Changelog

[0.3.0] - 2026-10-09 - First public release

• GPU BPE encoder captured as one CUDA graph per power-of-two input-size class: GPU pre-tokenization, warp-cooperative BPE merge with an overflow path for long pre-tokens, and assembly.
• Tokenizers: GPT-2, Llama 3, Qwen 2.5 / Qwen 3, DeepSeek-V3, Gemma 3. Token IDs match tiktoken (GPT-2) and Hugging Face tokenizers.
• Batched encode of many documents in one graph replay, and GPU decode.
• CPU route for small inputs (gigatoken or the built-in host encoder) and a load-adaptive dispatcher that batches concurrent requests.
• Interfaces: CLI gpu_bpe_tokenize, Python module gpu_bpe_tokenizer (Tokenizer, Dispatcher), C library libgtok.
• Serving integrations: vLLM general plugin and an NVIDIA Dynamo 1.3.0 patch.