r"""Load value head parameters from Hugging Face Hub or local disk. Returns: dict with keys `v_head.summary.weight` and `v_head.summary.bias`.

Compute the log_probs from logits with rmpad input_ids and logits. Note that logits_rmpad = model(input_ids_rmpad). For each sentences, there is a

A memory efficient implementation of logprobs_from_logits

使用大语言模型生成基金分析摘要 参数: prompt (str): 输入提示词 返回: str: 模型生成的内容

Create a mapping function to process dataset examples. Args: split: Dataset split name ('train' or 'test') Returns: Function

Make minibatch iterator for updating the actor Args: data (DataProto): a DataProto containing keys ``input_ids``

Extracts function signature.

Exact match of math if and only if: 1. numerical equal: both can convert to float and are equal 2. symbolic equal: both can convert to sy

Exact match of math if and only if: 1. numerical equal: both can convert to float and are equal 2. symbolic equal: both can convert to sy

r"""Detach `tensors` (even if it's a nested list/tuple/dict of tensors).

提取并标准化选择题答案中的字母

Normalize the pp vpp params into a complete named parameters. This is useful when gather parameters from pp ranks and passed to a model witho

解析YAML格式的响应

r"""Expand 2d attention mask to 4d attention mask. Expand the attention mask with indices from (batch_size, seq_len) to (batch_size, 1, seq_len,

AllGather data from sp region This is because the data is first sharded along the FSDP dimension as we utilize the DP_COMPUTE

r"""Get arguments from the command line or a config file.

Register an agent with the environment

Register all tools with the server.

This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch, num_key_value_heads, seqlen, he

r"""Replace the default/reward modules in the model. The model is already unwrapped.

Run the battle environment with the given research report.

Execute the agent's main loop asynchronously. Args: request: Optional initial user request to process. Returns:

处理单个batch文件，返回其与所有相关规则的相关性判断列表。

r"""Save the training configuration to config path.

保存压缩后的数据集

将财务报表保存为CSV文件 Args: financial_statements (Dict): 包含财务报表的字典 stock_code (str): 股票代码，用于文件命名 market (str): 股票市场，"HK"

保存HTML报告

将数据保存到向量库中，会自动排除重复数据

r"""Set up a filesystem object based on the path protocol.

r"""EMA implementation according to TensorBoard.

spawn to a dictionary of worker groups, each with a subset of method with prefix.

Split the elements in a tuple/interval, while handling well-formatted commas in large numbers

Split the elements in a tuple/interval, while handling well-formatted commas in large numbers

Convert tool to function call format.

Create a tool message

Update the policy with an iterator of DataProto Args: data (DataProto): an iterator over the DataProto that returns by

verify the batch and save as ``acc`` tensor

将分块内容写入文件，并放入对应的子文件夹

Writes an iterable of dictionaries to jsonl

Adapts the dataset for evaluation using the specified model and parameters. Args: dataset: The input dataset to process

Compress the CoT outputs with LLMLingua-2.

Initialize the volcengine TTS client. Args: appid: Platform application ID access_token: Access token for au