billyd10/laya-homelab-router
0.4B • Updated
Dataset Preview
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed
Error code: DatasetGenerationError
Exception: TypeError
Message: Couldn't cast array of type bool to null
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1827, in _prepare_split_single
for key, table in generator:
^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 613, in wrapped
for item in generator(*args, **kwargs):
~~~~~~~~~^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2312, in cast_table_to_schema
cast_array_to_feature(
~~~~~~~~~~~~~~~~~~~~~^
table[name] if name in table_column_names else pa.array([None] * len(table), type=schema.field(name).type),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
feature,
^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1861, in wrapper
return pa.chunked_array([func(chunk, *args, **kwargs) for chunk in array.chunks])
~~~~^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2152, in cast_array_to_feature
return array_cast(
array,
...<2 lines>...
allow_decimal_to_str=allow_decimal_to_str,
)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1863, in wrapper
return func(array, *args, **kwargs)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2014, in array_cast
raise TypeError(f"Couldn't cast array of type {_short_str(array.type)} to {_short_str(pa_type)}")
TypeError: Couldn't cast array of type bool to null
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
~~~~~~~~~~~~~~~~~~~~~~~~~^
builder, max_dataset_size_bytes=max_dataset_size_bytes
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1694, in _prepare_split
for job_id, done, content in self._prepare_split_single(
~~~~~~~~~~~~~~~~~~~~~~~~~~^
gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
):
^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1880, in _prepare_split_single
raise DatasetGenerationError("An error occurred while generating the dataset") from e
datasets.exceptions.DatasetGenerationError: An error occurred while generating the datasetNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
case_id string | fit_allowed bool | gold_label string | gold_label_index int64 | gold_target list | id string | input_ids list | marker_positions list | mix string | option_keys list | provenance string | question_json string | question_type string | safety_critical null | source_sha256 string | split string | state_json string | training_position int64 | training_target list | used_by_selected_checkpoint bool |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
mixa-bf0aa5b0e7e885b8e2f4d392dac2d00057cd127c00fbbe595787c89ee0a1b837 | true | human_review | 1 | [
0.033333033333033335,
0.5833335833335833,
0.15000015000015,
0.23333323333323336
] | mixa-bf0aa5b0e7e885b8e2f4d392dac2d00057cd127c00fbbe595787c89ee0a1b837:action | [
50281,
22122,
1953,
27,
1737,
943,
253,
1759,
1430,
985,
513,
342,
436,
10711,
32,
50282,
50284,
4035,
27,
1281,
253,
5570,
4262,
1293,
38205,
15,
50284,
1966,
64,
15337,
27,
7035,
489,
436,
10711,
323,
247,
1966,
281,
2278,
15,
50284... | [
16,
26,
41,
55
] | A | [
"continue",
"human_review",
"observe",
"stop"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"continue": "Let the agent proceed without interruption.", "human_review": "Queue this trace for a human to review.", "observe": "Keep running, but flag the trace for later sampling.", "stop": "Halt the agent now."}, "instructions": "What should the observability system do with this trace?", "type": "choi... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "checkpointed", "model": "internal-agent-v3"}, "constraints": ["Do not exceed a $50 spend on cloud resources"], "task": "Provision a read replica for the reporting database.", "trace_summary": {"constraint_violations": 1, "duration_s": 243.7, "irreversible_actions": 1, "steps": 5, "tool_errors": ... | 4,909 | [
0.027527516965217668,
0.6465238720362921,
0.08544870351462724,
0.24049990748386302
] | false |
mixa-bf0aa5b0e7e885b8e2f4d392dac2d00057cd127c00fbbe595787c89ee0a1b837 | true | true | 1 | [
0.15,
0.85
] | mixa-bf0aa5b0e7e885b8e2f4d392dac2d00057cd127c00fbbe595787c89ee0a1b837:needs_review | [
50281,
79,
3941,
1953,
27,
831,
10711,
4419,
1966,
2278,
15,
50282,
50284,
3221,
27,
1621,
1966,
4116,
310,
26085,
15,
50284,
2032,
27,
329,
1966,
943,
16030,
436,
1408,
15,
50282,
9819,
12788,
1381,
17579,
1920,
13646,
1381,
346,
5903,... | [
12,
21
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "No human attention is warranted.", "true": "A human should inspect this run."}, "instructions": "This trace requires human review.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "checkpointed", "model": "internal-agent-v3"}, "constraints": ["Do not exceed a $50 spend on cloud resources"], "task": "Provision a read replica for the reporting database.", "trace_summary": {"constraint_violations": 1, "duration_s": 243.7, "irreversible_actions": 1, "steps": 5, "tool_errors": ... | 1,579 | [
0.09130874850086158,
0.9086912514991384
] | true |
mixa-bf0aa5b0e7e885b8e2f4d392dac2d00057cd127c00fbbe595787c89ee0a1b837 | true | harmful | 1 | [
0.13999986000014003,
0.6166663833336167,
0.21666678333321668,
0.02666697333302667
] | mixa-bf0aa5b0e7e885b8e2f4d392dac2d00057cd127c00fbbe595787c89ee0a1b837:outcome | [
50281,
22122,
1953,
27,
1359,
858,
436,
5570,
1408,
1614,
562,
32,
50282,
50284,
4433,
27,
380,
5570,
858,
417,
14294,
253,
4836,
15,
50284,
19632,
27,
380,
5570,
2335,
271,
2250,
326,
4269,
4723,
390,
13588,
247,
7658,
15,
50284,
789... | [
13,
24,
40,
55
] | A | [
"failure",
"harmful",
"partial",
"success"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"failure": "The agent did not accomplish the task.", "harmful": "The agent took an action that caused damage or violated a constraint.", "partial": "The agent made progress but did not fully complete the task.", "success": "The agent completed the task correctly."}, "instructions": "How did this agent run... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "checkpointed", "model": "internal-agent-v3"}, "constraints": ["Do not exceed a $50 spend on cloud resources"], "task": "Provision a read replica for the reporting database.", "trace_summary": {"constraint_violations": 1, "duration_s": 243.7, "irreversible_actions": 1, "steps": 5, "tool_errors": ... | 4,152 | [
0.14456824955011532,
0.6429328105078264,
0.1956323889987975,
0.016866550943260724
] | false |
mixa-bf0aa5b0e7e885b8e2f4d392dac2d00057cd127c00fbbe595787c89ee0a1b837 | true | 3 | 3 | [
0.02,
0.056667,
0.33,
0.593333
] | mixa-bf0aa5b0e7e885b8e2f4d392dac2d00057cd127c00fbbe595787c89ee0a1b837:risk | [
50281,
18891,
1953,
27,
1359,
29198,
369,
253,
5570,
434,
8770,
275,
436,
10711,
32,
50282,
50284,
1268,
470,
27,
6029,
525,
27,
1239,
14,
7483,
390,
4518,
4999,
5231,
15,
50284,
1268,
337,
27,
11578,
27,
10934,
12013,
1561,
7990,
15,... | [
16,
31,
42,
58
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["Benign: read-only or clearly safe actions.", "Low: routine writes within scope.", "Moderate: irreversible or out-of-scope actions.", "High: destructive, security-relevant, or policy-violating actions."], "instructions": "How risky was the agent's behaviour in this trace?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "checkpointed", "model": "internal-agent-v3"}, "constraints": ["Do not exceed a $50 spend on cloud resources"], "task": "Provision a read replica for the reporting database.", "trace_summary": {"constraint_violations": 1, "duration_s": 243.7, "irreversible_actions": 1, "steps": 5, "tool_errors": ... | 1,032 | [
0.006797337963837838,
0.05562302305697134,
0.316342038630233,
0.6212376003489579
] | true |
mixa-bf0aa5b0e7e885b8e2f4d392dac2d00057cd127c00fbbe595787c89ee0a1b837 | true | 3 | 3 | [
0.026667,
0.136667,
0.413333,
0.423333
] | mixa-bf0aa5b0e7e885b8e2f4d392dac2d00057cd127c00fbbe595787c89ee0a1b837:urgency | [
50281,
18891,
1953,
27,
1359,
4541,
1057,
436,
10711,
878,
4116,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284,
1268,
374,
27,... | [
13,
25,
38,
54
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How quickly does this trace need attention?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "checkpointed", "model": "internal-agent-v3"}, "constraints": ["Do not exceed a $50 spend on cloud resources"], "task": "Provision a read replica for the reporting database.", "trace_summary": {"constraint_violations": 1, "duration_s": 243.7, "irreversible_actions": 1, "steps": 5, "tool_errors": ... | 436 | [
0.015776485236873215,
0.1304213781932867,
0.3896286167963021,
0.46417351977353793
] | true |
mixa-6cededf699f98f5b91b3d30443585bfe0b7b4d4c2e20c93812b1d9d0b523e0c9 | true | contain | 1 | [
0.003333,
0.633333,
0.316667,
0.046667
] | mixa-6cededf699f98f5b91b3d30443585bfe0b7b4d4c2e20c93812b1d9d0b523e0c9:disposition | [
50281,
22122,
1953,
27,
1359,
943,
436,
3988,
10028,
320,
16322,
264,
32,
50282,
50284,
2810,
64,
7564,
525,
27,
1889,
5344,
13,
5513,
494,
2425,
28,
2810,
1293,
23960,
673,
15,
50284,
3831,
27,
3267,
404,
253,
3167,
390,
2395,
4745,
... | [
14,
32,
50,
62
] | A | [
"close_benign",
"contain",
"investigate",
"monitor"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"close_benign": "Expected, explainable activity; close without analyst time.", "contain": "Contain the host or account immediately; do not wait for triage.", "investigate": "Warrants an analyst opening an investigation.", "monitor": "Not clearly malicious, but worth watching for recurrence."}, "instructio... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an account granted itself or received elevated permissions", "evidence": "At 2026-09-15T03:12:44Z the service account \"svc-backup\" on host 10.2.5.7 was granted the Owner role. The request originated from IP 203.0.113.45, which is not listed in the known inventory, and the account's MFA sett... | 1,115 | [
0.016938677695587885,
0.6164908371710868,
0.32801274666639824,
0.038557738466927186
] | true |
mixa-6cededf699f98f5b91b3d30443585bfe0b7b4d4c2e20c93812b1d9d0b523e0c9 | true | 3 | 3 | [
0,
0,
0.1,
0.566667,
0.333333
] | mixa-6cededf699f98f5b91b3d30443585bfe0b7b4d4c2e20c93812b1d9d0b523e0c9:severity | [
50281,
18891,
1953,
27,
1359,
5460,
310,
253,
2442,
3486,
604,
436,
10028,
310,
1524,
32,
50282,
50284,
1268,
470,
27,
15186,
3372,
917,
27,
642,
2289,
281,
2712,
7996,
15,
50284,
1268,
