update readme system pipeline

README.md

@@ -49,24 +49,27 @@ import torch
 torch.cuda.empty_cache()
 torch.cuda.reset_peak_memory_stats()
 
-7/12
-# 1 Fine‑tune image model LoRA+QLoRA
+pipeline:
+# 1 Fully Fine‑tune image model with ZeRO
 accelerate launch --deepspeed_config_file=ds_config_zero3.json train_lora.py
 python train_lora.py
 
-# 2 SFT 语言模型
+# 2 SFT 120B OSS 语言模型 with QLoRA
 python sft_train.py
 
-# 3 Build RAG index
+# 3 distill 120B OSS模型给20B OSS模型
+用 Teacher 生成 Response，student模型用LoRA fine tuning
+
+# 4 Build RAG index
 python build_embeddings.py
 
-# 4 (可选) 收集偏好 → 训练 reward model 
+# 5 收集偏好 → 训练 reward model 
 python reward_model.py
 
-# 5 PPO RLHF 微调LLM
+# 6 PPO RLHF 微调20B OSS model
 python ppo_tune.py
 
-# 6 Inference with RAG
+# 7 Inference with RAG
 python rag_infer.py