# CPM

## Overview

CPM モデルは、Zhengyan Zhang、Xu Han、Hao Zhou、Pei Ke、Yuxian Gu によって [CPM: A Large-scale Generative Chinese Pre-trained Language Model](https://huggingface.co/papers/2012.00413) で提案されました。葉徳明、秦裕佳、
Yusheng Su、Haozhe Ji、Jian Guan、Fanchao Qi、Xiaozi Wang、Yanan Zheng、Guoyang Zeng、Huanqi Cao、Shengqi Chen、
Daixuan Li、Zhenbo Sun、Zhiyuan Liu、Minlie Huang、Wentao Han、Jie Tang、Juanzi Li、Xiaoyan Zhu、Maosong Sun。

論文の要約は次のとおりです。

*事前トレーニングされた言語モデル (PLM) は、さまざまな下流の NLP タスクに有益であることが証明されています。最近ではGPT-3、
1,750億個のパラメータと570GBの学習データを備え、数回の撮影（1枚でも）の容量で大きな注目を集めました
ゼロショット）学習。ただし、GPT-3 を適用して中国語の NLP タスクに対処することは依然として困難です。
GPT-3 の言語は主に英語であり、パラメーターは公開されていません。この技術レポートでは、
大規模な中国語トレーニング データに対する生成的事前トレーニングを備えた中国語事前トレーニング済み言語モデル (CPM)。最高に
私たちの知識の限りでは、26 億のパラメータと 100GB の中国語トレーニング データを備えた CPM は、事前トレーニングされた中国語としては最大のものです。
言語モデルは、会話、エッセイの作成、
クローゼテストと言語理解。広範な実験により、CPM が多くの環境で優れたパフォーマンスを達成できることが実証されています。
少数ショット (ゼロショットでも) 学習の設定での NLP タスク。*

このモデルは [canwenxu](https://huggingface.co/canwenxu) によって提供されました。オリジナルの実装が見つかります
ここ: https://github.com/TsinghuaAI/CPM-Generate

CPM のアーキテクチャは、トークン化方法を除いて GPT-2 と同じです。詳細については、[GPT-2 ドキュメント](openai-community/gpt2) を参照してください。
API リファレンス情報。

## CpmTokenizer[[transformers.CpmTokenizer]]

#### transformers.CpmTokenizer[[transformers.CpmTokenizer]]

```python
transformers.CpmTokenizer(vocab_file, do_lower_case = False, remove_space = True, keep_accents = False, bos_token = '<s>', eos_token = '</s>', unk_token = '<unk>', sep_token = '<sep>', pad_token = '<pad>', cls_token = '<cls>', mask_token = '<mask>', additional_special_tokens = ['<eop>', '<eod>'], sp_model_kwargs: dict[str, typing.Any] | None = None, **kwargs)
```

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/cpm/tokenization_cpm.py#L34)

Runs pre-tokenization with Jieba-RS segmentation tool. It is used in CPM models.

#### build_inputs_with_special_tokens[[transformers.CpmTokenizer.build_inputs_with_special_tokens]]

```python
build_inputs_with_special_tokens(token_ids_0: list, token_ids_1: list[int] | None = None)
```

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/cpm/tokenization_cpm.py#L230)

**Parameters:**

token_ids_0 (`list[int]`) : List of IDs to which the special tokens will be added.

token_ids_1 (`list[int]`, *optional*) : Optional second list of IDs for sequence pairs.

**Returns:** `list[int]`

List of [input IDs](../glossary#input-ids) with the appropriate special tokens.

Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
adding special tokens. An XLNet sequence has the following format:

- single sequence: `X <sep> <cls>`
- pair of sequences: `A <sep> B <sep> <cls>`

#### convert_tokens_to_string[[transformers.CpmTokenizer.convert_tokens_to_string]]

```python
convert_tokens_to_string(tokens)
```

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/cpm/tokenization_cpm.py#L225)

Converts a sequence of tokens (strings for sub-words) in a single string.