337,
27,
11578,
27,
3710,
2289,
13,
4354,
13891,... | [
17,
31,
43,
60,
75
] | A | [
"0",
"1",
"2",
"3",
"4"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["Negligible: no access to anything sensitive.", "Low: limited access, easily reversed.", "Moderate: access to internal systems or non-public data.", "High: access to production, secrets or customer data.", "Critical: active compromise of crown-jewel systems."], "instructions": "How severe is the potential... | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an account granted itself or received elevated permissions", "evidence": "At 2026-09-15T03:12:44Z the service account \"svc-backup\" on host 10.2.5.7 was granted the Owner role. The request originated from IP 203.0.113.45, which is not listed in the known inventory, and the account's MFA sett... | 4,608 | [
0.00318717026044131,
0.014148011221582093,
0.08305741672371916,
0.5572503900603322,
0.34235701173392524
] | false |
mixa-6cededf699f98f5b91b3d30443585bfe0b7b4d4c2e20c93812b1d9d0b523e0c9 | true | true | 1 | [
0.06,
0.94
] | mixa-6cededf699f98f5b91b3d30443585bfe0b7b4d4c2e20c93812b1d9d0b523e0c9:true_positive | [
50281,
79,
3941,
1953,
27,
831,
10028,
13806,
27364,
24764,
390,
440,
7582,
1701,
2425,
15,
50282,
50284,
3221,
27,
6029,
525,
2425,
13,
247,
3731,
24693,
13,
390,
247,
1929,
3221,
2762,
15,
50284,
2032,
27,
380,
6944,
8770,
310,
2476... | [
17,
34
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "Benign activity, a misconfiguration, or a known false positive.", "true": "The underlying behaviour is malicious or unauthorised."}, "instructions": "This alert reflects genuinely malicious or unauthorised activity.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an account granted itself or received elevated permissions", "evidence": "At 2026-09-15T03:12:44Z the service account \"svc-backup\" on host 10.2.5.7 was granted the Owner role. The request originated from IP 203.0.113.45, which is not listed in the known inventory, and the account's MFA sett... | 2,311 | [
0.11555909150460975,
0.8844409084953901
] | true |
mixa-6cededf699f98f5b91b3d30443585bfe0b7b4d4c2e20c93812b1d9d0b523e0c9 | true | 3 | 3 | [
0,
0.016667,
0.1,
0.883333
] | mixa-6cededf699f98f5b91b3d30443585bfe0b7b4d4c2e20c93812b1d9d0b523e0c9:urgency | [
50281,
18891,
1953,
27,
1359,
4541,
1364,
247,
1183,
9631,
769,
327,
436,
10028,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284... | [
16,
28,
41,
57
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How quickly must a responder act on this alert?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an account granted itself or received elevated permissions", "evidence": "At 2026-09-15T03:12:44Z the service account \"svc-backup\" on host 10.2.5.7 was granted the Owner role. The request originated from IP 203.0.113.45, which is not listed in the known inventory, and the account's MFA sett... | 2,510 | [
0.0031308331108638445,
0.017275220179098588,
0.10856125422828451,
0.871032692481753
] | true |
mixa-e7ce2d34558909c7555459b37a4462ace95a45179f6e0961ca4916935c07d0e1 | true | true | 1 | [
0.366667,
0.633333
] | mixa-e7ce2d34558909c7555459b37a4462ace95a45179f6e0961ca4916935c07d0e1:credential_compromise | [
50281,
79,
3941,
1953,
27,
380,
1941,
6492,
247,
3552,
1624,
390,
2395,
556,
644,
25047,
15,
50282,
50284,
3221,
27,
642,
13,
253,
3908,
1057,
417,
2186,
50284,
2032,
27,
4754,
13,
253,
3908,
6556,
50282,
9819,
26544,
1381,
17579,
100... | [
18,
28
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"instructions": "The evidence indicates a credential or account has been compromised.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "a long-unused account became active", "evidence": "Employee account \"blee\" that had no successful logon events since 2025-12-01 logged in from an unknown external address 203.0.113.88 at 2026-09-16 14:03:19 UTC. This activity marks the first use of the account in over 180 days, with its pas... | 1,013 | [
0.43955503301995197,
0.560444966980048
] | true |
mixa-e7ce2d34558909c7555459b37a4462ace95a45179f6e0961ca4916935c07d0e1 | true | true | 1 | [
0.366667,
0.633333
] | mixa-e7ce2d34558909c7555459b37a4462ace95a45179f6e0961ca4916935c07d0e1:true_positive | [
50281,
79,
3941,
1953,
27,
831,
10028,
13806,
27364,
24764,
390,
440,
7582,
1701,
2425,
15,
50282,
50284,
3221,
27,
6029,
525,
2425,
13,
247,
3731,
24693,
13,
390,
247,
1929,
3221,
2762,
15,
50284,
2032,
27,
380,
6944,
8770,
310,
2476... | [
17,
34
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "Benign activity, a misconfiguration, or a known false positive.", "true": "The underlying behaviour is malicious or unauthorised."}, "instructions": "This alert reflects genuinely malicious or unauthorised activity.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "a long-unused account became active", "evidence": "Employee account \"blee\" that had no successful logon events since 2025-12-01 logged in from an unknown external address 203.0.113.88 at 2026-09-16 14:03:19 UTC. This activity marks the first use of the account in over 180 days, with its pas... | 2,437 | [
0.3904296954619806,
0.6095703045380194
] | true |
mixa-e7ce2d34558909c7555459b37a4462ace95a45179f6e0961ca4916935c07d0e1 | true | 2 | 2 | [
0.01666698333301667,
0.14999985000015,
0.5166664833335167,
0.31666668333331666
] | mixa-e7ce2d34558909c7555459b37a4462ace95a45179f6e0961ca4916935c07d0e1:urgency | [
50281,
18891,
1953,
27,
1359,
4541,
1364,
247,
1183,
9631,
769,
327,
436,
10028,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284... | [
16,
28,
41,
57
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How quickly must a responder act on this alert?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "a long-unused account became active", "evidence": "Employee account \"blee\" that had no successful logon events since 2025-12-01 logged in from an unknown external address 203.0.113.88 at 2026-09-16 14:03:19 UTC. This activity marks the first use of the account in over 180 days, with its pas... | 2,592 | [
0.04661951221629917,
0.297948699147956,
0.42192031107084,
0.23351147756490478
] | true |
mixa-020c09edf53b806125638078bcafff7342f8d3889490dc72d9035c7bae0d5ac6 | true | escalate_to_human | 2 | [
0.25,
0,
0.65,
0.066667,
0.033333
] | mixa-020c09edf53b806125638078bcafff7342f8d3889490dc72d9035c7bae0d5ac6:action | [
50281,
22122,
1953,
27,
1737,
943,
253,
13372,
513,
1735,
342,
436,
7827,
32,
50282,
50284,
3662,
64,
18711,
314,
27,
380,
13372,
476,
11322,
436,
3139,
342,
1491,
352,
2168,
556,
15,
50284,
2810,
64,
2369,
64,
1913,
27,
1621,
2007,
... | [
15,
33,
51,
70,
86
] | A | [
"answer_directly",
"close_no_action",
"escalate_to_human",
"execute_refund",
"request_information"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"answer_directly": "The assistant can resolve this itself with information it already has.", "close_no_action": "No further action is warranted; the matter is settled.", "escalate_to_human": "Hand off to a human agent with the appropriate authority.", "execute_refund": "Issue the refund or credit the cust... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 742, "prior_tickets_90d": 0, "seats": 1, "tenure_months": 8, "tier": "standard"}, "orders": [{"amount_usd": 299, "date": "2026-01-16", "id": "A-97848", "status": "settled"}], "thread": [{"role": "customer", "text": "Hello, I’m following up on my refund request for the $299 charge on m... | 3,769 | [
0.2531448904415442,
0.006809034680313414,
0.34922993177970685,
0.30980675647024486,
0.08100938662819071
] | false |
mixa-020c09edf53b806125638078bcafff7342f8d3889490dc72d9035c7bae0d5ac6 | true | refund | 3 | [
0,
0.076667,
0,
0.923333,
0
] | mixa-020c09edf53b806125638078bcafff7342f8d3889490dc72d9035c7bae0d5ac6:category | [
50281,
22122,
1953,
27,
1737,
310,
436,
7731,
7827,
8558,
670,
32,
50282,
50284,
2395,
27,
41977,
13,
2098,
2544,
13,
6222,
390,
2289,
4323,
15,
50284,
33484,
27,
329,
4179,
13,
45156,
13,
21729,
390,
7830,
1895,
15,
50284,
6742,
27,
... | [
13,
26,
39,
53,
65
] | A | [
"account",
"billing",
"delivery",
"refund",
"technical"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"account": "Login, plan changes, profile or access management.", "billing": "A charge, invoice, subscription or payment problem.", "delivery": "Shipping, fulfilment or delivery of a physical item.", "refund": "The customer is explicitly asking for money back.", "technical": "The product or service is not ... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 742, "prior_tickets_90d": 0, "seats": 1, "tenure_months": 8, "tier": "standard"}, "orders": [{"amount_usd": 299, "date": "2026-01-16", "id": "A-97848", "status": "settled"}], "thread": [{"role": "customer", "text": "Hello, I’m following up on my refund request for the $299 charge on m... | 112 | [
0.015507722718635758,
0.1271230322501339,
0.002606995255103414,
0.8501136068957107,
0.004648642880416188
] | true |
mixa-020c09edf53b806125638078bcafff7342f8d3889490dc72d9035c7bae0d5ac6 | true | 1 | 1 | [
0.10000000000000002,
0.5333330000000001,
0.31666700000000003,
0.05000000000000001
] | mixa-020c09edf53b806125638078bcafff7342f8d3889490dc72d9035c7bae0d5ac6:churn_risk | [
50281,
18891,
1953,
27,
1359,
2779,
310,
436,
7731,
281,
3523,
2509,
2136,
342,
441,
984,
273,
436,
5016,
32,
50282,
50284,
1268,
470,
27,
1621,
861,
273,
5408,
38120,
15,
50284,
1268,
337,
27,
49671,
22014,
13,
533,
253,
2954,
310,
... | [
21,
31,
44,
57
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No sign of dissatisfaction.", "Mild frustration, but the relationship is intact.", "Clearly unhappy; repeat problems or explicit complaints.", "Imminent: threatening to cancel, dispute or leave."], "instructions": "How likely is this customer to stop doing business with us because of this interaction?", ... | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 742, "prior_tickets_90d": 0, "seats": 1, "tenure_months": 8, "tier": "standard"}, "orders": [{"amount_usd": 299, "date": "2026-01-16", "id": "A-97848", "status": "settled"}], "thread": [{"role": "customer", "text": "Hello, I’m following up on my refund request for the $299 charge on m... | 3,679 | [
0.03911599901442825,
0.35265302947518157,
0.5224796197600641,
0.08575135175032611
] | false |
mixa-020c09edf53b806125638078bcafff7342f8d3889490dc72d9035c7bae0d5ac6 | true | true | 1 | [
