Dataset Viewer
Auto-converted to Parquet Duplicate
conversations
list
agent
string
model
string
model_provider
string
date
string
task
string
episode
string
run_id
string
trial_name
string
result
string
instruction
string
verifier_output
string
prompt_token_ids
list
completion_token_ids
list
logprobs
list
[{"content":"","role":"user"},{"content":"<think>The user wants me to implement a Kedro-like Node ab(...TRUNCATED)
opencode
vllm/9647069541629285
vllm
2026-07-11T19:48:30.499648Z
5k-adversarial-1415
episode-10
05a75ff4-f5e3-4156-bae1-e2b718da890b
5k-adversarial-1415__QQ97v8U
AgentTimeoutError
Let me search for test files that might be related to this task:
null
[[248045,8678,198,2,13455,271,2523,599,2528,310,279,2614,5568,25,271,27,15449,29,198,4754,1267,763,3(...TRUNCATED)
[[760,1156,6587,728,310,4072,264,155954,298,12317,5829,56764,440,264,8369,709,321,5829,523,13,6558,7(...TRUNCATED)
[[-0.21058163046836853,-0.0009220149368047714,-0.003677153727039695,-0.000014781810023123398,-7.1525(...TRUNCATED)
[{"content":"","role":"user"},{"content":"<think>The user wants me to implement a FileAlchemy utilit(...TRUNCATED)
opencode
vllm/9647069541629285
vllm
2026-07-12T08:04:51.265313Z
5k-adversarial-0420
episode-4
05a75ff4-f5e3-4156-bae1-e2b718da890b
5k-adversarial-0420__6SysvBx
AgentTimeoutError
"I see the /app directory is empty. I need to create the Flask-FileAlchemy implementation from scrat(...TRUNCATED)
null
[]
[]
[]
[{"content":"","role":"user"},{"content":"<think>The user wants me to implement an asynchronous emai(...TRUNCATED)
opencode
vllm/9647069541629285
vllm
2026-07-09T10:19:12.465723Z
5k-adversarial-4792
episode-18
05a75ff4-f5e3-4156-bae1-e2b718da890b
5k-adversarial-4792__MDtRPTV
0.0
(tool use)
"Creating virtual environment...\nInstalling pytest...\nRunning pytest...\n=========================(...TRUNCATED)
[[248045,8678,198,2,13455,271,2523,599,2528,310,279,2614,5568,25,271,27,15449,29,198,4754,1267,763,3(...TRUNCATED)
[[760,1156,6587,728,310,4072,449,37715,2469,1876,4540,709,13,6558,728,1118,12804,279,1970,3051,310,3(...TRUNCATED)
[[-0.06260738521814346,-0.0007328674546442926,-0.04323504492640495,-0.000067708439019043,-5.96045083(...TRUNCATED)
[{"content":"","role":"user"},{"content":"<think>The user wants me to implement a greedy coloring ut(...TRUNCATED)
opencode
vllm/9647069541629285
vllm
2026-07-12T07:29:26.116868Z
5k-adversarial-0474
episode-6
05a75ff4-f5e3-4156-bae1-e2b718da890b
5k-adversarial-0474__SCsVHkS
AgentTimeoutError
"The workspace is empty. I need to create the entire project structure from scratch. Let me implemen(...TRUNCATED)
null
[[248045,8678,198,2,13455,271,2523,599,2528,310,279,2614,5568,25,271,27,15449,29,198,4754,1267,763,3(...TRUNCATED)
[[760,1156,6587,728,310,4072,264,54102,33730,15092,440,5081,11597,23065,286,14400,13,6558,728,1118,1(...TRUNCATED)
[[-0.3022925555706024,-0.0003048905055038631,-0.03374980390071869,-0.000012874488675151952,-2.384180(...TRUNCATED)
[{"content":"","role":"user"},{"content":"<think>The user wants me to implement a BoxCoxTransformer (...TRUNCATED)
opencode
vllm/9647069541629285
vllm
2026-07-10T09:13:44.468069Z
5k-adversarial-3530
episode-3
05a75ff4-f5e3-4156-bae1-e2b718da890b
5k-adversarial-3530__2yoKYvA
AgentTimeoutError
null
null
[[248045,8678,198,2,13455,271,2523,599,2528,310,279,2614,5568,25,271,27,15449,29,198,4754,1267,763,3(...TRUNCATED)
[[760,1156,6587,728,310,4072,264,8003,34,4967,44812,523,303,4416,24038,77358,1570,6971,13,6558,728,1(...TRUNCATED)
[[-0.02488027699291706,-0.00025078444741666317,-0.11315203458070755,-0.00010322968591935933,-2.38418(...TRUNCATED)
[{"content":"","role":"user"},{"content":"<think>The user wants me to implement a Python module for (...TRUNCATED)
opencode
vllm/9647069541629285
vllm
2026-07-10T09:00:14.320169Z
5k-adversarial-3544
episode-3
05a75ff4-f5e3-4156-bae1-e2b718da890b
5k-adversarial-3544__6ZVS9UU
AgentTimeoutError
null
null
[[248045,8678,198,2,13455,271,2523,599,2528,310,279,2614,5568,25,271,27,15449,29,198,4754,1267,763,3(...TRUNCATED)
[[760,1156,6587,728,310,4072,264,12654,4536,364,53304,9672,627,53,18,17066,421,15734,264,3050,1228,1(...TRUNCATED)
[[-0.08348855376243591,-0.00018869027553591877,-0.20245057344436646,-0.00004863608774030581,-3.69547(...TRUNCATED)
