Zobrazeno 1 - 10
of 611
pro vyhledávání: '"Van Durme, P."'
From Models to Microtheories: Distilling a Model's Topical Knowledge for Grounded Question Answering
Autor:
Weir, Nathaniel, Mishra, Bhavana Dalvi, Weller, Orion, Tafjord, Oyvind, Hornstein, Sam, Sabol, Alexander, Jansen, Peter, Van Durme, Benjamin, Clark, Peter
Recent reasoning methods (e.g., chain-of-thought, entailment reasoning) help users understand how language models (LMs) answer a single question, but they do little to reveal the LM's overall understanding, or "theory," about the question's topic, ma
Externí odkaz:
http://arxiv.org/abs/2412.17701
The decompose-then-verify strategy for verification of Large Language Model (LLM) generations decomposes claims that are then independently verified. Decontextualization augments text (claims) to ensure it can be verified outside of the original cont
Externí odkaz:
http://arxiv.org/abs/2412.13175
Autor:
Cheng, Jeffrey, Van Durme, Benjamin
Chain-of-thought (CoT) decoding enables language models to improve reasoning performance at the cost of high generation latency in decoding. Recent proposals have explored variants of contemplation tokens, a term we introduce that refers to special t
Externí odkaz:
http://arxiv.org/abs/2412.13171
Autor:
Chen, Tong, Fang, Hao, Xia, Patrick, Liu, Xiaodong, Van Durme, Benjamin, Zettlemoyer, Luke, Gao, Jianfeng, Cheng, Hao
Large language models (LMs) are typically adapted to improve performance on new contexts (\eg text prompts that define new tasks or domains) through fine-tuning or prompting. However, there is an accuracy compute tradeoff -- fine-tuning incurs signif
Externí odkaz:
http://arxiv.org/abs/2411.05877
Document retrieval for tasks such as search and retrieval-augmented generation typically involves datasets that are unstructured: free-form text without explicit internal structure in each document. However, documents can have a structured form, cons
Externí odkaz:
http://arxiv.org/abs/2410.20056
Autor:
Kriz, Reno, Sanders, Kate, Etter, David, Murray, Kenton, Carpenter, Cameron, Van Ochten, Kelly, Recknor, Hannah, Guallar-Blasco, Jimena, Martin, Alexander, Colaianni, Ronald, King, Nolan, Yang, Eugene, Van Durme, Benjamin
Efficiently retrieving and synthesizing information from large-scale multimodal collections has become a critical challenge. However, existing video retrieval datasets suffer from scope limitations, primarily focusing on matching descriptive but vagu
Externí odkaz:
http://arxiv.org/abs/2410.11619
The current paradigm for safety alignment of large language models (LLMs) follows a one-size-fits-all approach: the model refuses to interact with any content deemed unsafe by the model provider. This approach lacks flexibility in the face of varying
Externí odkaz:
http://arxiv.org/abs/2410.08968
Autor:
Sanders, Kate, Kriz, Reno, Etter, David, Recknor, Hannah, Martin, Alexander, Carpenter, Cameron, Lin, Jingyang, Van Durme, Benjamin
How are we able to learn about complex current events just from short snippets of video? While natural language enables straightforward ways to represent under-specified, partially observable events, visual data does not facilitate analogous methods
Externí odkaz:
http://arxiv.org/abs/2410.05267
Autor:
Jiang, Dongwei, Wang, Guoxuan, Lu, Yining, Wang, Andrew, Zhang, Jingyu, Liu, Chuyu, Van Durme, Benjamin, Khashabi, Daniel
The reasoning steps generated by LLMs might be incomplete, as they mimic logical leaps common in everyday communication found in their pre-training data: underlying rationales are frequently left implicit (unstated). To address this challenge, we int
Externí odkaz:
http://arxiv.org/abs/2410.01044
Autor:
Weller, Orion, Van Durme, Benjamin, Lawrie, Dawn, Paranjape, Ashwin, Zhang, Yuhao, Hessel, Jack
Instruction-tuned language models (LM) are able to respond to imperative commands, providing a more natural user interface compared to their base counterparts. In this work, we present Promptriever, the first retrieval model able to be prompted like
Externí odkaz:
http://arxiv.org/abs/2409.11136