ob-e4-k5-cap.summary.json 4.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217
  1. {
  2. "base": "http://10.0.0.1:8888",
  3. "reps": 5,
  4. "max_tokens": 512,
  5. "phases": {
  6. "c1-prose": {
  7. "start": "2026-09-24T03:10:08.813109Z",
  8. "end": "2026-09-24T03:11:34.837253Z",
  9. "decode_tps": {
  10. "median": 30.20463244637247,
  11. "min": 30.19835447555504,
  12. "max": 30.304643603392023,
  13. "stdev": 0.04002213600933091,
  14. "n": 5
  15. },
  16. "e2e_tps": {
  17. "median": 29.746680927229757,
  18. "min": 29.720619419455584,
  19. "max": 29.83782536992468,
  20. "stdev": 0.0404729459511774,
  21. "n": 5
  22. },
  23. "ttft_s": {
  24. "median": 0.2951492109859828,
  25. "min": 0.29373712997767143,
  26. "max": 0.30916249300935306,
  27. "stdev": 0.005771756515360771,
  28. "n": 5
  29. },
  30. "completion_tokens": [
  31. 512,
  32. 512,
  33. 512,
  34. 512,
  35. 512
  36. ],
  37. "distinct_outputs": 3,
  38. "errors": []
  39. },
  40. "c1-code": {
  41. "start": "2026-09-24T03:11:34.837672Z",
  42. "end": "2026-09-24T03:12:24.922433Z",
  43. "decode_tps": {
  44. "median": 52.665543069350676,
  45. "min": 52.26552116906319,
  46. "max": 52.82127123086595,
  47. "stdev": 0.19816605286333822,
  48. "n": 5
  49. },
  50. "e2e_tps": {
  51. "median": 51.13923264592087,
  52. "min": 50.794177380080534,
  53. "max": 51.27085965033531,
  54. "stdev": 0.17083597753909913,
  55. "n": 5
  56. },
  57. "ttft_s": {
  58. "median": 0.31065292897983454,
  59. "min": 0.302895508997608,
  60. "max": 0.31204641101066954,
  61. "stdev": 0.0036150490863715865,
  62. "n": 5
  63. },
  64. "completion_tokens": [
  65. 512,
  66. 512,
  67. 512,
  68. 512,
  69. 512
  70. ],
  71. "distinct_outputs": 4,
  72. "errors": []
  73. },
  74. "c1-chat_sampled": {
  75. "start": "2026-09-24T03:12:24.922745Z",
  76. "end": "2026-09-24T03:13:47.283789Z",
  77. "decode_tps": {
  78. "median": 31.765500659857278,
  79. "min": 29.697373265373784,
  80. "max": 32.89356615436822,
  81. "stdev": 1.081278254239538,
  82. "n": 5
  83. },
  84. "e2e_tps": {
  85. "median": 31.24830403377015,
  86. "min": 29.29132630025729,
  87. "max": 32.36481422791053,
  88. "stdev": 1.037611049399527,
  89. "n": 5
  90. },
  91. "ttft_s": {
  92. "median": 0.28469609600142576,
  93. "min": 0.2726681989734061,
  94. "max": 0.29825468600029126,
  95. "stdev": 0.009478380429013247,
  96. "n": 5
  97. },
  98. "completion_tokens": [
  99. 512,
  100. 512,
  101. 512,
  102. 512,
  103. 512
  104. ],
  105. "distinct_outputs": 5,
  106. "errors": []
  107. },
  108. "c1-prose2": {
  109. "start": "2026-09-24T03:13:47.284098Z",
  110. "end": "2026-09-24T03:15:19.360703Z",
  111. "decode_tps": {
  112. "median": 28.096832911046235,
  113. "min": 27.975949300176225,
  114. "max": 28.457373954477696,
  115. "stdev": 0.16877691121756025,
  116. "n": 5
  117. },
  118. "e2e_tps": {
  119. "median": 27.727376467069007,
  120. "min": 27.602561546637236,
  121. "max": 28.093428111831376,
  122. "stdev": 0.1694924207250327,
  123. "n": 5
  124. },
  125. "ttft_s": {
  126. "median": 0.2751621979987249,
  127. "min": 0.26610920898383483,
  128. "max": 0.2833137319830712,
  129. "stdev": 0.00636920423840279,
  130. "n": 5
  131. },
  132. "completion_tokens": [
  133. 512,
  134. 512,
  135. 512,
  136. 512,
  137. 512
  138. ],
  139. "distinct_outputs": 4,
  140. "errors": []
  141. },
  142. "c4": {
  143. "start": "2026-09-24T03:15:19.360900Z",
  144. "end": "2026-09-24T03:17:48.375919Z",
  145. "agg_tps": {
  146. "median": 68.90043919413297,
  147. "min": 67.26018174004479,
  148. "max": 70.38741791972843,
  149. "stdev": 1.0250982333140672,
  150. "n": 5
  151. },
  152. "per_stream_decode_median": 22.819343452595632,
  153. "waves": [
  154. {
  155. "wall_s": 30.029465629020706,
  156. "tokens": 2048,
  157. "agg_tps": 68.19968178257548,
  158. "per_stream_decode": [
  159. 17.330983415469433,
  160. 29.674530897233026,
  161. 17.239953136699377,
  162. 27.87157712372412
  163. ],
  164. "errors": []
  165. },
  166. {
  167. "wall_s": 29.715141758002574,
  168. "tokens": 2048,
  169. "agg_tps": 68.92109136408389,
  170. "per_stream_decode": [
  171. 18.017023748034518,
  172. 29.3220631361003,
  173. 17.426993049078842,
  174. 27.621663157156743
  175. ],
  176. "errors": []
  177. },
  178. {
  179. "wall_s": 29.096109227015404,
  180. "tokens": 2048,
  181. "agg_tps": 70.38741791972843,
  182. "per_stream_decode": [
  183. 17.99971657795235,
  184. 28.320648117565863,
  185. 17.80305666705549,
  186. 29.365599983227597
  187. ],
  188. "errors": []
  189. },
  190. {
  191. "wall_s": 30.44892159101437,
  192. "tokens": 2048,
  193. "agg_tps": 67.26018174004479,
  194. "per_stream_decode": [
  195. 17.926219994823803,
  196. 27.923457787314774,
  197. 17.006494004776034,
  198. 30.995281662385445
  199. ],
  200. "errors": []
  201. },
  202. {
  203. "wall_s": 29.724048554024193,
  204. "tokens": 2048,
  205. "agg_tps": 68.90043919413297,
  206. "per_stream_decode": [
  207. 17.92787249161996,
  208. 29.444913146434587,
  209. 17.42699541032832,
  210. 27.722769512862573
  211. ],
  212. "errors": []
  213. }
  214. ]
  215. }
  216. }
  217. }