Hugging Face's logo

LeoChiuu
/
all-MiniLM-L6-v2

Sentence Similarity
sentence-transformers
Safetensors
bert
feature-extraction
Generated from Trainer
dataset_size:560
loss:CoSENTLoss
Eval Results (legacy)
text-embeddings-inference
Model card Files
xet
 Community
LeoChiuu commited on
Commit
f5b391a
·
verified ·
1 Parent(s): 8a7276b

Add new SentenceTransformer model.

Browse files
Files changed (3) hide show
  1. README.md +144 -145
  2. config_sentence_transformers.json +1 -1
  3. model.safetensors +1 -1
README.md CHANGED
@@ -45,34 +45,34 @@ tags:
45
  - sentence-similarity
46
  - feature-extraction
47
  - generated_from_trainer
48
- - dataset_size:216
49
- - loss:MultipleNegativesRankingLoss
50
  widget:
51
- - source_sentence: Sophie why are you pressured?
52
  sentences:
53
- - Sophie Are you pressured?
54
- - Did you place the scarf in the fireplace?
55
- - A marked Globe.
56
- - source_sentence: Because of the red stain from the dish
57
  sentences:
58
- - Are you using my slippers?
59
- - Do you know this book?
60
- - There was a red stain on the dish
61
- - source_sentence: Outside
62
  sentences:
63
- - To grant the wish of having adventure
64
- - Let's look inside
65
- - Let's go outside
66
- - source_sentence: Actually I want a candle
67
  sentences:
68
- - Is that a cloth on the tree?
69
- - Did you have a beef stew for dinner?
70
- - Give me a candle
71
- - source_sentence: I found a flower pot.
72
  sentences:
73
- - Last night?
74
- - I found flowers.
75
- - Do you know this picture?
76
  model-index:
77
  - name: SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2
78
  results:
@@ -84,109 +84,109 @@ model-index:
84
  type: custom-arc-semantics-data
85
  metrics:
86
  - type: cosine_accuracy
87
- value: 0.9818181818181818
88
  name: Cosine Accuracy
89
  - type: cosine_accuracy_threshold
90
- value: 0.26917901635169983
91
  name: Cosine Accuracy Threshold
92
  - type: cosine_f1
93
- value: 0.9908256880733944
94
  name: Cosine F1
95
  - type: cosine_f1_threshold
96
- value: 0.26917901635169983
97
  name: Cosine F1 Threshold
98
  - type: cosine_precision
99
- value: 1.0
100
  name: Cosine Precision
101
  - type: cosine_recall
102
- value: 0.9818181818181818
103
  name: Cosine Recall
104
  - type: cosine_ap
105
- value: 1.0
106
  name: Cosine Ap
107
  - type: dot_accuracy
108
- value: 0.9818181818181818
109
  name: Dot Accuracy
110
  - type: dot_accuracy_threshold
111
- value: 0.2691790461540222
112
  name: Dot Accuracy Threshold
113
  - type: dot_f1
114
- value: 0.9908256880733944
115
  name: Dot F1
116
  - type: dot_f1_threshold
117
- value: 0.2691790461540222
118
  name: Dot F1 Threshold
119
  - type: dot_precision
120
- value: 1.0
121
  name: Dot Precision
122
  - type: dot_recall
123
- value: 0.9818181818181818
124
  name: Dot Recall
125
  - type: dot_ap
126
- value: 1.0
127
  name: Dot Ap
128
  - type: manhattan_accuracy
129
- value: 0.9818181818181818
130
  name: Manhattan Accuracy
131
  - type: manhattan_accuracy_threshold
132
- value: 18.48493194580078
133
  name: Manhattan Accuracy Threshold
134
  - type: manhattan_f1
135
- value: 0.9908256880733944
136
  name: Manhattan F1
137
  - type: manhattan_f1_threshold
138
- value: 18.48493194580078
139
  name: Manhattan F1 Threshold
140
  - type: manhattan_precision
141
- value: 1.0
142
  name: Manhattan Precision
143
  - type: manhattan_recall
144
- value: 0.9818181818181818
145
  name: Manhattan Recall
146
  - type: manhattan_ap
147
- value: 1.0
148
  name: Manhattan Ap
149
  - type: euclidean_accuracy
150
- value: 0.9818181818181818
151
  name: Euclidean Accuracy
152
  - type: euclidean_accuracy_threshold
153
- value: 1.2088721990585327
154
  name: Euclidean Accuracy Threshold
155
  - type: euclidean_f1
156
- value: 0.9908256880733944
157
  name: Euclidean F1
158
  - type: euclidean_f1_threshold
159
- value: 1.2088721990585327
160
  name: Euclidean F1 Threshold
161
  - type: euclidean_precision
162
- value: 1.0
163
  name: Euclidean Precision
164
  - type: euclidean_recall
165
- value: 0.9818181818181818
166
  name: Euclidean Recall
167
  - type: euclidean_ap
168
- value: 1.0
169
  name: Euclidean Ap
170
  - type: max_accuracy
171
- value: 0.9818181818181818
172
  name: Max Accuracy
173
  - type: max_accuracy_threshold
174
- value: 18.48493194580078
175
  name: Max Accuracy Threshold
176
  - type: max_f1
177
- value: 0.9908256880733944
178
  name: Max F1
179
  - type: max_f1_threshold
180
- value: 18.48493194580078
181
  name: Max F1 Threshold
182
  - type: max_precision
183
- value: 1.0
184
  name: Max Precision
185
  - type: max_recall
186
- value: 0.9818181818181818
187
  name: Max Recall
188
  - type: max_ap
189
- value: 1.0
190
  name: Max Ap
191
  ---
192
 
