Train the RL agent to discover the best vectorized symbolic expression. Parameters ---------- X : np.ndarray, shape

r"""Use a trained model to make predictions. Args: X (torch.Tensor): Data that needs to be predicted. Returns:

r"""Fit to data, then transform it. Args: X (array-like of shape (n_samples, n_features)): Input samples. y (array-li

r"""Load a model from the specified path with the given filename. Args: path: The location of the trained model. file

r"""Set the random seed for reproducibility of experiments. Args: seed: Random seed.

Test model saving and loading. Args: model: The model to test. input_args: Input tensor(s) to pass to forward. hx_ar

r"""Update the plots for loss and accuracy across epochs. Args: epoch: Epochs during training. loss: Loss value durin

初始化 LLM 客户端。 :param api_key: API 密钥 :param model: 模型名称 :param base_url: API 的基础 URL :param verbose: 控制输出详细程度

Generic save-load test. Args: model: The model to test. args: Positional arguments to pass to forward. kwargs: Keyw

make resnet layers(by layer i didnt mean this 'layer' was the same as a neuron netowork layer, ex. conv layer), one layer may contain

Return a function that converts input to a tuple of length n.

r"""Constructs the neural network model based on the specified architecture. Args: input_dim: The input dimension of the network.

Clean up control characters in LLM output that would cause the Python parser to crash directly.

r"""Save the current model to the specified path with the given filename. Args: path: The path where the model weights are saved.

r"""Perform standardization by centering and scaling. Args: X ({array-like, sparse matrix} of shape (n_samples, n_features)): The

r"""Method to select a random node within a given tree structure. Args: selected: The selected node. parent: The pare

Safely evaluate a basis function on a 1D torch tensor. Returns a numpy array of shape (n_samples,).

Parses a symbolic expression and returns a list of callable basis functions. The expression is split at the top‑level '+' and '-' operators

Yield each token with a bool indicating whether it is a function call.

r"""Constructs the neural network model based on the specified architecture. Args: input_dim: The input dimension of the network.

r"""Method to evaluate an expression using input data. Args: expr: Symbolic expression. x_data: Input data for evalua

r"""Get the corresponding PyTorch activation function by name. Args: name: A string name of the activation function. Returns:

r"""Get the corresponding PyTorch loss function by name. Args: name: A string name of the loss function. Returns: The corres

r"""Get the corresponding PyTorch optimizer by name. Args: name: The name of the optimizer (e.g., 'adam', 'sgd'). parameters: The

r"""Method to select a random parent from a population based on fitness. Args: popu: Total population. fitne: Fitness

r"""Generate a random coefficient within a specified range.

r"""Method to generate a random program/tree structure. Returns: A symbol tree structure.

r"""Update the plots for loss and accuracy across epochs. Args: loss: Loss value during training.

r"""Method to simplify a tree-based expression into a string. Uses sympify and expand functions to simplify and convert the expression.

Augments the input by applying all basis functions and concatenating.

Build a human‑readable expression string in the format: intercept + c1@phi1 + c2@phi2 + ... where phi_k are the basis fun

Subclasses override this method to call the corresponding F.conv1d/2d/3d.

Subclasses override this method to call the corresponding F.conv_transpose1d/2d/3d.

Evaluate a set of selected basis functions in a vectorized (homogeneous) way. For each selected basis function phi_k, we compute:

从原始符号表达式中提取所有 '@' 之后的变量表达式（丢弃系数）。 例如 "2@x**3 + 3@torch.sin(x) + x" -> ['x**3', 'torch.sin(x)', 'x']

从内存中的 database 获取最佳方程体（用于 save=False 时）

Parse a symbolic expression to extract polynomial and sine terms.

Extract all sub‑expressions containing 'x', ignoring pure constant terms.

Sample up to max_terms basis functions (with replacement) and remove duplicates.

按顶层加减号分割表达式，正确保留括号内的符号。

Split expression by top-level '+' and '-' while respecting parentheses.

Transform `code` into Python tokens.

Transform a list of Python tokens into code.