Mooler0410/LLMsPracticalGuide

Organizes LLM resources across model architectures (encoder-decoder vs. decoder-only), data pipeline stages (pretraining, finetuning, evaluation), and downstream NLP task categories, with an evolutionary tree mapping model lineages and relationships. Includes practical guidance on model selection, licensing restrictions for commercial/research use, and emerging concerns like safety alignment, truthfulness, and computational efficiency. Grounded in a peer-reviewed survey paper analyzing ChatGPT and successor models with direct links to implementation papers and blogs on production deployment considerations.

10,160 stars. No commits in the last 6 months.