SakanaAI

SakanaAI

Explained projects

SakanaAI/google-code-golf-2025: The Repo That Turns ARC Reasoning Into Byte-Count Combat
Benchmarks & Eval
SakanaAI/google-code-golf-2025: The Repo That Turns ARC Reasoning Into Byte-Count Combat
A deep dive into the judge, minifier, and prompt loop that squeeze correct solutions down to the smallest possible Python.
8 min read · Apr 7, 2026
RePo teaches LLMs to rearrange their own context
Local LLM Inference
RePo teaches LLMs to rearrange their own context
Sakana AI’s open-source stack turns token order into a learnable layer. The payoff is strongest when the input is messy, the signal is uneven, and rigid sequence order gets in the way.
10 min read · Apr 7, 2026
SakanaAI/sparser-faster-llms: How TwELL Makes Sparse Transformers Run Like They Were Built for Hopper
Local LLM Inference
SakanaAI/sparser-faster-llms: How TwELL Makes Sparse Transformers Run Like They Were Built for Hopper
A deep dive into the repo that couples activation sparsity with H100-tuned CUDA kernels, so the model’s zeros become real speed, lower memory use, and lower energy.
8 min read · Apr 7, 2026
SakanaAI/neuroevolution-for-ai: The README That Tries to Organize a Research Field
Benchmarks & Eval
SakanaAI/neuroevolution-for-ai: The README That Tries to Organize a Research Field
A curated Markdown map of labs, libraries, benchmarks, and learning resources shows how neuroevolution becomes usable when someone turns a field into a commons.
8 min read · Apr 7, 2026
TransEvalnia Makes Translation Scores Explain Themselves
Benchmarks & Eval
TransEvalnia Makes Translation Scores Explain Themselves
Sakana AI's evaluation framework asks an LLM to critique, compare, and rank translations before it ever emits a final verdict.
9 min read · Apr 7, 2026
DiffusionBlocks: When Transformer Depth Becomes a Noise Schedule
Local LLM Inference
DiffusionBlocks: When Transformer Depth Becomes a Noise Schedule
This research repo treats training as denoising, splits ViT into independently trainable blocks, and uses diffusion math to rethink the memory wall.
8 min read · Apr 7, 2026
Doc-to-LoRA Turns Documents Into Temporary Model Weights
Agent Memory & Skills
Doc-to-LoRA Turns Documents Into Temporary Model Weights
Sakana AI’s D2L skips the long prompt and generates LoRA adapters from the document itself.
10 min read · Apr 7, 2026
ike: Sakana AI’s Modular Shell Around DeepSpeed
Local LLM Inference
ike: Sakana AI’s Modular Shell Around DeepSpeed
A closer look at the framework that turns distributed LLM training into a plug-in architecture for data, loss functions, configs, and checkpoints.
8 min read · Apr 7, 2026
drq: When the benchmark fights back
Benchmarks & Eval
drq: When the benchmark fights back
Digital Red Queen uses an LLM to evolve Core War warriors in an adversarial loop, turning a retro programming game into a lab for open-ended adaptation.
10 min read · Apr 7, 2026
rl-razor-mnist: How a Tiny Task Flag Reveals Why RL Forgets Less
Benchmarks & Eval
rl-razor-mnist: How a Tiny Task Flag Reveals Why RL Forgets Less
A minimal replication of Sakana AI’s RL’s Razor experiment, where a single extra input bit, a KL-minimal oracle, and group-relative RL turn catastrophic forgetting into something you can inspect, compare, and predict.
8 min read · Apr 6, 2026
petri-dish-nca: When Neural Cellular Automata Start Competing for Space
3D Generative AI
petri-dish-nca: When Neural Cellular Automata Start Competing for Space
Sakana AI's PD-NCA turns growth into an ecosystem, where multiple learners fight, adapt, and survive inside one shared dish.
11 min read · Apr 6, 2026
DroPE: The LLM Context Trick That Works Better After You Drop the Map
Local LLM Inference
DroPE: The LLM Context Trick That Works Better After You Drop the Map
Sakana AI’s repo extends pretrained models by removing positional embeddings, then recalibrating just enough to unlock longer context windows without brute-force long-context fine-tuning.
12 min read · Apr 6, 2026
Shachi: A Laboratory for LLM Societies
Multi-Agent Frameworks
Shachi: A Laboratory for LLM Societies
Sakana AI’s framework turns brittle agent-based modeling into repeatable experiments, with tools, memory, and a second parsing pass when models return messy output.
11 min read · Apr 6, 2026
fast-weight-product-key-memory: FwPKM: The Memory Layer That Writes Back
Agent Memory & Skills
fast-weight-product-key-memory: FwPKM: The Memory Layer That Writes Back
Sakana AI’s open-source experiment fuses product-key lookup with fast-weight updates, so an LLM can retrieve from a huge memory and revise what it remembers mid-stream.
10 min read · Apr 6, 2026
robust-kbench: The benchmark that refuses to be gamed
Benchmarks & Eval
robust-kbench: The benchmark that refuses to be gamed
Sakana AI’s robust-kbench turns CUDA evaluation into a robustness test, not a leaderboard trick.
11 min read · Apr 6, 2026
IASC: The Compiler for Constructed Languages
Multi-Agent Frameworks
IASC: The Compiler for Constructed Languages
Sakana AI's open-source pipeline turns phonology, grammar, lexicon, and orthography into a reproducible language build, not a one-shot prompt.
8 min read · Apr 6, 2026
ShinkaEvolve: The Survival of the Fittest Code
Benchmarks & Eval
ShinkaEvolve: The Survival of the Fittest Code
Sakana AI's framework for autonomous discovery uses "Islands of Intelligence" and a ruthless Novelty Judge to evolve programs that outperform human engineers.
· Mar 28, 2026