Sign in to view source links and access this dataset
Description
Momento is a benchmark dataset for evaluating LLM-based agents on persistent, tool-mediated task completion across multiple conversational sessions. The dataset, created by adrilmanurung, is grounded in a restaurant service domain and was last updated on May 30, 2026. Tasks require agents to recall past user preferences and resolve goals spanning multiple interactions.
Use Cases
Benchmarking agent performance on persistent memory tasks based on the multi-session conversation structure.
Evaluating tool invocation and sequencing accuracy based on the tool-mediated task completion requirement.
Testing an agent's ability to recall user preferences across sessions based on the described restaurant service domain.
Assessing an agent's adherence to domain policies during extended goal-oriented dialogues.
Strengths
Designed specifically for evaluating persistent memory and reasoning in multi-session agentic conversations.
Tasks are grounded in a concrete restaurant service domain, providing a realistic evaluation scenario.
Last updated on 2026-05-30, suggesting recent maintenance.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment for large-scale training.
Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
Source
huggingface, author adrilmanurung
Collection Method
Likely constructed for benchmarking purposes, but the specific collection method is not detailed.
Freshness
Last updated 2026-05-30 13:06:34.
License is unknown; terms of use must be verified before application.