task_id large_stringlengths 11 11 | domain large_stringclasses 1
value | question_type large_stringclasses 3
values | question large_stringlengths 222 1.06k | gold_answer large_stringlengths 1 8 | context large_stringlengths 640k 640k |
|---|---|---|---|---|---|
lme-test-00 | enterprise | static-environment | "I am working with a few forms in our ServiceNow portal. Among these five forms (change request/prob(...TRUNCATED) | true | "TRAJ e62532d7: goal=Retrieve information from the chart with the title #CAT088191232 and perform th(...TRUNCATED) |
lme-test-01 | enterprise | dynamic-environment | "I am working with our ServiceNow service catalog and comparing the order pages for the `Standard La(...TRUNCATED) | C | "TRAJ acad20ab: goal=Allocate investments to maximize returns\n\nCreate a filter for the list to ext(...TRUNCATED) |
lme-test-02 | enterprise | dynamic-environment | "I am working with our ServiceNow portal. When we remove a user record, which option exactly matches(...TRUNCATED) | D | "TRAJ 1dd9a032: goal=Find the warranty expiration date for Karen-Caroline Patrick-Jackson's laptop a(...TRUNCATED) |
lme-test-03 | enterprise | procedure | "I am working with our ServiceNow portal. For the task of navigating and creating problem requests f(...TRUNCATED) | G | "TRAJ c700854e: goal=The schedule for your change requests with hashtag #SERIES-7ecd8a83-6 is curren(...TRUNCATED) |
lme-test-04 | enterprise | static-environment | "I am working with our ServiceNow portal. When we order a loaner laptop, we are able to choose from (...TRUNCATED) | false | "TRAJ a302bc89: goal=Create a new \"Normal\" change request\n\nCreate a new change request with a va(...TRUNCATED) |
lme-test-05 | enterprise | dynamic-environment | "I am working with our ServiceNow portal. When filtering the Hardware list by the \"Assigned to\" fi(...TRUNCATED) | B | "TRAJ 46da7f42: goal=You have to retrieve some information from a dashboard chart based on the descr(...TRUNCATED) |
lme-test-06 | enterprise | procedure | "I am working with our ServiceNow Portal. I am tasked with balancing the workload for problems with (...TRUNCATED) | Reports | "TRAJ 92810c4f: goal=Allocate investments to maximize returns\n\nCreate a filter for the list to ext(...TRUNCATED) |
lme-test-07 | enterprise | procedure | "I am working with our ServiceNow portal. When we create a user record, which option contains only i(...TRUNCATED) | A | "TRAJ f9a23d3a: goal=Find the warranty expiration date for Alexandra-Dylan Johnson-Martinez's laptop(...TRUNCATED) |
lme-test-08 | enterprise | procedure | "I am working with our ServiceNow Portal. I need to offboard a user. What should I do on that user's(...TRUNCATED) | C | "TRAJ 1c4bb7ab: goal=Sort the \"assets\" list by the following fields:\n - Warranty expiration (asce(...TRUNCATED) |
lme-test-09 | enterprise | procedure | "I am working with our ServiceNow Portal. I am tasked with balancing the workload for problems with (...TRUNCATED) | E | "TRAJ 8ecdb045: goal=Managing Your Existing Expenses\n\nCreate a filter for the list to extract all (...TRUNCATED) |
End of preview. Expand in Data Studio
CompactionBench LME Text Subset
20 questions from LongMemEval-V2 filtered for text-answerable content.
Each task has ~160k tokens of web agent trajectory context (thoughts, actions, page states). All answers are confirmed present in the text context.
Used for testing context compaction in long-running agents.
Format
JSONL with one task per line. Each line is a CompactionBench TaskRow with:
- task_id
- context (~160k tokens)
- question
- gold_answer
- metadata (domain, question_type, trajectories)
Source
Questions from LongMemEval-V2. Trajectories rendered from agent thought logs and action histories.
- Downloads last month
- 76