FAR AI
FAR AI is an alignment research non-profit working to ensure AI systems are trustworthy and beneficial to society.
Pinned Repositories
epic
Implements the Equivalent-Policy Invariant Comparison (EPIC) distance for reward functions.
go_attack
gpt-4-novel-apis-attacks
KataGo-custom
Child repository of https://github.com/HumanCompatibleAI/go_attack.
KataGoVisualizer
learned-planners-stable-baselines3
PyTorch version of Stable Baselines, reliable implementations of reinforcement learning algorithms.
magicwormhole-docker
Dockerfile for Magic Wormhole
stable-baselines3
PyTorch version of Stable Baselines, reliable implementations of reinforcement learning algorithms.
tuned-lens
Tools for understanding how transformer predictions are built layer-by-layer
vlmrm
FAR AI's Repositories
AlignmentResearch/tuned-lens
Tools for understanding how transformer predictions are built layer-by-layer
AlignmentResearch/go_attack
AlignmentResearch/vlmrm
AlignmentResearch/gpt-4-novel-apis-attacks
AlignmentResearch/KataGo-custom
Child repository of https://github.com/HumanCompatibleAI/go_attack.
AlignmentResearch/KataGoVisualizer
AlignmentResearch/epic
Implements the Equivalent-Policy Invariant Comparison (EPIC) distance for reward functions.
AlignmentResearch/learned-planners-stable-baselines3
PyTorch version of Stable Baselines, reliable implementations of reinforcement learning algorithms.
AlignmentResearch/alpaca-lora
Instruct-tune LLaMA on consumer hardware
AlignmentResearch/farconf
Easy dataclass-based configuration for ML projects
AlignmentResearch/gogui
Graphical user interface for the game of Go, and other similar board games
AlignmentResearch/magicwormhole-docker
Dockerfile for Magic Wormhole
AlignmentResearch/pgx
A collection of highly-parallel RL game environments written in JAX
AlignmentResearch/stable-baselines3
PyTorch version of Stable Baselines, reliable implementations of reinforcement learning algorithms.
AlignmentResearch/ELF
ELF: a platform for game research with AlphaGoZero/AlphaZero reimplementation
AlignmentResearch/envpool
C++-based high-performance parallel environment execution engine (vectorized env) for general RL environments.
AlignmentResearch/gym-sokoban
Sokoban environment for Gym
AlignmentResearch/kueue
Kubernetes-native Job Queueing
AlignmentResearch/leela-zero
Go engine with no human-provided knowledge, modeled after the AlphaGo Zero paper.
AlignmentResearch/MambaLens
Mamba support for transformer lens
AlignmentResearch/mats_sae_training
Training Sparse Autoencoders on Language Models
AlignmentResearch/polygames
AlignmentResearch/sae-k-sparse-mamba
K-Sparse Autoencoders for Mamba
AlignmentResearch/SimpleParsing
Simple, Elegant, Typed Argument Parsing with argparse