RAG vs Fine-Tuning: Which AI Customization to Choose

RAG vs Fine-Tuning: Which AI Customization to Choose

💡 What You Will Learn

RAG vs Fine-Tuning: Which AI Customization to Choose

from langchain_community.vectorstores import Chroma
from langchain_community.embeddings import HuggingFaceEmbeddings

# 1. Load documents
docs = ["30", "20020010"]

# 2. 
embeddings = HuggingFaceEmbeddings(model_name="BAAI/bge-small-zh-v1.5")
vectorstore = Chroma.from_texts(docs, embeddings)

# 3. Retrieve + 
query = ""
results = vectorstore.similarity_search(query, k=2)
context = "\n".join([r.page_content for r in results])
prompt = f"\n{context}\n{query}"

git clone https://github.com/hiyouga/LLaMA-Factory.git
cd LLaMA-Factory

CUDA_VISIBLE_DEVICES=0 python src/train.py \
  --model_name_or_path Qwen/Qwen2.5-7B \
  --dataset train_data.json \
  --output_dir ./qwen-finetuned \
  --num_train_epochs 3 \
  --per_device_train_batch_size 1 \
  --learning_rate 2e-4

|:----|:---:|:----:|