init commit

.gitattributes

@@ -26,3 +26,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zstandard filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+pytorch_model.bin filter=lfs diff=lfs merge=lfs -text

1_Pooling/config.json

@@ -0,0 +1,7 @@
+{
+  "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
+}

README.md

@@ -0,0 +1,106 @@
+---
+pipeline_tag: sentence-similarity
+tags:
+- sentence-transformers
+- feature-extraction
+- sentence-similarity
+- transformers
+- semantic-search
+- chinese
+---
+
+# DMetaSoul/sbert-chinese-qmc-domain-v1
+
+此模型是基于之前开源[问题匹配模型](https://huggingface.co/DMetaSoul/sbert-chinese-qmc-domain-v1)的蒸馏轻量化版本（仅含4层 BERT），适用于**开放领域的问题匹配**场景，比如：
+
+
+- 洗澡用什么香皂好？vs. 洗澡用什么香皂好
+- 大连哪里拍婚纱照好点？ vs. 大连哪里拍婚纱照比较好
+- 银行卡怎样挂失？vs. 银行卡丢了怎么挂失啊？
+
+离线训练好的大模型如果直接用于线上推理，对计算资源有苛刻的需求，而且难以满足业务环境对延迟、吞吐量等性能指标的要求，这里我们使用蒸馏手段来把大模型轻量化。从 12 层 BERT 蒸馏为 4 层后，模型参数量缩小到 44%，大概 latency 减半、throughput 翻倍、精度下降 4% 左右（具体结果详见下文评估小节）。
+
+# Usage
+
+## 1. Sentence-Transformers
+
+通过  [sentence-transformers](https://www.SBERT.net) 框架来使用该模型，首先进行安装：
+
+```
+pip install -U sentence-transformers
+```
+
+然后使用下面的代码来载入该模型并进行文本表征向量的提取：
+
+```python
+from sentence_transformers import SentenceTransformer
+sentences = ["我的儿子！他猛然间喊道，我的儿子在哪儿？", "我的儿子呢！他突然喊道，我的儿子在哪里？"]
+
+model = SentenceTransformer('DMetaSoul/sbert-chinese-qmc-domain-v1')
+embeddings = model.encode(sentences)
+print(embeddings)
+```
+
+## 2. HuggingFace Transformers
+
+如果不想使用   [sentence-transformers](https://www.SBERT.net) 的话，也可以通过 HuggingFace Transformers 来载入该模型并进行文本向量抽取：
+
+```python
+from transformers import AutoTokenizer, AutoModel
+import torch
+
+
+#Mean Pooling - Take attention mask into account for correct averaging
+def mean_pooling(model_output, attention_mask):
+    token_embeddings = model_output[0] #First element of model_output contains all token embeddings
+    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
+    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)
+
+
+# Sentences we want sentence embeddings for
+sentences = ["我的儿子！他猛然间喊道，我的儿子在哪儿？", "我的儿子呢！他突然喊道，我的儿子在哪里？"]
+
+# Load model from HuggingFace Hub
+tokenizer = AutoTokenizer.from_pretrained('DMetaSoul/sbert-chinese-qmc-domain-v1')
+model = AutoModel.from_pretrained('DMetaSoul/sbert-chinese-qmc-domain-v1')
+
+# Tokenize sentences
+encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
+
+# Compute token embeddings
+with torch.no_grad():
+    model_output = model(**encoded_input)
+
+# Perform pooling. In this case, mean pooling.
+sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
+
+print("Sentence embeddings:")
+print(sentence_embeddings)
+```
+
+## Evaluation
+
+这里主要跟蒸馏前对应的 teacher 模型作了对比
+
+*性能：*
+
+|            | Teacher               | Student             | Gap   |
+| ---------- | --------------------- | ------------------- | ----- |
+| Model      | BERT-12-layers (102M) | BERT-4-layers (45M) | 0.44x |
+| Cost       | 23s                   | 12s                 | -47%  |
+| Latency    | 38ms                  | 20ms                | -47%  |
+| Throughput | 421 sentence/s        | 791 sentence/s      | 1.9x  |
+
+*精度：*
+
+|                | **csts_dev** | **csts_test** | **afqmc** | **lcqmc** | **bqcorpus** | **pawsx** | **xiaobu** | **Avg** |
+| -------------- | ------------ | ------------- | --------- | --------- | ------------ | --------- | ---------- | ------- |
+| **Teacher**    | 80.90%       | 76.62%        | 34.51%    | 77.05%    | 52.95%       | 12.97%    | 59.47%     | 56.35%  |
+| **Student**    | 79.89%       | 76.34%        | 27.59%    | 69.26%    | 49.40%       | 9.06%     | 53.52%     | 52.15%  |
+| **Gap** (abs.) | -            | -             | -         | -         | -            | -         | -          | -4.2%   |
+
+*基于1万条数据测试，GPU设备是V100，batch_size=16，max_seq_len=256*
+
+## Citing & Authors
+
+E-mail: xiaowenbin@dmetasoul.com

config.json

@@ -0,0 +1,31 @@
+{
+  "_name_or_path": "releases/sbert-chinese-qmc-domain-v1-distill/",
+  "architectures": [
+    "BertModel"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "classifier_dropout": null,
+  "directionality": "bidi",
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 4,
+  "pad_token_id": 0,
+  "pooler_fc_size": 768,
+  "pooler_num_attention_heads": 12,
+  "pooler_num_fc_layers": 3,
+  "pooler_size_per_head": 128,
+  "pooler_type": "first_token_transform",
+  "position_embedding_type": "absolute",
+  "torch_dtype": "float32",
+  "transformers_version": "4.16.0",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 21128
+}

config_sentence_transformers.json

@@ -0,0 +1,7 @@
+{
+  "__version__": {
+    "sentence_transformers": "2.1.0",
+    "transformers": "4.16.0",
+    "pytorch": "1.10.2"
+  }
+}

modules.json

@@ -0,0 +1,14 @@
+[
+  {
+    "idx": 0,
+    "name": "0",
+    "path": "",
+    "type": "sentence_transformers.models.Transformer"
+  },
+  {
+    "idx": 1,
+    "name": "1",
+    "path": "1_Pooling",
+    "type": "sentence_transformers.models.Pooling"
+  }
+]

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad505e52d15c2e6c396da1e3ff39c4707368d1c9e91a5a8ed18c18a46e590b24
+size 182288973

sentence_bert_config.json

@@ -0,0 +1,4 @@
+{
+  "max_seq_length": 256,
+  "do_lower_case": false
+}

special_tokens_map.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": false, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "model_max_length": 512, "special_tokens_map_file": null, "name_or_path": "releases/sbert-chinese-qmc-domain-v1-distill/", "tokenizer_class": "BertTokenizer"}