Amal Zouaq

Payel Das

There is a growing interest in training domain-expert LLMs that excel in specific technical fields compared to their general-purpose instruc… (see more)tion-tuned counterparts. However, these expert models often experience a loss in their safety abilities in the process, making them capable of generating harmful content. As a solution, we introduce an efficient and effective merging-based alignment method called \textsc{MergeAlign} that interpolates the domain and alignment vectors, creating safer domain-specific models while preserving their utility. We apply \textsc{MergeAlign} on Llama3 variants that are experts in medicine and finance, obtaining substantial alignment improvements with minimal to no degradation on domain-specific benchmarks. We study the impact of model merging through model similarity metrics and contributions of individual models being merged. We hope our findings open new research avenues and inspire more efficient development of safe expert LLMs.

2024-11-11

ArXiv (preprint)

Combining Domain and Alignment Vectors to Achieve Better Knowledge-Safety Trade-offs in LLMs

Megh Thakkar

Yash More

Quentin Fournier

Matthew D Riemer

Pin-Yu Chen

Payel Das

There is a growing interest in training domain-expert LLMs that excel in specific technical fields compared to their general-purpose instruc… (see more)tion-tuned counterparts. However, these expert models often experience a loss in their safety abilities in the process, making them capable of generating harmful content. As a solution, we introduce an efficient and effective merging-based alignment method called \textsc{MergeAlign} that interpolates the domain and alignment vectors, creating safer domain-specific models while preserving their utility. We apply \textsc{MergeAlign} on Llama3 variants that are experts in medicine and finance, obtaining substantial alignment improvements with minimal to no degradation on domain-specific benchmarks. We study the impact of model merging through model similarity metrics and contributions of individual models being merged. We hope our findings open new research avenues and inspire more efficient development of safe expert LLMs.

2024-11-11

ArXiv (preprint)

Combining Domain and Alignment Vectors to Achieve Better Knowledge-Safety Trade-offs in LLMs

Megh Thakkar

Yash More

Quentin Fournier

Matthew D Riemer

Pin-Yu Chen

Payel Das

Chandar Research Lab

Mila - Québec

AI Institute

Ibm Research

Polytechnique Montréal

There is a growing interest in training domain-expert LLMs that excel in specific technical fields compared to their general-purpose instruc… (see more)tion-tuned counterparts. However, these expert models often experience a loss in their safety abilities in the process, making them capable of generating harmful content. As a solution, we introduce an efficient and effective merging-based alignment method called \textsc{MergeAlign} that interpolates the domain and alignment vectors, creating safer domain-specific models while preserving their utility. We apply \textsc{MergeAlign} on Llama3 variants that are experts in medicine and finance, obtaining substantial alignment improvements with minimal to no degradation on domain-specific benchmarks. We study the impact of model merging through model similarity metrics and contributions of individual models being merged. We hope our findings open new research avenues and inspire more efficient development of safe expert LLMs.

2024-10-10

NeurIPS.cc/2024/Workshop/AFM (poster)

openreview.net

A Deep Dive into the Trade-Offs of Parameter-Efficient Preference Alignment Techniques

Megh Thakkar

Quentin Fournier

Matthew D Riemer

Pin-Yu Chen

Payel Das

Large language models are first pre-trained on trillions of tokens and then instruction-tuned or aligned to specific preferences. While pre-… (see more)training remains out of reach for most researchers due to the compute required, fine-tuning has become affordable thanks to parameter-efficient methods such as LoRA and QLoRA. Alignment is known to be sensitive to the many factors involved, including the quantity and quality of data, the alignment method, and the adapter rank. However, there has not yet been an extensive study of their effect on downstream performance. To address this gap, we conduct an in-depth investigation of the impact of popular choices for three crucial axes: (i) the alignment dataset (HH-RLHF and BeaverTails), (ii) the alignment technique (SFT and DPO), and (iii) the model (LLaMA-1, Vicuna-v1.3, Mistral-7b, and Mistral-7b-Instruct). Our extensive setup spanning over 300 experiments reveals consistent trends and unexpected findings. We observe how more informative data helps with preference alignment, cases where supervised fine-tuning outperforms preference optimization, and how aligning to a distinct preference boosts performance on downstream tasks. Through our in-depth analyses, we put forward key guidelines to help researchers perform more effective parameter-efficient LLM alignment.

