Returns: entropy: # (bs, response_len) log_probs: # (bs, response_len)

r"""Get the path to user config.

Get appropriate error message for agent

r"""Return the jinja template.

r"""Return the merged datasets in the standard format.

r"""Compute video sample indices according to fps.

Initialize session and populate tool map.

判断预测答案是否正确。 参数: predicted: 预测答案 reference: 参考答案 返回: 是否正确

r"""Start the training process.

Normalize answer expressions.

Normalize answer expressions.

r"""Build and validate the evaluation arguments.

r"""Parse eval example. input: a dict with keys {"question", "A", "B", "C", "D", "answer"} output: a tuple of (prompt, response).

r"""Build and validate the training arguments.

r"""Pre-process a single image.

r"""Preview the training commands.

设置matplotlib中文字体显示

预导入常用库

r"""Asynchronously get a list of responses of the chat model.

Execute decided actions

Add event to history and return the event

为Run中的图片添加灰色边框

Add multiple tools to the collection. If any tool has a name conflict with an existing tool, it will be skipped and a warning will be logged.

r"""Asynchronously get a list of scores of the reward model.

allgather params of all pp ranks. Return a list of handles

分析多个内容片段之间的一致性 参数: module_name: 模块名称 content_list: 待分析的内容片段列表 返回: 分析结果文本

分析结果并生成报告。 参数: save_path: 保存报告的路径，如果为None则不保存 返回: 包含分析结果的字典

r"""Asynchronously get the response token-by-token of the chat model.

构建基金分析提示词 参数: fund_code (int): 基金代码 生成: tuple: (提示词字符串, 日期)

计算准确率。 返回: 准确率

计算最终得分。 返回: 最终得分（所有问题COT长度总和的相反数）

r"""Calculate effective tokens per second.

Call Gemini LLM on Vertex AI with retry logic. Args: prompt: Text prompt to send to the model system_prompt: System instruction f

Process a query using the appropriate server based on tool name

Check correctness of code generation with a global timeout. The global timeout is to catch some extreme/rare cases not handled by the timeouts

清理过期报告

r"""Compute the sum log probabilities of the labels under given logits if loss_type is not IPO, ORPO or SimPO. Otherwise the average log prob

Connect to an MCP server using SSE transport.

r"""Return the number of trainable parameters and number of all parameters in the model.

This function should return a class instance that delegates the calls to every cls in cls_dict

Create GSM8K math problem prompts with token budget constraints. Args: budget: Token limit for response (default: 512)