Calculate global F1 accuracy score (field-level, micro-averaged) by counting all true positives, false negatives and false positives

Calculate the IoU of two lists of intervals. Each list contains intervals represented as [start, end].

Re-calculate model performance summary statistics across core and open tasks. Args: task_results: List of task results with scor

Check if the power expression can be computed. Parameters: expr (sympy expression): The expression to check. Return

Return True if the predicted denotation is correct. Args: target_values (list[Value]) predicted_values (list[Value]) Returns:

Transform a sequence of chess moves encoded in SAN into a set.

Collect task metadata for a model's results using the all_task_meta.json file

Convert a dictionary to a list of tuples, handling both simple and nested dictionaries. Args: data (dict): The input diction

Calculate keyword-based statistics for skills, input_format, and output_format.

Encode an image into VQ indices using the loaded VQ-VAE model.

Modify ``x`` in place to replace ``float("-inf")`` with the minimum value of the dtype when ``check_neg_inf`` is ``True`` and to replace ``fl

Modify ``x`` in place to replace ``float("-inf")`` with the minimum value of the dtype when ``check_neg_inf`` is ``True`` and to replace ``fl

The main entry point to evaluate all tasks in self.data based on the HF dataset’s metric info.

(默认 expression1 为 Ground_Truth) 函数: 判断两个表达式是否在数学意义上等价 步骤1: 提取表达式, 防止有的模型会给出"x=1"而不是"1" 步骤2: 使用 sympy 库进行等价判断

Convert Dictionary into Non-nested Dictionary Example: input(dict) { "menu": [ {"name

Function to generate all possible combinations of values from a dictionary.

MaskGit parallel decoding to generate VQ tokens Args: model: Model prompt: Prompt tensor seq_len: Sequence

The inner function to generate the answer. Returns: tuple(int, str, str): ret_code, response, log

Generate prompt for text-to-image generation Args: prompt_text: User input text prompt templates: Optional prompt t