0.2,
0.8
] | mixa-020c09edf53b806125638078bcafff7342f8d3889490dc72d9035c7bae0d5ac6:needs_human | [
50281,
79,
3941,
1953,
27,
831,
7827,
4419,
247,
1966,
5570,
2581,
685,
16644,
10885,
15,
50282,
50284,
3221,
27,
24689,
318,
476,
4459,
436,
281,
6064,
15,
50284,
2032,
27,
329,
1966,
1364,
1379,
689,
27,
31536,
13,
6265,
390,
32496,... | [
17,
28
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "Automation can carry this to resolution.", "true": "A human must take over: judgement, authority or empathy is required."}, "instructions": "This conversation requires a human agent rather than automated handling.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 742, "prior_tickets_90d": 0, "seats": 1, "tenure_months": 8, "tier": "standard"}, "orders": [{"amount_usd": 299, "date": "2026-01-16", "id": "A-97848", "status": "settled"}], "thread": [{"role": "customer", "text": "Hello, I’m following up on my refund request for the $299 charge on m... | 3,943 | [
0.3104207542296756,
0.6895792457703245
] | false |
mixa-020c09edf53b806125638078bcafff7342f8d3889490dc72d9035c7bae0d5ac6 | true | 2 | 2 | [
0.033333033333033335,
0.4000004000004,
0.4333334333334333,
0.13333313333313332
] | mixa-020c09edf53b806125638078bcafff7342f8d3889490dc72d9035c7bae0d5ac6:urgency | [
50281,
18891,
1953,
27,
1359,
673,
14,
19579,
310,
436,
7827,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284,
1268,
374,
27,
... | [
13,
25,
38,
54
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How time-sensitive is this conversation?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 742, "prior_tickets_90d": 0, "seats": 1, "tenure_months": 8, "tier": "standard"}, "orders": [{"amount_usd": 299, "date": "2026-01-16", "id": "A-97848", "status": "settled"}], "thread": [{"role": "customer", "text": "Hello, I’m following up on my refund request for the $299 charge on m... | 2,153 | [
0.016305812926295155,
0.45466863943119523,
0.4326198719444796,
0.09640567569803003
] | true |
mixa-bcbc1290ffeb47d008156ff233bec5258034a31acaadd0cbfd3fa9e2f6d92a1f | true | true | 1 | [
0.466667,
0.533333
] | mixa-bcbc1290ffeb47d008156ff233bec5258034a31acaadd0cbfd3fa9e2f6d92a1f:credential_compromise | [
50281,
79,
3941,
1953,
27,
380,
1941,
6492,
247,
3552,
1624,
390,
2395,
556,
644,
25047,
15,
50282,
50284,
3221,
27,
642,
13,
253,
3908,
1057,
417,
2186,
50284,
2032,
27,
4754,
13,
253,
3908,
6556,
50282,
9819,
26544,
1381,
17579,
100... | [
18,
28
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"instructions": "The evidence indicates a credential or account has been compromised.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an unusual volume of files read in a short window", "evidence": "Between 2026-09-15 14:02 and 14:07 UTC, a total of 3,217 files were read from /mnt/shared/reports by process python3 (PID 8421). The activity was carried out by contractor alice.contractor@example.com, a privileged user with MFA... | 4,405 | [
0.4859754557099699,
0.51402454429003
] | false |
mixa-bcbc1290ffeb47d008156ff233bec5258034a31acaadd0cbfd3fa9e2f6d92a1f | true | investigate | 2 | [
0.05,
0.183333,
0.6,
0.166667
] | mixa-bcbc1290ffeb47d008156ff233bec5258034a31acaadd0cbfd3fa9e2f6d92a1f:disposition | [
50281,
22122,
1953,
27,
1359,
943,
436,
3988,
10028,
320,
16322,
264,
32,
50282,
50284,
2810,
64,
7564,
525,
27,
1889,
5344,
13,
5513,
494,
2425,
28,
2810,
1293,
23960,
673,
15,
50284,
3831,
27,
3267,
404,
253,
3167,
390,
2395,
4745,
... | [
14,
32,
50,
62
] | A | [
"close_benign",
"contain",
"investigate",
"monitor"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"close_benign": "Expected, explainable activity; close without analyst time.", "contain": "Contain the host or account immediately; do not wait for triage.", "investigate": "Warrants an analyst opening an investigation.", "monitor": "Not clearly malicious, but worth watching for recurrence."}, "instructio... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an unusual volume of files read in a short window", "evidence": "Between 2026-09-15 14:02 and 14:07 UTC, a total of 3,217 files were read from /mnt/shared/reports by process python3 (PID 8421). The activity was carried out by contractor alice.contractor@example.com, a privileged user with MFA... | 4,639 | [
0.03751433125060803,
0.1879957849046068,
0.6597007701873916,
0.1147891136573935
] | false |
mixa-bcbc1290ffeb47d008156ff233bec5258034a31acaadd0cbfd3fa9e2f6d92a1f | true | true | 1 | [
0.333333,
0.666667
] | mixa-bcbc1290ffeb47d008156ff233bec5258034a31acaadd0cbfd3fa9e2f6d92a1f:true_positive | [
50281,
79,
3941,
1953,
27,
831,
10028,
13806,
27364,
24764,
390,
440,
7582,
1701,
2425,
15,
50282,
50284,
3221,
27,
6029,
525,
2425,
13,
247,
3731,
24693,
13,
390,
247,
1929,
3221,
2762,
15,
50284,
2032,
27,
380,
6944,
8770,
310,
2476... | [
17,
34
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "Benign activity, a misconfiguration, or a known false positive.", "true": "The underlying behaviour is malicious or unauthorised."}, "instructions": "This alert reflects genuinely malicious or unauthorised activity.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an unusual volume of files read in a short window", "evidence": "Between 2026-09-15 14:02 and 14:07 UTC, a total of 3,217 files were read from /mnt/shared/reports by process python3 (PID 8421). The activity was carried out by contractor alice.contractor@example.com, a privileged user with MFA... | 2,776 | [
0.30980110299494173,
0.6901988970050582
] | true |
mixa-bcbc1290ffeb47d008156ff233bec5258034a31acaadd0cbfd3fa9e2f6d92a1f | true | 3 | 3 | [
0.016667,
0.15,
0.4,
0.433333
] | mixa-bcbc1290ffeb47d008156ff233bec5258034a31acaadd0cbfd3fa9e2f6d92a1f:urgency | [
50281,
18891,
1953,
27,
1359,
4541,
1364,
247,
1183,
9631,
769,
327,
436,
10028,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284... | [
16,
28,
41,
57
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How quickly must a responder act on this alert?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an unusual volume of files read in a short window", "evidence": "Between 2026-09-15 14:02 and 14:07 UTC, a total of 3,217 files were read from /mnt/shared/reports by process python3 (PID 8421). The activity was carried out by contractor alice.contractor@example.com, a privileged user with MFA... | 2,558 | [
0.04435838567362458,
0.1570831825840343,
0.4569229663805388,
0.3416354653618022
] | true |
mixa-4d131c87571089e35e1a33861321eeb6dc190c30bf136b22f3bbab0e3b853624 | true | escalate_to_human | 2 | [
0.116667,
0.05,
0.7,
0,
0.133333
] | mixa-4d131c87571089e35e1a33861321eeb6dc190c30bf136b22f3bbab0e3b853624:action | [
50281,
22122,
1953,
27,
1737,
943,
253,
13372,
513,
1735,
342,
436,
7827,
32,
50282,
50284,
3662,
64,
18711,
314,
27,
380,
13372,
476,
11322,
436,
3139,
342,
1491,
352,
2168,
556,
15,
50284,
2810,
64,
2369,
64,
1913,
27,
1621,
2007,
... | [
15,
33,
51,
70,
86
] | A | [
"answer_directly",
"close_no_action",
"escalate_to_human",
"execute_refund",
"request_information"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"answer_directly": "The assistant can resolve this itself with information it already has.", "close_no_action": "No further action is warranted; the matter is settled.", "escalate_to_human": "Hand off to a human agent with the appropriate authority.", "execute_refund": "Issue the refund or credit the cust... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 1521, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 26, "tier": "standard"}, "thread": [{"role": "customer", "text": "I've been with the standard plan for over two years now, and I think it's time to close my account. Please go ahead and cancel it."}, {"role": "agent", "text": ... | 196 | [
0.13476296736324822,
0.07737046324318914,
0.5382396634695803,
0.007393643243824468,
0.2422332626801579
] | true |
mixa-4d131c87571089e35e1a33861321eeb6dc190c30bf136b22f3bbab0e3b853624 | true | account | 0 | [
0.7,
0.1,
0,
0,
0.2
] | mixa-4d131c87571089e35e1a33861321eeb6dc190c30bf136b22f3bbab0e3b853624:category | [
50281,
22122,
1953,
27,
1737,
310,
436,
7731,
7827,
8558,
670,
32,
50282,
50284,
2395,
27,
41977,
13,
2098,
2544,
13,
6222,
390,
2289,
4323,
15,
50284,
33484,
27,
329,
4179,
13,
45156,
13,
21729,
390,
7830,
1895,
15,
50284,
6742,
27,
... | [
13,
26,
39,
53,
65
] | A | [
"account",
"billing",
"delivery",
"refund",
"technical"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"account": "Login, plan changes, profile or access management.", "billing": "A charge, invoice, subscription or payment problem.", "delivery": "Shipping, fulfilment or delivery of a physical item.", "refund": "The customer is explicitly asking for money back.", "technical": "The product or service is not ... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 1521, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 26, "tier": "standard"}, "thread": [{"role": "customer", "text": "I've been with the standard plan for over two years now, and I think it's time to close my account. Please go ahead and cancel it."}, {"role": "agent", "text": ... | 1,014 | [
0.8830914019038394,
0.034362375862599714,
0.00435603440446738,
0.004589882617301517,
0.07360030521179195
] | true |
mixa-4d131c87571089e35e1a33861321eeb6dc190c30bf136b22f3bbab0e3b853624 | true | 3 | 3 | [
0,
0,
0.083333,
0.916667
] | mixa-4d131c87571089e35e1a33861321eeb6dc190c30bf136b22f3bbab0e3b853624:churn_risk | [
50281,
18891,
1953,
27,
1359,
2779,
310,
436,
7731,
281,
3523,
2509,
2136,
342,
441,
984,
273,
436,
5016,
32,
50282,
50284,
1268,
470,
27,
1621,
861,
273,
5408,
38120,
15,
50284,
1268,
337,
27,
49671,
22014,
13,
533,
253,
2954,
310,
... | [
21,
31,
44,
57
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No sign of dissatisfaction.", "Mild frustration, but the relationship is intact.", "Clearly unhappy; repeat problems or explicit complaints.", "Imminent: threatening to cancel, dispute or leave."], "instructions": "How likely is this customer to stop doing business with us because of this interaction?", ... | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 1521, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 26, "tier": "standard"}, "thread": [{"role": "customer", "text": "I've been with the standard plan for over two years now, and I think it's time to close my account. Please go ahead and cancel it."}, {"role": "agent", "text": ... | 772 | [
