Perouz Taslakian

In-context learning (ICL) approaches typically leverage prompting to condition decoder-only language model generation on reference informati… (see more)on. Just-in-time processing of a context is inefficient due to the quadratic cost of self-attention operations, and caching is desirable. However, caching transformer states can easily require almost as much space as the model parameters. When the right context isn't known in advance, caching ICL can be challenging. This work addresses these limitations by introducing models that, inspired by the encoder-decoder architecture, use cross-attention to condition generation on reference text without the prompt. More precisely, we leverage pre-trained decoder-only models and only train a small number of added layers. We use Question-Answering (QA) as a testbed to evaluate the ability of our models to perform conditional generation and observe that they outperform ICL, are comparable to fine-tuned prompted LLMs, and drastically reduce the space footprint relative to standard KV caching by two orders of magnitude.

2024-04-23

ArXiv (preprint)

XC-Cache: Cross-Attending to Cached Context for Efficient LLM Inference

Jo˜ao Monteiro

Étienne Marcotte

Pierre-Andre Noel

Valentina Zantedeschi

David Vazquez

In-context learning (ICL) approaches typically leverage prompting to condition decoder-only language model generation on reference informati… (see more)on. Just-in-time processing of a context is inefficient due to the quadratic cost of self-attention operations, and caching is desirable. However, caching transformer states can easily require almost as much space as the model parameters. When the right context isn't known in advance, caching ICL can be challenging. This work addresses these limitations by introducing models that, inspired by the encoder-decoder architecture, use cross-attention to condition generation on reference text without the prompt. More precisely, we leverage pre-trained decoder-only models and only train a small number of added layers. We use Question-Answering (QA) as a testbed to evaluate the ability of our models to perform conditional generation and observe that they outperform ICL, are comparable to fine-tuned prompted LLMs, and drastically reduce the space footprint relative to standard KV caching by two orders of magnitude.

2024-04-23

ArXiv (preprint)

XC-Cache: Cross-Attending to Cached Context for Efficient LLM Inference

Jo˜ao Monteiro

Étienne Marcotte

Pierre-Andre Noel

Valentina Zantedeschi

David Vazquez

In-context learning (ICL) approaches typically leverage prompting to condition decoder-only language model generation on reference informati… (see more)on. Just-in-time processing of a context is inefficient due to the quadratic cost of self-attention operations, and caching is desirable. However, caching transformer states can easily require almost as much space as the model parameters. When the right context isn't known in advance, caching ICL can be challenging. This work addresses these limitations by introducing models that, inspired by the encoder-decoder architecture, use cross-attention to condition generation on reference text without the prompt. More precisely, we leverage pre-trained decoder-only models and only train a small number of added layers. We use Question-Answering (QA) as a testbed to evaluate the ability of our models to perform conditional generation and observe that they outperform ICL, are comparable to fine-tuned prompted LLMs, and drastically reduce the space footprint relative to standard KV caching by two orders of magnitude.

2024-01-01

EMNLP (Findings) (published)

Capture the Flag: Uncovering Data Insights with Large Language Models

Issam Hadj Laradji

Sai Rajeswar

Valentina Zantedeschi

Alexandre Lacoste

David Vazquez

Alexandre Drouin

The extraction of a small number of relevant insights from vast amounts of data is a crucial component of data-driven decision-making. Howev… (see more)er, accomplishing this task requires considerable technical skills, domain expertise, and human labor. This study explores the potential of using Large Language Models (LLMs) to automate the discovery of insights in data, leveraging recent advances in reasoning and code generation techniques. We propose a new evaluation methodology based on a"capture the flag"principle, measuring the ability of such models to recognize meaningful and pertinent information (flags) in a dataset. We further propose two proof-of-concept agents, with different inner workings, and compare their ability to capture such flags in a real-world sales dataset. While the work reported here is preliminary, our results are sufficiently interesting to mandate future exploration by the community.

2023-12-21

ArXiv (preprint)

Capture the Flag: Uncovering Data Insights with Large Language Models

Issam Hadj Laradji

Sai Rajeswar

Valentina Zantedeschi

Alexandre Lacoste

David Vazquez

Alexandre Drouin

The extraction of a small number of relevant insights from vast amounts of data is a crucial component of data-driven decision-making. Howev… (see more)er, accomplishing this task requires considerable technical skills, domain expertise, and human labor. This study explores the potential of using Large Language Models (LLMs) to automate the discovery of insights in data, leveraging recent advances in reasoning and code generation techniques. We propose a new evaluation methodology based on a"capture the flag"principle, measuring the ability of such models to recognize meaningful and pertinent information (flags) in a dataset. We further propose two proof-of-concept agents, with different inner workings, and compare their ability to capture such flags in a real-world sales dataset. While the work reported here is preliminary, our results are sufficiently interesting to mandate future exploration by the community.

