+ {%- if ns.multi_step_tool %}
79
+ {{- raise_exception('No user query found in messages.') }}
80
+ {%- endif %}
81
+ {%- for message in messages %}
82
+ {%- set content = render_content(message.content, true)|trim %}
83
+ {%- if message.role == "system" %}
84
+ {%- if not loop.first %}
85
+ {{- raise_exception('System message must be at the beginning.') }}
86
+ {%- endif %}
87
+ {%- elif message.role == "user" %}
88
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
89
+ {%- elif message.role == "assistant" %}
90
+ {%- set reasoning_content = '' %}
91
+ {%- if message.reasoning_content is string %}
92
+ {%- set reasoning_content = message.reasoning_content %}
93
+ {%- else %}
94
+ {%- if '</think>' in content %}
95
+ {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
96
+ {%- set content = content.split('</think>')[-1].lstrip('\n') %}
97
+ {%- endif %}
98
+ {%- endif %}
99
+ {%- set reasoning_content = reasoning_content|trim %}
100
+ {%- if loop.index0 > ns.last_query_index %}
101
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
102
+ {%- else %}
103
+ {{- '<|im_start|>' + message.role + '\n' + content }}
104
+ {%- endif %}
105
+ {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
106
+ {%- for tool_call in message.tool_calls %}
107
+ {%- if tool_call.function is defined %}
108
+ {%- set tool_call = tool_call.function %}
109
+ {%- endif %}
110
+ {%- if loop.first %}
111
+ {%- if content|trim %}
112
+ {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
113
+ {%- else %}
114
+ {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
115
+ {%- endif %}
116
+ {%- else %}
117
+ {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
118
+ {%- endif %}
119
+ {%- if tool_call.arguments is defined %}
120
+ {%- for args_name, args_value in tool_call.arguments|items %}
121
+ {{- '<parameter=' + args_name + '>\n' }}
122
+ {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
123
+ {{- args_value }}
124
+ {{- '\n</parameter>\n' }}
125
+ {%- endfor %}
126
+ {%- endif %}
127
+ {{- '</function>\n</tool_call>' }}
128
+ {%- endfor %}
129
+ {%- endif %}
130
+ {{- '<|im_end|>\n' }}
131
+ {%- elif message.role == "tool" %}
132
+ {%- if loop.previtem and loop.previtem.role != "tool" %}
133
+ {{- '<|im_start|>user' }}
134
+ {%- endif %}
135
+ {{- '\n<tool_response>\n' }}
136
+ {{- content }}
137
+ {{- '\n</tool_response>' }}
138
+ {%- if not loop.last and loop.nextitem.role != "tool" %}
139
+ {{- '<|im_end|>\n' }}
140
+ {%- elif loop.last %}
141
+ {{- '<|im_end|>\n' }}
142
+ {%- endif %}
143
+ {%- else %}
144
+ {{- raise_exception('Unexpected message role.') }}
145
+ {%- endif %}
146
+ {%- endfor %}
147
+ {%- if add_generation_prompt %}
148
+ {{- '<|im_start|>assistant\n' }}
149
+ {%- if enable_thinking is defined and enable_thinking is false %}
150
+ {{- '<think>\n\n</think>\n\n' }}
151
+ {%- else %}
152
+ {{- '<think>\n' }}
153
+ {%- endif %}
154
+ {%- endif %}