dynamo_llm/tokenizers/
hf.rs

1// SPDX-FileCopyrightText: Copyright (c) 2024-2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
2// SPDX-License-Identifier: Apache-2.0
3
4use tokenizers::tokenizer::Tokenizer as HfTokenizer;
5
6use super::{
7    Encoding, Error, Result, TokenIdType,
8    traits::{Decoder, Encoder, Tokenizer},
9};
10
11pub struct HuggingFaceTokenizer {
12    tokenizer: HfTokenizer,
13}
14
15impl HuggingFaceTokenizer {
16    pub fn from_file(model_name: &str) -> Result<Self> {
17        let tokenizer = HfTokenizer::from_file(model_name)
18            .map_err(|err| Error::msg(format!("Error loading tokenizer: {}", err)))?;
19
20        Ok(HuggingFaceTokenizer { tokenizer })
21    }
22
23    pub fn from_tokenizer(tokenizer: HfTokenizer) -> Self {
24        HuggingFaceTokenizer { tokenizer }
25    }
26}
27
28impl Encoder for HuggingFaceTokenizer {
29    fn encode(&self, input: &str) -> Result<Encoding> {
30        // This self.tokenizer is the library
31        let encoding = self
32            .tokenizer
33            .encode(input, false)
34            .map_err(|err| Error::msg(format!("Error tokenizing input: {err}")))?;
35
36        Ok(Encoding::Hf(Box::new(encoding)))
37    }
38
39    fn encode_batch(&self, inputs: &[&str]) -> Result<Vec<Encoding>> {
40        let hf_encodings = self
41            .tokenizer
42            .encode_batch(inputs.to_vec(), false)
43            .map_err(|err| Error::msg(format!("Error batch tokenizing input: {err}")))?;
44
45        let encodings = hf_encodings
46            .into_iter()
47            .map(|enc| Encoding::Hf(Box::new(enc)))
48            .collect();
49
50        Ok(encodings)
51    }
52}
53
54impl Decoder for HuggingFaceTokenizer {
55    fn decode(&self, token_ids: &[TokenIdType], skip_special_tokens: bool) -> Result<String> {
56        // This calls into the library
57        let text = self
58            .tokenizer
59            .decode(token_ids, skip_special_tokens)
60            .map_err(|err| Error::msg(format!("Error de-tokenizing input: {err}")))?;
61
62        Ok(text)
63    }
64}
65
66impl Tokenizer for HuggingFaceTokenizer {}
67
68impl From<HfTokenizer> for HuggingFaceTokenizer {
69    fn from(tokenizer: HfTokenizer) -> Self {
70        HuggingFaceTokenizer { tokenizer }
71    }
72}