Fix GPT modle error link. (#487)

Author: ZHUI

docs/transformers.md

@@ -5,15 +5,15 @@
 
 ## Transformer预训练模型汇总
 
-下表汇总了目前PaddleNLP支持的各类预训练模型。用户可以使用PaddleNLP提供的模型，完成问答、文本分类、序列标注、文本生成等任务。同时我们提供了48种预训练的参数权重供用户使用，其中包含了23种中文语言模型的预训练权重。
+下表汇总了目前PaddleNLP支持的各类预训练模型。用户可以使用PaddleNLP提供的模型，完成问答、文本分类、序列标注、文本生成等任务。同时我们提供了48种预训练的参数权重供用户使用，其中包含了24种中文语言模型的预训练权重。
 
 | Model | Tokenizer | Supported Task | Pretrained Weight|
 |---|---|---|---|
 |[ALBERT](https://arxiv.org/abs/1909.11942)| AlbertTokenizer| AlbertModel<br> AlbertForMaskedLM<br> AlbertForQuestionAnswering<br> AlbertForMultipleChoice<br> AlbertForSequenceClassification<br> AlbertForTokenClassification |`albert-base-v1`<br> `albert-large-v1`<br> `albert-xlarge-v1`<br> `albert-xxlarge-v1`<br> `albert-base-v2`<br> `albert-large-v2`<br> `albert-xlarge-v2`<br> `albert-xxlarge-v2`<br> `albert-chinese-tiny`<br> `albert-chinese-small`<br> `albert-chinese-base`<br> `albert-chinese-large`<br> `albert-chinese-xlarge`<br> `albert-chinese-xxlarge` |
 |[BERT](https://arxiv.org/abs/1810.04805) | BertTokenizer|BertModel<br> BertForQuestionAnswering<br> BertForSequenceClassification<br>BertForTokenClassification| `bert-base-uncased`<br> `bert-large-uncased` <br>`bert-base-multilingual-uncased` <br>`bert-base-cased`<br> `bert-base-chinese`<br> `bert-base-multilingual-cased`<br> `bert-large-cased`<br> `bert-wwm-chinese`<br> `bert-wwm-ext-chinese` |
 |[ERNIE](https://arxiv.org/abs/1904.09223)|ErnieTokenizer<br>ErnieTinyTokenizer|ErnieModel<br> ErnieForQuestionAnswering<br> ErnieForSequenceClassification<br> ErnieForTokenClassification | `ernie-1.0`<br> `ernie-tiny`<br> `ernie-2.0-en`<br> `ernie-2.0-large-en`|
 |[ERNIE-GEN](https://arxiv.org/abs/2001.11314)|ErnieTokenizer| ErnieForGeneration|`ernie-gen-base-en`<br>`ernie-gen-large-en`<br>`ernie-gen-large-en-430g`|
-|[GPT](https://cdn.openai.com/better-language-models/language_models_are_unsupervised_multitask_learners.pdf)| GPTTokenizer<br> GPTChineseTokenizer| GPTForGreedyGeneration| `gpt-cpm-large-cn` <br> `gpt2-medium-en`|
+|[GPT](https://cdn.openai.com/better-language-models/language_models_are_unsupervised_multitask_learners.pdf)| GPTTokenizer<br> GPTChineseTokenizer| GPTForGreedyGeneration| `gpt-cpm-large-cn` <br> `gpt-cpm-small-cn-distill` <br> `gpt2-medium-en`|
 |[RoBERTa](https://arxiv.org/abs/1907.11692)|RobertaTokenizer| RobertaModel<br>RobertaForQuestionAnswering<br>RobertaForSequenceClassification<br>RobertaForTokenClassification| `roberta-wwm-ext`<br> `roberta-wwm-ext-large`<br> `rbt3`<br> `rbtl3`|
 |[ELECTRA](https://arxiv.org/abs/2003.10555) | ElectraTokenizer| ElectraModel<br>ElectraForSequenceClassification<br>ElectraForTokenClassification<br>|`electra-small`<br> `electra-base`<br> `electra-large`<br> `chinese-electra-small`<br> `chinese-electra-base`<br>|
 |[XLNet](https://arxiv.org/abs/1906.08237)| XLNetTokenizer| XLNetModel<br> XLNetForSequenceClassification<br> XLNetForTokenClassification |`xlnet-base-cased`<br> `xlnet-large-cased`<br> `chinese-xlnet-base`<br> `chinese-xlnet-mid`<br> `chinese-xlnet-large`|
@@ -22,7 +22,7 @@
 |[TinyBERT](https://arxiv.org/abs/1909.10351) |TinyBertTokenizer | TinyBertModel<br>TinyBertForPretraining<br>TinyBertForSequenceClassification | `tinybert-4l-312d`<br>`tinybert-6l-768d`<br>`tinybert-4l-312d-v2`<br>`tinybert-6l-768d-v2` |
 |[Transformer](https://arxiv.org/abs/1706.03762) |- | TransformerModel | - |
 
