Verifiable Harbor tasks for training and evaluating small agents (shell, file editing, software engineering). Ready for RL with palingenesis.
🔄 In a Training Loop
Edoardo Federici
efederici
AI & ML interests
llms, ir, graphs & co, NAACL 2025
Recent Activity
liked a dataset 3 days ago
nvidia/Open-SWE-Traces upvoted a paper 5 days ago
Score Centering Stabilizes Off-policy Reinforcement Learning updated a collection 8 days ago
Harbor agent tasksOrganizations
-
Self-Rewarding Language Models
Paper • 2401.10020 • Published • 153 -
Exponentially Faster Language Modelling
Paper • 2311.10770 • Published • 119 -
Fine-tuning Language Models for Factuality
Paper • 2311.08401 • Published • 29 -
NEFTune: Noisy Embeddings Improve Instruction Finetuning
Paper • 2310.05914 • Published • 14
Verifiable Harbor tasks for training and evaluating small agents (shell, file editing, software engineering). Ready for RL with palingenesis.
-
Self-Rewarding Language Models
Paper • 2401.10020 • Published • 153 -
Exponentially Faster Language Modelling
Paper • 2311.10770 • Published • 119 -
Fine-tuning Language Models for Factuality
Paper • 2311.08401 • Published • 29 -
NEFTune: Noisy Embeddings Improve Instruction Finetuning
Paper • 2310.05914 • Published • 14