microsoft/onnxruntime-extensions
Publicmirrored fromhttps://github.com/microsoft/onnxruntime-extensionsAvailable
operators/tokenizer/bert_tokenizer_decoder.hpp
52lines · modecode
| 1 | // Copyright (c) Microsoft Corporation. All rights reserved. |
| 2 | // Licensed under the MIT License. |
| 3 | |
| 4 | #pragma once |
| 5 | |
| 6 | #include <unordered_map> |
| 7 | #include <vector> |
| 8 | #include "ocos.h" |
| 9 | #include "ustring.h" |
| 10 | #include "string_utils.h" |
| 11 | #include "string_tensor.h" |
| 12 | |
| 13 | |
| 14 | class BertTokenizerDecoder { |
| 15 | public: |
| 16 | BertTokenizerDecoder(std::string vocab, std::string unk_token, std::string sep_token, std::string pad_token, |
| 17 | std::string cls_token,std::string mask_token,std::string suffix_indicator); |
| 18 | std::string Decode(const std::vector<int64_t>& ids, bool skip_special_tokens, bool clean_up_tokenization_spaces); |
| 19 | |
| 20 | private: |
| 21 | std::string unk_token_; |
| 22 | int32_t unk_token_id_ = -1; |
| 23 | int32_t sep_token_id_ = -1; |
| 24 | int32_t pad_token_id_ = -1; |
| 25 | int32_t cls_token_id_ = -1; |
| 26 | int32_t mask_token_id_ = -1; |
| 27 | std::string suffix_indicator_; |
| 28 | std::vector<std::string_view> vocab_; |
| 29 | std::string raw_vocab_; |
| 30 | std::vector<bool> is_substr_; |
| 31 | |
| 32 | bool RemoveTokenizeSpace(int64_t pre_token_id, int64_t new_token_id); |
| 33 | }; |
| 34 | |
| 35 | struct KernelBertTokenizerDecoder : BaseKernel { |
| 36 | KernelBertTokenizerDecoder(const OrtApi& api, const OrtKernelInfo* info); |
| 37 | void Compute(OrtKernelContext* context); |
| 38 | private: |
| 39 | std::shared_ptr<BertTokenizerDecoder> decoder_; |
| 40 | bool use_indices_; |
| 41 | bool skip_special_tokens_; |
| 42 | bool clean_up_tokenization_spaces_; |
| 43 | }; |
| 44 | |
| 45 | struct CustomOpBertTokenizerDecoder : OrtW::CustomOpBase<CustomOpBertTokenizerDecoder, KernelBertTokenizerDecoder> { |
| 46 | void* CreateKernel(const OrtApi& api, const OrtKernelInfo* info) const; |
| 47 | const char* GetName() const; |
| 48 | size_t GetInputTypeCount() const; |
| 49 | ONNXTensorElementDataType GetInputType(size_t index) const; |
| 50 | size_t GetOutputTypeCount() const; |
| 51 | ONNXTensorElementDataType GetOutputType(size_t index) const; |
| 52 | }; |