Skip to main content

load_hf_from_json

Function load_hf_from_json 

Source
pub fn load_hf_from_json(json: &str) -> Result<BpeTokenizer, AprenderError>
Expand description

Load tokenizer from HuggingFace tokenizer.json format.

Uses the optimized loading path (GH-378): pre-sized HashMaps, owned-string moves, and fast merge loading. Applies to all vocab sizes via config_from_vocab_size dispatch (Qwen2, Whisper, GPT-2, LLaMA).

§Arguments

  • json - JSON string of tokenizer configuration

§Returns

Loaded tokenizer with vocabulary and merge rules

§Errors

Returns error if parsing fails.