-**NOTE**：其中中文的预训练模型有`albert-chinese-tiny, albert-chinese-small, albert-chinese-base, albert-chinese-large, albert-chinese-xlarge, albert-chinese-xxlarge, bert-base-chinese, bert-wwm-chinese, bert-wwm-ext-chinese, ernie-1.0, ernie-tiny, gpt-cpm-large-cn, roberta-wwm-ext, roberta-wwm-ext-large, rbt3, rbtl3, chinese-electra-base, chinese-electra-small, chinese-xlnet-base, chinese-xlnet-mid, chinese-xlnet-large, unified_transformer-12L-cn, unified_transformer-12L-cn-luge`。
+**NOTE**：其中中文的预训练模型有`albert-chinese-tiny, albert-chinese-small, albert-chinese-base, albert-chinese-large, albert-chinese-xlarge, albert-chinese-xxlarge, bert-base-chinese, bert-wwm-chinese, bert-wwm-ext-chinese, ernie-1.0, ernie-tiny, gpt-cpm-large-cn, gpt-cpm-small-cn-distill, roberta-wwm-ext, roberta-wwm-ext-large, rbt3, rbtl3, chinese-electra-base, chinese-electra-small, chinese-xlnet-base, chinese-xlnet-mid, chinese-xlnet-large, unified_transformer-12L-cn, unified_transformer-12L-cn-luge`。
 
 ## 预训练模型使用方法
 

examples/language_model/gpt/README.md

@@ -53,7 +53,7 @@ bash decompress.sh
 
 ```shell
 python create_pretraining_data.py --input_path raw_data \
- --model_name gpt2-medium-en \
+ --model_name gpt2-en \
  --append_eod \
  --workers 8
 ```

examples/language_model/gpt/create_pretraining_data.py

@@ -46,8 +46,8 @@ class Converter(object):
     def __init__(self, model_name, append_eod):
         self.append_eod = append_eod
         self.tokenizer = GPTTokenizer.from_pretrained(model_name)
-        self.eod_id = tokenizer.eod_token_id
-        self.vocab_size = len(tokenizer)
+        self.eod_id = self.tokenizer.eod_token_id
+        self.vocab_size = len(self.tokenizer)
 
     def encode(self, text):
         tokens = self.tokenizer(text)["input_ids"]

paddlenlp/transformers/gpt/tokenizer.py

@@ -81,14 +81,17 @@ class GPTChineseTokenizer(PretrainedTokenizer):
         "model_file": "sentencepiece.model"
     }  # for save_pretrained
 
-    cpm_modle_link = "https://paddlenlp.bj.bcebos.com/models/transformers/gpt/gpt-cpm-cn-sentencepiece.model"
+    cpm_model_link = "https://paddlenlp.bj.bcebos.com/models/transformers/gpt/gpt-cpm-cn-sentencepiece.model"
     pretrained_resource_files_map = {
         "model_file": {
-            "gpt-cpm-large-cn": cpm_modle_link,
-            "gpt-cpm-small-cn-distill": cpm_modle_link,
+            "gpt-cpm-large-cn": cpm_model_link,
+            "gpt-cpm-small-cn-distill": cpm_model_link,
         }
     }
-    pretrained_init_configuration = {"gpt-cpm-large-cn": {}, "gpt-cpm-small-cn-distill": {}, }
+    pretrained_init_configuration = {
+        "gpt-cpm-large-cn": {},
+        "gpt-cpm-small-cn-distill": {},
+    }
 
     def __init__(
             self,