ob-f2-final-reuse.summary.json 4.4 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": "2026-09-24T06:13:07.456951Z",
  8. "end": "2026-09-24T06:14:23.613837Z",
  9. "decode_tps": {
  10. "median": 34.14652405666999,
  11. "min": 33.85988724492192,
  12. "max": 34.52475906528059,
  13. "stdev": 0.2501525493349125,
  14. "n": 5
  15. },
  16. "e2e_tps": {
  17. "median": 33.60508109290937,
  18. "min": 33.32614655198231,
  19. "max": 33.940932149480815,
  20. "stdev": 0.24448188130552415,
  21. "n": 5
  22. },
  23. "ttft_s": {
  24. "median": 0.27170867199311033,
  25. "min": 0.26574828498996794,
  26. "max": 0.2913744800025597,
  27. "stdev": 0.009468896702495939,
  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": "2026-09-24T06:14:23.614195Z",
  42. "end": "2026-09-24T06:15:06.080114Z",
  43. "decode_tps": {
  44. "median": 62.24104837155688,
  45. "min": 61.46504315335072,
  46. "max": 62.85357491360117,
  47. "stdev": 0.5375414964436913,
  48. "n": 5
  49. },
  50. "e2e_tps": {
  51. "median": 60.363319584036006,
  52. "min": 59.54126223094562,
  53. "max": 60.83859443499137,
  54. "stdev": 0.4785512899536748,
  55. "n": 5
  56. },
  57. "ttft_s": {
  58. "median": 0.2854100909898989,
  59. "min": 0.27195627099717967,
  60. "max": 0.3009387729980517,
  61. "stdev": 0.009287248547467045,
  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": "2026-09-24T06:15:06.080389Z",
  76. "end": "2026-09-24T06:16:16.192505Z",
  77. "decode_tps": {
  78. "median": 37.20330727837108,
  79. "min": 36.248058082039485,
  80. "max": 37.763801676830276,
  81. "stdev": 0.5304518189123368,
  82. "n": 5
  83. },
  84. "e2e_tps": {
  85. "median": 36.58426333185751,
  86. "min": 35.680963039867294,
  87. "max": 37.133553600396496,
  88. "stdev": 0.5076432023434098,
  89. "n": 5
  90. },
  91. "ttft_s": {
  92. "median": 0.2520815610187128,
  93. "min": 0.24551763600902632,
  94. "max": 0.2597504220029805,
  95. "stdev": 0.0048272690688732165,
  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": "2026-09-24T06:16:16.192701Z",
  110. "end": "2026-09-24T06:17:36.240490Z",
  111. "decode_tps": {
  112. "median": 32.55656103431811,
  113. "min": 31.997337897613313,
  114. "max": 32.659714023506226,
  115. "stdev": 0.24978272656672335,
  116. "n": 5
  117. },
  118. "e2e_tps": {
  119. "median": 32.10533596989721,
  120. "min": 31.56760094132209,
  121. "max": 32.19531328660531,
  122. "stdev": 0.2373185390593803,
  123. "n": 5
  124. },
  125. "ttft_s": {
  126. "median": 0.2517438460199628,
  127. "min": 0.24810012700618245,
  128. "max": 0.2567485769977793,
  129. "stdev": 0.0032167999252467367,
  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": "2026-09-24T06:17:36.240735Z",
  144. "end": "2026-09-24T06:19:52.427243Z",
  145. "agg_tps": {
  146. "median": 75.79343951270069,
  147. "min": 73.36471371141046,
  148. "max": 76.60329488892638,
  149. "stdev": 1.243935341968832,
  150. "n": 5
  151. },
  152. "per_stream_decode_median": 24.312903080764492,
  153. "waves": [
  154. {
  155. "wall_s": 26.890230299002724,
  156. "tokens": 2048,
  157. "agg_tps": 76.16148977630564,
  158. "per_stream_decode": [
  159. 19.260773789147976,
  160. 28.777115672636146,
  161. 19.78929918090842,
  162. 29.966202353618012
  163. ],
  164. "errors": []
  165. },
  166. {
  167. "wall_s": 26.735142437013565,
  168. "tokens": 2048,
  169. "agg_tps": 76.60329488892638,
  170. "per_stream_decode": [
  171. 19.37607595673969,
  172. 32.29823161107042,
  173. 19.76544782545669,
  174. 32.298844416036566
  175. ],
  176. "errors": []
  177. },
  178. {
  179. "wall_s": 27.915327360999072,
  180. "tokens": 2048,
  181. "agg_tps": 73.36471371141046,
  182. "per_stream_decode": [
  183. 18.5472707733846,
  184. 33.242378311279985,
  185. 19.070615912900447,
  186. 29.635868061268702
  187. ],
  188. "errors": []
  189. },
  190. {
  191. "wall_s": 27.623401710006874,
  192. "tokens": 2048,
  193. "agg_tps": 74.14003610055347,
  194. "per_stream_decode": [
  195. 18.74881061590412,
  196. 33.23032340382344,
  197. 18.963252856430834,
  198. 29.60866152452925
  199. ],
  200. "errors": []
  201. },
  202. {
  203. "wall_s": 27.02080830698833,
  204. "tokens": 2048,
  205. "agg_tps": 75.79343951270069,
  206. "per_stream_decode": [
  207. 19.259836858334495,
  208. 33.25823915382336,
  209. 19.848690488892842,
  210. 29.501133611971337
  211. ],
  212. "errors": []
  213. }
  214. ]
  215. }
  216. }
  217. }