Hugging Face's logo

thegenerativegeneration
/
stay_or_go_conversation_classifier_s_v2

Text Classification
setfit
Safetensors
sentence-transformers
bert
generated_from_setfit_trainer
Eval Results
Model card Files Community
thegenerativegeneration commited on
Commit
11152e7
·
verified ·
1 Parent(s): cca2b4d

Upload 13 files

Browse files
Files changed (6) hide show
  1. README.md +48 -60
  2. config.json +1 -1
  3. model.safetensors +1 -1
  4. model_head.pkl +1 -1
  5. tokenizer.json +2 -2
  6. tokenizer_config.json +7 -0
README.md CHANGED
@@ -9,12 +9,11 @@ base_model: intfloat/multilingual-e5-small
9
  metrics:
10
  - accuracy
11
  widget:
12
- - text: 'query: Interessant. Hast du das schon mal ausprobiert?'
13
- - text: 'query: はい、持っていますよ。すぐにメールで送りますね。'
14
- - text: 'query: Va bene ci sentiamo dopo Marco buona giornata'
15
- - text: 'query: Ζητώ συγγνώμη, πρέπει να αποχωρήσω τώρα.'
16
- - text: 'query: Guten Morgen, Maria! Hast du die Präsentation für das Meeting heute
17
- fertig?'
18
  pipeline_tag: text-classification
19
  inference: true
20
  ---
@@ -47,10 +46,10 @@ The model has been trained using an efficient few-shot learning technique that i
47
  - **Blogpost:** [SetFit: Efficient Few-Shot Learning Without Prompts](https://huggingface.co/blog/setfit)
48
 
49
  ### Model Labels
50
- | Label | Examples |
51
- |:------|:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
52
- | 0 | <ul><li>'query: สวัสดีค่ะ วันนี้เป็นอย่างไรบ้าง?'</li><li>'query: Jag förstår. Vad tycker du att vi ska göra nu?'</li><li>'query: Hej, wszystko w porządku. Właśnie dostałam nową pracę.'</li></ul> |
53
- | 1 | <ul><li>'query: Чудесно, доскоро!'</li><li>'query: Mama mă cheamă, trebuie să mă întorc acasă, pa.'</li><li>'query: Perdó, ja he de marxar.'</li></ul> |
54
 
