Skip to content

Instantly share code, notes, and snippets.

View ulysze's full-sized avatar
🌐

Ulysse É. O. Forest ulysze

🌐
  • Swiss Institute of Technology (EPFL)
  • Lausanne, Switzerland
View GitHub Profile
@ulysze
ulysze / grpo_demo.py
Created January 31, 2025 17:09 — forked from willccbb/grpo_demo.py
GRPO Llama-1B
# train_grpo.py
import re
import torch
from datasets import load_dataset, Dataset
from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import LoraConfig
from trl import GRPOConfig, GRPOTrainer
# Load and prep dataset