Reformer matches standard Transformer accuracy on long sequences while using far less memory and running faster via LSH attention and reversible residual layers.
Large-scale Simple Question Answering with Memory Networks
8 Pith papers cite this work, alongside 565 external citations. Polarity classification is still indexing.
8
Pith papers citing it
565
external citations · Pith
abstract
Training large-scale question answering systems is complicated because training sources usually cover a small portion of the range of possible questions. This paper studies the impact of multitask and transfer learning for simple question answering; a setting for which the reasoning required to answer is quite easy, as long as one can retrieve the correct evidence given a question, which can be difficult in large-scale conditions. To this end, we introduce a new dataset of 100k questions that we use in conjunction with existing benchmarks. We conduct our study within the framework of Memory Networks (Weston et al., 2015) because this perspective allows us to eventually scale up to more complex reasoning, and show that Memory Networks can be successfully trained to achieve excellent performance.
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
TriviaQA is a new large-scale dataset for reading comprehension that features complex compositional questions, high lexical variability, and cross-sentence reasoning requirements, where current baselines reach only 40% while humans reach 80%.
The authors release the Wikidata Query Logs dataset containing 335k real question-query pairs constructed via an agent-based de-anonymization process from query service logs.
Introduces 2WikiMultiHopQA, a multi-hop QA dataset with explicit evidence chains generated via templates and Wikidata logical rules to force and evaluate multi-hop reasoning.
A GNN trained on bipartite alignment graphs between references and LLM generations reports state-of-the-art hallucination detection across four datasets, beating prior methods and GPT-4o.
KoRe compresses one-hop knowledge-graph subgraphs into 20 discrete tokens that, injected into Qwen3-8B, match or beat text-based knowledge injection on three QA benchmarks while using up to 10x fewer tokens.
KG-R1 trains a single RL agent to retrieve from and reason over knowledge graphs in one loop, achieving higher accuracy with fewer tokens than multi-module baselines and transferring to unseen graphs.
Fine-tuned models under 1B parameters reach micro-F1 of 0.83 on general-domain RE versus 0.69 for GPT-5.4 zero-shot, with similar gains on literary benchmarks.
citing papers explorer
-
Reformer: The Efficient Transformer
Reformer matches standard Transformer accuracy on long sequences while using far less memory and running faster via LSH attention and reversible residual layers.
-
TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension
TriviaQA is a new large-scale dataset for reading comprehension that features complex compositional questions, high lexical variability, and cross-sentence reasoning requirements, where current baselines reach only 40% while humans reach 80%.
-
The Wikidata Query Logs Dataset
The authors release the Wikidata Query Logs dataset containing 335k real question-query pairs constructed via an agent-based de-anonymization process from query service logs.
-
Constructing A Multi-hop QA Dataset for Comprehensive Evaluation of Reasoning Steps
Introduces 2WikiMultiHopQA, a multi-hop QA dataset with explicit evidence chains generated via templates and Wikidata logical rules to force and evaluate multi-hop reasoning.
-
Graph Alignment Topology as an Inductive Bias for Grounding Detection
A GNN trained on bipartite alignment graphs between references and LLM generations reports state-of-the-art hallucination detection across four datasets, beating prior methods and GPT-4o.
-
KoRe: Compact Knowledge Representations for Large Language Models
KoRe compresses one-hop knowledge-graph subgraphs into 20 discrete tokens that, injected into Qwen3-8B, match or beat text-based knowledge injection on three QA benchmarks while using up to 10x fewer tokens.
-
Efficient and Transferable Agentic Knowledge Graph RAG via Reinforcement Learning
KG-R1 trains a single RL agent to retrieve from and reason over knowledge graphs in one loop, achieving higher accuracy with fewer tokens than multi-module baselines and transferring to unseen graphs.
-
Sub-Billion, Super-Frontier: Small Language Models Rival Zero-Shot Frontier LLMs on General and Literary Relation Extraction
Fine-tuned models under 1B parameters reach micro-F1 of 0.83 on general-domain RE versus 0.69 for GPT-5.4 zero-shot, with similar gains on literary benchmarks.