Victor Gallego's picture

Victor Gallego

vicgalle

·

https://github.com/vicgalle

vicgalle_

vicgalle

AI & ML interests

Preference fine-tuning, alignment & synthetic data. Building LLMs in general!

Organizations

vicgalle's activity

upvoted an article 10 days ago

Article

Can we create pedagogically valuable multi-turn synthetic datasets from Cosmopedia?

By

•

11 days ago

• 6

upvoted a paper 18 days ago

Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse Models

Paper • 2404.18796 • Published 19 days ago • 62

upvoted an article 22 days ago

Article

🦙⚗️ Using Llama3 and distilabel to build fine-tuning datasets

By

•

21 days ago

• 54

upvoted a collection about 1 month ago

Configurable Safety Tuning ⚙️

CST allows for configurable inference-time control of LLM safety levels, so users can dictate model behavior based on the system prompt • 7 items • Updated 16 days ago • 2

upvoted 2 papers about 2 months ago

Configurable Safety Tuning of Language Models with Synthetic Preference Data

Paper • 2404.00495 • Published Mar 30 • 2

Evolutionary Optimization of Model Merging Recipes

Paper • 2403.13187 • Published Mar 19 • 44

upvoted a paper 2 months ago

Large Language Models are Null-Shot Learners

Paper • 2401.08273 • Published Jan 16 • 1

upvoted a collection 2 months ago

Quantized Models (GGUF, IQ, Imatrix)

Various quantizations of models in the GGUF format. Models with a "checkmark" are personal favorites. An "orange arrow" means it's being uploaded. • 76 items • Updated 4 days ago • 32

upvoted a paper 2 months ago

GRATH: Gradual Self-Truthifying for Large Language Models

Paper • 2401.12292 • Published Jan 22 • 2

upvoted 4 papers 3 months ago

A Survey on Data Selection for LLM Instruction Tuning

Paper • 2402.05123 • Published Feb 4 • 3

Refined Direct Preference Optimization with Synthetic Data for Behavioral Alignment of LLMs

Paper • 2402.08005 • Published Feb 12 • 1

Suppressing Pink Elephants with Direct Principle Feedback

Paper • 2402.07896 • Published Feb 12 • 7

Vision Superalignment: Weak-to-Strong Generalization for Vision Foundation Models

Paper • 2402.03749 • Published Feb 6 • 9

upvoted a collection 3 months ago

🛰️🌍 Geospatial Datasets

A curated collections of diverse geospatial and satellite imagery datasets. • 54 items • Updated Mar 6 • 10

upvoted 4 collections 4 months ago

Exotic Frankenmerges 🥨

Merges of models of different architectures and sizes that end up working surprisingly well • 1 item • Updated Jan 21 • 1

Usable Models

5 items • Updated Dec 10, 2023 • 2

Upscaled Models ⏫

A collection of my frankenmerges, upscaling several models. All of them have the corresponding GGUF variants. • 4 items • Updated Jan 20 • 2

Model Merging

Model Merging is a very popular technique nowadays in LLM. Here is a chronological list of papers on the space that will help you get started with it! • 28 items • Updated Mar 23 • 180

upvoted a paper 5 months ago

Distilled Self-Critique of LLMs with Synthetic Data: a Bayesian Perspective

Paper • 2312.01957 • Published Dec 4, 2023 • 1

upvoted a collection 7 months ago

Optimised Translation Models 🌍

A collection of optimised and quantised multilingual translation models • 6 items • Updated Nov 7, 2023 • 3

upvoted 3 papers 9 months ago

Fast Adaptation with Bradley-Terry Preference Models in Text-To-Image Classification and Generation

Paper • 2308.07929 • Published Jul 15, 2023 • 1

Personalizing Text-to-Image Generation via Aesthetic Gradients

Paper • 2209.12330 • Published Sep 25, 2022 • 1

ZYN: Zero-Shot Reward Models with Yes-No Questions

Paper • 2308.06385 • Published Aug 11, 2023 • 1

upvoted a paper 10 months ago

LoraHub: Efficient Cross-Task Generalization via Dynamic LoRA Composition

Paper • 2307.13269 • Published Jul 25, 2023 • 29