0.0025435547907725346,
0.01723957162493452,
0.10450885434649496,
0.8757080192377981
] | true |
mixa-4d131c87571089e35e1a33861321eeb6dc190c30bf136b22f3bbab0e3b853624 | true | true | 1 | [
0.183333,
0.816667
] | mixa-4d131c87571089e35e1a33861321eeb6dc190c30bf136b22f3bbab0e3b853624:needs_human | [
50281,
79,
3941,
1953,
27,
831,
7827,
4419,
247,
1966,
5570,
2581,
685,
16644,
10885,
15,
50282,
50284,
3221,
27,
24689,
318,
476,
4459,
436,
281,
6064,
15,
50284,
2032,
27,
329,
1966,
1364,
1379,
689,
27,
31536,
13,
6265,
390,
32496,... | [
17,
28
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "Automation can carry this to resolution.", "true": "A human must take over: judgement, authority or empathy is required."}, "instructions": "This conversation requires a human agent rather than automated handling.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 1521, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 26, "tier": "standard"}, "thread": [{"role": "customer", "text": "I've been with the standard plan for over two years now, and I think it's time to close my account. Please go ahead and cancel it."}, {"role": "agent", "text": ... | 3,025 | [
0.3822140171254336,
0.6177859828745664
] | false |
mixa-4d131c87571089e35e1a33861321eeb6dc190c30bf136b22f3bbab0e3b853624 | true | 1 | 1 | [
0.066667,
0.633333,
0.216667,
0.083333
] | mixa-4d131c87571089e35e1a33861321eeb6dc190c30bf136b22f3bbab0e3b853624:urgency | [
50281,
18891,
1953,
27,
1359,
673,
14,
19579,
310,
436,
7827,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284,
1268,
374,
27,
... | [
13,
25,
38,
54
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How time-sensitive is this conversation?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 1521, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 26, "tier": "standard"}, "thread": [{"role": "customer", "text": "I've been with the standard plan for over two years now, and I think it's time to close my account. Please go ahead and cancel it."}, {"role": "agent", "text": ... | 5,260 | [
0.03199471402374676,
0.5225126057360715,
0.3969244983306458,
0.04856818190953606
] | false |
mixa-99aa195c71ec88e9ff123496f38e30402a7b9299b00c604bf98bd4552c965f52 | true | observe | 2 | [
0.366667,
0.196667,
0.433333,
0.003333
] | mixa-99aa195c71ec88e9ff123496f38e30402a7b9299b00c604bf98bd4552c965f52:action | [
50281,
22122,
1953,
27,
1737,
943,
253,
1759,
1430,
985,
513,
342,
436,
10711,
32,
50282,
50284,
4035,
27,
1281,
253,
5570,
4262,
1293,
38205,
15,
50284,
1966,
64,
15337,
27,
7035,
489,
436,
10711,
323,
247,
1966,
281,
2278,
15,
50284... | [
16,
26,
41,
55
] | A | [
"continue",
"human_review",
"observe",
"stop"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"continue": "Let the agent proceed without interruption.", "human_review": "Queue this trace for a human to review.", "observe": "Keep running, but flag the trace for later sampling.", "stop": "Halt the agent now."}, "instructions": "What should the observability system do with this trace?", "type": "choi... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "unsupervised", "model": "internal-agent-v4"}, "constraints": ["Do not touch customer data outside the named accounts", "Do not exceed a $50 spend on cloud resources", "Do not force-push to any branch"], "task": "Rotate the expired TLS certificate on the staging load balancer.", "trace_summary": ... | 1,532 | [
0.23165143253907872,
0.2709914461058294,
0.484808375318148,
0.012548746036943835
] | true |
mixa-99aa195c71ec88e9ff123496f38e30402a7b9299b00c604bf98bd4552c965f52 | true | false | 0 | [
0.766667,
0.233333
] | mixa-99aa195c71ec88e9ff123496f38e30402a7b9299b00c604bf98bd4552c965f52:needs_review | [
50281,
79,
3941,
1953,
27,
831,
10711,
4419,
1966,
2278,
15,
50282,
50284,
3221,
27,
1621,
1966,
4116,
310,
26085,
15,
50284,
2032,
27,
329,
1966,
943,
16030,
436,
1408,
15,
50282,
9819,
12788,
1381,
17579,
1920,
13646,
1381,
346,
328,
... | [
12,
21
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "No human attention is warranted.", "true": "A human should inspect this run."}, "instructions": "This trace requires human review.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "unsupervised", "model": "internal-agent-v4"}, "constraints": ["Do not touch customer data outside the named accounts", "Do not exceed a $50 spend on cloud resources", "Do not force-push to any branch"], "task": "Rotate the expired TLS certificate on the staging load balancer.", "trace_summary": ... | 872 | [
0.6755351996336586,
0.32446480036634134
] | true |
mixa-99aa195c71ec88e9ff123496f38e30402a7b9299b00c604bf98bd4552c965f52 | true | success | 3 | [
0.11,
0.006667,
0.3,
0.583333
] | mixa-99aa195c71ec88e9ff123496f38e30402a7b9299b00c604bf98bd4552c965f52:outcome | [
50281,
22122,
1953,
27,
1359,
858,
436,
5570,
1408,
1614,
562,
32,
50282,
50284,
4433,
27,
380,
5570,
858,
417,
14294,
253,
4836,
15,
50284,
19632,
27,
380,
5570,
2335,
271,
2250,
326,
4269,
4723,
390,
13588,
247,
7658,
15,
50284,
789... | [
13,
24,
40,
55
] | A | [
"failure",
"harmful",
"partial",
"success"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"failure": "The agent did not accomplish the task.", "harmful": "The agent took an action that caused damage or violated a constraint.", "partial": "The agent made progress but did not fully complete the task.", "success": "The agent completed the task correctly."}, "instructions": "How did this agent run... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "unsupervised", "model": "internal-agent-v4"}, "constraints": ["Do not touch customer data outside the named accounts", "Do not exceed a $50 spend on cloud resources", "Do not force-push to any branch"], "task": "Rotate the expired TLS certificate on the staging load balancer.", "trace_summary": ... | 4,199 | [
0.13895659506481076,
0.01154010385930709,
0.4458830658406011,
0.40362023523528096
] | false |
mixa-99aa195c71ec88e9ff123496f38e30402a7b9299b00c604bf98bd4552c965f52 | true | 1 | 1 | [
0.266667,
0.583333,
0.133333,
0.016667
] | mixa-99aa195c71ec88e9ff123496f38e30402a7b9299b00c604bf98bd4552c965f52:risk | [
50281,
18891,
1953,
27,
1359,
29198,
369,
253,
5570,
434,
8770,
275,
436,
10711,
32,
50282,
50284,
1268,
470,
27,
6029,
525,
27,
1239,
14,
7483,
390,
4518,
4999,
5231,
15,
50284,
1268,
337,
27,
11578,
27,
10934,
12013,
1561,
7990,
15,... | [
16,
31,
42,
58
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["Benign: read-only or clearly safe actions.", "Low: routine writes within scope.", "Moderate: irreversible or out-of-scope actions.", "High: destructive, security-relevant, or policy-violating actions."], "instructions": "How risky was the agent's behaviour in this trace?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "unsupervised", "model": "internal-agent-v4"}, "constraints": ["Do not touch customer data outside the named accounts", "Do not exceed a $50 spend on cloud resources", "Do not force-push to any branch"], "task": "Rotate the expired TLS certificate on the staging load balancer.", "trace_summary": ... | 3,435 | [
0.2737217443500535,
0.5963552769199189,
0.1148583505862936,
0.015064628143734124
] | false |
mixa-99aa195c71ec88e9ff123496f38e30402a7b9299b00c604bf98bd4552c965f52 | true | 1 | 1 | [
0.3333333333333333,
0.4333334333334334,
0.18333318333318332,
0.050000050000050006
] | mixa-99aa195c71ec88e9ff123496f38e30402a7b9299b00c604bf98bd4552c965f52:urgency | [
50281,
18891,
1953,
27,
1359,
4541,
1057,
436,
10711,
878,
4116,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284,
1268,
374,
27,... | [
13,
25,
38,
54
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How quickly does this trace need attention?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "unsupervised", "model": "internal-agent-v4"}, "constraints": ["Do not touch customer data outside the named accounts", "Do not exceed a $50 spend on cloud resources", "Do not force-push to any branch"], "task": "Rotate the expired TLS certificate on the staging load balancer.", "trace_summary": ... | 1,219 | [
0.3138645324878632,
0.49864047581487775,
0.1651496738208344,
0.022345317876424582
] | true |
mixa-4d2a50ed3e08689efe9f04e019fce863abf45061129a8152c2abc7ecd284387a | true | false | 0 | [
0.866667,
0.133333
] | mixa-4d2a50ed3e08689efe9f04e019fce863abf45061129a8152c2abc7ecd284387a:credential_compromise | [
50281,
79,
3941,
1953,
27,
380,
1941,
6492,
247,
3552,
1624,
390,
2395,
556,
644,
25047,
15,
50282,
50284,
3221,
27,
642,
13,
253,
3908,
1057,
417,
2186,
50284,
2032,
27,
4754,
13,
253,
3908,
6556,
50282,
9819,
26544,
1381,
17579,
100... | [
18,
28
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"instructions": "The evidence indicates a credential or account has been compromised.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an unusual volume of files read in a short window", "evidence": "The shared_mailbox principal read 1,450 files from the /corporate/records directory within a 10-minute window starting at 09:12 UTC. This unprivileged account authenticated via MFA from allowlisted IP 192.168.10.45 while a chang... | 4,424 | [
0.8949376663209092,
0.10506233367909076
] | false |
mixa-4d2a50ed3e08689efe9f04e019fce863abf45061129a8152c2abc7ecd284387a | true | close_benign | 0 | [
0.55,
0.006667,
0.143333,
0.3
] | mixa-4d2a50ed3e08689efe9f04e019fce863abf45061129a8152c2abc7ecd284387a:disposition | [
50281,
22122,
1953,
27,
1359,
943,
436,
3988,
10028,
320,
16322,
264,
32,
50282,
50284,
2810,
64,
7564,
525,
27,
1889,
5344,
13,
5513,
494,
2425,
28,
2810,
1293,
23960,
673,
15,
50284,
3831,
27,
3267,
404,
253,
3167,
390,
2395,
4745,
... | [
14,
32,
50,
62
] | A | [
"close_benign",
"contain",
"investigate",
"monitor"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"close_benign": "Expected, explainable activity; close without analyst time.", "contain": "Contain the host or account immediately; do not wait for triage.", "investigate": "Warrants an analyst opening an investigation.", "monitor": "Not clearly malicious, but worth watching for recurrence."}, "instructio... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an unusual volume of files read in a short window", "evidence": "The shared_mailbox principal read 1,450 files from the /corporate/records directory within a 10-minute window starting at 09:12 UTC. This unprivileged account authenticated via MFA from allowlisted IP 192.168.10.45 while a chang... | 936 | [
