Tags
A tag marks what a post is about: a model family, a technique, or a failure mode.
Evaluating and benchmarking model output.
Function/tool calling and tool design.
Designing, testing, and refining prompts.
Reducing token spend and latency.
Anthropic's Claude family of models.
OpenAI GPT models.
JSON mode, schemas, and parsing.
Detecting and reducing fabricated output.
Retrieval-augmented generation.
Long context, chunking, and context management.
Vector embeddings and semantic search.
Autonomous and semi-autonomous LLM agents.
Fine-tuning and post-training.
Google Gemini models.
Text-to-image models and workflows.
Reproducing and remixing visual styles.
Reusable packaged agent skills.
Model Context Protocol servers and clients.
System prompt design and behavior steering.
Vision, audio, and mixed-modality models.