ezyang / ai-blindspotsLinks
Blindspots in LLMs I've noticed while AI coding. Sonnet family emphasis.
☆13Updated 4 months ago
Alternatives and similar repositories for ai-blindspots
Users that are interested in ai-blindspots are comparing it to the libraries listed below
Sorting:
- Pivotal Token Search☆119Updated 3 weeks ago
- ☆38Updated last year
- Editor with LLM generation tree exploration☆73Updated 6 months ago
- Approximating the joint distribution of language models via MCTS☆21Updated 9 months ago
- Because it's there.☆16Updated 10 months ago
- explore token trajectory trees on instruct and base models☆134Updated 2 months ago
- Synthetic data derived by templating, few shot prompting, transformations on public domain corpora, and monte carlo tree search.☆32Updated 5 months ago
- Lego for GRPO☆28Updated 2 months ago
- look how they massacred my boy☆63Updated 9 months ago
- never forget anything again! combine AI and intelligent tooling for a local knowledge base to track catalogue, annotate, and plan for you…☆37Updated last year
- papers.day☆91Updated last year
- ☆23Updated last month
- LLM Divergent Thinking Creativity Benchmark. LLMs generate 25 unique words that start with a given letter with no connections to each oth…☆33Updated 4 months ago
- Small, simple agent task environments for training and evaluation☆18Updated 9 months ago
- Turing machines, Rule 110, and A::B reversal using Claude 3 Opus.☆58Updated last year
- alternative way to calculating self attention☆18Updated last year
- Modify Entropy Based Sampling to work with Mac Silicon via MLX☆49Updated 9 months ago
- An easy-to-understand framework for LLM samplers that rewind and revise generated tokens☆146Updated 5 months ago
- LLMs represent numbers on a helix and manipulate that helix to do addition.☆25Updated 6 months ago
- anything you want can be built with morph cloud☆20Updated 3 months ago
- Pytorch script hot swap: Change code without unloading your LLM from VRAM☆126Updated 3 months ago
- ☆57Updated last month
- Benchmark that evaluates LLMs using 651 NYT Connections puzzles extended with extra trick words☆136Updated this week
- MLX port for xjdr's entropix sampler (mimics jax implementation)☆62Updated 9 months ago
- A framework for pitting LLMs against each other in an evolving library of games ⚔☆34Updated 3 months ago
- A tree-based prefix cache library that allows rapid creation of looms: hierarchal branching pathways of LLM generations.☆72Updated 6 months ago
- Nexusflow function call, tool use, and agent benchmarks.☆28Updated 7 months ago
- RAG Agent for the ARC AGI Challenge☆21Updated last year
- Lightweight toolkit package to train and fine-tune 1.58bit Language models☆83Updated 2 months ago
- rot13 version of claudd code☆42Updated 4 months ago