2024-06-07

ArXiv (preprint)

A Deep Dive into the Trade-Offs of Parameter-Efficient Preference Alignment Techniques

Megh Thakkar

Quentin Fournier

Matthew D Riemer

Pin-Yu Chen

Payel Das

2024-06-07

ArXiv (preprint)

A Deep Dive into the Trade-Offs of Parameter-Efficient Preference Alignment Techniques

Megh Thakkar

Quentin Fournier

Matthew D Riemer

Pin-Yu Chen

Payel Das

Large language models are first pre-trained on trillions of tokens and then instruction-tuned or aligned to specific preferences. While pre-… (see more)training remains out of reach for most researchers due to the compute required, fine-tuning has become affordable thanks to parameter-efficient methods such as LoRA and QLoRA. Alignment is known to be sensitive to the many factors involved, including the quantity and quality of data, the alignment method, and the adapter rank. However, there has not yet been an extensive study of their effect on downstream performance. To address this gap, we conduct an in-depth investigation of the impact of popular choices for three crucial axes: (i) the alignment dataset (HH-RLHF and BeaverTails), (ii) the alignment technique (SFT and DPO), and (iii) the model (LLaMA-1, Vicuna-v1.3, Mistral-7b, and Mistral-7b-Instruct). Our extensive setup spanning over 300 experiments reveals consistent trends and unexpected findings. We observe how more informative data helps with preference alignment, cases where supervised fine-tuning outperforms preference optimization, and how aligning to a distinct preference boosts performance on downstream tasks. Through our in-depth analyses, we put forward key guidelines to help researchers perform more effective parameter-efficient LLM alignment.

2024-06-07

ArXiv (preprint)

MVP: Minimal Viable Phrase for Long Text Understanding.

Louis Clouâtre

2024-01-01

International Conference on Language Resources and Evaluation (published)

dblp.uni-trier.de

Assessing the Generalization Capabilities of Neural Machine Translation Models for SPARQL Query Generation

Samuel Reyd

2023-10-27

The Semantic Web – ISWC 2023 (published)

SORBET: A Siamese Network for Ontology Embeddings Using a Distance-Based Regression Loss and BERT

Francis Gosselin

2023-10-27

The Semantic Web – ISWC 2023 (published)

SORBETmatcher results for OAEI 2023.

Francis Gosselin

2023-01-01

OM@ISWC (published)

dblp.uni-trier.de

Local Structure Matters Most: Perturbation Study in NLU

Louis Clouâtre

Prasanna Parthasarathi

Recent research analyzing the sensitivity of natural language understanding models to word-order perturbations has shown that neural models … (see more)are surprisingly insensitive to the order of words.In this paper, we investigate this phenomenon by developing order-altering perturbations on the order of words, subwords, and characters to analyze their effect on neural models’ performance on language understanding tasks.We experiment with measuring the impact of perturbations to the local neighborhood of characters and global position of characters in the perturbed texts and observe that perturbation functions found in prior literature only affect the global ordering while the local ordering remains relatively unperturbed.We empirically show that neural models, invariant of their inductive biases, pretraining scheme, or the choice of tokenization, mostly rely on the local structure of text to build understanding and make limited use of the global structure.

2022-05-01

Findings of the Association for Computational Linguistics: ACL 2022 (published)

Local Structure Matters Most: Perturbation Study in NLU

Louis Clouâtre

Prasanna Parthasarathi

Recent research analyzing the sensitivity of natural language understanding models to word-order perturbations has shown that neural models … (see more)are surprisingly insensitive to the order of words.In this paper, we investigate this phenomenon by developing order-altering perturbations on the order of words, subwords, and characters to analyze their effect on neural models’ performance on language understanding tasks.We experiment with measuring the impact of perturbations to the local neighborhood of characters and global position of characters in the perturbed texts and observe that perturbation functions found in prior literature only affect the global ordering while the local ordering remains relatively unperturbed.We empirically show that neural models, invariant of their inductive biases, pretraining scheme, or the choice of tokenization, mostly rely on the local structure of text to build understanding and make limited use of the global structure.

2021-07-29

ArXiv (preprint)