Skip to content

Instantly share code, notes, and snippets.

View lshm's full-sized avatar
💭
I may be slow to respond.

lshm

💭
I may be slow to respond.
View GitHub Profile
@lshm
lshm / grpo_demo.py
Created March 21, 2025 15:10 — forked from willccbb/grpo_demo.py
GRPO Llama-1B
# train_grpo.py
#
# See https://github.com/willccbb/verifiers for ongoing developments
#
import re
import torch
from datasets import load_dataset, Dataset
from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import LoraConfig
from trl import GRPOConfig, GRPOTrainer