Skip to content
The Chat Archive
llama-4-scout-17b-16e-instruct· Meta LMArena Human Preference 140K

outdir = "BasicMotionscsv" os.makedirs(outdir, existok=True) for classid, classsamples…

  1. User

    out_dir = "BasicMotions_csv" os.makedirs(out_dir, exist_ok=True)

    for class_id, class_samples in enumerate(data): for sample_id, sample_arr in enumerate(class_samples): fname = f"{out_dir}/class{class_id}_sample{sample_id}.csv" pd.DataFrame(sample_arr).to_csv(fname, index=False, header=False) print("已按样本拆分为多个 CSV") 我想知道如果我想保留原始形状的话,我应该修改下面这段代码的哪里,也就是说我要把上面这段代码加在什么位置: import pickle import os import pandas as pd import numpy as np

    1. 读取 pkl

    pkl_path = "shapeformer-main/store/BasicMotions_100.pkl" with open(pkl_path, 'rb') as f: data = pickle.load(f)

    2. 打平并构造 DataFrame

    records = [] for class_id, class_samples in enumerate(data): # 4 for sample_id, sample_arr in enumerate(class_samples): # 6 for time_step, row in enumerate(sample_arr): records.append( [class_id, sample_id, time_step] + row.tolist() )

    3. 生成 DataFrame

    columns = ['class_id', 'sample_id', 'time_step', 'col0', 'col1', 'col2', 'col3', 'col4', 'col5'] df = pd.DataFrame(records, columns=columns)

    4. 保存 CSV

    csv_path = "BasicMotions_100.csv" df.to_csv(csv_path, index=False) print(f"已导出 {len(df)} 行到 {csv_path}")

  2. Assistant

    如果你想在导出打平的 CSV 之外,还要保留原始形状的 CSV 文件,可以把以下代码块加在 保存 CSV 之后:

    out_dir = "BasicMotions_csv"
    os.makedirs(out_dir, exist_ok=True)
    
    for class_id, class_samples in enumerate(data):
        for sample_id, sample_arr in enumerate(class_samples):
            fname = f"{out_dir}/class{class_id}_sample{sample_id}.csv"
            pd.DataFrame(sample_arr).to_csv(fname, index=False, header=False)
    print("已按样本拆分为多个 CSV")
    

    加在 第 4 步之后 即可:

    import pickle
    import os
    import pandas as pd
    import numpy as np
    
    # 1. 读取 pkl
    pkl_path = "shapeformer-main/store/BasicMotions_100.pkl"
    with open(pkl_path, 'rb') as f:
        data = pickle.load(f)
    
    # 2. 打平并构造 DataFrame
    records = []
    for class_id, class_samples in enumerate(data):        # 4
        for sample_id, sample_arr in enumerate(class_samples):  # 6
            for time_step, row in enumerate(sample_arr):
                records.append(
                    [class_id, sample_id, time_step] + row.tolist()
                )
    
    # 3. 生成 DataFrame
    columns = ['class_id', 'sample_id', 'time_step', 'col0', 'col1', 'col2', 'col3', 'col4', 'col5']
    df = pd.DataFrame(records, columns=columns)
    
    # 4. 保存 CSV
    csv_path = "BasicMotions_100.csv"
    df.to_csv(csv_path, index=False)
    print(f"已导出 {len(df)} 行到 {csv_path}")
    
    # 5. 保存为单独的 CSV(保留原始形状)
    out_dir = "BasicMotions_csv"
    os.makedirs(out_dir, exist_ok=True)
    
    for class_id, class_samples in enumerate(data):
        for sample_id, sample_arr in enumerate(class_samples):
            fname = f"{out_dir}/class{class_id}_sample{sample_id}.csv"
            pd.DataFrame(sample_arr).to_csv(fname, index=False, header=False)
    print("已按样本拆分为多个 CSV")
    

    这样,你会得到:

    1. 一个 打平的 BasicMotions_100.csv,包含所有数据。
    2. 一个 目录 BasicMotions_csv,里面按类别和样本 ID 拆分,每个文件保存一个 ndarray。

    运行后检查输出目录,应该会有两个结果:

    • BasicMotions_100.csv(打平)
    • BasicMotions_csv/(保留形状)

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.