#ai #aiagent #chatgpt #python #chatbot #llm #machinelearning
What are Hybrid RAG + Fine-tuning Systems?
Hybrid systems blend the benefits of RAG and fine-tuning in a single pipeline:
The retriever provides fast access to expansive, fresh external data.
Fine-tuning deeply specializes in model behavior for the domain.
The generator produces outputs utilizing both external context and fine-tuned domain knowledge.
Timestamp:
00:00: Understand the problem
00:33: Example (problem)
02:30: What is Fine-tuning
02:50: Why Fine-tuning The Model?
03:44: Supervised Fine-Tuning
04:30: Supervised Fine-Tuning limitations
05:30: Full Fine-Tuning
06:22: Parameter Efficient Fine-tuning
06:57: LoRA with example
07:57: QLoRA with example
09:13: Choose the open-source fine-tuning framework
10:15: Strengths framework
11:11: Fine-Tuning LLama3.1 - 8B
12:08: Data Preparation for Fine-tuning
13:29: Format Data
14:12: Model Configuration
14:55: LoRA Configuration :
15:45: Training Configuration :
16:50: Training Dataset :
17:20: Loading The Model
20:02: Test fine-tuned a model
20:52: LLM-as-a-judge
22:27: Evaluating using BERTScore
23:22: Hybrid RAG + Fine-tuning
24:03: integrate Hybrid RAG + Fine-tuning
25:28: Experiments and Results
27:16: Conclusion
FOLLOW ME :
Subscribe Newsletter for free: substack.com/@gaodalie
Follow me on Twitter: GaoDalie_AI
Follow me on Linkedin: shorturl.at/dnvEX
Follow me on Medium: medium.com/@GaoDalie_AI
More Ideas On My Page: quickaitutorial.com/
Buy me a Coffee: www.buymeacoffee.com/gaodalie98d
[Must Watch]
RAG Playlist: ua-cam.com/video/nbiPmETb9u4/v-deo.html