v1.0: dataset card with full metadata, precision info, coverage sparsity, croissant alignment

README.md

@@ -14,6 +14,7 @@ tags:
 size_categories:
   - 1K<n<10K
 pretty_name: AgentPerfBench
+version: "1.0"
 configs:
   - config_name: trace_replay
     data_files:
@@ -23,37 +24,172 @@ configs:
     data_files:
       - split: summary
         path: distributional/summary.parquet
+dataset_info:
+  - config_name: trace_replay
+    features:
+      - name: run_id
+        dtype: string
+      - name: model
+        dtype: string
+      - name: model_family
+        dtype: string
+      - name: hardware
+        dtype: string
+      - name: engine
+        dtype: string
+      - name: tensor_parallelism
+        dtype: int64
+      - name: profile
+        dtype: string
+      - name: concurrency
+        dtype: int64
+      - name: num_requests
+        dtype: int64
+      - name: duration_s
+        dtype: float64
+      - name: successful_requests
+        dtype: int64
+      - name: failed_requests
+        dtype: int64
+      - name: request_throughput
+        dtype: float64
+      - name: input_token_throughput
+        dtype: float64
+      - name: output_token_throughput
+        dtype: float64
+      - name: total_token_throughput
+        dtype: float64
+      - name: mean_ttft_ms
+        dtype: float64
+      - name: median_ttft_ms
+        dtype: float64
+      - name: p90_ttft_ms
+        dtype: float64
+      - name: p99_ttft_ms
+        dtype: float64
+      - name: mean_tpot_ms
+        dtype: float64
+      - name: median_tpot_ms
+        dtype: float64
+      - name: p90_tpot_ms
+        dtype: float64
+      - name: p99_tpot_ms
+        dtype: float64
+      - name: mean_itl_ms
+        dtype: float64
+      - name: median_itl_ms
+        dtype: float64
+      - name: p90_itl_ms
+        dtype: float64
+      - name: p99_itl_ms
+        dtype: float64
+      - name: mean_e2el_ms
+        dtype: float64
+      - name: median_e2el_ms
+        dtype: float64
+      - name: p90_e2el_ms
+        dtype: float64
+      - name: p99_e2el_ms
+        dtype: float64
+    splits:
+      - name: summary
+        num_rows: 3147
+        num_bytes: 694254
+  - config_name: distributional
+    features:
+      - name: run_id
+        dtype: string
+      - name: model
+        dtype: string
+      - name: model_family
+        dtype: string
+      - name: hardware
+        dtype: string
+      - name: engine
+        dtype: string
+      - name: tensor_parallelism
+        dtype: int64
+      - name: profile
+        dtype: string
+      - name: concurrency
+        dtype: int64
+      - name: num_requests
+        dtype: int64
+      - name: duration_s
+        dtype: float64
+      - name: successful_requests
+        dtype: int64
+      - name: failed_requests
+        dtype: int64
+      - name: request_throughput
+        dtype: float64
+      - name: input_token_throughput
+        dtype: float64
+      - name: output_token_throughput
+        dtype: float64
+      - name: total_token_throughput
+        dtype: float64
+      - name: mean_ttft_ms
+        dtype: float64
+      - name: median_ttft_ms
+        dtype: float64
+      - name: p90_ttft_ms
+        dtype: float64
+      - name: p99_ttft_ms
+        dtype: float64
+      - name: mean_tpot_ms
+        dtype: float64
+      - name: median_tpot_ms
+        dtype: float64
+      - name: p90_tpot_ms
+        dtype: float64
+      - name: p99_tpot_ms
+        dtype: float64
+      - name: mean_itl_ms
+        dtype: float64
+      - name: median_itl_ms
+        dtype: float64
+      - name: p90_itl_ms
+        dtype: float64
+      - name: p99_itl_ms
+        dtype: float64
+      - name: mean_e2el_ms
+        dtype: float64
+      - name: median_e2el_ms
+        dtype: float64
+      - name: p90_e2el_ms
+        dtype: float64
+      - name: p99_e2el_ms
+        dtype: float64
+    splits:
+      - name: summary
+        num_rows: 245
+        num_bytes: 70836
 ---
 
 # AgentPerfBench
 
-LLM inference benchmark measuring TTFT, TPOT, ITL, and throughput across 9 models, up to 14 GPU configurations, 2 engines, and 21 workload profiles.
+LLM inference benchmark: 3,392 runs measuring TTFT, TPOT, ITL, and throughput across 9 models, up to 14 GPU configurations, and 2 serving engines (vLLM 0.19.0, SGLang 0.5.9). All models served in BF16 except gpt-oss, which uses mxfp4 for projection weights.
 
 ## Dataset configurations
 
-Two configs with different data collection methods.
+The dataset provides two configurations. *trace_replay* replays exact input/output sequences from recorded agent sessions. *distributional* samples from statistical distributions fitted to those same workloads, trading fidelity for faster sweeps across the hardware matrix.
 
