ngc1024/tokenledger-forecast
Tabular Regression • 1.44k • Updated • 3
Datasets:
thread_id string | turn int64 | features list | target_output float64 | target_burst int64 |
|---|---|---|---|---|
5316b20a1803 | 0 | [
0,
3.0333,
0,
0,
1,
0.1517
] | 159 | 0 |
5316b20a1803 | 1 | [
0.02,
3.1916,
0.159,
1.6094,
0.8737,
0.1596
] | 91 | 0 |
5316b20a1803 | 2 | [
0.04,
3.3008,
0.091,
1.3863,
0.9066,
0.165
] | 95 | 0 |
5316b20a1803 | 3 | [
0.06,
3.3712,
0.095,
1.6094,
0.9237,
0.1686
] | 360 | 0 |
5316b20a1803 | 4 | [
0.08,
3.7654,
0.36,
1.7918,
0.9183,
0.1883
] | 74 | 0 |
5316b20a1803 | 5 | [
0.1,
3.8771,
0.074,
1.3863,
0.8955,
0.1939
] | 127 | 0 |
5316b20a1803 | 6 | [
0.12,
4.0517,
0.127,
1.6094,
0.9184,
0.2026
] | 129 | 0 |
5316b20a1803 | 7 | [
0.14,
4.1479,
0.129,
1.7918,
0.9311,
0.2074
] | 1,299 | 0 |
5316b20a1803 | 8 | [
0.16,
4.3937,
1.299,
2.4849,
0.9275,
0.2197
] | 1,194 | 0 |
5316b20a1803 | 9 | [
0.18,
4.5398,
1.194,
2.4849,
0.9441,
0.227
] | 423 | 0 |
5316b20a1803 | 10 | [
0.2,
4.5948,
0.423,
1.9459,
0.9422,
0.2297
] | 298 | 0 |
5316b20a1803 | 11 | [
0.22,
4.6417,
0.298,
1.6094,
0.9435,
0.2321
] | 237 | 0 |
5316b20a1803 | 12 | [
0.24,
4.704,
0.237,
2.9957,
0.9458,
0.2352
] | 704 | 0 |
5316b20a1803 | 13 | [
0.26,
4.7873,
0.704,
2.3026,
0.9432,
0.2394
] | 187 | 0 |
5316b20a1803 | 14 | [
0.28,
4.8731,
0.187,
1.6094,
0.9566,
0.2437
] | 855 | 0 |
5316b20a1803 | 15 | [
0.3,
5.0317,
0.855,
2.3026,
0.9571,
0.2516
] | 1,102 | 0 |
5316b20a1803 | 16 | [
0.32,
5.3213,
1.102,
2.3979,
0.9551,
0.2661
] | 84 | 0 |
5316b20a1803 | 17 | [
0.34,
5.5121,
0.084,
1.6094,
0.9457,
0.2756
] | 237 | 0 |
5316b20a1803 | 18 | [
0.36,
5.6387,
0.237,
1.9459,
0.9284,
0.2819
] | 1,035 | 0 |
5316b20a1803 | 19 | [
0.38,
5.7548,
1.035,
2.3979,
0.9413,
0.2877
] | 408 | 0 |
5316b20a1803 | 20 | [
0.4,
5.8105,
0.408,
2.0794,
0.9619,
0.2905
] | 307 | 0 |
5316b20a1803 | 21 | [
0.42,
5.858,
0.307,
1.9459,
0.9552,
0.2929
] | 210 | 0 |
5316b20a1803 | 22 | [
0.44,
5.9973,
0.21,
1.6094,
0.9569,
0.2999
] | 242 | 0 |
5316b20a1803 | 23 | [
0.46,
6.0365,
0.242,
1.7918,
0.9571,
0.3018
] | 196 | 0 |
5316b20a1803 | 24 | [
0.48,
6.283,
0.196,
1.7918,
0.9596,
0.3141
] | 181 | 0 |
5316b20a1803 | 25 | [
0.5,
6.4534,
0.181,
1.7918,
0.9608,
0.3227
] | 222 | 0 |
5316b20a1803 | 26 | [
0.52,
6.5379,
0.222,
1.9459,
0.942,
0.3269
] | 153 | 0 |
5316b20a1803 | 27 | [
0.54,
6.1042,
0.153,
0,
0.9638,
0.3052
] | 1,290 | 0 |
5316b20a1803 | 28 | [
0.56,
6.2956,
1.29,
2.8332,
0.9558,
0.3148
] | 459 | 0 |
5316b20a1803 | 29 | [
0.58,
6.4263,
0.459,
2.0794,
0.9558,
0.3213
] | 336 | 0 |
