lynae-1219 commited on
Commit
b479602
·
verified ·
1 Parent(s): ade7ccf

Remove workload trace entries from Croissant metadata

Browse files
Files changed (1) hide show
  1. croissant.json +1 -87
croissant.json CHANGED
@@ -49,7 +49,7 @@
49
  },
50
  "@type": "sc:Dataset",
51
  "name": "AgentPerfBench",
52
- "description": "LLM inference benchmark dataset measuring serving performance (TTFT, TPOT, throughput) across 9 models, 4 GPU platforms, 2 serving engines under agentic and chat workloads. Seven configurations: trace_replay, distributional, kernels_labeled, roofline_quadrant, coding_agent_prompts, osworld_trajectories, and predictions.",
53
  "url": "https://huggingface.co/datasets/agent-perf-bench/AgentPerfBench",
54
  "license": "https://spdx.org/licenses/Apache-2.0.html",
55
  "conformsTo": "http://mlcommons.org/croissant/1.1",
@@ -91,20 +91,6 @@
91
  "contentUrl": "https://huggingface.co/datasets/agent-perf-bench/AgentPerfBench/resolve/main/kernel_profiles/roofline_quadrant.parquet",
92
  "encodingFormat": "application/x-parquet"
93
  },
94
- {
95
- "@type": "cr:FileObject",
96
- "@id": "coding-agent-prompts-parquet",
97
- "name": "workload_traces/coding_agent_prompts.parquet",
98
- "contentUrl": "https://huggingface.co/datasets/agent-perf-bench/AgentPerfBench/resolve/main/workload_traces/coding_agent_prompts.parquet",
99
- "encodingFormat": "application/x-parquet"
100
- },
101
- {
102
- "@type": "cr:FileObject",
103
- "@id": "osworld-trajectories-parquet",
104
- "name": "workload_traces/osworld_trajectories.parquet",
105
- "contentUrl": "https://huggingface.co/datasets/agent-perf-bench/AgentPerfBench/resolve/main/workload_traces/osworld_trajectories.parquet",
106
- "encodingFormat": "application/x-parquet"
107
- },
108
  {
109
  "@type": "cr:FileObject",
110
  "@id": "predictions-parquet",
@@ -349,78 +335,6 @@
349
  }
350
  ]
351
  },
352
- {
353
- "@type": "cr:RecordSet",
354
- "@id": "coding-agent-prompts",
355
- "name": "Coding Agent Prompts",
356
- "description": "System/user prompt pairs with output token counts from SWE-Bench coding agent sessions.",
357
- "field": [
358
- {
359
- "@type": "cr:Field",
360
- "@id": "coding-agent-prompts/system",
361
- "name": "system",
362
- "dataType": "sc:Text",
363
- "source": { "fileObject": { "@id": "coding-agent-prompts-parquet" }, "extract": { "column": "system" } }
364
- },
365
- {
366
- "@type": "cr:Field",
367
- "@id": "coding-agent-prompts/user",
368
- "name": "user",
369
- "dataType": "sc:Text",
370
- "source": { "fileObject": { "@id": "coding-agent-prompts-parquet" }, "extract": { "column": "user" } }
371
- },
372
- {
373
- "@type": "cr:Field",
374
- "@id": "coding-agent-prompts/osl_tokens",
375
- "name": "osl_tokens",
376
- "dataType": "sc:Integer",
377
- "source": { "fileObject": { "@id": "coding-agent-prompts-parquet" }, "extract": { "column": "osl_tokens" } }
378
- }
379
- ]
380
- },
381
- {
382
- "@type": "cr:RecordSet",
383
- "@id": "osworld-trajectories",
384
- "name": "OSWorld Trajectories",
385
- "description": "Multi-turn OSWorld sessions with per-turn action/observation data (up to 30 turns per session).",
386
- "field": [
387
- {
388
- "@type": "cr:Field",
389
- "@id": "osworld-trajectories/session_id",
390
- "name": "session_id",
391
- "dataType": "sc:Text",
392
- "source": { "fileObject": { "@id": "osworld-trajectories-parquet" }, "extract": { "column": "session_id" } }
393
- },
394
- {
395
- "@type": "cr:Field",
396
- "@id": "osworld-trajectories/source",
397
- "name": "source",
398
- "dataType": "sc:Text",
399
- "source": { "fileObject": { "@id": "osworld-trajectories-parquet" }, "extract": { "column": "source" } }
400
- },
401
- {
402
- "@type": "cr:Field",
403
- "@id": "osworld-trajectories/num_turns",
404
- "name": "num_turns",
405
- "dataType": "sc:Integer",
406
- "source": { "fileObject": { "@id": "osworld-trajectories-parquet" }, "extract": { "column": "num_turns" } }
407
- },
408
- {
409
- "@type": "cr:Field",
410
- "@id": "osworld-trajectories/turns",
411
- "name": "turns",
412
- "dataType": "sc:Text",
413
- "source": { "fileObject": { "@id": "osworld-trajectories-parquet" }, "extract": { "column": "turns" } }
414
- },
415
- {
416
- "@type": "cr:Field",
417
- "@id": "osworld-trajectories/run",
418
- "name": "run",
419
- "dataType": "sc:Text",
420
- "source": { "fileObject": { "@id": "osworld-trajectories-parquet" }, "extract": { "column": "run" } }
421
- }
422
- ]
423
- },
424
  {
425
  "@type": "cr:RecordSet",
426
  "@id": "predictions",
 
49
  },
50
  "@type": "sc:Dataset",
51
  "name": "AgentPerfBench",
52
+ "description": "LLM inference benchmark dataset measuring serving performance (TTFT, TPOT, throughput) across 9 models, 4 GPU platforms, 2 serving engines under agentic and chat workloads. Five configurations: trace_replay, distributional, kernels_labeled, roofline_quadrant, and predictions.",
53
  "url": "https://huggingface.co/datasets/agent-perf-bench/AgentPerfBench",
54
  "license": "https://spdx.org/licenses/Apache-2.0.html",
55
  "conformsTo": "http://mlcommons.org/croissant/1.1",
 
91
  "contentUrl": "https://huggingface.co/datasets/agent-perf-bench/AgentPerfBench/resolve/main/kernel_profiles/roofline_quadrant.parquet",
92
  "encodingFormat": "application/x-parquet"
93
  },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
94
  {
95
  "@type": "cr:FileObject",
96
  "@id": "predictions-parquet",
 
335
  }
336
  ]
337
  },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
338
  {
339
  "@type": "cr:RecordSet",
340
  "@id": "predictions",