#### create_token_type_ids_from_sequences[[transformers.CpmTokenizer.create_token_type_ids_from_sequences]]

```python
create_token_type_ids_from_sequences(token_ids_0: list, token_ids_1: list[int] | None = None)
```

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/cpm/tokenization_cpm.py#L283)

**Parameters:**

token_ids_0 (`list[int]`) : List of IDs.

token_ids_1 (`list[int]`, *optional*) : Optional second list of IDs for sequence pairs.

**Returns:** `list[int]`

List of [token type IDs](../glossary#token-type-ids) according to the given sequence(s).

Create a mask from the two sequences passed to be used in a sequence-pair classification task. An XLNet

sequence pair mask has the following format:

```
0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1
| first sequence    | second sequence |
```

If `token_ids_1` is `None`, this method only returns the first portion of the mask (0s).

#### get_special_tokens_mask[[transformers.CpmTokenizer.get_special_tokens_mask]]

```python
get_special_tokens_mask(token_ids_0: list, token_ids_1: list[int] | None = None, already_has_special_tokens: bool = False)
```

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/cpm/tokenization_cpm.py#L255)

**Parameters:**

token_ids_0 (`list[int]`) : List of IDs.

token_ids_1 (`list[int]`, *optional*) : Optional second list of IDs for sequence pairs.

already_has_special_tokens (`bool`, *optional*, defaults to `False`) : Whether or not the token list is already formatted with special tokens for the model.

**Returns:** `list[int]`

A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.

Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding
special tokens using the tokenizer `prepare_for_model` method.

## CpmTokenizerFast[[transformers.CpmTokenizerFast]]

#### transformers.CpmTokenizerFast[[transformers.CpmTokenizerFast]]

```python
transformers.CpmTokenizerFast(vocab_file = None, tokenizer_file = None, do_lower_case = False, remove_space = True, keep_accents = False, bos_token = '<s>', eos_token = '</s>', unk_token = '<unk>', sep_token = '<sep>', pad_token = '<pad>', cls_token = '<cls>', mask_token = '<mask>', additional_special_tokens = ['<eop>', '<eod>'], **kwargs)
```

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/cpm/tokenization_cpm_fast.py#L28)

Runs pre-tokenization with Jieba-RS segmentation tool. It is used in CPM models.

#### build_inputs_with_special_tokens[[transformers.CpmTokenizerFast.build_inputs_with_special_tokens]]

```python
build_inputs_with_special_tokens(token_ids_0: list, token_ids_1: list[int] | None = None)
```

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/cpm/tokenization_cpm_fast.py#L145)

**Parameters:**

token_ids_0 (`list[int]`) : List of IDs to which the special tokens will be added.

token_ids_1 (`list[int]`, *optional*) : Optional second list of IDs for sequence pairs.

**Returns:** `list[int]`

List of [input IDs](../glossary#input-ids) with the appropriate special tokens.

Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
adding special tokens. An XLNet sequence has the following format:

- single sequence: `X <sep> <cls>`
- pair of sequences: `A <sep> B <sep> <cls>`

#### create_token_type_ids_from_sequences[[transformers.CpmTokenizerFast.create_token_type_ids_from_sequences]]

```python
create_token_type_ids_from_sequences(token_ids_0: list, token_ids_1: list[int] | None = None)
```

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/cpm/tokenization_cpm_fast.py#L170)

**Parameters:**

token_ids_0 (`list[int]`) : List of IDs.

token_ids_1 (`list[int]`, *optional*) : Optional second list of IDs for sequence pairs.

**Returns:** `list[int]`

List of [token type IDs](../glossary#token-type-ids) according to the given sequence(s).

Create a mask from the two sequences passed to be used in a sequence-pair classification task. An XLNet

sequence pair mask has the following format:

```
0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1
| first sequence    | second sequence |
```

If `token_ids_1` is `None`, this method only returns the first portion of the mask (0s).