Creates LoCoMo benchmark dataset for very long-term LLM conversational memory and shows current models struggle with lengthy dialogues and long-range temporal dynamics.
hub
BlenderBot 3: a deployed conversational agent that continually learns to responsibly engage
13 Pith papers cite this work, alongside 98 external citations. Polarity classification is still indexing.
13
Pith papers citing it
98
external citations · Pith
abstract
We present BlenderBot 3, a 175B parameter dialogue model capable of open-domain conversation with access to the internet and a long-term memory, and having been trained on a large number of user defined tasks. We release both the model weights and code, and have also deployed the model on a public web page to interact with organic users. This technical report describes how the model was built (architecture, model and training scheme), and details of its deployment, including safety mechanisms. Human evaluations show its superiority to existing open-domain dialogue agents, including its predecessors (Roller et al., 2021; Komeili et al., 2022). Finally, we detail our plan for continual learning using the data collected from deployment, which will also be publicly released. The goal of this research program is thus to enable the community to study ever-improving responsible agents that learn through interaction.
hub tools
citation-role summary
background 3
citation-polarity summary
roles
background 3representative citing papers
GAIA benchmark shows humans at 92% accuracy on simple real-world questions far outperform current AI systems at 15%, proposing this gap as a key milestone for general AI.
PMVisAgent uses multi-turn progressive interactions and a validation agent with ReAct-style verification to achieve up to 23.21% higher execution accuracy on the new PMVisBench dataset for text-to-vis tasks.
Optimus mitigates toxicity during LLM fine-tuning by combining repurposed LLM safety alignments for detection with synthetic data and DPO alignment, remaining effective even with highly biased classifiers and against attacks.
Chain-of-Verification reduces hallucinations in large language models by drafting responses, planning independent verification questions, answering them separately, and generating a final verified output.
Gorilla is a fine-tuned LLM that surpasses GPT-4 in accurate API call generation and uses retrieval to handle documentation updates.
CAMEL proposes a role-playing framework with inception prompting that enables autonomous multi-agent cooperation among LLMs and generates conversational data for studying their behaviors.
On 635 synthetic BOP applications, multi-agent Agentic RAG reaches 86.5% decision accuracy versus 77.6% single-LLM and 76.9% naive RAG, with largest gains on multi-step and missing-information cases.
DebiasRAG uses a three-stage RAG process to generate and rerank query-specific debiasing contexts that act as fairness constraints for LLM outputs.
A user study with 20 participants shows that closeness between sketches, annotations, and language in a shared space helps disambiguate multimodal queries, leading to the concept of proximity semantics for data exploration systems.
A survey classifying RAG foundations for AIGC, summarizing enhancements, cross-modal applications, benchmarks, limitations, and future directions.
StreamMUSE performs frame-synchronous streaming inference for language models by having a client send high-frequency requests and a server return outputs aligned to an external clock, shown on live music accompaniment with open-source code.
WhiteTesseract integrates XR diminished reality and LLM dialogue to increase viewing duration and interaction depth in physical cultural heritage exhibitions, shown in a 26-participant Monet exhibition study with statistically significant results.
citing papers explorer
-
Evaluating Very Long-Term Conversational Memory of LLM Agents
Creates LoCoMo benchmark dataset for very long-term LLM conversational memory and shows current models struggle with lengthy dialogues and long-range temporal dynamics.
-
GAIA: a benchmark for General AI Assistants
GAIA benchmark shows humans at 92% accuracy on simple real-world questions far outperform current AI systems at 15%, proposing this gap as a key milestone for general AI.
-
Towards Reliable Agentic Progressive Text-to-Visualization with Verification Rules
PMVisAgent uses multi-turn progressive interactions and a validation agent with ReAct-style verification to achieve up to 23.21% higher execution accuracy on the new PMVisBench dataset for text-to-vis tasks.
-
Optimus: A Robust Defense Framework for Mitigating Toxicity while Fine-Tuning Conversational AI
Optimus mitigates toxicity during LLM fine-tuning by combining repurposed LLM safety alignments for detection with synthetic data and DPO alignment, remaining effective even with highly biased classifiers and against attacks.
-
Chain-of-Verification Reduces Hallucination in Large Language Models
Chain-of-Verification reduces hallucinations in large language models by drafting responses, planning independent verification questions, answering them separately, and generating a final verified output.
-
Gorilla: Large Language Model Connected with Massive APIs
Gorilla is a fine-tuned LLM that surpasses GPT-4 in accurate API call generation and uses retrieval to handle documentation updates.
-
CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Society
CAMEL proposes a role-playing framework with inception prompting that enables autonomous multi-agent cooperation among LLMs and generates conversational data for studying their behaviors.
-
Agentic AI and Retrieval-Augmented Models in Straight-Through Underwriting
On 635 synthetic BOP applications, multi-agent Agentic RAG reaches 86.5% decision accuracy versus 77.6% single-LLM and 76.9% naive RAG, with largest gains on multi-step and missing-information cases.
-
DebiasRAG: A Tuning-Free Path to Fair Generation in Large Language Models through Retrieval-Augmented Generation
DebiasRAG uses a three-stage RAG process to generate and rerank query-specific debiasing contexts that act as fairness constraints for LLM outputs.
-
From 'Here' to 'There': Exploring Proximity Semantics in Multimodal Data Exploration
A user study with 20 participants shows that closeness between sketches, annotations, and language in a shared space helps disambiguate multimodal queries, leading to the concept of proximity semantics for data exploration systems.
-
Retrieval-Augmented Generation for AI-Generated Content: A Survey
A survey classifying RAG foundations for AIGC, summarizing enhancements, cross-modal applications, benchmarks, limitations, and future directions.
-
Real-Time Language Model Jamming: A Case Study for Live Music Accompaniment Generation
StreamMUSE performs frame-synchronous streaming inference for language models by having a client send high-frequency requests and a server return outputs aligned to an external clock, shown on live music accompaniment with open-source code.
-
WhiteTesseract: Reframing the Interpretation of Cultural Heritage through XR and Conversational AI
WhiteTesseract integrates XR diminished reality and LLM dialogue to increase viewing duration and interaction depth in physical cultural heritage exhibitions, shown in a 26-participant Monet exhibition study with statistically significant results.