@@ -240,9 +240,9 @@ from sentence_transformers import SentenceTransformer
240
  model = SentenceTransformer("LeoChiuu/all-MiniLM-L6-v2")
241
  # Run inference
242
  sentences = [
243
- 'I found a flower pot.',
244
- 'I found flowers.',
245
- 'Do you know this picture?',
246
  ]
247
  embeddings = model.encode(sentences)
248
  print(embeddings.shape)
@@ -286,43 +286,43 @@ You can finetune this model on your own dataset.
286
  * Dataset: `custom-arc-semantics-data`
287
  * Evaluated with [<code>BinaryClassificationEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.BinaryClassificationEvaluator)
288
 
289
- | Metric | Value |
290
- |:-----------------------------|:--------|
291
- | cosine_accuracy | 0.9818 |
292
- | cosine_accuracy_threshold | 0.2692 |
293
- | cosine_f1 | 0.9908 |
294
- | cosine_f1_threshold | 0.2692 |
295
- | cosine_precision | 1.0 |
296
- | cosine_recall | 0.9818 |
297
- | cosine_ap | 1.0 |
298
- | dot_accuracy | 0.9818 |
299
- | dot_accuracy_threshold | 0.2692 |
300
- | dot_f1 | 0.9908 |
301
- | dot_f1_threshold | 0.2692 |
302
- | dot_precision | 1.0 |
303
- | dot_recall | 0.9818 |
304
- | dot_ap | 1.0 |
305
- | manhattan_accuracy | 0.9818 |
306
- | manhattan_accuracy_threshold | 18.4849 |
307
- | manhattan_f1 | 0.9908 |
308
- | manhattan_f1_threshold | 18.4849 |
309
- | manhattan_precision | 1.0 |
310
- | manhattan_recall | 0.9818 |
311
- | manhattan_ap | 1.0 |
312
- | euclidean_accuracy | 0.9818 |
313
- | euclidean_accuracy_threshold | 1.2089 |
314
- | euclidean_f1 | 0.9908 |
315
- | euclidean_f1_threshold | 1.2089 |
316
- | euclidean_precision | 1.0 |
317
- | euclidean_recall | 0.9818 |
318
- | euclidean_ap | 1.0 |
319
- | max_accuracy | 0.9818 |
320
- | max_accuracy_threshold | 18.4849 |
321
- | max_f1 | 0.9908 |
322
- | max_f1_threshold | 18.4849 |
323
- | max_precision | 1.0 |
324
- | max_recall | 0.9818 |
325
- | **max_ap** | **1.0** |
326
 