5316b20a1803 | 30 | [
0.6,
6.5089,
0.336,
2.3979,
0.9617,
0.3254
] | 185 | 0 |
5316b20a1803 | 31 | [
0.62,
6.5852,
0.185,
1.6094,
0.9558,
0.3293
] | 120 | 0 |
5316b20a1803 | 32 | [
0.64,
6.664,
0.12,
1.6094,
0.9573,
0.3332
] | 2,095 | 1 |
5316b20a1803 | 33 | [
0.66,
6.9191,
2.095,
3.1355,
0.9691,
0.346
] | 156 | 0 |
5316b20a1803 | 34 | [
0.68,
6.9867,
0.156,
2.7081,
0.9631,
0.3493
] | 341 | 0 |
5316b20a1803 | 35 | [
0.7,
7.0807,
0.341,
2.4849,
0.9701,
0.354
] | 752 | 0 |
5316b20a1803 | 36 | [
0.72,
7.2233,
0.752,
2.6391,
0.9705,
0.3612
] | 797 | 0 |
5316b20a1803 | 37 | [
0.74,
7.3934,
0.797,
2.5649,
0.9599,
0.3697
] | 1,140 | 0 |
5316b20a1803 | 38 | [
0.76,
7.515,
1.14,
2.4849,
0.9678,
0.3757
] | 105 | 0 |
5316b20a1803 | 39 | [
0.78,
7.542,
0.105,
1.7918,
0.9717,
0.3771
] | 1,245 | 0 |
5316b20a1803 | 40 | [
0.8,
7.6741,
1.245,
2.7726,
0.9693,
0.3837
] | 88 | 0 |
5316b20a1803 | 41 | [
0.82,
7.8056,
0.088,
2.7081,
0.9709,
0.3903
] | 1,119 | 0 |
5316b20a1803 | 42 | [
0.84,
7.9251,
1.119,
2.3026,
0.9692,
0.3963
] | 122 | 0 |
5316b20a1803 | 43 | [
0.86,
8.0928,
0.122,
1.6094,
0.9694,
0.4046
] | 207 | 0 |
5316b20a1803 | 44 | [
0.88,
8.1255,
0.207,
3.5553,
0.9698,
0.4063
] | 648 | 0 |
5316b20a1803 | 45 | [
0.9,
8.2025,
0.648,
2.0794,
0.9708,
0.4101
] | 95 | 0 |
5316b20a1803 | 46 | [
0.92,
8.2824,
0.095,
2.7081,
0.9657,
0.4141
] | 95 | 0 |
5316b20a1803 | 47 | [
0.94,
8.5791,
0.095,
1.6094,
0.9657,
0.429
] | 914 | 0 |
5316b20a1803 | 48 | [
0.96,
8.6832,
0.914,
2.3026,
0.9655,
0.4342
] | 116 | 0 |
5316b20a1803 | 49 | [
0.98,
8.7068,
0.116,
1.3863,
0.9645,
0.4353
] | 152 | 0 |
5316b20a1803 | 50 | [
1,
8.7404,
0.152,
1.6094,
0.9738,
0.437
] | 232 | 0 |
5316b20a1803 | 51 | [
1.02,
8.792,
0.232,
1.6094,
0.9727,
0.4396
] | 121 | 0 |
5316b20a1803 | 52 | [
1.04,
8.8303,
0.121,
1.3863,
0.9709,
0.4415
] | 1,282 | 0 |
5316b20a1803 | 53 | [
1.06,
8.9777,
1.282,
2.3979,
0.9725,
0.4489
] | 94 | 0 |
5316b20a1803 | 54 | [
1.08,
9.0569,
0.094,
1.3863,
0.976,
0.4528
] | 219 | 0 |
5316b20a1803 | 55 | [
1.1,
9.1001,
0.219,
1.7918,
0.9687,
0.455
] | 122 | 0 |
5316b20a1803 | 56 | [
1.12,
9.1259,
0.122,
1.6094,
0.9728,
0.4563
] | 70 | 0 |
5316b20a1803 | 57 | [
1.14,
9.2004,
0.07,
1.3863,
0.9747,
0.46
] | 1,299 | 0 |
5316b20a1803 | 58 | [
1.16,
9.3407,
1.299,
2.3979,
0.9766,
0.467
] | 122 | 0 |
5316b20a1803 | 59 | [
1.18,
9.4144,
0.122,
1.6094,
0.9777,
0.4707
] | 836 | 0 |
5316b20a1803 | 60 | [
1.2,
9.5056,
0.836,
2.0794,
0.9783,
0.4753
] | 138 | 0 |
5316b20a1803 | 61 | [
1.22,
9.5778,
0.138,
1.6094,
0.9689,
0.4789
] | 592 | 0 |
5316b20a1803 | 62 | [
