https://monostate.ai/blog Insights on AI training, model optimization, and AGI research from the Monostate team. Learn about no-code AI, entropy-based learning, and practical machine learning. en-us Thu, 09 Oct 2025 00:00:00 GMT https://monostate.ai/blog/training https://monostate.ai/blog/training Mon, 02 Sep 2024 00:00:00 GMT Andrew Correa A reproducible pipeline for compact LLMs: search and curate datasets on Hugging Face, upload seed examples, generate synthetic data on demand, run LoRA training with live W&B, publish to the Hub, and export GGUF for on-device use. https://monostate.ai/blog/entropy-refinement-blog https://monostate.ai/blog/entropy-refinement-blog Thu, 15 Aug 2024 00:00:00 GMT Andrew Correa and Ana Carolina H. Matos We discovered how to make language models fix their own mistakes by showing them exactly where they're uncertain - achieving reasoning-model quality at a fraction of the cost. https://monostate.ai/blog/child-benchmark-agi https://monostate.ai/blog/child-benchmark-agi Mon, 20 May 2024 00:00:00 GMT Monostate Research Team Why AI that beats humans at chess can't answer 'What does chocolate taste like?' properly. Introducing a benchmark that reveals the kindergarten-level reasoning gaps in frontier models. https://monostate.ai/blog/hallucinations-entropy-llms https://monostate.ai/blog/hallucinations-entropy-llms Thu, 28 Mar 2024 00:00:00 GMT Andrew Correa Why language models make stuff up, what entropy has to do with it, and how companies are trying to fix it.