Topic/Strategy/checkpoint-26250/plot.py

import json
import matplotlib.pyplot as plt
from pathlib import Path

# TODO: 换成你的 checkpoint 路径
ckpt_dir = Path(r"D:\Task_design\Topic\strategy_train\outputs\qwen7b-lora-topic_strategy\checkpoint-26250")
state_path = ckpt_dir / "trainer_state.json"

with open(state_path, "r", encoding="utf-8") as f:
    state = json.load(f)

log_history = state["log_history"]

train_steps, train_loss = [], []
eval_steps, eval_loss = [], []

for log in log_history:
    # 训练过程中的 loss 记录
    if "loss" in log and "step" in log:
        train_steps.append(log["step"])
        train_loss.append(log["loss"])
    # 在 dev/validation 上的 eval_loss 记录
    if "eval_loss" in log and "step" in log:
        eval_steps.append(log["step"])
        eval_loss.append(log["eval_loss"])

plt.figure()
if train_steps:
    plt.plot(train_steps, train_loss, label="train_loss")
if eval_steps:
    plt.plot(eval_steps, eval_loss, label="eval_loss")

plt.xlabel("step")
plt.ylabel("loss")
plt.title("Training / Eval Loss Curve")
plt.legend()
plt.grid(True)


plt.ylim(0, 10)
# 保存成图片
out_path = ckpt_dir / "loss_curve.png"
plt.savefig(out_path, dpi=200)
print(f"保存训练曲线到: {out_path}")