ob-f3-serving.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": "2026-09-24T06:49:15.701488Z",
  8. "end": "2026-09-24T06:50:31.494958Z",
  9. "decode_tps": {
  10. "median": 34.274566645912266,
  11. "min": 34.15239604761976,
  12. "max": 34.61999059533001,
  13. "stdev": 0.15861288647835778,
  14. "n": 5
  15. },
  16. "e2e_tps": {
  17. "median": 33.735942983559,
  18. "min": 33.6132004510652,
  19. "max": 34.05840317296785,
  20. "stdev": 0.15363224762153205,
  21. "n": 5
  22. },
  23. "ttft_s": {
  24. "median": 0.27274253897485323,
  25. "min": 0.2676773499988485,
  26. "max": 0.28290049300994724,
  27. "stdev": 0.0052269384369985055,
  28. "n": 5
  29. },
  30. "completion_tokens": [
  31. 512,
  32. 512,
  33. 512,
  34. 512,
  35. 512
  36. ],
  37. "distinct_outputs": 5,
  38. "errors": []
  39. },
  40. "c1-code": {
  41. "start": "2026-09-24T06:50:31.495221Z",
  42. "end": "2026-09-24T06:51:13.779439Z",
  43. "decode_tps": {
  44. "median": 62.67873542107646,
  45. "min": 62.15725972313302,
  46. "max": 62.70511991309862,
  47. "stdev": 0.21729854579888436,
  48. "n": 5
  49. },
  50. "e2e_tps": {
  51. "median": 60.67628431945975,
  52. "min": 60.209015956415236,
  53. "max": 60.70866672528289,
  54. "stdev": 0.19499229615079605,
  55. "n": 5
  56. },
  57. "ttft_s": {
  58. "median": 0.28273768100189045,
  59. "min": 0.2822947039967403,
  60. "max": 0.2882653340057004,
  61. "stdev": 0.002225666041263451,
  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": "2026-09-24T06:51:13.779676Z",
  76. "end": "2026-09-24T06:52:24.485822Z",
  77. "decode_tps": {
  78. "median": 36.55333723543977,
  79. "min": 36.05265632779633,
  80. "max": 37.695192388650085,
  81. "stdev": 0.665646571176813,
  82. "n": 5
  83. },
  84. "e2e_tps": {
  85. "median": 35.9915604885383,
  86. "min": 35.43762882300372,
  87. "max": 37.06824163205108,
  88. "stdev": 0.6433389754703916,
  89. "n": 5
  90. },
  91. "ttft_s": {
  92. "median": 0.25625738099915907,
  93. "min": 0.24539374498999678,
  94. "max": 0.27420636199531145,
  95. "stdev": 0.010851814511473563,
  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": "2026-09-24T06:52:24.486012Z",
  110. "end": "2026-09-24T06:53:45.039624Z",
  111. "decode_tps": {
  112. "median": 32.161525481733015,
  113. "min": 31.94363849920592,
  114. "max": 32.64346689095394,
  115. "stdev": 0.2401527004325526,
  116. "n": 5
  117. },
  118. "e2e_tps": {
  119. "median": 31.73084971687867,
  120. "min": 31.517133489014412,
  121. "max": 32.18621313290993,
  122. "stdev": 0.23558031092045953,
  123. "n": 5
  124. },
  125. "ttft_s": {
  126. "median": 0.24820686201564968,
  127. "min": 0.24301303600077517,
  128. "max": 0.2676815149898175,
  129. "stdev": 0.008561569874572779,
  130. "n": 5
  131. },
  132. "completion_tokens": [
  133. 512,
  134. 512,
  135. 512,
  136. 512,
  137. 512
  138. ],
  139. "distinct_outputs": 5,
  140. "errors": []
  141. },
  142. "c4": {
  143. "start": "2026-09-24T06:53:45.039859Z",
  144. "end": "2026-09-24T06:55:58.384592Z",
  145. "agg_tps": {
  146. "median": 76.79886287951686,
  147. "min": 75.83009707326711,
  148. "max": 77.99304801104851,
  149. "stdev": 0.8709029832289641,
  150. "n": 5
  151. },
  152. "per_stream_decode_median": 24.95974639371618,
  153. "waves": [
  154. {
  155. "wall_s": 26.258750648004934,
  156. "tokens": 2048,
  157. "agg_tps": 77.99304801104851,
  158. "per_stream_decode": [
  159. 20.633933912670766,
  160. 31.6252234357315,
  161. 19.98553245976985,
  162. 32.13830218294725
  163. ],
  164. "errors": []
  165. },
  166. {
  167. "wall_s": 26.66706150601385,
  168. "tokens": 2048,
  169. "agg_tps": 76.79886287951686,
  170. "per_stream_decode": [
  171. 19.433704217755388,
  172. 33.64849653262657,
  173. 20.282094879618136,
  174. 29.35379844363607
  175. ],
  176. "errors": []
  177. },
  178. {
  179. "wall_s": 27.007745987997623,
  180. "tokens": 2048,
  181. "agg_tps": 75.83009707326711,
  182. "per_stream_decode": [
  183. 19.176872243065873,
  184. 33.40189541914637,
  185. 19.85652974982934,
  186. 29.285558874761595
  187. ],
  188. "errors": []
  189. },
  190. {
  191. "wall_s": 26.412572906992864,
  192. "tokens": 2048,
  193. "agg_tps": 77.53882998114817,
  194. "per_stream_decode": [
  195. 19.624028724543226,
  196. 32.13836932980318,
  197. 19.85861150345344,
  198. 32.410098611037014
  199. ],
  200. "errors": []
  201. },
  202. {
  203. "wall_s": 26.99689267500071,
  204. "tokens": 2048,
  205. "agg_tps": 75.8605823512593,
  206. "per_stream_decode": [
  207. 19.199059807942152,
  208. 31.48728255206798,
  209. 19.540177088311744,
  210. 32.265380648920534
  211. ],
  212. "errors": []
  213. }
  214. ]
  215. }
  216. }
  217. }