Dataset Viewer
Auto-converted to Parquet Duplicate
agent
stringclasses
1 value
conversations
listlengths
2
278
date
stringlengths
32
32
episode
stringclasses
63 values
model
stringclasses
1 value
model_provider
stringclasses
1 value
result
stringclasses
9 values
run_id
stringclasses
152 values
source_dataset
stringclasses
9 values
task
stringlengths
8
25
trace_source
stringclasses
15 values
trial_name
stringlengths
17
34
terminus-2
[ { "content": "You are an AI assistant tasked with solving command-line tasks in a Linux environment. You will be given a task description and the output from previously executed commands. Your goal is to solve the task by providing batches of shell commands.\n\nFormat your response as JSON with the following st...
2026-04-16T19:51:17.711808+00:00
episode-5
hosted_vllm/glm
hosted_vllm
null
d99d40a0-de73-4c84-ad52-67f36ae80114
superuser
superuser-007279_copy0003
main
superuser-007279_copy0003__CXiP9FM
terminus-2
[ { "content": "You are an AI assistant tasked with solving command-line tasks in a Linux environment. You will be given a task description and the output from previously executed commands. Your goal is to solve the task by providing batches of shell commands.\n\nFormat your response as JSON with the following st...
2026-04-26T17:49:33.248846+00:00
episode-16
hosted_vllm/glm
hosted_vllm
AgentTimeoutError
9fbae20e-96dd-4e16-9615-4b7f9fa2e5a1
r2egym
r2egym-4250_src1_copy3
main
r2egym-4250_src1_copy3__FaFkt9W
terminus-2
[ { "content": "You are an AI assistant tasked with solving command-line tasks in a Linux environment. You will be given a task description and the output from previously executed commands. Your goal is to solve the task by providing batches of shell commands.\n\nFormat your response as JSON with the following st...
2026-04-26T11:57:44.611481+00:00
episode-55
hosted_vllm/glm
hosted_vllm
null
142680b4-dcc9-4863-b369-2a5fb5fc62f3
tor
tor-1437_src2
main
tor-1437_src2__HLMGtFu
terminus-2
[{"content":"You are an AI assistant tasked with solving command-line tasks in a Linux environment. (...TRUNCATED)
2026-04-26T18:14:20.001930+00:00
episode-10
hosted_vllm/glm
hosted_vllm
AgentTimeoutError
ac1d2896-55eb-4348-9d61-ab08013501e5
r2egym
r2egym-0844_src1_copy2
main
r2egym-0844_src1_copy2__ARnXrgw
terminus-2
[{"content":"You are an AI assistant tasked with solving command-line tasks in a Linux environment. (...TRUNCATED)
2026-04-18T03:35:29.877549+00:00
episode-7
hosted_vllm/glm
hosted_vllm
null
b655f6d2-a2ef-4340-a508-044b4adf8d41
swesmith
swesmith-26319_copy0000
main
swesmith-26319_copy0000__XaHzTBL
terminus-2
[{"content":"You are an AI assistant tasked with solving command-line tasks in a Linux environment. (...TRUNCATED)
2026-04-01T05:32:04.070206+00:00
episode-18
hosted_vllm/glm
hosted_vllm
AgentTimeoutError
1b2d6488-c8f7-4f85-ba55-2bb34c0a29d8
swegym
swegym-1471
main
swegym-1471__t55n7HH
terminus-2
[{"content":"You are an AI assistant tasked with solving command-line tasks in a Linux environment. (...TRUNCATED)
2026-04-28T01:36:05.968514+00:00
episode-17
hosted_vllm/glm
hosted_vllm
AgentTimeoutError
04ebc312-4318-438d-9903-1bd03425a51d
r2egym
r2egym-2542_src1_copy1
main
r2egym-2542_src1_copy1__hnJTAAM
terminus-2
[{"content":"You are an AI assistant tasked with solving command-line tasks in a Linux environment. (...TRUNCATED)
2026-04-19T02:03:21.144029+00:00
episode-14
hosted_vllm/glm
hosted_vllm
AgentTimeoutError
d73854f4-d271-4c79-a154-0fa2523ca37a
swesmith
swesmith-12410_copy0002
main
swesmith-12410_copy0002__r5gHgKR
terminus-2
[{"content":"You are an AI assistant tasked with solving command-line tasks in a Linux environment. (...TRUNCATED)
2026-04-28T06:33:14.786835+00:00
episode-16
hosted_vllm/glm
hosted_vllm
AgentTimeoutError
cf50e27a-6720-4bc4-9c2a-994d427f292e
scaffold
scaffold-0957
main
scaffold-0957__ogPF8Au
terminus-2
[{"content":"You are an AI assistant tasked with solving command-line tasks in a Linux environment. (...TRUNCATED)
2026-04-26T12:22:19.973811+00:00
episode-5
hosted_vllm/glm
hosted_vllm
null
91baddd0-bfc9-47cb-bd67-98cae502c3f2
tor
tor-0293_src3_copy2
main
tor-0293_src3_copy2__vx3XXXq
End of preview. Expand in Data Studio

DCAgent2/g1_top8_3160_decontam_glm47_traces

3,160 rows. Stratified, nested subsample of DCAgent2/g1_top8_100k_decontam_glm47_traces. Subset proportions preserved within ±0.05pp.

This dataset is a subset of the 100,110-row DCAgent2/g1_top8_100k_decontam_glm47_traces (which itself replaces the 2,442 SWE-bench Verified-derived rows in DCAgent/g1_min_episodes_top8_100k_glm47_traces with non-Verified issue traces).

Decontamination

metric value
Rows from SWE-bench Verified 0
Unique Verified instance_ids hit 0 / 500

Nesting

This sample is a subset of every larger sample in the cascade: 1k ⊂ 3.16k ⊂ 10k ⊂ 31.6k ⊂ 100k (parent).

Constructed by stratified sampling: each subset (swesmith, r2egym, swegym, issue, scaffold, tezos, superuser, tor) was shuffled with seed=42, then the first proportional N indices were taken at each size. Larger samples are strict supersets of smaller ones.

Subset breakdown

The same proportions as the 100,110-row parent (within ±0.05pp):

subset %
swesmith 16.25%
r2egym 15.40%
swegym 14.96%
issue 13.77%
scaffold 13.64%
tezos 9.85%
superuser 9.83%
tor 6.29%
Downloads last month
18