327
  <!--
328
  ## Bias, Risks and Limitations
@@ -343,24 +343,24 @@ You can finetune this model on your own dataset.
343
  #### Unnamed Dataset
344
 
345
 
346
- * Size: 216 training samples
347
  * Columns: <code>text1</code>, <code>text2</code>, and <code>label</code>
348
  * Approximate statistics based on the first 1000 samples:
349
- | | text1 | text2 | label |
350
- |:--------|:---------------------------------------------------------------------------------|:---------------------------------------------------------------------------------|:-----------------------------|
351
- | type | string | string | int |
352
- | details | <ul><li>min: 3 tokens</li><li>mean: 7.19 tokens</li><li>max: 13 tokens</li></ul> | <ul><li>min: 3 tokens</li><li>mean: 7.49 tokens</li><li>max: 13 tokens</li></ul> | <ul><li>1: 100.00%</li></ul> |
353
  * Samples:
354
- | text1 | text2 | label |
355
- |:-------------------------------------------------|:---------------------------------------------------|:---------------|
356
- | <code>Let's search inside</code> | <code>Let's look inside</code> | <code>1</code> |
357
- | <code>Do you see your scarf in the wagon?</code> | <code>Is your scarf in the wagon?</code> | <code>1</code> |
358
- | <code>Scarf on the tree.</code> | <code>Is that a scarf, the one on the tree?</code> | <code>1</code> |
359
- * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:
360
  ```json
361
  {
362
  "scale": 20.0,
363
- "similarity_fct": "cos_sim"
364
  }
365
  ```
366
 
@@ -369,24 +369,24 @@ You can finetune this model on your own dataset.
369
  #### Unnamed Dataset
370
 
371
 
372
- * Size: 55 evaluation samples
373
  * Columns: <code>text1</code>, <code>text2</code>, and <code>label</code>
374
  * Approximate statistics based on the first 1000 samples:
375
- | | text1 | text2 | label |
376
- |:--------|:---------------------------------------------------------------------------------|:---------------------------------------------------------------------------------|:-----------------------------|
377
- | type | string | string | int |
378
- | details | <ul><li>min: 3 tokens</li><li>mean: 7.04 tokens</li><li>max: 12 tokens</li></ul> | <ul><li>min: 3 tokens</li><li>mean: 7.55 tokens</li><li>max: 13 tokens</li></ul> | <ul><li>1: 100.00%</li></ul> |
379
  * Samples:
380
- | text1 | text2 | label |
381
- |:---------------------------------|:-----------------------------------|:---------------|
382
- | <code>A candle</code> | <code>I want a candle</code> | <code>1</code> |
383
- | <code>I did </code> | <code>I did it</code> | <code>1</code> |
384
- | <code>When you had dinner</code> | <code>Before cooking dinner</code> | <code>1</code> |
385
- * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:
386
  ```json
387
  {
388
  "scale": 20.0,
389
- "similarity_fct": "cos_sim"
390
  }
391
  ```
392
 
