Interactive chat session with the agent

Run interactive chat mode with optional streaming

Run the agent in interactive mode

Run the agent in interactive mode Args: api_key: API key for the LLM provider provider: LLM provider to use mode

交互模式

Interactive mode for sending custom events

Run the agent in interactive mode with separated architecture

Kill process and all its children.

List directories that are accessible. Returns: Dictionary with allowed directories

List contents of a directory. Args: directory: Directory to list (relative to current) Retu

List directory contents with file sizes. Args: directory_path: Path to directory Returns:

List documents with pagination.

List documents in the store.

List all documents in the store. Args: limit: Maximum number of documents to return Returns:

List all available test cases. Args: category: Optional filter by category (layer1, layer2, layer3)

Load all workflows from storage into memory.

Load and extract content if in extract mode

Load and index conversation histories

加载并准备多语言推理数据集 使用 HuggingFaceH4/Multilingual-Thinking 数据集，该数据集包含： - 多种语言的推理链（思维链） - 支持英语、西班牙语、法语、德语、意大利语等 Returns:

加载基础模型和分词器 使用 Mxfp4Config 进行量化，这是专门为 OpenAI 模型优化的 4-bit 浮点格式。 Args: model_name: 模型名称或路径 Returns: t

Load the original Mistral model (before any training)

Load configuration from YAML file.

Load LOCOMO dataset

Load evaluation dataset

Load evaluation dataset

Load the dataset for voice context examples.

Load the dataset for voice context examples. Args: dataset_name: Name of the dataset split: Dataset split to use

Load experiences from file.

加载 GSM8K 数据集的金标答案 返回一个字典，键是数据集中的原始索引（0-1318），值是标准化后的答案

Load conversation history from storage

Load chunks from disk and re-index in retrieval pipeline Args: path: Path to load index from (uses config defaul

Load chunks from disk and re-index in retrieval pipeline Args: path: Path to load index from (uses config defaul

Load training data from JSONL file. Args: file_path: Path to JSONL file with messages format Returns: Datas

Load MCP configuration.

Load MCP tools asynchronously

Load MCP tools asynchronously

Load memory cards from storage

Load a trained LoRA model

Load the saved LoRA model

Load the base model and optionally apply LoRA adapters.

Load the base model and optionally apply LoRA adapters. Args: base_model_name: Name or path of the base model lora_path:

Load the fine-tuned model with LoRA adapters.

Load all results from the results directory Returns: Dictionary mapping experiment names to lists of rewards

Load texts from a JSON file. Expected format: [ {"text": "Hello world", "speaker_id": 0, "output": "hello.wav"}, {"t

加载训练好的模型进行推理 Args: base_model_name: 基础模型名称 peft_model_path: LoRA 权重路径 Returns: tuple: (model, token

Log document deletion.

Log embedding generation details.

Log index update operations.

Log the start of document indexing.

Log search results with detailed information.

Log the start of search operation.

Main function for standalone execution.

主函数：完整的训练流程

Example usage

Run a simple coding agent interaction

Entry point

Example usage

Main entry point.

Main entry point.

Main menu.

Main entry point.

Main entry point for demo.

Run benchmark comparison.

Run model leaderboard calculation. Args: method: 'bradley-terry' (default, official) or 'online-elo' (simple)

Main analysis function

Main entry point

Run a quick demo comparing correct vs incorrect implementation

Main entry point

Main function to run log sanitization Args: test_id: Specific test case ID to process (optional) limit: Maximum number o

Main entry point

Main entry point

Main entry point

Main entry point

Main function to start vLLM server

Main compatibility check