Prepare data for D3.js bar chart race animation. Args: history_df: DataFrame with columns: date, model, rating, rank top

Load model and tokenizer, optionally with LoRA. Args: model_name: Model name or path use_lora: Whether to use LoRA for e

配置 LoRA（低秩适应）进行高效微调 LoRA 只训练少量参数，大大减少内存使用和训练时间。 针对 openai/gpt-oss-20b 的 MoE（混合专家）架构，除了注意力层外， 还需要特别指定 MLP 专家层进行训练。 Args:

Prepare a test case by chunking and indexing its conversations Args: test_id: The test case ID to prepare

Prepare a test case by chunking and indexing its conversations Args: test_id: The test case ID to prepare

Print demo banner

打印欢迎横幅

Print current configuration (hiding sensitive data)

Pretty print confusion matrix and analysis.

Print a formatted table of results

Print results in a formatted table Args: results: List of test results

Print a colored section header

Print a colored section header

Print a formatted section header.

Print agent status

Print evaluation summary

Print evaluation summary

* Process audio chunk and detect voice activity using Silero VAD * @param {Buffer} audioChunk - Raw audio data * @returns {Promise<Array>} VAD r

* Process audio chunk through VAD * @param {Buffer} audioChunk - Raw audio data

* Process speech segment with STT * @param {Buffer} audioData - Speech audio data * @param {number} duration - Duration of speech in ms

Process all legal documents with contextual chunking

Process all legal documents

Process multiple conversation contexts in batch Args: conversation_contexts: List of conversation contexts

Process a document using contextual chunking with optional immediate indexing

Process Elo updates using vectorized NumPy operations with Numba JIT. This is the core hot loop optimized with Numba for maximum perform

Process a single file with the agent

Process all matches using vectorized NumPy operations and Numba JIT. This is the fastest way to compute Elo ratings for larg

Process all conversations in a test case

Provide recommendations based on system

Process query with contextual retrieval and detailed logging. This override adds educational logging to show the retrieval p

Query RAPTOR and/or GraphRAG indexes.

Quick integration test with real browser-use.

Quick test with simple task.

Run a quick test with context-aware citations strategy

Quick demonstration of the execution tools.

Run a quick demonstration of the memory system with separated architecture

Read and extract content from documents (PDF, DOCX, PPTX). Args: file_path: Path to the document file (or URL) extract_i

Read a legal document from disk

Read a legal document from disk

Read a file from current directory. Args: file_path: File path relative to current directory encodin

Read and extract content from webpage. Args: url: URL to read extract_links: Whether to extract links Retur

Build/rebuild the ANNOY index.

HNSW doesn't require explicit rebuild.

Register a background process for monitoring

Register a new tool

Recursively remove all description fields from a nested object Args: obj: The object to process Returns: Object

Request input from a human administrator. Args: prompt: Question or prompt for the admin input_type: Type of input expected (

Reset agent state

Reset all ratings to initial values.

Reset the agent's state

Reset the agent state

Reset the conversation history

Reset the conversation history

Admin response to an approval request. Args: request_id: ID of the request to respond to approved: Whether the request is app

Run agent with streaming output Args: user_message: User's input message max_iterations: Maximum num

Run the interactive evaluation session

Run the interactive session

* Run Silero VAD inference on audio chunk * @param {Float32Array} audioSamples - Normalized audio samples * @returns {Promise<number>} Speech pr

Run ablation study to test importance of context Args: api_key: API key for the LLM provider provider: LLM provider to u

Run ablation study across specified context modes Args: context_modes: List of context modes to test (defaults t

Run the agent with user input and display results

Run all ablation experiments and compare results

Run all compression strategies

Run experiments for all compression strategies

Run all tests.

Run all validation tests.

Run batch evaluation on multiple test cases.

Run batch queries from file

Run batch queries from file

Run comparison across all modes Args: api_key: API key for Kimi task: Custom task (optional) root_dir: Root dire

Run comparison between agentic and non-agentic modes

Run both modes and compare results

Run both modes and compare results

Run a complex multi-step task

Run the interactive demo Args: enable_streaming: Whether to enable streaming output (default: True)

Run a demonstration of all collaboration tools.

Run a comprehensive demo of the vector search service.

Run interactive demo

Run a demonstration evaluation with sample agent responses.

Run educational test cases.

Run all evaluation tests

Run full evaluation