AI fundamentals — Trust is earned, not givenhttps://www.bobhuang.com/series/ai-fundamentals/The groundwork under the AI series: how a model reads text, attention, a tiny GPT written in NumPy, scaling laws, inference, evaluation, and alignment.en10 Sep 2025 12:00:00 GMTAI fundamentals, part 8: From lab to product — context, tools, and failure modeshttps://www.bobhuang.com/blog/748/ai-fundamentals-building-products-with-llms/https://www.bobhuang.com/blog/748/ai-fundamentals-building-products-with-llms/10 Sep 2025 12:00:00 GMTThe series closer. Everything so far — tokenization, attention, training, scaling, sampling, evaluation, alignment — describes the model. Shipping it in a…AI fundamentals, part 7: Alignment — RLHF and why models behave (mostly)https://www.bobhuang.com/blog/747/ai-fundamentals-alignment-rlhf/https://www.bobhuang.com/blog/747/ai-fundamentals-alignment-rlhf/14 May 2025 12:00:00 GMTParts 1–6 produced a model that predicts text disturbingly well. But raw next-token prediction gives you an internet simulator, not an assistant: ask it a…AI fundamentals, part 6: Evaluation — measuring what models can do (and how benchmarks lie)https://www.bobhuang.com/blog/746/ai-fundamentals-evaluation-benchmarks/https://www.bobhuang.com/blog/746/ai-fundamentals-evaluation-benchmarks/22 Jan 2025 12:00:00 GMTYou've built, scaled and served a model (parts 1–5). Now the hardest question in applied AI: is it any good? This part covers what benchmarks measure, how…AI fundamentals, part 5: Inference — how an LLM writes one token at a timehttps://www.bobhuang.com/blog/745/ai-fundamentals-inference-sampling-kv-cache/https://www.bobhuang.com/blog/745/ai-fundamentals-inference-sampling-kv-cache/25 Sep 2024 12:00:00 GMTParts 1–4 built and scaled a model. Part 5 runs it: the generation loop, temperature and top-p sampling, the KV cache that makes it 100× faster, and why your…AI fundamentals, part 4: Scaling laws — the economics of intelligencehttps://www.bobhuang.com/blog/744/ai-fundamentals-scaling-laws-economics/https://www.bobhuang.com/blog/744/ai-fundamentals-scaling-laws-economics/19 Jun 2024 12:00:00 GMTPart 4 is the one practitioners should read twice: the mathematics that decides whether a model costs \$100 or \$100 million. Neural loss follows strikingly…AI fundamentals, part 3: Building a tiny GPT — the transformer block in NumPyhttps://www.bobhuang.com/blog/743/ai-fundamentals-building-a-tiny-gpt-numpy/https://www.bobhuang.com/blog/743/ai-fundamentals-building-a-tiny-gpt-numpy/14 Feb 2024 12:00:00 GMTParts 1–2 gave us tokens and attention. Now we assemble a complete GPT-style transformer block — embeddings, attention, the feed-forward network, residual…AI fundamentals, part 2: Attention — the idea that made transformers winhttps://www.bobhuang.com/blog/742/ai-fundamentals-attention-mechanism/https://www.bobhuang.com/blog/742/ai-fundamentals-attention-mechanism/30 Aug 2023 12:00:00 GMTPart 2 of the series. With text turned into token IDs, the model must decide which previous tokens matter for the next one. In "The key to the cabinet is…AI fundamentals, part 1: From bytes to BPE — how language models read texthttps://www.bobhuang.com/blog/741/ai-fundamentals-tokenization-bytes-to-bpe/https://www.bobhuang.com/blog/741/ai-fundamentals-tokenization-bytes-to-bpe/08 Jun 2023 12:00:00 GMTThis begins a series on how large language models actually work, written for programmers who know Python but not the internals of AI. We follow the same arc as…