draw_wer.py

import json
from json import JSONDecodeError
import matplotlib.pyplot as plt

jsonl_path = "/workspace/intern/pangkaiyu/Kimi-Audio/Kimi-Audio-Evalkit/lora_-5_to_10__7B/Qwen2.5-Omni-7B/kimi_10k_noise_-5_to_10_linear_val5_abs/Qwen2.5-Omni-7B_kimi_10k_noise_-5_to_10_linear_val5_abs_wer_details.jsonl"  # 改成你的文件
out_png = "/workspace/intern/pangkaiyu/Kimi-Audio/Kimi-Audio-Evalkit/pic/-5_to_10_val5.png"            # 保存路径

dec = json.JSONDecoder()

def iter_json_from_line(line: str):
    line = line.strip()
    if not line:
        return
    try:
        yield json.loads(line)
        return
    except JSONDecodeError as e:
        # 典型：一行里粘了多个 JSON（Extra data）
        i, n = 0, len(line)
        while i < n:
            while i < n and line[i].isspace():
                i += 1
            if i >= n:
                break
            obj, j = dec.raw_decode(line, i)
            yield obj
            i = j

xs, ys = [], []
cnt_obj = 0

with open(jsonl_path, "r", encoding="utf-8") as f:
    for ln, line in enumerate(f, 1):
        try:
            for obj in iter_json_from_line(line):
                cnt_obj += 1
                x = obj.get("index", cnt_obj)

                y = obj.get("utt_wer", None)
                if y is None:
                    y = (obj.get("wer_details") or {}).get("utt_wer", None)
                if y is None:
                    continue

                y = float(y)
                # 兼容：0~1 or 0~100
                if y <= 1.0:
                    y *= 100.0

                xs.append(x)
                ys.append(y)
        except Exception as e:
            # 定位到底是哪一行坏了
            print(f"[ERROR] line {ln} parse failed: {repr(e)}")
            print("line snippet:", repr(line[:200]))
            raise

print(f"parsed {cnt_obj} json objects, plotted {len(xs)} points")

plt.figure()
plt.scatter(xs, ys, s=8)
plt.xlabel("index")
plt.ylabel("utt_wer (%)")
plt.ylim(0, 120)
plt.title("Per-utterance WER scatter")
plt.savefig(out_png, dpi=300, bbox_inches="tight")
plt.close()

print("saved to:", out_png)