Prof. Randall Balestriero - LLMs without pretraining and SSL

123 snips

Apr 23, 2025

Randall Balestriero, an AI researcher renowned for his work on self-supervised learning and geographic bias, explores fascinating findings in AI training. He reveals that large language models can perform well even without extensive pre-training. Randall also highlights the similarities between self-supervised and supervised learning, emphasizing their potential for improvement. Additionally, he discusses biases in climate models, demonstrating the risks of relying on their predictions, particularly for vulnerable regions, which has significant policy implications.

Ask episode

AI Snips

Chapters

Transcript

Episode notes

INSIGHT

LLMs Train Well From Scratch

Large language models (LLMs) with billions of parameters can be trained from random initialization on small supervised datasets without severe overfitting.
This challenges the idea that massive pre-training on huge data is always necessary for good performance on specific tasks.

INSIGHT

Pre-training Benefits Limited Tasks

For specific classification tasks, randomly initialized models trained from scratch can achieve performance comparable to costly pre-trained LLMs.
Pre-training mainly benefits open-ended generation tasks rather than all discriminative ones.

INSIGHT

Generalization vs Specialized Models

Task-specific models trained from scratch excel on narrow, in-distribution challenges but cannot generalize well out-of-distribution.
Pre-trained LLMs offer broader reasoning ability for open-ended or novel tasks beyond training data.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

Randall Balestriero joins the show to discuss some counterintuitive findings in AI. He shares research showing that huge language models, even when started from scratch (randomly initialized) without massive pre-training, can learn specific tasks like sentiment analysis surprisingly well, train stably, and avoid severe overfitting, sometimes matching the performance of costly pre-trained models. This raises questions about when giant pre-training efforts are truly worth it.

He also talks about how self-supervised learning (where models learn from data structure itself) and traditional supervised learning (using labeled data) are fundamentally similar, allowing researchers to apply decades of supervised learning theory to improve newer self-supervised methods.

Finally, Randall touches on fairness in AI models used for Earth data (like climate prediction), revealing that these models can be biased, performing poorly in specific locations like islands or coastlines even if they seem accurate overall, which has important implications for policy decisions based on this data.

SPONSOR MESSAGES:

***

Tufa AI Labs is a brand new research lab in Zurich started by Benjamin Crouzier focussed on o-series style reasoning and AGI. They are hiring a Chief Engineer and ML engineers. Events in Zurich.

Goto https://tufalabs.ai/

***

TRANSCRIPT + SHOWNOTES:

https://www.dropbox.com/scl/fi/n7yev71nsjso71jyjz1fy/RANDALLNEURIPS.pdf?rlkey=0dn4injp1sc4ts8njwf3wfmxv&dl=0

TOC:

1. Model Training Efficiency and Scale

[00:00:00] 1.1 Training Stability of Large Models on Small Datasets

[00:04:09] 1.2 Pre-training vs Random Initialization Performance Comparison

[00:07:58] 1.3 Task-Specific Models vs General LLMs Efficiency

2. Learning Paradigms and Data Distribution

[00:10:35] 2.1 Fair Language Model Paradox and Token Frequency Issues

[00:12:02] 2.2 Pre-training vs Single-task Learning Spectrum

[00:16:04] 2.3 Theoretical Equivalence of Supervised and Self-supervised Learning

[00:19:40] 2.4 Self-Supervised Learning and Supervised Learning Relationships

[00:21:25] 2.5 SSL Objectives and Heavy-tailed Data Distribution Challenges

3. Geographic Representation in ML Systems

[00:25:20] 3.1 Geographic Bias in Earth Data Models and Neural Representations

[00:28:10] 3.2 Mathematical Limitations and Model Improvements

[00:30:24] 3.3 Data Quality and Geographic Bias in ML Datasets

REFS:

[00:01:40] Research on training large language models from scratch on small datasets, Randall Balestriero et al.

https://openreview.net/forum?id=wYGBWOjq1Q

[00:10:35] The Fair Language Model Paradox (2024), Andrea Pinto, Tomer Galanti, Randall Balestriero

https://arxiv.org/abs/2410.11985

[00:12:20] Muppet: Massive Multi-task Representations with Pre-Finetuning (2021), Armen Aghajanyan et al.

https://arxiv.org/abs/2101.11038

[00:14:30] Dissociating language and thought in large language models (2023), Kyle Mahowald et al.

https://arxiv.org/abs/2301.06627

[00:16:05] The Birth of Self-Supervised Learning: A Supervised Theory, Randall Balestriero et al.

https://openreview.net/forum?id=NhYAjAAdQT

[00:21:25] VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning, Adrien Bardes, Jean Ponce, Yann LeCun

https://arxiv.org/abs/2105.04906

[00:25:20] No Location Left Behind: Measuring and Improving the Fairness of Implicit Representations for Earth Data (2025), Daniel Cai, Randall Balestriero, et al.

https://arxiv.org/abs/2502.06831

[00:33:45] Mark Ibrahim et al.'s work on geographic bias in computer vision datasets, Mark Ibrahim

https://arxiv.org/pdf/2304.12210