ob-e6b-prefetch.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": "2026-09-24T04:09:27.692497Z",
  8. "end": "2026-09-24T04:10:49.812175Z",
  9. "decode_tps": {
  10. "median": 31.7278133583062,
  11. "min": 31.50968162055648,
  12. "max": 31.762688242567467,
  13. "stdev": 0.09303928495967567,
  14. "n": 5
  15. },
  16. "e2e_tps": {
  17. "median": 31.227463781339413,
  18. "min": 31.013296182618113,
  19. "max": 31.264856760721198,
  20. "stdev": 0.09087139615100134,
  21. "n": 5
  22. },
  23. "ttft_s": {
  24. "median": 0.28833950401167385,
  25. "min": 0.2824551460216753,
  26. "max": 0.29181034601060674,
  27. "stdev": 0.0031900284768870087,
  28. "n": 5
  29. },
  30. "completion_tokens": [
  31. 512,
  32. 512,
  33. 512,
  34. 512,
  35. 512
  36. ],
  37. "distinct_outputs": 4,
  38. "errors": []
  39. },
  40. "c1-code": {
  41. "start": "2026-09-24T04:10:49.812422Z",
  42. "end": "2026-09-24T04:11:36.713622Z",
  43. "decode_tps": {
  44. "median": 56.328264799633075,
  45. "min": 56.05789912936029,
  46. "max": 56.482173446614915,
  47. "stdev": 0.1440660855699417,
  48. "n": 5
  49. },
  50. "e2e_tps": {
  51. "median": 54.653951269495145,
  52. "min": 54.28040425096555,
  53. "max": 54.86378671864269,
  54. "stdev": 0.20063167244189237,
  55. "n": 5
  56. },
  57. "ttft_s": {
  58. "median": 0.3007574660005048,
  59. "min": 0.2851008139841724,
  60. "max": 0.31692631301120855,
  61. "stdev": 0.012005561019250247,
  62. "n": 5
  63. },
  64. "completion_tokens": [
  65. 512,
  66. 512,
  67. 512,
  68. 512,
  69. 512
  70. ],
  71. "distinct_outputs": 1,
  72. "errors": []
  73. },
  74. "c1-chat_sampled": {
  75. "start": "2026-09-24T04:11:36.713914Z",
  76. "end": "2026-09-24T04:12:55.853918Z",
  77. "decode_tps": {
  78. "median": 32.49810366046333,
  79. "min": 31.20885034262006,
  80. "max": 35.0137645967413,
  81. "stdev": 1.3364338358472703,
  82. "n": 5
  83. },
  84. "e2e_tps": {
  85. "median": 32.02722417544107,
  86. "min": 30.76258203981316,
  87. "max": 34.43333008553438,
  88. "stdev": 1.2936996694320673,
  89. "n": 5
  90. },
  91. "ttft_s": {
  92. "median": 0.2700358850124758,
  93. "min": 0.2531899609894026,
  94. "max": 0.27505363000091165,
  95. "stdev": 0.007794632833261645,
  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": "2026-09-24T04:12:55.854105Z",
  110. "end": "2026-09-24T04:14:24.800461Z",
  111. "decode_tps": {
  112. "median": 29.105996363526554,
  113. "min": 28.878276772067856,
  114. "max": 29.657962860712324,
  115. "stdev": 0.2712110667179917,
  116. "n": 5
  117. },
  118. "e2e_tps": {
  119. "median": 28.730578432969978,
  120. "min": 28.524910416525206,
  121. "max": 29.284066162400332,
  122. "stdev": 0.2653458779871224,
  123. "n": 5
  124. },
  125. "ttft_s": {
  126. "median": 0.25725750499987043,
  127. "min": 0.2541366899968125,
  128. "max": 0.2642143919947557,
  129. "stdev": 0.0037565974900274704,
  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": "2026-09-24T04:14:24.800671Z",
  144. "end": "2026-09-24T04:16:52.685283Z",
  145. "agg_tps": {
  146. "median": 69.27184425077319,
  147. "min": 68.02500626731297,
  148. "max": 70.07557816749221,
  149. "stdev": 0.7180666037391091,
  150. "n": 5
  151. },
  152. "per_stream_decode_median": 23.460183366513448,
  153. "waves": [
  154. {
  155. "wall_s": 29.564681324001867,
  156. "tokens": 2048,
  157. "agg_tps": 69.27184425077319,
  158. "per_stream_decode": [
  159. 17.62965177557744,
  160. 31.12099037428442,
  161. 17.503397528741406,
  162. 29.10053902537144
  163. ],
  164. "errors": []
  165. },
  166. {
  167. "wall_s": 29.66353382100351,
  168. "tokens": 2048,
  169. "agg_tps": 69.04099870089978,
  170. "per_stream_decode": [
  171. 17.442054974443455,
  172. 29.15994746580009,
  173. 17.64855878950813,
  174. 28.28101432047258
  175. ],
  176. "errors": []
  177. },
  178. {
  179. "wall_s": 30.10657569000614,
  180. "tokens": 2048,
  181. "agg_tps": 68.02500626731297,
  182. "per_stream_decode": [
  183. 17.1883854863466,
  184. 29.019777942391453,
  185. 17.972981967428588,
  186. 30.023506955902356
  187. ],
  188. "errors": []
  189. },
  190. {
  191. "wall_s": 29.225588336994406,
  192. "tokens": 2048,
  193. "agg_tps": 70.07557816749221,
  194. "per_stream_decode": [
  195. 17.711308709935153,
  196. 29.956639862518017,
  197. 18.639352412554313,
  198. 30.705619286999106
  199. ],
  200. "errors": []
  201. },
  202. {
  203. "wall_s": 29.32287099401583,
  204. "tokens": 2048,
  205. "agg_tps": 69.84309280008608,
  206. "per_stream_decode": [
  207. 17.662443151267883,
  208. 29.05710987890815,
  209. 17.830646387014045,
  210. 28.330242931451835
  211. ],
  212. "errors": []
  213. }
  214. ]
  215. }
  216. }
  217. }