4d ago · 11 min read · Here is the single most common mistake I see teams make when they start building with LLMs. Someone says "the model doesn't know about our stuff." A smart-sounding engineer nods and says "we should fine-tune it on our data." A project gets scoped. A ...
Join discussion
4d ago · 11 min read · Here is a question you have probably never thought to ask. When you talk to ChatGPT or Claude, and the thing on the other end is unfailingly polite, stays on task, refuses things that seem dangerous, and generally behaves like an assistant — where di...
Join discussion
Mar 30 · 6 min read · Last week, we were building an AI agent to answer questions about our company's internal documentation. The team split into two camps: one pushing for fine-tuning GPT-4 on our docs, the other advocating for a RAG system. Sound familiar? We see this d...
Join discussionMar 24 · 10 min read · Not all fine-tuning is the same. The method you pick determines your compute bill, your results, and whether your model forgets everything it already knew. Fine-tuning is not a single button you press
Join discussion
Feb 19 · 9 min read · Predicting whether a clinical trial will succeed or fail is notoriously difficult. Even experienced pharmaceutical analysts struggle with accuracy rates much better than a coin flip. But what if we co
Join discussion
Feb 11 · 11 min read · Prompt injections and their new rival Prompt injection turns “trusted inputs” (emails, web pages, retrieved docs) into an execution surface. If your agent can browse, read email, or call tools, a buried instruction can hijack actions. The dirty secre...
Join discussion
Feb 12 · 8 min read · If you trained a frontier LLM today the way we trained them in 2021—pretrain, do a little instruction tuning, ship—you’d get crushed in production. Not because the base model can’t write or reason, but because users don’t experience “capability”; the...
Join discussionFeb 10 · 3 min read · In my journey to dive deeper into multimodal AI systems, I decided to fine-tune BLIP-2, a powerful vision-language model trained on the Flickr8k dataset to generate image captions. What made this more exciting was integrating LoRA (Low-Rank Adaptatio...
Join discussion
Feb 2 · 9 min read · This blog post guides you through the process of fine-tuning the DeepSeek Math model to solve math problems expressed using emojis. This is a fun and engaging way to explore the capabilities of large language models (LLMs) and their ability to unders...
Join discussion
