Update README.md
Browse files
README.md
CHANGED
@@ -1,4 +1,14 @@
|
|
1 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
|
3 |
结构为[sentence-transformers](https://github.com/UKPLab/sentence-transformers),使用其**distiluse-base-multilingual-cased-v2**预训练权重,以5e-5的学习率在动漫相关语句对数据集下进行微调,损失函数为MultipleNegativesRankingLoss。
|
4 |
|
@@ -34,4 +44,86 @@
|
|
34 |
* 动画中文名的简介-简介
|
35 |
* 动画中文名+小标题-对应内容
|
36 |
|
37 |
-
在进行爬取,清洗,处理后得到510w对文本对(还在持续增加),batchzise=80训练了20个epoch,使st的权重能够适应该问题空间,生成融合了领域知识的文本特征向量(体现为有关的文本距离更加接近,例如作品与登场人物,或者来自同一作品的登场人物)。
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
pipeline_tag: sentence-similarity
|
3 |
+
tags:
|
4 |
+
- sentence-transformers
|
5 |
+
- feature-extraction
|
6 |
+
- sentence-similarity
|
7 |
+
|
8 |
+
|
9 |
+
---
|
10 |
+
|
11 |
+
# {acgvoc2vec}
|
12 |
|
13 |
结构为[sentence-transformers](https://github.com/UKPLab/sentence-transformers),使用其**distiluse-base-multilingual-cased-v2**预训练权重,以5e-5的学习率在动漫相关语句对数据集下进行微调,损失函数为MultipleNegativesRankingLoss。
|
14 |
|
|
|
44 |
* 动画中文名的简介-简介
|
45 |
* 动画中文名+小标题-对应内容
|
46 |
|
47 |
+
在进行爬取,清洗,处理后得到510w对文本对(还在持续增加),batchzise=80训练了20个epoch,使st的权重能够适应该问题空间,生成融合了领域知识的文本特征向量(体现为有关的文本距离更加接近,例如作品与登场人物,或者来自同一作品的登场人物)。
|
48 |
+
|
49 |
+
## Usage (Sentence-Transformers)
|
50 |
+
|
51 |
+
Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
|
52 |
+
|
53 |
+
```
|
54 |
+
pip install -U sentence-transformers
|
55 |
+
```
|
56 |
+
|
57 |
+
Then you can use the model like this:
|
58 |
+
|
59 |
+
```python
|
60 |
+
from sentence_transformers import SentenceTransformer
|
61 |
+
sentences = ["This is an example sentence", "Each sentence is converted"]
|
62 |
+
|
63 |
+
model = SentenceTransformer('{MODEL_NAME}')
|
64 |
+
embeddings = model.encode(sentences)
|
65 |
+
print(embeddings)
|
66 |
+
```
|
67 |
+
|
68 |
+
|
69 |
+
|
70 |
+
## Evaluation Results
|
71 |
+
|
72 |
+
<!--- Describe how your model was evaluated -->
|
73 |
+
|
74 |
+
For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name={MODEL_NAME})
|
75 |
+
|
76 |
+
|
77 |
+
## Training
|
78 |
+
|
79 |
+
The model was trained with the parameters:
|
80 |
+
|
81 |
+
**DataLoader**:
|
82 |
+
|
83 |
+
`torch.utils.data.dataloader.DataLoader` of length 64769 with parameters:
|
84 |
+
|
85 |
+
```
|
86 |
+
{'batch_size': 80, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}
|
87 |
+
```
|
88 |
+
|
89 |
+
**Loss**:
|
90 |
+
|
91 |
+
`sentence_transformers.losses.MultipleNegativesRankingLoss.MultipleNegativesRankingLoss` with parameters:
|
92 |
+
|
93 |
+
```
|
94 |
+
{'scale': 20.0, 'similarity_fct': 'cos_sim'}
|
95 |
+
```
|
96 |
+
|
97 |
+
Parameters of the fit()-Method:
|
98 |
+
|
99 |
+
```
|
100 |
+
{
|
101 |
+
"epochs": 20,
|
102 |
+
"evaluation_steps": 0,
|
103 |
+
"evaluator": "NoneType",
|
104 |
+
"max_grad_norm": 1,
|
105 |
+
"optimizer_class": "<class 'torch.optim.adamw.AdamW'>",
|
106 |
+
"optimizer_params": {
|
107 |
+
"lr": 2e-05
|
108 |
+
},
|
109 |
+
"scheduler": "WarmupLinear",
|
110 |
+
"steps_per_epoch": null,
|
111 |
+
"warmup_steps": 129538,
|
112 |
+
"weight_decay": 0.01
|
113 |
+
}
|
114 |
+
```
|
115 |
+
|
116 |
+
|
117 |
+
## Full Model Architecture
|
118 |
+
|
119 |
+
```
|
120 |
+
SentenceTransformer(
|
121 |
+
(0): Transformer({'max_seq_length': 128, 'do_lower_case': False}) with Transformer model: DistilBertModel
|
122 |
+
(1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
|
123 |
+
(2): Dense({'in_features': 768, 'out_features': 512, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
|
124 |
+
)
|
125 |
+
```
|
126 |
+
|
127 |
+
## Citing & Authors
|
128 |
+
|
129 |
+
<!--- Describe where people can find more information -->
|