模型微调与适配
What is fine-tuning, and when should you fine-tune an LLM?
原题Explain the difference between full fine-tuning and parameter-efficient fine-tuning (PEFT).
原题What is LoRA (Low-Rank Adaptation), and how does it work?
原题What is QLoRA, and how does it enable fine-tuning on consumer hardware?
原题How does fine-tuning work?
原题Explain Prefix Tuning and Prompt Tuning. How are they different from LoRA?
原题What is adapter-based fine-tuning?
原题Explain the difference between pre-training, supervised fine-tuning (SFT), and preference optimization (RLHF/DPO).
原题What is RLHF (Reinforcement Learning from Human Feedback), and how is it used to align LLMs?
原题What is Deep RL from Human Preferences, the paper that started RLHF?
原题Why did DPO displace PPO-based RLHF at many labs? When is online RL still better?
原题What is instruction tuning, and why is it important for chat models?
原题How do you prepare a dataset for fine-tuning an LLM?
原题What is catastrophic forgetting, and how do you prevent it during fine-tuning?
原题When should you choose fine-tuning over RAG over prompt engineering?
原题How do you evaluate a fine-tuned model's performance?
原题What is synthetic data generation, and how do you use it for fine-tuning?
原题What are the key hyperparameters for fine-tuning (learning rate, epochs, batch size, LoRA rank)?
原题Do the GPU memory math for full fine-tuning a 7B model in bf16 with Adam. Now with LoRA and QLoRA.
原题How do you fine-tune a model for a specific domain (legal, medical, finance)?
原题What is continual pre-training, and when would you use it?
原题How do you merge multiple LoRA adapters?
原题What is the difference between SFT (Supervised Fine-Tuning) and alignment training?
原题What is RLAIF (RL from AI Feedback), and how does it differ from RLHF?
原题What is Constitutional AI, and how does it differ from RLHF?
原题What is RLVR (Reinforcement Learning with Verifiable Rewards), and when does it beat a learned reward model?
原题What is knowledge distillation for fine-tuning, and what are the legal considerations?
原题Your fine-tuned LLM produces factually wrong outputs due to training data quality issues. How do you fix it?
原题You must choose between LoRA and full fine-tuning for a domain-specific assistant. How do you decide?
原题Your fine-tuned model memorized training data verbatim instead of learning patterns. How do you fix overfitting?
原题Your fine-tuned LLM forgot its general capabilities after domain-specific fine-tuning. How do you fix catastrophic forgetting?
原题Your RLHF preference data has low annotator agreement. How do you ensure data quality?
原题
暂无题目