Search

Trích xuất bảng từ ảnh/PDF thành Markdown cho RAG pipeline

Tokenizer + detokenizer for the Codec binary transport protocol — decode streaming token IDs from vLLM/SGLang, encode text into IDs for the bidirectional path. The Python twin of @codecai/web and Codec.Net.

3-tier distributed KV cache for LLM inference — preserve evicted KV across cluster nodes

Extract CO2 emissions data from PDF sustainability reports using LLMs

Automatic observability for LLM API calls