Direct Preference Heads (Preprint) - a Avelina Collection

Avelina 's Collections

Direct Preference Heads (Preprint)

Direct Preference Heads (Preprint)

updated May 31

This collection contains the pre-trained, fine-tuned and aligned models for the Direct Preference Heads paper.

Avelina/lovelace-medium-alpha1

Text Generation • Updated May 31 • 14 • 1

Note Pretrained Transformer-XL model with 550M parameters, trained on 100B tokens from The Pile.
Avelina/lovelace-medium-alpha1-sft

Text Generation • Updated May 29 • 13
Avelina/lovelace-medium-alpha1-dph

Updated May 29 • 2 • 1
Would I Lie To You? Inference Time Alignment of Language Models using Direct Preference Heads

Paper • 2405.20053 • Published May 30 • 2