55
  ## Uses
56
 
@@ -70,7 +69,7 @@ from setfit import SetFitModel
70
  # Download from the 🤗 Hub
71
  model = SetFitModel.from_pretrained("setfit_model_id")
72
  # Run inference
73
- preds = model("query: はい、持っていますよ。すぐにメールで送りますね。")
74
  ```
75
 
76
  <!--
@@ -102,23 +101,23 @@ preds = model("query: はい、持っていますよ。すぐにメールで送
102
  ### Training Set Metrics
103
  | Training set | Min | Median | Max |
104
  |:-------------|:----|:-------|:----|
105
- | Word count | 2 | 7.3663 | 21 |
106
 
107
  | Label | Training Sample Count |
108
  |:------|:----------------------|
109
- | 0 | 286 |
110
- | 1 | 290 |
111
 
112
  ### Training Hyperparameters
113
  - batch_size: (16, 2)
114
  - num_epochs: (1, 16)
115
- - max_steps: 2000
116
  - sampling_strategy: undersampling
117
  - body_learning_rate: (1e-05, 1e-05)
118
  - head_learning_rate: 0.001
119
  - loss: CosineSimilarityLoss
120
  - distance_metric: cosine_distance
121
- - margin: 0.1
122
  - end_to_end: False
123
  - use_amp: False
124
  - warmup_proportion: 0.1
@@ -128,50 +127,39 @@ preds = model("query: はい、持っていますよ。すぐにメールで送
128
  - load_best_model_at_end: True
129
 
130
  ### Training Results
131
- | Epoch | Step | Training Loss | Validation Loss |
132
- |:------:|:----:|:-------------:|:---------------:|
133
- | 0.0002 | 1 | 0.3683 | - |
134
- | 0.0125 | 50 | 0.3256 | - |
135
- | 0.0250 | 100 | 0.211 | 0.1998 |
136
- | 0.0375 | 150 | 0.1668 | - |
137
- | 0.0500 | 200 | 0.0788 | 0.0571 |
138
- | 0.0625 | 250 | 0.0644 | - |
139
- | 0.0750 | 300 | 0.0232 | 0.0286 |
140
- | 0.0875 | 350 | 0.0024 | - |
141
- | 0.1000 | 400 | 0.0014 | 0.0945 |
142
- | 0.1125 | 450 | 0.0007 | - |
143
- | 0.1250 | 500 | 0.0008 | 0.1036 |
144
- | 0.1375 | 550 | 0.0005 | - |
145
- | 0.1500 | 600 | 0.0005 | 0.098 |
146
- | 0.1625 | 650 | 0.0003 | - |
147
- | 0.1750 | 700 | 0.0005 | 0.1056 |
148
- | 0.1875 | 750 | 0.0004 | - |
149
- | 0.2000 | 800 | 0.0006 | 0.1044 |
150
- | 0.2124 | 850 | 0.0005 | - |
151
- | 0.2249 | 900 | 0.0004 | 0.1072 |
152
- | 0.2374 | 950 | 0.0003 | - |
153
- | 0.2499 | 1000 | 0.0001 | 0.0993 |
154
- | 0.2624 | 1050 | 0.0003 | - |
155
- | 0.2749 | 1100 | 0.0003 | 0.1114 |
156
- | 0.2874 | 1150 | 0.0002 | - |
157
- | 0.2999 | 1200 | 0.0002 | 0.1078 |
158
- | 0.3124 | 1250 | 0.0001 | - |
159
- | 0.3249 | 1300 | 0.0002 | 0.0908 |
160
- | 0.3374 | 1350 | 0.0002 | - |
161
- | 0.3499 | 1400 | 0.0002 | 0.1019 |
162
- | 0.3624 | 1450 | 0.0001 | - |
163
- | 0.3749 | 1500 | 0.0002 | 0.11 |
164
- | 0.3874 | 1550 | 0.0002 | - |
165
- | 0.3999 | 1600 | 0.0001 | 0.1031 |
166
- | 0.4124 | 1650 | 0.0001 | - |
167
- | 0.4249 | 1700 | 0.0001 | 0.0996 |
168
- | 0.4374 | 1750 | 0.0002 | - |
169
- | 0.4499 | 1800 | 0.0001 | 0.0903 |
170
- | 0.4624 | 1850 | 0.0002 | - |
171
- | 0.4749 | 1900 | 0.0001 | 0.0901 |
172
- | 0.4874 | 1950 | 0.0002 | - |
173
- | 0.4999 | 2000 | 0.0001 | 0.0854 |
174
-
175
  ### Framework Versions
176
  - Python: 3.10.11
177
  - SetFit: 1.0.3
 
9
  metrics:
10
  - accuracy
11
  widget:
12
+ - text: 'query: Baiklah, kita cakap lagi nanti, Mark. Selamat hari!'
13
+ - text: 'query: Tôi xin lỗi nhưng tôi phải đi'
14
+ - text: 'query: 次回行くときは、私を連れて行ってください。もっと自然の中で活動したいと思っています。'
15
+ - text: 'query: Entschuldigung, ich muss jetzt gehen.'
16
+ - text: 'query: Buenos días, ¿cómo están ustedes?'
 
17
  pipeline_tag: text-classification
18
  inference: true
19
  ---
 
46
  - **Blogpost:** [SetFit: Efficient Few-Shot Learning Without Prompts](https://huggingface.co/blog/setfit)
47
 
48
  ### Model Labels
49
+ | Label | Examples |
50
+ |:------|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------|
51
+ | 0 | <ul><li>'query: Értem. Mit csinálunk most?'</li><li>'query: Ola Luca, que tal? Rematache o traballo?'</li><li>'query: Lijepo je. Hvala.'</li></ul> |
52
+ | 1 | <ul><li>'query: Жөнейін, кейін кездесеміз.'</li><li>'query: Така, ќе се видиме повторно.'</li><li>'query: ठीक है बाद में बात करते हैं मार्क अच्छा दिन'</li></ul> |
53
 
54
  ## Uses
55
 
 
69
  # Download from the 🤗 Hub
70
  model = SetFitModel.from_pretrained("setfit_model_id")
71
  # Run inference
72
+ preds = model("query: Tôi xin lỗi nhưng tôi phải đi")
73
  ```
74
 
75
  <!--
 
101
  ### Training Set Metrics
102
  | Training set | Min | Median | Max |
103
  |:-------------|:----|:-------|:----|
104
+ | Word count | 2 | 7.2168 | 25 |
105
 
106
  | Label | Training Sample Count |
107
  |:------|:----------------------|
108
+ | 0 | 346 |
109
+ | 1 | 346 |
110
 
111
  ### Training Hyperparameters
112
  - batch_size: (16, 2)
113
  - num_epochs: (1, 16)
114
+ - max_steps: 1400
115
  - sampling_strategy: undersampling
116
  - body_learning_rate: (1e-05, 1e-05)
117
  - head_learning_rate: 0.001
118
  - loss: CosineSimilarityLoss
119
  - distance_metric: cosine_distance
120
+ - margin: 0.05
121
  - end_to_end: False
122
  - use_amp: False
123
  - warmup_proportion: 0.1
 
