Topic Graph 기반 합성 데이터 파이프라인으로 저자원 언어 LLM 성능 최적화
Designing a Synthetic Data Pipeline for Persian LLM Fine Tuning: From Topic Graphs to QLoRA Evaluation
Designing a Synthetic Data Pipeline for Persian LLM Fine Tuning: From Topic Graphs to QLoRA Evaluation
Foundation vs. Instruct vs. Chat Models: One Question, Three Answers
Your CLAUDE.md Rules Are Probabilistic: Why Claude Quietly Deprioritizes Some Instructions
95. Fine-Tuning LLMs: Make a General Model Do Your Specific Job
LLM Model Names Decoded: A Developer's Guide to Parameters, Quantization & Formats
카카오가 Pre-training과 Post-training 사이에 Mid-training 단계를 도입하고 Pre-training 데이터를 50B 토큰 규모로 리플레이해 한국어 성능 저하를 방지하면서 수학 벤치마크 AIME24에서 9.21%에서 53.21%로 성능 향상
Visual Salamandra: Pushing the Boundaries of Multimodal Understanding
StarCoder2-Instruct: Fully Transparent and Permissive Self-Alignment for Code Generation
Instruction-tuning Stable Diffusion with InstructPix2Pix