2023-12-21

ArXiv (preprint)

Capture the Flag: Uncovering Data Insights with Large Language Models

Issam Hadj Laradji

Sai Rajeswar

Valentina Zantedeschi

Alexandre Lacoste

David Vazquez

Alexandre Drouin

2023-11-07

NeurIPS.cc/2023/Workshop/FMDM (published)

openreview.net

OC-NMN: Object-centric Compositional Neural Module Network for Generative Visual Analogical Reasoning

Rim Assouel

Pau Rodriguez

David Vazquez

Yoshua Bengio

2023-10-28

ArXiv (preprint)

OCIM : Object-centric Compositional Imagination for Visual Abstract Reasoning

Rim Assouel

Pau Rodriguez

David Vazquez

Yoshua Bengio

A long-sought property of machine learning systems is the ability to compose learned concepts in novel ways that would enable them to m… (see more)ake sense of new situations. Such capacity for imagination -- a core aspect of human intelligence -- is not yet attained for machines. In this work, we show that object-centric inductive biases can be leveraged to derive an imagination-based learning framework that achieves compositional generalization on a series of tasks. Our method, denoted Object-centric Compositional IMagination (OCIM), decomposes visual reasoning tasks into a series of primitives applied to objects without using a domain-specific language. We show that these primitives can be recomposed to generate new imaginary tasks. By training on such imagined tasks, the model learns to reuse the previously-learned concepts to systematically generalize at test time. We test our model on a series of arithmetic tasks where the model has to infer the sequence of operations (programs) applied to a series of inputs. We find that imagination is key for the model to find the correct solution for unseen combinations of operations.

2023-02-01

ICLR.cc/2023/Conference (rejected)

openreview.net

Explaining Graph Neural Networks Using Interpretable Local Surrogates

Farzaneh Heidari

Guillaume Rabusseau

We propose an interpretable local surrogate (ILS) method for understanding the predictions of black-box graph models. Explainability methods… (see more) are commonly employed to gain insights into black-box models and, given the widespread adoption of GNNs in diverse applications, understanding the underlying reasoning behind their decision-making processes becomes crucial. Our ILS method approximates the behavior of a black-box graph model by fitting a simple surrogate model in the local neighborhood of a given input example. Leveraging the interpretability of the surrogate, ILS is able to identify the most relevant nodes contributing to a specific prediction. To efficiently identify these nodes, we utilize group sparse linear models as local surrogates. Through empirical evaluations on explainability benchmarks, our method consistently outperforms state-of-the-art graph explainability methods. This demonstrates the effectiveness of our approach in providing enhanced interpretability for GNN predictions.

2023-01-01

TAG-ML (published)

proceedings.mlr.press

Using Graph Algorithms to Pretrain Graph Completion Transformers

Jonathan Pilault

Mikhail Galkin

Bahare Fatemi

David Vasquez

Recent work on Graph Neural Networks has demonstrated that self-supervised pretraining can further enhance performance on downstream graph, … (see more)link, and node classification tasks. However, the efficacy of pretraining tasks has not been fully investigated for downstream large knowledge graph completion tasks. Using a contextualized knowledge graph embedding approach, we investigate five different pretraining signals, constructed using several graph algorithms and no external data, as well as their combination. We leverage the versatility of our Transformer-based model to explore graph structure generation pretraining tasks (i.e. path and k-hop neighborhood generation), typically inapplicable to most graph embedding methods. We further propose a new path-finding algorithm guided by information gain and find that it is the best-performing pretraining task across three downstream knowledge graph completion datasets. While using our new path-finding algorithm as a pretraining signal provides 2-3% MRR improvements, we show that pretraining on all signals together gives the best knowledge graph completion results. In a multitask setting that combines all pretraining tasks, our method surpasses the latest and strong performing knowledge graph embedding methods on all metrics for FB15K-237, on MRR and Hit@1 for WN18RRand on MRR and hit@10 for JF17K (a knowledge hypergraph dataset).

2022-10-14

ArXiv (preprint)

Object-centric Compositional Imagination for Visual Abstract Reasoning

Rim Assouel

Pau Rodriguez