Not every agent task needs a general LLM — specialized small models by task type
ziv_ravid · x · 2026-11-20
Ziv Ravid's thread (posts 10-11): model choice varies a lot by problem, and it doesn't have to be a general-purpose LLM. For classification problems, use a classification model or combine such models with a text encoder. There are also small models specializing in agentic tasks — a heavily invested direction right now — worth trying when your agent has a limited set of tasks; similar specialized options exist for vision.
More from coding & agent
- NormGym benchmark shows agents learn from own norm violations but not from others — xuanalogue · 2026-11-20
- Every adopted 'compound engineering' company-wide so its codebase gets easier to maintain — every · 2026-11-20
- Start custom LLMs at 30B: runs on a single GPU and works with LoRA — ziv_ravid · 2026-11-20
- Before fine-tuning, distill traces into skills in-context to test if training is worth it — ziv_ravid · 2026-11-20
- Fine-tuning is mostly iteration: cleaning data beats tuning hyperparameters — ziv_ravid · 2026-11-20
- Classification problems don't need a general LLM — use classifiers plus text encoders — ziv_ravid · 2026-11-20