Personal AI Lab of Fernando Velasco Lozano — Senior Applied Data Scientist and AI Tech Lead. Deep-dives on LLMs, neuro-symbolic AI, agentic orchestration, and production ML.https://fermaat.github.io/https://fermaat.github.io/blog/au-net/https://fermaat.github.io/blog/au-net/A byte-level language modeling proposal where autoregressive U-Nets learn a hierarchical token representation — no tokenizer required.Fri, 27 Jun 2025 00:00:00 GMThttps://fermaat.github.io/blog/helmet/https://fermaat.github.io/blog/helmet/A benchmark system for evaluating LLMs across long contexts — summarization, long-doc QA, RAG, re-ranking, and generation with citations — beyond synthetic needle-in-a-haystack tests.Mon, 23 Jun 2025 00:00:00 GMThttps://fermaat.github.io/blog/titans/https://fermaat.github.io/blog/titans/A paper that proposes test-time learnable long-term memory for LLMs, with a 'Surprise' metric that decides what in the context is worth remembering.Fri, 20 Jun 2025 00:00:00 GMThttps://fermaat.github.io/blog/ctc-loss/https://fermaat.github.io/blog/ctc-loss/A hands-on look at Connectionist Temporal Classification — the soft-alignment loss that powers speech and OCR models, and the humble 'blank' label that makes it work.Mon, 16 Jun 2025 00:00:00 GMThttps://fermaat.github.io/blog/crnns-you-dont-always-need-a-transformer/https://fermaat.github.io/blog/crnns-you-dont-always-need-a-transformer/Why Convolutional Recurrent Neural Networks still matter for scene text recognition, handwriting transcription, and speech — even in the VLM era.Fri, 13 Jun 2025 00:00:00 GMThttps://fermaat.github.io/blog/lora-vs-full-fine-tuning/https://fermaat.github.io/blog/lora-vs-full-fine-tuning/A critical look at how LoRA fine-tuning actually differs from full fine-tuning — and the 'intruder dimensions' that formalize how LoRA silently forgets part of the pre-training.Wed, 11 Jun 2025 00:00:00 GMT