ThinkLLM, A knowledge graph of AI models (HTTPS://thinkllm.dev)
Hugging Face but organized by use case instead of architecture, with model comparisons.

Resource blocked by 403 error, so no code or paper is actually accessible to verify.
Machine learning researchers
Hugging Face but organized by use case instead of architecture, with model comparisons.
336× faster tree model inference; compiles sklearn/XGBoost to C99, serves like Ollama.
Wavelet-based attention-free architecture beats GPT-2 Medium with 80x less training data.
Multi-model debate on research hypotheses, but Z3 can't verify the actual claims.
Compile-time model capability validation beats runtime errors every time.
TikTok for papers sounds fun, but researchers need citations, not streaks.