1.24,
9.647,
0.592,
2.1972,
0.9711,
0.4824
] | 3,849 | 1 |
5316b20a1803 | 63 | [
1.26,
10.0395,
3.849,
3.2958,
0.9779,
0.502
] | 3,008 | 1 |
5316b20a1803 | 64 | [
1.28,
10.3479,
3.008,
3.091,
0.961,
0.5174
] | 510 | 0 |
5316b20a1803 | 65 | [
1.3,
10.4585,
0.51,
2.0794,
0.9703,
0.5229
] | 91 | 0 |
5316b20a1803 | 66 | [
1.32,
10.4792,
0.091,
2.3979,
0.9699,
0.524
] | 128 | 0 |
5316b20a1803 | 67 | [
1.34,
10.8152,
0.128,
1.7918,
0.9785,
0.5408
] | 96 | 0 |
ade726522329 | 0 | [
0,
3.4107,
0,
0,
1,
0.1705
] | 227 | 0 |
ade726522329 | 1 | [
0.02,
3.6758,
0.227,
1.7918,
0,
0.1838
] | 203 | 0 |
ade726522329 | 2 | [
0.04,
3.9366,
0.203,
1.7918,
0.9279,
0.1968
] | 199 | 0 |
ade726522329 | 3 | [
0.06,
4.2193,
0.199,
1.6094,
0.7973,
0.211
] | 1,203 | 0 |
ade726522329 | 4 | [
0.08,
4.3795,
1.203,
2.5649,
0.933,
0.219
] | 77 | 0 |
ade726522329 | 5 | [
0.1,
4.441,
0.077,
1.3863,
0.9167,
0.222
] | 74 | 0 |
ade726522329 | 6 | [
0.12,
4.6103,
0.074,
2.8904,
0.94,
0.2305
] | 817 | 0 |
ade726522329 | 7 | [
0.14,
4.74,
0.817,
2.3979,
0.9515,
0.237
] | 75 | 0 |
ade726522329 | 8 | [
0.16,
4.8473,
0.075,
1.6094,
0.9295,
0.2424
] | 4,785 | 1 |
ade726522329 | 9 | [
0.18,
5.3334,
4.785,
3.7842,
0.9357,
0.2667
] | 1,440 | 0 |
ade726522329 | 10 | [
0.2,
5.4869,
1.44,
2.8332,
0.6891,
0.2743
] | 3,663 | 1 |
ade726522329 | 11 | [
0.22,
5.8608,
3.663,
3.8501,
0.8836,
0.293
] | 2,305 | 1 |
ade726522329 | 12 | [
0.24,
6.0989,
2.305,
3.0445,
0.9363,
0.3049
] | 1,572 | 0 |
ade726522329 | 13 | [
0.26,
6.2637,
1.572,
2.7081,
0.961,
0.3132
] | 1,266 | 0 |
ade726522329 | 14 | [
0.28,
6.3979,
1.266,
2.3979,
0.9411,
0.3199
] | 769 | 0 |
ade726522329 | 15 | [
0.3,
6.4894,
0.769,
2.3979,
0.9564,
0.3245
] | 554 | 0 |
ade726522329 | 16 | [
0.32,
12.2669,
0.554,
0,
0.9544,
0.6133
] | 2,676 | 1 |
ade726522329 | 17 | [
0.34,
12.5591,
2.676,
2.9957,
0.9808,
0.628
] | 630 | 0 |
ade726522329 | 18 | [
0.36,
12.6362,
0.63,
2.1972,
0.9767,
0.6318
] | 373 | 0 |
ade726522329 | 19 | [
0.38,
12.7013,
0.373,
1.7918,
0.9777,
0.6351
] | 329 | 0 |
ade726522329 | 20 | [
0.4,
12.7537,
0.329,
1.7918,
0.9787,
0.6377
] | 467 | 0 |
ade726522329 | 21 | [
0.42,
12.8297,
0.467,
2.0794,
0.9798,
0.6415
] | 865 | 0 |
ade726522329 | 22 | [
0.44,
12.9586,
0.865,
2.3026,
0.9781,
0.6479
] | 2,039 | 1 |
ade726522329 | 23 | [
0.46,
13.2593,
2.039,
2.9444,
0.9743,
0.663
] | 662 | 0 |
ade726522329 | 24 | [
0.48,
13.3473,
0.662,
2.1972,
0.9773,
0.6674
] | 771 | 0 |
ade726522329 | 25 | [
0.5,
13.5145,
0.771,
2.0794,
0.9781,
0.6757
] | 651 | 0 |
ade726522329 | 26 | [
0.52,
13.7307,
0.651,