@@ -520,28 +520,28 @@ You can finetune this model on your own dataset.
520
  ### Training Logs
521
  | Epoch | Step | Training Loss | loss | custom-arc-semantics-data_max_ap |
522
  |:-----:|:----:|:-------------:|:------:|:--------------------------------:|
523
- | None | 0 | - | - | 1.0 |
524
- | 1.0 | 27 | 0.2251 | 0.1920 | 1.0 |
525
- | 2.0 | 54 | 0.1218 | 0.1768 | 1.0 |
526
- | 3.0 | 81 | 0.0466 | 0.1644 | 1.0 |
527
- | 4.0 | 108 | 0.0231 | 0.1514 | 1.0 |
528
- | 5.0 | 135 | 0.0161 | 0.1374 | 1.0 |
529
- | 6.0 | 162 | 0.0119 | 0.1339 | 1.0 |
530
- | 7.0 | 189 | 0.0091 | 0.1331 | 1.0 |
531
- | 8.0 | 216 | 0.0074 | 0.1292 | 1.0 |
532
- | 9.0 | 243 | 0.0054 | 0.1265 | 1.0 |
533
- | 10.0 | 270 | 0.0059 | 0.1244 | 1.0 |
534
- | 11.0 | 297 | 0.0055 | 0.1254 | 1.0 |
535
- | 12.0 | 324 | 0.0068 | 0.1236 | 1.0 |
536
- | 13.0 | 351 | 0.0035 | 0.1234 | 1.0 |
537
 
538
 
539
  ### Framework Versions
540
  - Python: 3.10.14
541
  - Sentence Transformers: 3.0.1
542
  - Transformers: 4.44.2
543
- - PyTorch: 2.4.0+cu121
544
- - Accelerate: 0.34.0
545
  - Datasets: 2.20.0
546
  - Tokenizers: 0.19.1
547
 
