Tag: LLM공부
All the articles with the tag "LLM공부".
-
Claude Vision — How Do You Send Images In, and Which Model Should Receive Them
I tried both ways of sending images — base64 encoding and URL — and found them accurate but more token-costly, hit limits with small text and complex shapes, and got a sense of when to move up to a bigger model for tasks that need accuracy.
-
Giving LLMs Tools — Claude Tool Use and the Agent Loop
LLMs don't know large-number arithmetic or today's weather. I tried out Tool Use, which patches that weakness with external function calls. Message flow, automatic multi-tool selection, and the agent loop.
-
Claude API Streaming — Why TTFT Matters, and How Events Actually Flow
Streaming isn't just a 'text appears letter by letter like an LLM' effect. Even with the same total time, the time until the user sees the first result (TTFT) differs, and that difference is the core of UX. I also traced the event order that delivers it, hands-on.
-
How System Prompts Differ from User Messages — Persona Consistency and Forcing JSON Output
A question I ran into while experimenting with personas in the Claude API: how is a system prompt actually different from just writing 'please do X' in a user message? I worked through this in terms of consistency, security, and output enforcement.