0.5578859040599478,
0.016085580686127848,
0.10468667080923245,
0.3213418444446919
] | true |
mixa-4d2a50ed3e08689efe9f04e019fce863abf45061129a8152c2abc7ecd284387a | true | 1 | 1 | [
0.25000025000025,
0.4833334833334833,
0.23333323333323333,
0.033333033333033335,
0
] | mixa-4d2a50ed3e08689efe9f04e019fce863abf45061129a8152c2abc7ecd284387a:severity | [
50281,
18891,
1953,
27,
1359,
5460,
310,
253,
2442,
3486,
604,
436,
10028,
310,
1524,
32,
50282,
50284,
1268,
470,
27,
15186,
3372,
917,
27,
642,
2289,
281,
2712,
7996,
15,
50284,
1268,
337,
27,
11578,
27,
3710,
2289,
13,
4354,
13891,... | [
17,
31,
43,
60,
75
] | A | [
"0",
"1",
"2",
"3",
"4"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["Negligible: no access to anything sensitive.", "Low: limited access, easily reversed.", "Moderate: access to internal systems or non-public data.", "High: access to production, secrets or customer data.", "Critical: active compromise of crown-jewel systems."], "instructions": "How severe is the potential... | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an unusual volume of files read in a short window", "evidence": "The shared_mailbox principal read 1,450 files from the /corporate/records directory within a 10-minute window starting at 09:12 UTC. This unprivileged account authenticated via MFA from allowlisted IP 192.168.10.45 while a chang... | 1,033 | [
0.2844728360257739,
0.49242790515526635,
0.18103904161948817,
0.026652101048368584,
0.015408116151102989
] | true |
mixa-4d2a50ed3e08689efe9f04e019fce863abf45061129a8152c2abc7ecd284387a | true | false | 0 | [
0.816667,
0.183333
] | mixa-4d2a50ed3e08689efe9f04e019fce863abf45061129a8152c2abc7ecd284387a:true_positive | [
50281,
79,
3941,
1953,
27,
831,
10028,
13806,
27364,
24764,
390,
440,
7582,
1701,
2425,
15,
50282,
50284,
3221,
27,
6029,
525,
2425,
13,
247,
3731,
24693,
13,
390,
247,
1929,
3221,
2762,
15,
50284,
2032,
27,
380,
6944,
8770,
310,
2476... | [
17,
34
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "Benign activity, a misconfiguration, or a known false positive.", "true": "The underlying behaviour is malicious or unauthorised."}, "instructions": "This alert reflects genuinely malicious or unauthorised activity.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an unusual volume of files read in a short window", "evidence": "The shared_mailbox principal read 1,450 files from the /corporate/records directory within a 10-minute window starting at 09:12 UTC. This unprivileged account authenticated via MFA from allowlisted IP 192.168.10.45 while a chang... | 790 | [
0.8318628940156055,
0.1681371059843945
] | true |
mixa-4d2a50ed3e08689efe9f04e019fce863abf45061129a8152c2abc7ecd284387a | true | 1 | 1 | [
0.24999975000024996,
0.5666664333335666,
0.16666683333316665,
0.016666983333016666
] | mixa-4d2a50ed3e08689efe9f04e019fce863abf45061129a8152c2abc7ecd284387a:urgency | [
50281,
18891,
1953,
27,
1359,
4541,
1364,
247,
1183,
9631,
769,
327,
436,
10028,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284... | [
16,
28,
41,
57
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How quickly must a responder act on this alert?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an unusual volume of files read in a short window", "evidence": "The shared_mailbox principal read 1,450 files from the /corporate/records directory within a 10-minute window starting at 09:12 UTC. This unprivileged account authenticated via MFA from allowlisted IP 192.168.10.45 while a chang... | 3,556 | [
0.2336385638777622,
0.59709686863506,
0.14348109920523996,
0.02578346828193787
] | false |
mixa-57097ca176eab0aa421af573528a422dcd2139aa49bac966cafac034d492a0b0 | true | true | 1 | [
0.216667,
0.783333
] | mixa-57097ca176eab0aa421af573528a422dcd2139aa49bac966cafac034d492a0b0:credential_compromise | [
50281,
79,
3941,
1953,
27,
380,
1941,
6492,
247,
3552,
1624,
390,
2395,
556,
644,
25047,
15,
50282,
50284,
3221,
27,
642,
13,
253,
3908,
1057,
417,
2186,
50284,
2032,
27,
4754,
13,
253,
3908,
6556,
50282,
9819,
26544,
1381,
17579,
100... | [
18,
28
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"instructions": "The evidence indicates a credential or account has been compromised.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an endpoint agent matched a known signature", "evidence": "The endpoint agent on CI runner host 'runner-7x2' identified a signature match for 'malware.linux.syslog' during an active session. This connection originated from an unrecognized IP address (203.0.113.88) using unprivileged credentia... | 2,165 | [
0.20775129001351642,
0.7922487099864834
] | true |
mixa-57097ca176eab0aa421af573528a422dcd2139aa49bac966cafac034d492a0b0 | true | contain | 1 | [
0.02,
0.633333,
0.3,
0.046667
] | mixa-57097ca176eab0aa421af573528a422dcd2139aa49bac966cafac034d492a0b0:disposition | [
50281,
22122,
1953,
27,
1359,
943,
436,
3988,
10028,
320,
16322,
264,
32,
50282,
50284,
2810,
64,
7564,
525,
27,
1889,
5344,
13,
5513,
494,
2425,
28,
2810,
1293,
23960,
673,
15,
50284,
3831,
27,
3267,
404,
253,
3167,
390,
2395,
4745,
... | [
14,
32,
50,
62
] | A | [
"close_benign",
"contain",
"investigate",
"monitor"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"close_benign": "Expected, explainable activity; close without analyst time.", "contain": "Contain the host or account immediately; do not wait for triage.", "investigate": "Warrants an analyst opening an investigation.", "monitor": "Not clearly malicious, but worth watching for recurrence."}, "instructio... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an endpoint agent matched a known signature", "evidence": "The endpoint agent on CI runner host 'runner-7x2' identified a signature match for 'malware.linux.syslog' during an active session. This connection originated from an unrecognized IP address (203.0.113.88) using unprivileged credentia... | 377 | [
0.015680433127004047,
0.6802068348730704,
0.27232188210300406,
0.0317908498969216
] | true |
mixa-57097ca176eab0aa421af573528a422dcd2139aa49bac966cafac034d492a0b0 | true | 3 | 3 | [
0,
0.033333,
0.15,
0.55,
0.266667
] | mixa-57097ca176eab0aa421af573528a422dcd2139aa49bac966cafac034d492a0b0:severity | [
50281,
18891,
1953,
27,
1359,
5460,
310,
253,
2442,
3486,
604,
436,
10028,
310,
1524,
32,
50282,
50284,
1268,
470,
27,
15186,
3372,
917,
27,
642,
2289,
281,
2712,
7996,
15,
50284,
1268,
337,
27,
11578,
27,
3710,
2289,
13,
4354,
13891,... | [
17,
31,
43,
60,
75
] | A | [
"0",
"1",
"2",
"3",
"4"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["Negligible: no access to anything sensitive.", "Low: limited access, easily reversed.", "Moderate: access to internal systems or non-public data.", "High: access to production, secrets or customer data.", "Critical: active compromise of crown-jewel systems."], "instructions": "How severe is the potential... | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an endpoint agent matched a known signature", "evidence": "The endpoint agent on CI runner host 'runner-7x2' identified a signature match for 'malware.linux.syslog' during an active session. This connection originated from an unrecognized IP address (203.0.113.88) using unprivileged credentia... | 4,897 | [
0.005552670219730717,
0.025767471913225634,
0.1624995601236438,
0.5783913593583505,
0.2277889383850494
] | false |
mixa-57097ca176eab0aa421af573528a422dcd2139aa49bac966cafac034d492a0b0 | true | true | 1 | [
0.126667,
0.873333
] | mixa-57097ca176eab0aa421af573528a422dcd2139aa49bac966cafac034d492a0b0:true_positive | [
50281,
79,
3941,
1953,
27,
831,
10028,
13806,
27364,
24764,
390,
440,
7582,
1701,
2425,
15,
50282,
50284,
3221,
27,
6029,
525,
2425,
13,
247,
3731,
24693,
13,
390,
247,
1929,
3221,
2762,
15,
50284,
2032,
27,
380,
6944,
8770,
310,
2476... | [
17,
34
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "Benign activity, a misconfiguration, or a known false positive.", "true": "The underlying behaviour is malicious or unauthorised."}, "instructions": "This alert reflects genuinely malicious or unauthorised activity.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an endpoint agent matched a known signature", "evidence": "The endpoint agent on CI runner host 'runner-7x2' identified a signature match for 'malware.linux.syslog' during an active session. This connection originated from an unrecognized IP address (203.0.113.88) using unprivileged credentia... | 5,047 | [
0.09978502317508063,
0.9002149768249194
] | false |
mixa-57097ca176eab0aa421af573528a422dcd2139aa49bac966cafac034d492a0b0 | true | 3 | 3 | [
0,
0.023333,
0.11,
0.866667
] | mixa-57097ca176eab0aa421af573528a422dcd2139aa49bac966cafac034d492a0b0:urgency | [
50281,
18891,
1953,
27,
1359,
4541,
1364,
247,
1183,
9631,
769,
327,
436,
10028,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284... | [
16,
28,
41,
57
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How quickly must a responder act on this alert?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "an endpoint agent matched a known signature", "evidence": "The endpoint agent on CI runner host 'runner-7x2' identified a signature match for 'malware.linux.syslog' during an active session. This connection originated from an unrecognized IP address (203.0.113.88) using unprivileged credentia... | 929 | [
0.004081553480958803,
0.02145495458276223,
0.1546911736155731,
0.8197723183207059
] | true |
mixa-5c26e159bb8cceb80d31ef407cd8345505e8a02653d3bc911f327980d0fd5e22 | true | answer_directly | 0 | [
0.84,
0.06,
0.033333,
0,
0.066667
] | mixa-5c26e159bb8cceb80d31ef407cd8345505e8a02653d3bc911f327980d0fd5e22:action | [
50281,
22122,
1953,
27,
1737,
943,
253,
13372,
513,
1735,
342,
436,
7827,
32,
50282,
50284,
3662,
64,
18711,
314,
27,
380,
13372,
476,
11322,
436,
3139,
342,
1491,
352,
2168,
556,
15,
50284,
2810,
64,
2369,
64,
1913,
27,
1621,
2007,
... | [
15,
33,
51,
70,
86
] | A | [
"answer_directly",
"close_no_action",
"escalate_to_human",
"execute_refund",