2.1972,
0.9837,
0.6865
] | 98 | 0 |
ade726522329 | 27 | [
0.54,
13.8448,
0.098,
1.9459,
0.9842,
0.6922
] | 84 | 0 |
ade726522329 | 28 | [
0.56,
14.0205,
0.084,
1.6094,
0.9808,
0.701
] | 1,116 | 0 |
ade726522329 | 29 | [
0.58,
14.1766,
1.116,
2.3979,
0.98,
0.7088
] | 202 | 0 |
ade726522329 | 30 | [
0.6,
14.2424,
0.202,
1.7918,
0.9746,
0.7121
] | 222 | 0 |
ade726522329 | 31 | [
0.62,
14.2788,
0.222,
1.7918,
0.981,
0.7139
] | 197 | 0 |
End of preview. Expand in Data Studio
tokenledger-events: Coding Agent Per-Turn Observability Dataset
A structured dataset of 21,617 per-turn token events collected from long-running autonomous AI coding agent workflows.
- Curator: ngc1024
- License: MIT
- Companion Model:
ngc1024/tokenledger-forecast - Interactive Space:
ngc1024/tokenledger-demo
Dataset Summary
Modern autonomous coding agents (Claude Code, Antigravity, Cursor, Devin) run multi-turn loops lasting tens or hundreds of steps. While short conversational turns require relatively few output tokens, complex reasoning, diff drafting, and tool invocation cause extreme long-tail output bursts. Furthermore, prompt caching behavior shifts drastically based on idle duration between turns.
This dataset provides real-world telemetry for:
- Predicting per-turn output generation lengths.
- Detecting high-cost tail burst risks.
- Modeling prompt cache expiration dynamics across idle gaps.
Privacy & Safety
Zero Text Content: This dataset strictly contains numerical and structural telemetry (token counts, timestamps, idle gaps, anonymous thread hashes). No user prompts, code snippets, or model response text are stored or distributed.
Splits & Structure
To ensure zero inter-turn data leakage, the dataset is partitioned strictly by session thread, never by random row selection:
- Train (
train.jsonl): 70% of threads - Validation (
val.jsonl): 15% of threads - Test (
test.jsonl): 15% of threads
Record Schema
{
"thread_id": "a9f8b4c2e10d",
"turn": 14,
"features": [
0.28, // turn_index scaled
14.25, // prompt_tokens in thousands (142.5k context)
0.32, // prev_output_tokens in thousands
3.81, // log(1 + gap_seconds)
1.0, // prev_cache_hit_ratio
0.71 // window_fill_ratio (against 200k context)
],
"target_output": 284.0, // Ground truth output tokens
"target_burst": 0 // 1 if output > 2000 tokens, else 0
}
How to Use with Hugging Face datasets
from datasets import load_dataset
dataset = load_dataset("ngc1024/tokenledger-events")
print(dataset["train"][0])
- Downloads last month
- -
Models trained or fine-tuned on ngc1024/tokenledger-events