From 62eedcf2464ca146b981864ed48d76bcedcda0bc Mon Sep 17 00:00:00 2001 From: Zeyu Chen Date: Wed, 9 Dec 2020 13:32:10 +0800 Subject: [PATCH] update readme --- PaddleNLP/README.md | 30 +++++++++++++++++++++++------- 1 file changed, 23 insertions(+), 7 deletions(-) diff --git a/PaddleNLP/README.md b/PaddleNLP/README.md index d0db257a..a4efb362 100644 --- a/PaddleNLP/README.md +++ b/PaddleNLP/README.md @@ -37,36 +37,52 @@ pip install paddlenlp==2.0.0a ### Quick Dataset Loading ```python -train_ds, test_ds = paddlenlp.datasets.ChnSentiCorp.get_datasets(['train','test']) + +from paddlenlp.datasets import ChnSentiCrop +train_ds, test_ds = ChnSentiCorp.get_datasets(['train','test']) ``` -### Reusable Text Emebdding +### Chinese Text Emebdding Loading ```python from paddlenlp.embeddings import TokenEmbedding wordemb = TokenEmbedding("word2vec.baike.300d") print(wordemb.search("中国")) ->>> [0.260801 0.1047 0.129453 ... 0.096542 0.0092513] +>>> [0.260801, 0.1047, 0.129453 ... 0.096542, 0.0092513] + +``` +### One-Line Classical Model Building + +```python +from paddlenlp.models import Ernie +ernie = Ernie(Ernie.Task.SeqCls) +ernie.forward(input_ids, segment_ids) ``` -### High Quality Chinsese Pre-trained Model +### Rich Chinsese Pre-trained Models ```python -from paddlenlp.transformers import ErnieModel -ernie = ErnieModel.from_pretrained("ernie") -sequence_output, pooled_output = ernie.forward(input_ids, segment_ids) +from paddlenlp.transformers import ErnieModel, BertModel, RobertaModel, ElectraModel +ernie = ErnieModel.from_pretrained('ernie-1.0') +bert = BertModel.from_pretrained('bert-wwm-ext-large') +electra = ElectraModel.from_pretrained('eclectra-chinese') +roberta = RobertaModel.from_pretrained('roberta-wwm-ext') ``` +For more pretrained model selection, please refer to [PretrainedModels](./paddlenlp/transformers/README.md) + ## Tutorials List our notebook tutorials based on AI Studio. +TBD ## Community * SIG for Pretrained Model Contribution * SIG for Dataset Integration +TBD ## FAQ -- GitLab