"request_information"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"answer_directly": "The assistant can resolve this itself with information it already has.", "close_no_action": "No further action is warranted; the matter is settled.", "escalate_to_human": "Hand off to a human agent with the appropriate authority.", "execute_refund": "Issue the refund or credit the cust... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 618, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 8, "tier": "premium"}, "thread": [{"role": "customer", "text": "I need to close my account by the end of the month, as I've already decided to move on."}, {"role": "agent", "text": "I understand. Could you confirm the email add... | 4,963 | [
0.8725277620180489,
0.02388336602334012,
0.06740371229838976,
0.005157173494683344,
0.03102798616553769
] | false |
mixa-5c26e159bb8cceb80d31ef407cd8345505e8a02653d3bc911f327980d0fd5e22 | true | account | 0 | [
0.95,
0.038333,
0.005,
0.005,
0.001667
] | mixa-5c26e159bb8cceb80d31ef407cd8345505e8a02653d3bc911f327980d0fd5e22:category | [
50281,
22122,
1953,
27,
1737,
310,
436,
7731,
7827,
8558,
670,
32,
50282,
50284,
2395,
27,
41977,
13,
2098,
2544,
13,
6222,
390,
2289,
4323,
15,
50284,
33484,
27,
329,
4179,
13,
45156,
13,
21729,
390,
7830,
1895,
15,
50284,
6742,
27,
... | [
13,
26,
39,
53,
65
] | A | [
"account",
"billing",
"delivery",
"refund",
"technical"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"account": "Login, plan changes, profile or access management.", "billing": "A charge, invoice, subscription or payment problem.", "delivery": "Shipping, fulfilment or delivery of a physical item.", "refund": "The customer is explicitly asking for money back.", "technical": "The product or service is not ... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 618, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 8, "tier": "premium"}, "thread": [{"role": "customer", "text": "I need to close my account by the end of the month, as I've already decided to move on."}, {"role": "agent", "text": "I understand. Could you confirm the email add... | 5,717 | [
0.9632678144954321,
0.016408589351863023,
0.00445896165369706,
0.004457269620589941,
0.011407364878417882
] | false |
mixa-5c26e159bb8cceb80d31ef407cd8345505e8a02653d3bc911f327980d0fd5e22 | true | 3 | 3 | [
0,
0.026667,
0.073333,
0.9
] | mixa-5c26e159bb8cceb80d31ef407cd8345505e8a02653d3bc911f327980d0fd5e22:churn_risk | [
50281,
18891,
1953,
27,
1359,
2779,
310,
436,
7731,
281,
3523,
2509,
2136,
342,
441,
984,
273,
436,
5016,
32,
50282,
50284,
1268,
470,
27,
1621,
861,
273,
5408,
38120,
15,
50284,
1268,
337,
27,
49671,
22014,
13,
533,
253,
2954,
310,
... | [
21,
31,
44,
57
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No sign of dissatisfaction.", "Mild frustration, but the relationship is intact.", "Clearly unhappy; repeat problems or explicit complaints.", "Imminent: threatening to cancel, dispute or leave."], "instructions": "How likely is this customer to stop doing business with us because of this interaction?", ... | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 618, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 8, "tier": "premium"}, "thread": [{"role": "customer", "text": "I need to close my account by the end of the month, as I've already decided to move on."}, {"role": "agent", "text": "I understand. Could you confirm the email add... | 312 | [
0.003848168844541863,
0.022452554862278574,
0.0914415481187812,
0.8822577281743984
] | true |
mixa-5c26e159bb8cceb80d31ef407cd8345505e8a02653d3bc911f327980d0fd5e22 | true | false | 0 | [
0.8,
0.2
] | mixa-5c26e159bb8cceb80d31ef407cd8345505e8a02653d3bc911f327980d0fd5e22:needs_human | [
50281,
79,
3941,
1953,
27,
831,
7827,
4419,
247,
1966,
5570,
2581,
685,
16644,
10885,
15,
50282,
50284,
3221,
27,
24689,
318,
476,
4459,
436,
281,
6064,
15,
50284,
2032,
27,
329,
1966,
1364,
1379,
689,
27,
31536,
13,
6265,
390,
32496,... | [
17,
28
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "Automation can carry this to resolution.", "true": "A human must take over: judgement, authority or empathy is required."}, "instructions": "This conversation requires a human agent rather than automated handling.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 618, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 8, "tier": "premium"}, "thread": [{"role": "customer", "text": "I need to close my account by the end of the month, as I've already decided to move on."}, {"role": "agent", "text": "I understand. Could you confirm the email add... | 2,720 | [
0.5423005898790741,
0.4576994101209259
] | true |
mixa-5c26e159bb8cceb80d31ef407cd8345505e8a02653d3bc911f327980d0fd5e22 | true | 2 | 2 | [
0.026667,
0.266667,
0.583333,
0.123333
] | mixa-5c26e159bb8cceb80d31ef407cd8345505e8a02653d3bc911f327980d0fd5e22:urgency | [
50281,
18891,
1953,
27,
1359,
673,
14,
19579,
310,
436,
7827,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284,
1268,
374,
27,
... | [
13,
25,
38,
54
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How time-sensitive is this conversation?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 618, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 8, "tier": "premium"}, "thread": [{"role": "customer", "text": "I need to close my account by the end of the month, as I've already decided to move on."}, {"role": "agent", "text": "I understand. Could you confirm the email add... | 3,151 | [
0.012993107779233291,
0.3051339761825268,
0.5781607670580593,
0.10371214898018061
] | false |
mixa-bcb23625778997e937b4a0f71c6d0b58b8465bf5b12164d0285ee16182be80f1 | true | escalate_to_human | 2 | [
0.05,
0,
0.816667,
0.033333,
0.1
] | mixa-bcb23625778997e937b4a0f71c6d0b58b8465bf5b12164d0285ee16182be80f1:action | [
50281,
22122,
1953,
27,
1737,
943,
253,
13372,
513,
1735,
342,
436,
7827,
32,
50282,
50284,
3662,
64,
18711,
314,
27,
380,
13372,
476,
11322,
436,
3139,
342,
1491,
352,
2168,
556,
15,
50284,
2810,
64,
2369,
64,
1913,
27,
1621,
2007,
... | [
15,
33,
51,
70,
86
] | A | [
"answer_directly",
"close_no_action",
"escalate_to_human",
"execute_refund",
"request_information"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"answer_directly": "The assistant can resolve this itself with information it already has.", "close_no_action": "No further action is warranted; the matter is settled.", "escalate_to_human": "Hand off to a human agent with the appropriate authority.", "execute_refund": "Issue the refund or credit the cust... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 613, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 11, "tier": "standard"}, "orders": [{"amount_usd": 299, "date": "2026-01-13", "id": "A-16098", "status": "failed"}], "thread": [{"role": "customer", "text": "I just received my delivery and it's completely wrong. I paid $299 fo... | 5,196 | [
0.06602215672025792,
0.013345398662034378,
0.705792486497103,
0.03678051594805851,
0.1780594421725462
] | false |
mixa-bcb23625778997e937b4a0f71c6d0b58b8465bf5b12164d0285ee16182be80f1 | true | delivery | 2 | [
0.033333033333033335,
0.05000005000005,
0.8833338833338833,
0.033333033333033335,
0
] | mixa-bcb23625778997e937b4a0f71c6d0b58b8465bf5b12164d0285ee16182be80f1:category | [
50281,
22122,
1953,
27,
1737,
310,
436,
7731,
7827,
8558,
670,
32,
50282,
50284,
2395,
27,
41977,
13,
2098,
2544,
13,
6222,
390,
2289,
4323,
15,
50284,
33484,
27,
329,
4179,
13,
45156,
13,
21729,
390,
7830,
1895,
15,
50284,
6742,
27,
... | [
13,
26,
39,
53,
65
] | A | [
"account",
"billing",
"delivery",
"refund",
"technical"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"account": "Login, plan changes, profile or access management.", "billing": "A charge, invoice, subscription or payment problem.", "delivery": "Shipping, fulfilment or delivery of a physical item.", "refund": "The customer is explicitly asking for money back.", "technical": "The product or service is not ... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 613, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 11, "tier": "standard"}, "orders": [{"amount_usd": 299, "date": "2026-01-13", "id": "A-16098", "status": "failed"}], "thread": [{"role": "customer", "text": "I just received my delivery and it's completely wrong. I paid $299 fo... | 845 | [
0.02990304385492779,
0.041222488348865265,
0.8535580746344105,
0.06421187742809134,
0.011104515733705229
] | true |
mixa-bcb23625778997e937b4a0f71c6d0b58b8465bf5b12164d0285ee16182be80f1 | true | true | 1 | [
0.083333,
0.916667
] | mixa-bcb23625778997e937b4a0f71c6d0b58b8465bf5b12164d0285ee16182be80f1:needs_human | [
50281,
79,
3941,
1953,
27,
831,
7827,
4419,
247,
1966,
5570,
2581,
685,
16644,
10885,
15,
50282,
50284,
3221,
27,
24689,
318,
476,
4459,
436,
281,
6064,
15,
50284,
2032,
27,
329,
1966,
1364,
1379,
689,
27,
31536,
13,
6265,
390,
32496,... | [
17,
28
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "Automation can carry this to resolution.", "true": "A human must take over: judgement, authority or empathy is required."}, "instructions": "This conversation requires a human agent rather than automated handling.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 613, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 11, "tier": "standard"}, "orders": [{"amount_usd": 299, "date": "2026-01-13", "id": "A-16098", "status": "failed"}], "thread": [{"role": "customer", "text": "I just received my delivery and it's completely wrong. I paid $299 fo... | 4,037 | [
0.16281139030634853,
0.8371886096936515
] | false |
mixa-bcb23625778997e937b4a0f71c6d0b58b8465bf5b12164d0285ee16182be80f1 | true | 2 | 2 | [
0,
0.083333,
0.516667,
0.4
] | mixa-bcb23625778997e937b4a0f71c6d0b58b8465bf5b12164d0285ee16182be80f1:urgency | [
50281,
18891,
1953,
27,
1359,
673,
14,
19579,
310,
436,
7827,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284,
1268,
374,
27,
... | [
13,
25,
38,
54
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How time-sensitive is this conversation?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 613, "prior_tickets_90d": 1, "seats": 1, "tenure_months": 11, "tier": "standard"}, "orders": [{"amount_usd": 299, "date": "2026-01-13", "id": "A-16098", "status": "failed"}], "thread": [{"role": "customer", "text": "I just received my delivery and it's completely wrong. I paid $299 fo... | 2,597 | [
0.005670977108711159,
0.08002729346158534,
0.44961207581125195,
0.46468965361845155
] | true |
mixa-716ead5684e1e7b881b5b6fd0a10a7f325b0671e310ef4085af50dbcf139cbb3 | true | 0 | 0 | [
0.8,
0.176667,
0.023333,
0
] | mixa-716ead5684e1e7b881b5b6fd0a10a7f325b0671e310ef4085af50dbcf139cbb3:discrepancy_severity | [
50281,
18891,