@@ -562,15 +562,14 @@ You can finetune this model on your own dataset.
562
  }
563
  ```
564
 
565
- #### MultipleNegativesRankingLoss
566
  ```bibtex
567
- @misc{henderson2017efficient,
568
- title={Efficient Natural Language Response Suggestion for Smart Reply},
569
- author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
570
- year={2017},
571
- eprint={1705.00652},
572
- archivePrefix={arXiv},
573
- primaryClass={cs.CL}
574
  }
575
  ```
576
 
 
45
  - sentence-similarity
46
  - feature-extraction
47
  - generated_from_trainer
48
+ - dataset_size:560
49
+ - loss:CoSENTLoss
50
  widget:
51
+ - source_sentence: Let's search inside
52
  sentences:
53
+ - Stuffed animal
54
+ - Let's look inside
55
+ - What is worse?
56
+ - source_sentence: I want a torch
57
  sentences:
58
+ - What do you think of Spike
59
+ - Actually I want a torch
60
+ - Why candle?
61
+ - source_sentence: Magic trace
62
  sentences:
63
+ - A sword.
64
+ - ' Why is he so tiny?'
65
+ - 'The flower is changed into flower. '
66
+ - source_sentence: Did you use illusion?
67
  sentences:
68
+ - Do you use illusion?
69
+ - You are a cat?
70
+ - It's Toby
71
+ - source_sentence: Do you see your scarf in the watering can?
72
  sentences:
73
+ - What is the Weeping Tree?
74
+ - Are these your footprints?
75
+ - Magic user
76
  model-index:
77
  - name: SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2
78
  results:
 
84
  type: custom-arc-semantics-data
85
  metrics:
86
  - type: cosine_accuracy
87
+ value: 0.9285714285714286
88
  name: Cosine Accuracy
89
  - type: cosine_accuracy_threshold
90
+ value: 0.42927420139312744
91
  name: Cosine Accuracy Threshold
92
  - type: cosine_f1
93
+ value: 0.9425287356321839
94
  name: Cosine F1
95
  - type: cosine_f1_threshold
96
+ value: 0.2269928753376007
97
  name: Cosine F1 Threshold
98
  - type: cosine_precision
99
+ value: 0.9111111111111111
100
  name: Cosine Precision
101
  - type: cosine_recall
102
+ value: 0.9761904761904762
103
  name: Cosine Recall
104
  - type: cosine_ap
105
+ value: 0.9720863676601571
106
  name: Cosine Ap
107
  - type: dot_accuracy
108
+ value: 0.9285714285714286
109
  name: Dot Accuracy
110
  - type: dot_accuracy_threshold
111
+ value: 0.42927438020706177
112
  name: Dot Accuracy Threshold
113
  - type: dot_f1
114
+ value: 0.9425287356321839
115
  name: Dot F1
116
  - type: dot_f1_threshold
117
+ value: 0.22699296474456787
118
  name: Dot F1 Threshold
119
  - type: dot_precision
120
+ value: 0.9111111111111111
121
  name: Dot Precision
122
  - type: dot_recall
123
+ value: 0.9761904761904762
124
  name: Dot Recall
125
  - type: dot_ap
126
+ value: 0.9720863676601571
127
  name: Dot Ap
128
  - type: manhattan_accuracy
129
+ value: 0.9285714285714286
130
  name: Manhattan Accuracy
131
  - type: manhattan_accuracy_threshold
132
+ value: 16.630834579467773
133
  name: Manhattan Accuracy Threshold
134
  - type: manhattan_f1
135
+ value: 0.9431818181818182
136
  name: Manhattan F1
137
  - type: manhattan_f1_threshold
138
+ value: 19.740108489990234
139
  name: Manhattan F1 Threshold
140
  - type: manhattan_precision
141
+ value: 0.9021739130434783
142
  name: Manhattan Precision
143
  - type: manhattan_recall
144
+ value: 0.9880952380952381
145
  name: Manhattan Recall
146
  - type: manhattan_ap
147
+ value: 0.9728353486982702
148
  name: Manhattan Ap
149
  - type: euclidean_accuracy
150
+ value: 0.9285714285714286
151
  name: Euclidean Accuracy
152
  - type: euclidean_accuracy_threshold
153
+ value: 1.068155288696289
154
  name: Euclidean Accuracy Threshold
155
  - type: euclidean_f1
156
+ value: 0.9425287356321839
157
  name: Euclidean F1
158
  - type: euclidean_f1_threshold
159
+ value: 1.2433418035507202
160
  name: Euclidean F1 Threshold
161
  - type: euclidean_precision
162
+ value: 0.9111111111111111
163
  name: Euclidean Precision
164