[{"content":"","role":"user"},{"content":"<think>The user wants me to implement a function `ridge_al(...TRUNCATED)
opencode
vllm/9647069541629285
vllm
2026-07-11T21:45:50.269623Z
5k-adversarial-1260
episode-7
05a75ff4-f5e3-4156-bae1-e2b718da890b
5k-adversarial-1260__8jiySDK
VerifierTimeoutError
"The `ridge_alpha_tuning` function has been implemented in `/app/src/analysis/alpha_tuning.py`. The (...TRUNCATED)
null
[[248045,8678,198,2,13455,271,2523,599,2528,310,279,2614,5568,25,271,27,15449,29,198,4754,1267,763,3(...TRUNCATED)
[[760,1156,6587,728,310,4072,264,709,1510,10858,25781,516,35971,63,303,1510,3431,14,33340,14,6918,51(...TRUNCATED)
[[-0.060820817947387695,-0.0011823319364339113,-0.00333550781942904,-2.8610108984139515e-6,0.0,-0.00(...TRUNCATED)
[{"content":"","role":"user"},{"content":"<think>The user is asking me to implement a composite_desc(...TRUNCATED)
opencode
vllm/9647069541629285
vllm
2026-07-11T14:23:59.142139Z
5k-adversarial-1888
episode-24
05a75ff4-f5e3-4156-bae1-e2b718da890b
5k-adversarial-1888__VA3wnoQ
AgentTimeoutError
"I'll implement the composite_descent utility for Dagster. Let me start by exploring the codebase st(...TRUNCATED)
null
[[248045,8678,198,2,13455,271,2523,599,2528,310,279,2614,5568,25,271,27,15449,29,198,4754,1267,763,3(...TRUNCATED)
[[760,1156,369,9859,728,310,4072,264,26816,15304,1129,15092,364,74393,3541,13,1061,369,264,6150,3274(...TRUNCATED)
[[-0.6106624603271484,-0.0017527445452287793,-0.975085437297821,-0.0011978112161159515,-0.0036684819(...TRUNCATED)
[{"content":"","role":"user"},{"content":"<think>The user wants me to implement a lightweight, panda(...TRUNCATED)
opencode
vllm/9647069541629285
vllm
2026-07-12T07:16:32.001559Z
5k-adversarial-0505
episode-4
05a75ff4-f5e3-4156-bae1-e2b718da890b
5k-adversarial-0505__RzR9mQE
AgentTimeoutError
"I'll start by exploring the test files to understand the expected behavior, then implement the modu(...TRUNCATED)
null
[[248045,8678,198,2,13455,271,2523,599,2528,310,279,2614,5568,25,271,27,15449,29,198,4754,1267,763,3(...TRUNCATED)
[[760,1156,6587,728,310,4072,264,28191,11,18056,12317,6317,1498,5165,13,6558,728,1118,12804,279,1228(...TRUNCATED)
[[-0.07503840327262878,-0.000633514893706888,-0.06217329576611519,-0.00005411946040112525,-1.1920901(...TRUNCATED)
[{"content":"","role":"user"},{"content":"<think>The user wants me to implement a Django Slack slash(...TRUNCATED)
opencode
vllm/9647069541629285
vllm
2026-07-11T17:17:46.221792Z
5k-adversarial-1642
episode-10
05a75ff4-f5e3-4156-bae1-e2b718da890b
5k-adversarial-1642__oAFwxE7
0.0
"The workspace is empty. I need to create the Django project structure from scratch. Let me create t(...TRUNCATED)
"Creating virtual environment...\nInstalling pytest...\nRunning pytest...\n=========================(...TRUNCATED)
[[248045,8678,198,2,13455,271,2523,599,2528,310,279,2614,5568,25,271,27,15449,29,198,4754,1267,763,3(...TRUNCATED)
[[760,1156,6587,728,310,4072,264,50832,55327,35414,3108,72703,6795,13,6558,728,1387,1441,279,8242,25(...TRUNCATED)
[[-0.05558769777417183,-0.0009258260834030807,-0.012650024145841599,-0.0001338628790108487,-1.788134(...TRUNCATED)
End of preview. Expand in Data Studio

Agent trace dataset

Decoding the literal token IDs

The prompt_token_ids / completion_token_ids / logprobs columns are the verbatim tokens the serving engine emitted, stored PER AGENT STEP as a list-of-lists (one inner list per turn). To turn them back into text you MUST use the exact tokenizer the model was served with — a generic same-family tokenizer will decode word tokens to garbage.

Served model / tokenizer source: Qwen/Qwen3.5-122B-A10B-FP8

from transformers import AutoTokenizer
# Use the served model's own tokenizer (pull from the ref above; if it is a
# gs:// mirror, copy tokenizer.json/tokenizer_config.json/vocab.json/merges.txt
# locally first and point AutoTokenizer at that dir).
tok = AutoTokenizer.from_pretrained("Qwen/Qwen3.5-122B-A10B-FP8")
# token_ids are list-of-lists (one list per turn) — decode each turn:
text = [tok.decode(turn, skip_special_tokens=False) for turn in completion_token_ids]

Engine-reported served model name: 9647069541629285

See tokenizer_provenance.json for a machine-readable version.

Downloads last month
91