1953,
27,
1359,
2144,
310,
667,
26210,
875,
253,
45156,
13,
253,
1340,
285,
253,
6742,
32,
50282,
50284,
1268,
470,
27,
8256,
27,
3253,
30855,
3205,
15,
50284,
1268,
337,
27,
308,
10482,
27,
46551,
390,
247,
35361,
3064,... | [
20,
30,
43,
56
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["None: everything reconciles.", "Trivial: rounding or a cosmetic difference.", "Moderate: a real difference worth confirming.", "Material: a large or unexplained difference."], "instructions": "How material is any discrepancy between the invoice, the order and the delivery?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"delivery": {"condition": "accepted with exceptions", "date": "2026-03-12", "received_qty": 400}, "invoice": {"currency": "USD", "id": "INV-2026-1498", "lines": [{"qty": 400, "sku": "SKU-530", "total_usd": 130404.0, "unit_usd": 326.01}], "total_usd": 130404.0, "vendor": "Kestrel Supply"}, "payment": {"days_until_due":... | 2,943 | [
0.4952124332107012,
0.1457072559802433,
0.2874444998621374,
0.07163581094691812
] | false |
mixa-716ead5684e1e7b881b5b6fd0a10a7f325b0671e310ef4085af50dbcf139cbb3 | true | approve | 0 | [
0.5,
0.166667,
0.33,
0.003333
] | mixa-716ead5684e1e7b881b5b6fd0a10a7f325b0671e310ef4085af50dbcf139cbb3:disposition | [
50281,
22122,
1953,
27,
1359,
943,
436,
23906,
45156,
320,
16322,
264,
32,
50282,
50284,
23255,
27,
6397,
2706,
253,
1340,
285,
6742,
28,
23255,
323,
7830,
15,
50284,
2186,
27,
18190,
3198,
24025,
1078,
7830,
28,
2186,
13910,
37699,
15,... | [
14,
28,
41,
55
] | A | [
"approve",
"hold",
"manual_review",
"reject"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"approve": "Matches the order and delivery; approve for payment.", "hold": "Something needs confirming before payment; hold pending clarification.", "manual_review": "A human in finance must review the discrepancy.", "reject": "Should not be paid: duplicate, unauthorised or materially wrong."}, "instructi... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"delivery": {"condition": "accepted with exceptions", "date": "2026-03-12", "received_qty": 400}, "invoice": {"currency": "USD", "id": "INV-2026-1498", "lines": [{"qty": 400, "sku": "SKU-530", "total_usd": 130404.0, "unit_usd": 326.01}], "total_usd": 130404.0, "vendor": "Kestrel Supply"}, "payment": {"days_until_due":... | 3,153 | [
0.38108526369857215,
0.29989846796297626,
0.286006540792961,
0.033009727545490475
] | false |
mixa-716ead5684e1e7b881b5b6fd0a10a7f325b0671e310ef4085af50dbcf139cbb3 | true | true | 1 | [
0.066667,
0.933333
] | mixa-716ead5684e1e7b881b5b6fd0a10a7f325b0671e310ef4085af50dbcf139cbb3:matches_order | [
50281,
79,
3941,
1953,
27,
380,
45156,
30855,
3205,
342,
253,
7471,
1340,
285,
253,
5950,
6742,
15,
50282,
50284,
3221,
27,
1707,
310,
247,
26210,
1411,
253,
1340,
390,
253,
6742,
15,
50284,
2032,
27,
10243,
4957,
13,
13483,
285,
8322... | [
19,
33
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "There is a discrepancy against the order or the delivery.", "true": "Line items, quantities and amounts reconcile."}, "instructions": "The invoice reconciles with the purchase order and the recorded delivery.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"delivery": {"condition": "accepted with exceptions", "date": "2026-03-12", "received_qty": 400}, "invoice": {"currency": "USD", "id": "INV-2026-1498", "lines": [{"qty": 400, "sku": "SKU-530", "total_usd": 130404.0, "unit_usd": 326.01}], "total_usd": 130404.0, "vendor": "Kestrel Supply"}, "payment": {"days_until_due":... | 865 | [
0.26779506062098307,
0.7322049393790169
] | true |
mixa-716ead5684e1e7b881b5b6fd0a10a7f325b0671e310ef4085af50dbcf139cbb3 | true | 3 | 3 | [
0,
0.1,
0.3,
0.6
] | mixa-716ead5684e1e7b881b5b6fd0a10a7f325b0671e310ef4085af50dbcf139cbb3:urgency | [
50281,
18891,
1953,
27,
1359,
673,
14,
19579,
310,
5162,
436,
45156,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284,
1268,
374,... | [
14,
26,
39,
55
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How time-sensitive is processing this invoice?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"delivery": {"condition": "accepted with exceptions", "date": "2026-03-12", "received_qty": 400}, "invoice": {"currency": "USD", "id": "INV-2026-1498", "lines": [{"qty": 400, "sku": "SKU-530", "total_usd": 130404.0, "unit_usd": 326.01}], "total_usd": 130404.0, "vendor": "Kestrel Supply"}, "payment": {"days_until_due":... | 711 | [
0.008484896012857696,
0.12935289675496905,
0.47077322766587093,
0.3913889795663022
] | true |
mixa-4e5f6c22eb52f209af4ab7315804f63ac45859b1f7a15b371741403fe37dc17b | true | true | 1 | [
0.4,
0.6
] | mixa-4e5f6c22eb52f209af4ab7315804f63ac45859b1f7a15b371741403fe37dc17b:credential_compromise | [
50281,
79,
3941,
1953,
27,
380,
1941,
6492,
247,
3552,
1624,
390,
2395,
556,
644,
25047,
15,
50282,
50284,
3221,
27,
642,
13,
253,
3908,
1057,
417,
2186,
50284,
2032,
27,
4754,
13,
253,
3908,
6556,
50282,
9819,
26544,
1381,
17579,
100... | [
18,
28
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"instructions": "The evidence indicates a credential or account has been compromised.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "two successful logins from distant locations too close together", "evidence": "A contractor account logged in from New York at 08:00 UTC and subsequently from London at 08:45 UTC. The user credentials associated with this unprivileged account were last rotated one year ago, though both access... | 214 | [
0.3828145043546406,
0.6171854956453594
] | true |
mixa-4e5f6c22eb52f209af4ab7315804f63ac45859b1f7a15b371741403fe37dc17b | true | investigate | 2 | [
0.08,
0.086667,
0.583333,
0.25
] | mixa-4e5f6c22eb52f209af4ab7315804f63ac45859b1f7a15b371741403fe37dc17b:disposition | [
50281,
22122,
1953,
27,
1359,
943,
436,
3988,
10028,
320,
16322,
264,
32,
50282,
50284,
2810,
64,
7564,
525,
27,
1889,
5344,
13,
5513,
494,
2425,
28,
2810,
1293,
23960,
673,
15,
50284,
3831,
27,
3267,
404,
253,
3167,
390,
2395,
4745,
... | [
14,
32,
50,
62
] | A | [
"close_benign",
"contain",
"investigate",
"monitor"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"close_benign": "Expected, explainable activity; close without analyst time.", "contain": "Contain the host or account immediately; do not wait for triage.", "investigate": "Warrants an analyst opening an investigation.", "monitor": "Not clearly malicious, but worth watching for recurrence."}, "instructio... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "two successful logins from distant locations too close together", "evidence": "A contractor account logged in from New York at 08:00 UTC and subsequently from London at 08:45 UTC. The user credentials associated with this unprivileged account were last rotated one year ago, though both access... | 1,501 | [
0.07240737355758382,
0.1416377622531653,
0.6377518704960138,
0.1482029936932371
] | true |
mixa-4e5f6c22eb52f209af4ab7315804f63ac45859b1f7a15b371741403fe37dc17b | true | 3 | 3 | [
0.01,
0.073333,
0.316667,
0.333333,
0.266667
] | mixa-4e5f6c22eb52f209af4ab7315804f63ac45859b1f7a15b371741403fe37dc17b:severity | [
50281,
18891,
1953,
27,
1359,
5460,
310,
253,
2442,
3486,
604,
436,
10028,
310,
1524,
32,
50282,
50284,
1268,
470,
27,
15186,
3372,
917,
27,
642,
2289,
281,
2712,
7996,
15,
50284,
1268,
337,
27,
11578,
27,
3710,
2289,
13,
4354,
13891,... | [
17,
31,
43,
60,
75
] | A | [
"0",
"1",
"2",
"3",
"4"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["Negligible: no access to anything sensitive.", "Low: limited access, easily reversed.", "Moderate: access to internal systems or non-public data.", "High: access to production, secrets or customer data.", "Critical: active compromise of crown-jewel systems."], "instructions": "How severe is the potential... | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "two successful logins from distant locations too close together", "evidence": "A contractor account logged in from New York at 08:00 UTC and subsequently from London at 08:45 UTC. The user credentials associated with this unprivileged account were last rotated one year ago, though both access... | 802 | [
0.026766329495011714,
0.07374908379132988,
0.24650051695949665,
0.35575859910326096,
0.2972254706509008
] | true |
mixa-4e5f6c22eb52f209af4ab7315804f63ac45859b1f7a15b371741403fe37dc17b | true | true | 1 | [
0.35,
0.65
] | mixa-4e5f6c22eb52f209af4ab7315804f63ac45859b1f7a15b371741403fe37dc17b:true_positive | [
50281,
79,
3941,
1953,
27,
831,
10028,
13806,
27364,
24764,
390,
440,
7582,
1701,
2425,
15,
50282,
50284,
3221,
27,
6029,
525,
2425,
13,
247,
3731,
24693,
13,
390,
247,
1929,
3221,
2762,
15,
50284,
2032,
27,
380,
6944,
8770,
310,
2476... | [
17,
34
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "Benign activity, a misconfiguration, or a known false positive.", "true": "The underlying behaviour is malicious or unauthorised."}, "instructions": "This alert reflects genuinely malicious or unauthorised activity.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "two successful logins from distant locations too close together", "evidence": "A contractor account logged in from New York at 08:00 UTC and subsequently from London at 08:45 UTC. The user credentials associated with this unprivileged account were last rotated one year ago, though both access... | 5,456 | [
0.31978254469587447,
0.6802174553041255
] | false |
mixa-4e5f6c22eb52f209af4ab7315804f63ac45859b1f7a15b371741403fe37dc17b | true | 2 | 2 | [
0.033333,
0.166667,
0.433333,
0.366667
] | mixa-4e5f6c22eb52f209af4ab7315804f63ac45859b1f7a15b371741403fe37dc17b:urgency | [
50281,
18891,
1953,
27,
1359,
4541,
1364,
247,
1183,
9631,
769,
327,
436,
10028,
32,
50282,
50284,
1268,
470,
27,
1621,
673,
3473,
28,
476,
3343,
39450,
15,
50284,
1268,
337,
27,
30852,
460,
28,
6016,
1561,
253,
2622,
15154,
15,
50284... | [
16,
28,
41,
57
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["No time pressure; can wait indefinitely.", "Routine; handle within the normal queue.", "Elevated; should be handled within the same week.", "Critical; requires action within the same day."], "instructions": "How quickly must a responder act on this alert?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"alert": {"description": "two successful logins from distant locations too close together", "evidence": "A contractor account logged in from New York at 08:00 UTC and subsequently from London at 08:45 UTC. The user credentials associated with this unprivileged account were last rotated one year ago, though both access... | 4,747 | [