  - type: euclidean_recall
165
+ value: 0.9761904761904762
166
  name: Euclidean Recall
167
  - type: euclidean_ap
168
+ value: 0.9720863676601571
169
  name: Euclidean Ap
170
  - type: max_accuracy
171
+ value: 0.9285714285714286
172
  name: Max Accuracy
173
  - type: max_accuracy_threshold
174
+ value: 16.630834579467773
175
  name: Max Accuracy Threshold
176
  - type: max_f1
177
+ value: 0.9431818181818182
178
  name: Max F1
179
  - type: max_f1_threshold
180
+ value: 19.740108489990234
181
  name: Max F1 Threshold
182
  - type: max_precision
183
+ value: 0.9111111111111111
184
  name: Max Precision
185
  - type: max_recall
186
+ value: 0.9880952380952381
187
  name: Max Recall
188
  - type: max_ap
189
+ value: 0.9728353486982702
190
  name: Max Ap
191
  ---
192
 
 
240
  model = SentenceTransformer("LeoChiuu/all-MiniLM-L6-v2")
241
  # Run inference
242
  sentences = [
243
+ 'Do you see your scarf in the watering can?',
244
+ 'Are these your footprints?',
245
+ 'Magic user',
246
  ]
247
  embeddings = model.encode(sentences)
248
  print(embeddings.shape)
 
286
  * Dataset: `custom-arc-semantics-data`
287
  * Evaluated with [<code>BinaryClassificationEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.BinaryClassificationEvaluator)
288
 
289
+ | Metric | Value |
290
+ |:-----------------------------|:-----------|
291
+ | cosine_accuracy | 0.9286 |
292
+ | cosine_accuracy_threshold | 0.4293 |
293
+ | cosine_f1 | 0.9425 |
294
+ | cosine_f1_threshold | 0.227 |
295
+ | cosine_precision | 0.9111 |
296
+ | cosine_recall | 0.9762 |
297
+ | cosine_ap | 0.9721 |
298
+ | dot_accuracy | 0.9286 |
299
+ | dot_accuracy_threshold | 0.4293 |
300
+ | dot_f1 | 0.9425 |
301
+ | dot_f1_threshold | 0.227 |
302
+ | dot_precision | 0.9111 |
303
+ | dot_recall | 0.9762 |
304
+ | dot_ap | 0.9721 |
305
+ | manhattan_accuracy | 0.9286 |
306
+ | manhattan_accuracy_threshold | 16.6308 |
307
+ | manhattan_f1 | 0.9432 |
308
+ | manhattan_f1_threshold | 19.7401 |
309
+ | manhattan_precision | 0.9022 |
310
+ | manhattan_recall | 0.9881 |
311
+ | manhattan_ap | 0.9728 |
312
+ | euclidean_accuracy | 0.9286 |
313
+ | euclidean_accuracy_threshold | 1.0682 |
314
+ | euclidean_f1 | 0.9425 |
315
+ | euclidean_f1_threshold | 1.2433 |
316
+ | euclidean_precision | 0.9111 |
317
+ | euclidean_recall | 0.9762 |
318
+ | euclidean_ap | 0.9721 |
319
+ | max_accuracy | 0.9286 |
320
+ | max_accuracy_threshold | 16.6308 |
321
+ | max_f1 | 0.9432 |
322
+ | max_f1_threshold | 19.7401 |
323
+ | max_precision | 0.9111 |
324
+ | max_recall | 0.9881 |
325
+ | **max_ap** | **0.9728** |
326
 
327
  <!--
328
  ## Bias, Risks and Limitations
 
343
  #### Unnamed Dataset
344
 
345
 
346
+ * Size: 560 training samples
347
  * Columns: <code>text1</code>, <code>text2</code>, and <code>label</code>
348
  * Approximate statistics based on the first 1000 samples:
349
+ | | text1 | text2 | label |
350
+ |:--------|:--------------------------------------------------------------------------------|:---------------------------------------------------------------------------------|:------------------------------------------------|
351
+ | type | string | string | int |
352
+ | details | <ul><li>min: 3 tokens</li><li>mean: 7.2 tokens</li><li>max: 18 tokens</li></ul> | <ul><li>min: 3 tokens</li><li>mean: 7.26 tokens</li><li>max: 18 tokens</li></ul> | <ul><li>0: ~36.07%</li><li>1: ~63.93%</li></ul> |
353
  * Samples:
354
+ | text1 | text2 | label |
355
+ |:-----------------------------------------------------|:--------------------------------------------------------------------------|:---------------|
356
+ | <code>When it was dinner</code> | <code>Dinner time</code> | <code>1</code> |
357
+ | <code>Did you cook chicken noodle last night?</code> | <code>Did you make chicken noodle for dinner?</code> | <code>1</code> |
358