127
  - load_best_model_at_end: True
128
 
129
  ### Training Results
130
+ | Epoch | Step | Training Loss | Validation Loss |
131
+ |:----------:|:--------:|:-------------:|:---------------:|
132
+ | 0.0004 | 1 | 0.3607 | - |
133
+ | 0.0179 | 50 | 0.3254 | - |
134
+ | 0.0357 | 100 | 0.2303 | 0.2049 |
135
+ | 0.0536 | 150 | 0.106 | - |
136
+ | 0.0714 | 200 | 0.1294 | 0.0748 |
137
+ | 0.0893 | 250 | 0.087 | - |
138
+ | 0.1071 | 300 | 0.0732 | 0.0787 |
139
+ | 0.1250 | 350 | 0.0019 | - |
140
+ | 0.1428 | 400 | 0.0027 | 0.1072 |
141
+ | 0.1607 | 450 | 0.0015 | - |
142
+ | 0.1785 | 500 | 0.0008 | 0.0999 |
143
+ | 0.1964 | 550 | 0.0016 | - |
144
+ | 0.2142 | 600 | 0.0004 | 0.1215 |
145
+ | 0.2321 | 650 | 0.0012 | - |
146
+ | 0.2499 | 700 | 0.0008 | 0.1267 |
147
+ | 0.2678 | 750 | 0.0005 | - |
148
+ | 0.2856 | 800 | 0.0003 | 0.1216 |
149
+ | 0.3035 | 850 | 0.0003 | - |
150
+ | 0.3213 | 900 | 0.0004 | 0.1142 |
151
+ | 0.3392 | 950 | 0.0004 | - |
152
+ | **0.3570** | **1000** | **0.0004** | **0.0616** |
153
+ | 0.3749 | 1050 | 0.0002 | - |
154
+ | 0.3927 | 1100 | 0.0004 | 0.0946 |
155
+ | 0.4106 | 1150 | 0.0002 | - |
156
+ | 0.4284 | 1200 | 0.0003 | 0.1091 |
157
+ | 0.4463 | 1250 | 0.0002 | - |
158
+ | 0.4641 | 1300 | 0.0003 | 0.1141 |
159
+ | 0.4820 | 1350 | 0.0004 | - |
160
+ | 0.4998 | 1400 | 0.0002 | 0.1209 |
161
+
162
+ * The bold row denotes the saved checkpoint.
 
 
 
 
 
 
 
 
 
 
 
163
  ### Framework Versions
164
  - Python: 3.10.11
165
  - SetFit: 1.0.3
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "intfloat/multilingual-e5-small",
3
  "architectures": [
4
  "BertModel"
5
  ],
 
1
  {
2
+ "_name_or_path": "checkpoints/step_1000",
3
  "architectures": [
4
  "BertModel"
5
  ],
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ecfce4dd8b2e3179e859bc278ca2390319e04a66f3179fbbeb1bf7b598a86307
3
  size 470637416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:27c89f801f10bb9afe5e4f308a41a0d7492b8725340318de1847eec8f6b84cf1
3
  size 470637416
model_head.pkl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:492fb3b7da876887807a7f0eb94fda6a77e65bbb7f72311fb8caaf601a46407c
3
  size 4608
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0b054fef0d715653a0dba9374d17ce2d5fa1a3fb6560f2768740890da80a0321
3
  size 4608
tokenizer.json CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:45b6ee00bc5023ac454b82c372ebe14b27866fa471b6dbb0d24e09b12909a1f4
3
- size 17083075
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:55ce1a4600af70b33f5a7fba12dbb41a504d3c08737c9b26b5e7fd6e437a9a23
3
+ size 17083087
tokenizer_config.json CHANGED
@@ -46,10 +46,17 @@
46
  "cls_token": "<s>",
47
  "eos_token": "</s>",
48
  "mask_token": "<mask>",
 
49
  "model_max_length": 512,
 
50
  "pad_token": "<pad>",
 
 
51
  "sep_token": "</s>",
52
  "sp_model_kwargs": {},
 
53
  "tokenizer_class": "XLMRobertaTokenizer",
 
 
54
  "unk_token": "<unk>"
55
  }
 
46
  "cls_token": "<s>",
47
  "eos_token": "</s>",
48
  "mask_token": "<mask>",
49
+ "max_length": 512,
50
  "model_max_length": 512,
51
+ "pad_to_multiple_of": null,
52
  "pad_token": "<pad>",
53
+ "pad_token_type_id": 0,
54
+ "padding_side": "right",
55
  "sep_token": "</s>",
56
  "sp_model_kwargs": {},
57
+ "stride": 0,
58
  "tokenizer_class": "XLMRobertaTokenizer",
59
+ "truncation_side": "right",
60
+ "truncation_strategy": "longest_first",
61
  "unk_token": "<unk>"
62
  }