Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

Locomo dataset and benchmarks – Page 9 · SOTA2 Research

Research Datasets

Locomo

Benchmarks

Task NameDataset NameSOTA MethodSort

Task NameDataset NameSOTA ResultTrendResults

Session-level Retrieval

LoCoMo (session-level)

62.5Hit@1

3

Long-context Conversational Question Answering

LoCoMo Adversarial

1.94G-EVAL Score

3

Long-context Conversational Question Answering

LoCoMo Open-domain

3.24G-EVAL

3

Long-context Conversational Question Answering

LoCoMo Temporal

3.08G-EVAL

3

Long-context Conversational Question Answering

LoCoMo Multi-hop

2.89G-EVAL

3

Long-context Conversational Question Answering

LoCoMo Single-hop

2.98G-EVAL Score

3

Long-context Conversational Question Answering

LoCoMo Overall

2.81G-EVAL

3

Conversational Memory Retrieval

LoCoMo V3.3 (Mode A)

80Single-Hop Accuracy

3

Refusal Prediction

LoCoMo full

73.3Precision

3

Conversational Question Answering

LoCoMo Adversarial

37EM

3

LoCoMo

0Harm Rate

2

Token Consumption Analysis

LoCoMo

56,170,922E2E Input Tokens

2

Long-term Memory Question Answering

LoCoMo single-hop

87.16Average Judge Score

2

Long-term Memory Question Answering

LoCoMo temporal

70.72Average Judge Score

2

Memory Evaluation

LoCoMo single episode

1.8Mean Score

2

Long-term Text Memory

LOCOMO

58.1Accuracy

2

Structured Reasoning

LoCoMo

61.8F1 Score

2

Long-context Memory Management

LOCOMO

52Overall QA F1

2

Memory Retrieval

LoCoMo

75.8Overall Score

2

Long-term conversation recall

LOCOMO Cross-LLM portability

80.8LLM-as-Judge Accuracy

1

Conversational Memory Retrieval

LoCoMo RESCUABLE_STALE_TOP1 alpha

0.309MRR

1

Conversational Memory Retrieval

LoCoMo CONV_TOP1_WRONG_GOLD_IN_POOL alpha

29.94MRR

1

Conversational Memory Retrieval

LoCoMo alpha (T_SUP_auto)

55.08MRR

1

Conversational Memory Retrieval

LoCoMo alpha (FULL)

56.38MRR

1

Long-Term Conversational Memory

LoCoMo 81 questions (conv-30)

64Single-Hop Accuracy

1

Page 9 of 10

Popular tasks

Question Answering Long-context Question Answering Conversational Question Answering Long-context Conversational Question Answering Long-horizon Question Answering

Follow for updates

Join SOTA2 Contact