 ### trace_replay (3,147 rows)
 
-Replays exact ISL/OSL sequences from recorded agent sessions (SWE-Bench, TerminalBench, OSWorld, ShareGPT).
+Replays exact ISL/OSL sequences from recorded agent sessions (SWE-Bench, TerminalBench, OSWorld, ShareGPT). Covers 77 unique (model, hardware, engine) combinations across 17 profiles and 6 concurrency levels. The full 5-dimensional matrix is 12.2% filled; not all models run on all hardware.
 
 17 profiles: `chat-medium`, `chat-multiturn-long`, `chat-multiturn-medium`, `chat-multiturn-short`, `chat-short`, `chat-singleturn`, `coding-singleturn`, `decode-heavy`, `osworld-multiturn-long`, `osworld-multiturn-medium`, `osworld-multiturn-short`, `prefill-heavy`, `random-1k`, `swebench-multiturn-medium`, `swebench-multiturn-short`, `terminalbench-multiturn-medium`, `terminalbench-multiturn-short`
 
 ### distributional (245 rows)
 
-Samples ISL/OSL from parameterized distributions (lognormal) fitted to real workload statistics. Validated against trace_replay.
-
-8 of 9 models covered (`gpt-oss-120b` excluded). 12 of 14 hardware configs (`3090x8`, `A100-40GBx8` excluded).
+Samples ISL/OSL from lognormal distributions fitted to real workload statistics. Covers 42 unique (model, hardware, engine) combinations across 6 profiles and 7 concurrency levels (3.0% matrix fill). `gpt-oss-120b`, `3090x8`, and `A100-40GBx8` are excluded from this configuration.
 
 6 profiles: `chat-multiturn`, `chat-singleturn`, `coding-singleturn`, `osworld-multiturn`, `swebench-multiturn`, `terminalbench-multiturn`
 
-### Why two configurations?
-
-trace_replay uses exact sequences from recorded sessions; distributional samples from fitted distributions for broader coverage with shorter runs.
-
 ### Concurrency filtering
 
-The benchmark harness capped actual concurrent connections at the session pool size. Rows where declared concurrency exceeded the pool were excluded.
+The benchmark harness capped actual concurrent connections at the session pool size. Rows where declared concurrency exceeded the pool were excluded:
 
 - trace_replay: concurrency > 100 removed (session pool was 100). Remaining values: {1, 5, 10, 20, 40, 80}.
 - distributional (pre-fix): concurrency > 10 removed (session pool was 10). Post-fix data has no cap. Remaining values: {1, 5, 10, 40, 80, 200, 320}.
@@ -66,34 +202,38 @@ The benchmark harness capped actual concurrent connections at the session pool s
 
 ### Failed requests
 
-Some configurations produce request failures, typically at high concurrency where the engine hits memory or timeout limits. 30.8% of trace_replay rows and 42% of distributional rows have `failed_requests > 0`. Summary metrics (TTFT, TPOT, throughput) are computed from successful requests only. The `failed_requests` column is included for transparency.
+Some runs produce request failures, typically at high concurrency where the engine hits memory or timeout limits. 30.8% of trace_replay rows and 42% of distributional rows have `failed_requests > 0`. Summary metrics (TTFT, TPOT, throughput) are computed from successful requests only.
 
 ## Coverage
 
 ### Hardware
 
+All benchmarks collected on CUDA 12.4.
+
 | GPU | VRAM | HBM bandwidth | Peak half-precision TFLOPS |
-|-----|------|---------------|------------------|
+|-----|------|---------------|---------------------------|
 | NVIDIA H100 SXM | 80 GB | 3.35 TB/s | 989 |
 | NVIDIA A100 SXM4 | 40 GB | 1.56 TB/s | 312 |
 | NVIDIA RTX 3090 | 24 GB | 936 GB/s | 71 |
 | NVIDIA RTX 2080 Ti | 11 GB | 616 GB/s | 27 |
 
-Multi-GPU configurations: 1, 2, 4, or 8 GPUs with tensor parallelism (TP degree depends on GPU and model).
+Multi-GPU configurations: 1, 2, 4, or 8 GPUs with tensor parallelism. TP degree depends on model size and available GPUs.
 
