Outlines-core 0.1.0 正式發布：支援 Rust 與 Python 的高效能結構化生成庫

In LLM application development, ensuring that a model outputs content that 100% conforms to a specific format — such as a JSON Schema, a regular expression, or a Pydantic model — has always been a major challenge. Outlines, a leading open-source tool in the structured generation space, addresses this by dynamically adjusting token probabilities (token masking) during the model's decoding phase, guaranteeing that outputs absolutely conform to the specification. However, the core logic implemented in pure Python in earlier versions was prone to CPU bottlenecks when handling complex schemas or high-concurrency requests, increasing inference latency.