Memory Injections: Correcting Multi-Hop Reasoning Failures during Inference in Transformer-Based Language Models.
Mansi SakarvadiaAswathy AjithArham KhanDaniel GrzendaNathaniel HudsonAndré BauerKyle ChardIan T. FosterPublished in: CoRR (2023)
Keyphrases
- language model
- multi hop
- language modeling
- wireless sensor networks
- wireless networks
- mobile ad hoc networks
- n gram
- data transmission
- energy efficient
- probabilistic model
- base station
- ad hoc networks
- document retrieval
- speech recognition
- end to end
- information retrieval
- retrieval model
- language modelling
- energy consumption
- query expansion
- routing protocol
- routing algorithm
- statistical language models
- wifi
- document ranking
- language models for information retrieval
- relevance model
- bayesian networks
- neural network
- test collection
- context sensitive
- sensor networks
- pseudo relevance feedback
- smoothing methods
- information retrieval systems
- wireless communication
- sensor nodes
- network topology