 ### Models
 
-| Model | Family | Parameters | Architecture |
-|-------|--------|-----------|--------------|
-| Llama-3.1-8B | Llama | 8B | Dense |
-| Llama-3.1-70B | Llama | 70B | Dense |
-| Llama-3.3-70B | Llama | 70B | Dense |
-| Qwen2.5-72B | Qwen | 72B | Dense |
-| Qwen3.5-9B | Qwen | 9B | Dense |
-| Qwen3.5-27B | Qwen | 27B | Dense |
-| Mixtral-8x7B | Mixtral | 46.7B (12.9B active) | MoE |
-| gpt-oss-20b | GPT-OSS | 21B (3.6B active) | MoE |
-| gpt-oss-120b | GPT-OSS | 117B (5.1B active) | MoE |
+All models served in BF16 unless noted.
+
+| Model | Family | Parameters | Architecture | Notes |
+|-------|--------|-----------|--------------|-------|
+| Llama-3.1-8B | Llama | 8B | Dense | |
+| Llama-3.1-70B | Llama | 70B | Dense | |
+| Llama-3.3-70B | Llama | 70B | Dense | |
+| Qwen2.5-72B | Qwen | 72B | Dense | |
+| Qwen3.5-9B | Qwen | 9B | Dense | |
+| Qwen3.5-27B | Qwen | 27B | Dense | |
+| Mixtral-8x7B | Mixtral | 46.7B (12.9B active) | MoE | |
+| gpt-oss-20b | GPT-OSS | 21B (3.6B active) | MoE | mxfp4 projections |
+| gpt-oss-120b | GPT-OSS | 117B (5.1B active) | MoE | mxfp4 projections |
 
 Model names in this table match the `model` column in the parquet files.
 
@@ -145,11 +285,14 @@ ds = load_dataset("agent-perf-bench/AgentPerfBench", "trace_replay")
 - 3-request warmup before each configuration.
 - Metrics: TTFT, TPOT, ITL, E2EL, request throughput, token throughput.
 - Summary statistics: mean, median, p90, p99.
+- Collection period: March 2026 onwards.
+- CUDA 12.4 on all machines. All models served in BF16 (gpt-oss: mxfp4 projection weights).
 
 ## Future releases
 
-- Per-request and multi-turn granularity data (pending raw JSON availability).
+- Per-request and multi-turn granularity data (pending raw JSON availability from collection infrastructure).
 - Per-kernel CUDA roofline profiles (PyTorch profiler, 2-layer forward passes, batch sizes 1/4/8/32/64).
+- This is version 1.0. Updates will be tagged with semantic versions.
 
 ## Intended uses
 
@@ -159,10 +302,11 @@ ds = load_dataset("agent-perf-bench/AgentPerfBench", "trace_replay")
 
 ## Limitations
 
-- Results are specific to tested hardware and software versions (vLLM 0.19.0, SGLang 0.5.9).
+- Results are specific to tested hardware and software versions (vLLM 0.19.0, SGLang 0.5.9, CUDA 12.4).
 - Distributional profiles approximate but do not replicate production traffic patterns.
 - No consumer GPUs beyond RTX 3090; no non-NVIDIA accelerators.
 - Closed-loop concurrency only; no open-loop (Poisson) arrivals.
+- The model-hardware-concurrency matrix is sparse (12.2% fill for trace_replay, 3.0% for distributional). Not all model-hardware combinations are represented.
 - No model quality metrics. This is a systems benchmark.
 
 ## Ethical considerations
@@ -171,7 +315,7 @@ No PII. Trace-replay profiles derive from open benchmarks (SWE-Bench MIT, Termin
 
 ## License
 
-The benchmark data is released under Apache-2.0. Source datasets retain their original licenses (see below).
+Benchmark data released under Apache-2.0. Source datasets retain their original licenses.
 
 ## Source datasets
 

croissant.json

@@ -54,7 +54,7 @@
   "license": "https://spdx.org/licenses/Apache-2.0.html",
   "conformsTo": "http://mlcommons.org/croissant/1.1",
   "datePublished": "2026-05-04",
-  "version": "2.0.0",
+  "version": "1.0",
   "citeAs": "@inproceedings{agentperfbench2026, title={AgentPerfBench: A Benchmarking and Evaluation Suite for Inference Performance of Agentic LLMs}, author={Anonymous}, booktitle={NeurIPS 2026 Evaluations and Datasets Track}, year={2026}}",
   "creator": {
     "@type": "sc:Organization",
@@ -217,7 +217,7 @@
       "@type": "cr:RecordSet",
       "@id": "distributional-summary",
       "name": "Distributional Summary",
-      "description": "One row per benchmark configuration from distributional runs (statistical sampling, validated against trace_replay).",
+      "description": "One row per benchmark configuration from distributional runs (statistical sampling from lognormal fits to recorded workload statistics).",
       "field": [
         {
           "@type": "cr:Field",
@@ -366,6 +366,10 @@
     {
       "@id": "https://github.com/xlang-ai/OSWorld",
       "name": "OSWorld"
+    },
+    {
+      "@id": "https://huggingface.co/datasets/Aeala/ShareGPT_Vicuna_unfiltered",
+      "name": "ShareGPT"
     }
   ],
   "prov:wasGeneratedBy": {