Build an LLM from scratch with transformers, attention, tokenization, dataset engineering, training, alignment, inference, multimodality, agents, and a runnable PyTorch mini language model.
-
Updated
Aug 2, 2026 - Python
Build an LLM from scratch with transformers, attention, tokenization, dataset engineering, training, alignment, inference, multimodality, agents, and a runnable PyTorch mini language model.
AI Training Data infrastructure, RLHF systems, and dataset engineering for high-performance AI models.
金融工单 Agent 的数据工程与后训练实验:数据/意图/工具/政策四张注册表 → SFT / DPO / tool-use SFT 导出 → Qwen2.5-3B LoRA;held-out 政策评测 88/91,加推理时 guard 后 91/91。
🌱 Community-driven dataset projects for inclusive, safe, and open AI engineering.
Self-study notes from Chip Huyen's AI Engineering (ch.1-10) + an internal talk on dataset engineering for LLMs
YOLOv8 road-object detection case study covering domain shift, dataset engineering, held-out evaluation, and real-world Turkish traffic scenes.
To associate your repository with the dataset-engineering topic, visit your repo's landing page and select "manage topics."