+ | <code>Someone who can change item</code> | <code>Someone who uses magic that turns something into something. </code> | <code>1</code> |
359
+ * Loss: [<code>CoSENTLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cosentloss) with these parameters:
360
  ```json
361
  {
362
  "scale": 20.0,
363
+ "similarity_fct": "pairwise_cos_sim"
364
  }
365
  ```
366
 
 
369
  #### Unnamed Dataset
370
 
371
 
372
+ * Size: 140 evaluation samples
373
  * Columns: <code>text1</code>, <code>text2</code>, and <code>label</code>
374
  * Approximate statistics based on the first 1000 samples:
375
+ | | text1 | text2 | label |
376
+ |:--------|:---------------------------------------------------------------------------------|:---------------------------------------------------------------------------------|:------------------------------------------------|
377
+ | type | string | string | int |
378
+ | details | <ul><li>min: 3 tokens</li><li>mean: 6.99 tokens</li><li>max: 18 tokens</li></ul> | <ul><li>min: 3 tokens</li><li>mean: 7.29 tokens</li><li>max: 18 tokens</li></ul> | <ul><li>0: ~40.00%</li><li>1: ~60.00%</li></ul> |
379
  * Samples:
380
+ | text1 | text2 | label |
381
+ |:-----------------------------------------|:-----------------------------------------|:---------------|
382
+ | <code>Let's check inside</code> | <code>Let's search inside</code> | <code>1</code> |
383
+ | <code>Sohpie, are you okay?</code> | <code>Sophie Are you pressured?</code> | <code>0</code> |
384
+ | <code>This wine glass is related.</code> | <code>This sword looks important.</code> | <code>0</code> |
385
+ * Loss: [<code>CoSENTLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cosentloss) with these parameters:
386
  ```json
387
  {
388
  "scale": 20.0,
389
+ "similarity_fct": "pairwise_cos_sim"
390
  }
391
  ```
392
 
 
520
  ### Training Logs
521
  | Epoch | Step | Training Loss | loss | custom-arc-semantics-data_max_ap |
522
  |:-----:|:----:|:-------------:|:------:|:--------------------------------:|
523
+ | None | 0 | - | - | 0.9254 |
524
+ | 1.0 | 70 | 2.9684 | 1.4087 | 0.9425 |
525
+ | 2.0 | 140 | 1.4461 | 1.0942 | 0.9629 |
526
+ | 3.0 | 210 | 0.6005 | 0.8398 | 0.9680 |
527
+ | 4.0 | 280 | 0.3021 | 0.7577 | 0.9703 |
528
+ | 5.0 | 350 | 0.2412 | 0.7216 | 0.9715 |
529
+ | 6.0 | 420 | 0.1816 | 0.7538 | 0.9722 |
530
+ | 7.0 | 490 | 0.1512 | 0.8049 | 0.9726 |
531
+ | 8.0 | 560 | 0.1208 | 0.7602 | 0.9726 |
532
+ | 9.0 | 630 | 0.0915 | 0.7286 | 0.9729 |
533
+ | 10.0 | 700 | 0.0553 | 0.7072 | 0.9729 |
534
+ | 11.0 | 770 | 0.0716 | 0.6984 | 0.9730 |
535
+ | 12.0 | 840 | 0.0297 | 0.7063 | 0.9725 |
536
+ | 13.0 | 910 | 0.0462 | 0.6997 | 0.9728 |
537
 
538
 
539
  ### Framework Versions
540
  - Python: 3.10.14
541
  - Sentence Transformers: 3.0.1
542
  - Transformers: 4.44.2
543
+ - PyTorch: 2.4.1+cu121
544
+ - Accelerate: 0.34.2
545
  - Datasets: 2.20.0
546
  - Tokenizers: 0.19.1
547
 
 
562
  }
563
  ```
564
 
565
+ #### CoSENTLoss
566
  ```bibtex
567
+ @online{kexuefm-8847,
568
+ title={CoSENT: A more efficient sentence vector scheme than Sentence-BERT},
569
+ author={Su Jianlin},
570
+ year={2022},
571
+ month={Jan},
572
+ url={https://kexue.fm/archives/8847},
 
573
  }
574
  ```
575
 
config_sentence_transformers.json CHANGED
@@ -2,7 +2,7 @@
2
  "__version__": {
3
  "sentence_transformers": "3.0.1",
4
  "transformers": "4.44.2",
5
- "pytorch": "2.4.0+cu121"
6
  },
7
  "prompts": {},
8
  "default_prompt_name": null,
 
2
  "__version__": {
3
  "sentence_transformers": "3.0.1",
4
  "transformers": "4.44.2",
5
+ "pytorch": "2.4.1+cu121"
6
  },
7
  "prompts": {},
8
  "default_prompt_name": null,
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a6cc633b22dd645d5916470915970b032282a71c48bdd15a97c49a86bb8bc83b
3
  size 90864192
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d9ab6b7472780e4b9271e02f535d125c33cef1b145ab2f8d3135ed97c72aea5
3
  size 90864192