Picture for Frederic Sala

Frederic Sala

Weak-to-Strong Generalization Through the Data-Centric Lens

Add code
Dec 05, 2024
Viaarxiv icon

MoRe Fine-Tuning with 10x Fewer Parameters

Add code
Aug 30, 2024
Figure 1 for MoRe Fine-Tuning with 10x Fewer Parameters
Figure 2 for MoRe Fine-Tuning with 10x Fewer Parameters
Figure 3 for MoRe Fine-Tuning with 10x Fewer Parameters
Figure 4 for MoRe Fine-Tuning with 10x Fewer Parameters
Viaarxiv icon

Evaluating Language Model Context Windows: A "Working Memory" Test and Inference-time Correction

Add code
Jul 04, 2024
Viaarxiv icon

Is Free Self-Alignment Possible?

Add code
Jun 05, 2024
Viaarxiv icon

Pretrained Hybrids with MAD Skills

Add code
Jun 02, 2024
Figure 1 for Pretrained Hybrids with MAD Skills
Figure 2 for Pretrained Hybrids with MAD Skills
Figure 3 for Pretrained Hybrids with MAD Skills
Figure 4 for Pretrained Hybrids with MAD Skills
Viaarxiv icon

Pearls from Pebbles: Improved Confidence Functions for Auto-labeling

Add code
Apr 24, 2024
Figure 1 for Pearls from Pebbles: Improved Confidence Functions for Auto-labeling
Figure 2 for Pearls from Pebbles: Improved Confidence Functions for Auto-labeling
Figure 3 for Pearls from Pebbles: Improved Confidence Functions for Auto-labeling
Figure 4 for Pearls from Pebbles: Improved Confidence Functions for Auto-labeling
Viaarxiv icon

OTTER: Improving Zero-Shot Classification via Optimal Transport

Add code
Apr 12, 2024
Viaarxiv icon

Product Manifold Representations for Learning on Biological Pathways

Add code
Jan 27, 2024
Viaarxiv icon

Train 'n Trade: Foundations of Parameter Markets

Add code
Dec 07, 2023
Figure 1 for Train 'n Trade: Foundations of Parameter Markets
Figure 2 for Train 'n Trade: Foundations of Parameter Markets
Figure 3 for Train 'n Trade: Foundations of Parameter Markets
Figure 4 for Train 'n Trade: Foundations of Parameter Markets
Viaarxiv icon

The Cost of Compression: Investigating the Impact of Compression on Parametric Knowledge in Language Models

Add code
Dec 01, 2023
Figure 1 for The Cost of Compression: Investigating the Impact of Compression on Parametric Knowledge in Language Models
Figure 2 for The Cost of Compression: Investigating the Impact of Compression on Parametric Knowledge in Language Models
Figure 3 for The Cost of Compression: Investigating the Impact of Compression on Parametric Knowledge in Language Models
Figure 4 for The Cost of Compression: Investigating the Impact of Compression on Parametric Knowledge in Language Models
Viaarxiv icon