[<Library component: Model|Core|etc...>] AutoModel Hyperparameter input_size_multiplier is wrong

[zllearn]

What happened + What you expected to happen

使用automodel模型时，其default_config提供的参数名input_size_multiplier错误，在实际使用时是input_size

Versions / Dependencies

neuralforecast库
python3.8 和 python3.10.9

Reproduction script

default_config = {
    "input_size_multiplier": [-1, 4, 16, 64],
    "inference_input_size_multiplier": [-1],
    "h": None,
    "encoder_hidden_size": tune.choice([50, 100, 200, 300]),
    "encoder_n_layers": tune.randint(1, 4),
    "context_size": tune.choice([5, 10, 50]),
    "decoder_hidden_size": tune.choice([64, 128, 256, 512]),
    "learning_rate": tune.loguniform(1e-4, 1e-1),
    "max_steps": tune.choice([500, 1000]),
    "batch_size": tune.choice([16, 32]),
    "loss": None,
    "random_seed": tune.randint(1, 20),
}

Issue Severity

Medium: It is a significant difficulty but I can work around it.

[marcopeix]

Hello! What is wrong exactly with the input_size_multiplier and what error do you see?

[JQGoh]

@zllearn 你有碰到什么报错吗？这是Neuralforecast 里采用的命名而已。实际上还是尝试不同input_size的参数来优化。

[zllearn]

您好！ 我最近较忙没有看见邮箱，不好意思！ 我之前反馈的问题是我自己看错了，我在传入AutoTimeMixerd的config参数时，给错了input_size的命名为input_size_multiplier 所以导致了后面的错误，后来改成了input_size就没有问题了 感谢您的回信！ 错误参数名： config_AutoTimeMixer = { &nbsp; &nbsp; "input_size_multiplier": tune.choice([INPUT_SIZE, INPUT_SIZE*2]),&nbsp; &nbsp; &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; "d_model": tune.choice([32, 64]),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; "e_layers": tune.choice([2, 4]),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "learning_rate": tune.loguniform(1e-4, 1e-2),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "batch_size": tune.choice([64, 128]),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; "max_steps": tune.choice([1500]), &nbsp; &nbsp; "random_seed": tune.randint(1, 10) } &nbsp; &nbsp;&nbsp; 正确参数名： config_AutoTimeMixer = { &nbsp; &nbsp; "input_size": tune.choice([INPUT_SIZE, INPUT_SIZE*2]),&nbsp; &nbsp; &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; "d_model": tune.choice([32, 64]),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; "e_layers": tune.choice([2, 4]),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "learning_rate": tune.loguniform(1e-4, 1e-2),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "batch_size": tune.choice([64, 128]),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; "max_steps": tune.choice([1500]), &nbsp; &nbsp; "random_seed": tune.randint(1, 10) } &nbsp; &nbsp;&nbsp; model = AutoTimeMixer( &nbsp; &nbsp; h=HORIZON,&nbsp; &nbsp; &nbsp; n_series=1,&nbsp; &nbsp; &nbsp; config=config_AutoTimeMixer,&nbsp; &nbsp; &nbsp; loss=MAE(),&nbsp; &nbsp; &nbsp; num_samples=20,&nbsp; &nbsp; &nbsp; verbose=False, &nbsp; &nbsp; ) 张乐 ***@***.*** &nbsp;

…

------------------ 原始邮件 ------------------ 发件人: "Nixtla/neuralforecast" ***@***.***&gt;; 发送时间:&nbsp;2024年12月24日(星期二) 凌晨0:18 ***@***.***&gt;; ***@***.******@***.***&gt;; 主题:&nbsp;Re: [Nixtla/neuralforecast] [<Library component: Model|Core|etc...&gt;] AutoModel Hyperparameter input_size_multiplier is wrong (Issue #1222) @zllearn 你有碰到什么报错吗？这是Neuralforecast 里采用的命名而已。实际上还是尝试不同input_size的参数来优化。 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***&gt;

[JQGoh]

@marcopeix You could resolve this as the reporter said that was the mistake not by Neuralforecast