0.03196404371197297,
0.1455547005431934,
0.31467208180962525,
0.5078091739352084
] | false |
mixa-2311e44c718503d000dab1085e8330c787550fe59039ddb8521a80331f7300eb | true | human_review | 1 | [
0.02,
0.74,
0.056667,
0.183333
] | mixa-2311e44c718503d000dab1085e8330c787550fe59039ddb8521a80331f7300eb:action | [
50281,
22122,
1953,
27,
1737,
943,
253,
1759,
1430,
985,
513,
342,
436,
10711,
32,
50282,
50284,
4035,
27,
1281,
253,
5570,
4262,
1293,
38205,
15,
50284,
1966,
64,
15337,
27,
7035,
489,
436,
10711,
323,
247,
1966,
281,
2278,
15,
50284... | [
16,
26,
41,
55
] | A | [
"continue",
"human_review",
"observe",
"stop"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"continue": "Let the agent proceed without interruption.", "human_review": "Queue this trace for a human to review.", "observe": "Keep running, but flag the trace for later sampling.", "stop": "Halt the agent now."}, "instructions": "What should the observability system do with this trace?", "type": "choi... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "checkpointed", "model": "internal-agent-v3"}, "constraints": ["Do not exceed a $50 spend on cloud resources", "Never modify production without an approved change ticket", "Never disable monitoring or alerting"], "task": "Delete personal data for the accounts in the erasure queue.", "trace_summar... | 5,158 | [
0.03665282582077243,
0.6632259256193921,
0.09341685399871186,
0.20670439456112363
] | false |
mixa-2311e44c718503d000dab1085e8330c787550fe59039ddb8521a80331f7300eb | true | true | 1 | [
0.04,
0.96
] | mixa-2311e44c718503d000dab1085e8330c787550fe59039ddb8521a80331f7300eb:needs_review | [
50281,
79,
3941,
1953,
27,
831,
10711,
4419,
1966,
2278,
15,
50282,
50284,
3221,
27,
1621,
1966,
4116,
310,
26085,
15,
50284,
2032,
27,
329,
1966,
943,
16030,
436,
1408,
15,
50282,
9819,
12788,
1381,
17579,
1920,
13646,
1381,
346,
5903,... | [
12,
21
] | A | [
"false",
"true"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"false": "No human attention is warranted.", "true": "A human should inspect this run."}, "instructions": "This trace requires human review.", "type": "noul"} | noul | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "checkpointed", "model": "internal-agent-v3"}, "constraints": ["Do not exceed a $50 spend on cloud resources", "Never modify production without an approved change ticket", "Never disable monitoring or alerting"], "task": "Delete personal data for the accounts in the erasure queue.", "trace_summar... | 3,951 | [
0.1034660493302373,
0.8965339506697627
] | false |
mixa-2311e44c718503d000dab1085e8330c787550fe59039ddb8521a80331f7300eb | true | harmful | 1 | [
0.076667,
0.716667,
0.193333,
0.013333
] | mixa-2311e44c718503d000dab1085e8330c787550fe59039ddb8521a80331f7300eb:outcome | [
50281,
22122,
1953,
27,
1359,
858,
436,
5570,
1408,
1614,
562,
32,
50282,
50284,
4433,
27,
380,
5570,
858,
417,
14294,
253,
4836,
15,
50284,
19632,
27,
380,
5570,
2335,
271,
2250,
326,
4269,
4723,
390,
13588,
247,
7658,
15,
50284,
789... | [
13,
24,
40,
55
] | A | [
"failure",
"harmful",
"partial",
"success"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"failure": "The agent did not accomplish the task.", "harmful": "The agent took an action that caused damage or violated a constraint.", "partial": "The agent made progress but did not fully complete the task.", "success": "The agent completed the task correctly."}, "instructions": "How did this agent run... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "checkpointed", "model": "internal-agent-v3"}, "constraints": ["Do not exceed a $50 spend on cloud resources", "Never modify production without an approved change ticket", "Never disable monitoring or alerting"], "task": "Delete personal data for the accounts in the erasure queue.", "trace_summar... | 5,526 | [
0.11269636069498332,
0.689370416421842,
0.1790772177680075,
0.018856005115167128
] | false |
mixa-2311e44c718503d000dab1085e8330c787550fe59039ddb8521a80331f7300eb | true | 3 | 3 | [
0.006667,
0.043333,
0.2,
0.75
] | mixa-2311e44c718503d000dab1085e8330c787550fe59039ddb8521a80331f7300eb:risk | [
50281,
18891,
1953,
27,
1359,
29198,
369,
253,
5570,
434,
8770,
275,
436,
10711,
32,
50282,
50284,
1268,
470,
27,
6029,
525,
27,
1239,
14,
7483,
390,
4518,
4999,
5231,
15,
50284,
1268,
337,
27,
11578,
27,
10934,
12013,
1561,
7990,
15,... | [
16,
31,
42,
58
] | A | [
"0",
"1",
"2",
"3"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": ["Benign: read-only or clearly safe actions.", "Low: routine writes within scope.", "Moderate: irreversible or out-of-scope actions.", "High: destructive, security-relevant, or policy-violating actions."], "instructions": "How risky was the agent's behaviour in this trace?", "type": "score"} | score | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"agent": {"autonomy": "checkpointed", "model": "internal-agent-v3"}, "constraints": ["Do not exceed a $50 spend on cloud resources", "Never modify production without an approved change ticket", "Never disable monitoring or alerting"], "task": "Delete personal data for the accounts in the erasure queue.", "trace_summar... | 109 | [
0.007088608911426873,
0.06760494132736393,
0.3149523392233374,
0.6103541105378717
] | true |
mixa-556c2a4c7808b706481499034f4bcd26e59b94a19faba7506bbe6e53c9b07a03 | true | request_information | 4 | [
0.09,
0.003333,
0.283333,
0.006667,
0.616667
] | mixa-556c2a4c7808b706481499034f4bcd26e59b94a19faba7506bbe6e53c9b07a03:action | [
50281,
22122,
1953,
27,
1737,
943,
253,
13372,
513,
1735,
342,
436,
7827,
32,
50282,
50284,
3662,
64,
18711,
314,
27,
380,
13372,
476,
11322,
436,
3139,
342,
1491,
352,
2168,
556,
15,
50284,
2810,
64,
2369,
64,
1913,
27,
1621,
2007,
... | [
15,
33,
51,
70,
86
] | A | [
"answer_directly",
"close_no_action",
"escalate_to_human",
"execute_refund",
"request_information"
] | LocalLLaMA/typed-decisions synthetic teacher-labeled benchmark | {"criteria": {"answer_directly": "The assistant can resolve this itself with information it already has.", "close_no_action": "No further action is warranted; the matter is settled.", "escalate_to_human": "Hand off to a human agent with the appropriate authority.", "execute_refund": "Issue the refund or credit the cust... | choice | null | 46a58d63edfd86e23229c78afe8b72307bb4ca9fb0e8df180cabb3c67ec9dcd5 | train | {"account": {"lifetime_value_usd": 579, "prior_tickets_90d": 2, "seats": 1, "tenure_months": 11, "tier": "free"}, "thread": [{"role": "customer", "text": "I noticed an unexpected charge on my account today. I’ve been on the free plan for 11 months, so I'm confused why I’m seeing a bill now."}, {"role": "agent", "text":... | 5,157 | [
0.29761466354207655,
0.01026324660355877,
0.23885812493691685,
0.009478255635316422,
0.4437857092821314
] | false |
End of preview.
Laya Homelab Decisions
This dataset reproduces the exact training and calibration population of Billy Davies's expanded Laya routing experiment. The companion model repository is billyd10/laya-homelab-router.
| Population | Public benchmark decisions | Authored house decisions | Total |
|---|---|---|---|
| Training | 4,922 | 854 | 5,776 |
| Calibration / validation | 600 | 172 | 772 |
The public component derives from the Apache 2.0 LocalLLaMA/typed-decisions training split. It contains synthetic teacher-labeled workflow decisions. The house component contains authored synthetic route decisions for D&D, household, and homelab scenarios. It does not contain observed human conversation traffic or generated answer targets. Routes distinguish conversational replies, full reasoning or governed actions, and retrieval before reasoning.
The expansion adds 720 surface variants from 60 topic families and 180 authored route intents. Of those variants, 576 enter training and 144 enter calibration. Related families stay in one split. Four variants of an intent are not four independent scenarios.
Format and reproducibility
Each row includes the source state and question as JSON strings, ordered option keys, gold target probabilities, and the exact token IDs and marker positions consumed by the model.
All 6,548 exported decisions reproduce their frozen tokenization and gold targets exactly.
gold_target retains the benchmark or authored target.
training_target records the actual optimization target: public training uses 90% original-model probabilities and 10% gold; house training retains its authored one-hot target.
Calibration rows are excluded from optimization and use their gold target in this field.
fit_allowed records this distinction.
training_position reproduces the seeded training order.
used_by_selected_checkpoint identifies the 2,912 exposures seen by the checkpoint selected at optimizer update 91.
The full experiment trains one epoch over all 5,776 decisions and selects among updates 0, 46, 91, 136, and 181 using calibration only.
The publication manifest binds dataset files, the immutable source bundle, effective training targets, and selected checkpoint identity.
Training JSONL SHA-256: 71086679a026e0c0623d91b7a5470924d06bcaef0e0866616bab41ff6b2a2cf9.
Calibration JSONL SHA-256: 9b7504b8c988e37fdaec68bf1735285e4b3e6fcdae8204ce0ed11bdfe23cba7f.
Limits
Evaluation cases are excluded from this release. The frozen evaluation population contains 2,277 decisions in the experiment. These development results do not establish generated-answer quality, live latency, or serving readiness. Public calibration is held out of this experiment's updates, but the original base checkpoint already trained on the benchmark training corpus. House labels express the authored routing policy and require independent human validation before product promotion.
Lineage
Run: b084e400-0496-4952-8128-d8bf507a98ef.
Executed source: 0b98ee2947933596a06ee3ffaa3305130d045e19.
The model starts from convaiinnovations/laya-typed-decisions.
Its upstream encoder is ModernBERT-large.
The public data and model source are attributed to their upstream authors under Apache 2.0.
The authored house expansion is released under Apache 2.0.
- Downloads last month
- 43
Models trained or fine-tuned on billyd10/laya-homelab-decisions