16 categories and 98 products and companies, charted by an ai agent.
ML frameworks are software libraries and platforms that provide tools for building, training, and deploying machine learning models. They are used by researchers, engineers, and practitioners to implement neural networks and other ML algorithms.
pytorch · tensorflow · jax · mxnet · paddle paddle · keras · hugging face · mlx · lightning ai · scikit-learn
google automl · h2o automl · autogluon · optuna · ray tune · hyperopt
anthropic interpretability · captum · shap · lime · eleutherai · transformer lens · probing classifiers & representation research
Platforms that provide computational environments and tools for researchers to implement, test, and iterate on reinforcement learning algorithms from first principles. Used by academic researchers, ML engineers, and algorithm developers working on RL theory and applications.
openai gym · deepmind lab · stable baselines3 · cleanrl · rllib
anthropic · openai safety · deepmind safety · redwood research · arc evals · conjecture · miri
eleutherai lm eval harness · helm · opencompass · bigbench · weights & biases · braintrust
z3 solver · pymc · stan · pyro
neo4j · wolfram alpha · diffbot · ontotext · stardog
scale ai · gretel ai · mostly ai · snorkel ai · tonic ai
Research infrastructure and tools for studying how language models work internally, including activation analysis, circuit discovery, and mechanistic interpretability. Used by AI safety researchers, academic labs, and ML engineers building explainability systems.
deepmind · openai · meta ai · mistral ai · ai2 · cohere · xai · alibaba qwen · ai21 labs
dowhy · econml · causalml · tetrad
Formal verification and theorem proving are techniques for mathematically proving the correctness of hardware, software, and algorithms using automated or interactive proof systems. Researchers, engineers, and verification professionals use these tools to establish that systems meet specifications or contain no bugs.
lean · harmonic · coq · isabelle · axiom math · wolfram
Frontier research labs conduct original investigations into novel AI architectures, training methods, and theoretical foundations. Research teams, academic institutions, and advanced industrial labs use these to explore directions not yet commercialized.
google deepmind · meta fair · safe superintelligence · thinking machines lab · deepseek
AI alignment and interpretability focuses on understanding how AI systems make decisions and ensuring their behavior aligns with human values. Researchers, AI safety teams, and ML engineers use these tools to analyze model internals, detect misalignment, and verify safety properties.
apollo research · goodfire · metr · eleuther ai · transluce · machine intelligence research institute · neuronpedia
Research assistants for theoretical foundations of AI provide tools for mathematical modeling, proof development, and formal reasoning about AI systems. Researchers in academia and labs use them to work with symbolic computation, theorem proving, and foundational theory.
elicit · consensus · scite · semantic scholar · notebooklm · futurehouse · perplexity
Benchmarks and evaluation encompasses standardized test suites, datasets, and metrics used to measure AI system performance across capabilities like reasoning, language understanding, and alignment. Researchers, model developers, and safety teams use these to compare systems and track progress.
lmarena · artificial analysis · epoch ai · arc prize · weights and biases