Hugging Face's logo

lv12
/
esci-nomic-embed-text-v1_5_4

Sentence Similarity
sentence-transformers
Safetensors
nomic_bert
feature-extraction
Generated from Trainer
dataset_size:1600000
loss:TripletLoss
custom_code
Eval Results
text-embeddings-inference
Inference Endpoints
Model card Files Community
lv12 commited on
Commit
b64e7e1
•
1 Parent(s): f4db65a

multi stage tuning

Browse files
Files changed (2) hide show
  1. README.md +322 -324
  2. model.safetensors +1 -1
README.md CHANGED
@@ -6,7 +6,7 @@ tags:
6
  - sentence-similarity
7
  - feature-extraction
8
  - generated_from_trainer
9
- - dataset_size:1500000
10
  - loss:TripletLoss
11
  datasets: []
12
  metrics:
@@ -38,52 +38,48 @@ metrics:
38
  - dot_mrr@10
39
  - dot_map@10
40
  widget:
41
- - source_sentence: 'search_query: speeder man slippers for boys size 12'
42
  sentences:
43
- - 'search_document: Marvel Boys Spider-Man Bootie Slippers (11-12 M US Little Kid),
44
- BBC International, Red/Blue'
45
- - 'search_document: MIXIN Little Kids Boys Spring Cozy Comfort Comfy Slippers Size
46
- Dark Blue 13 13.5 M, MIXIN, Blue'
47
- - 'search_document: Third Eye Blind, , '
48
- - source_sentence: 'search_query: water filter refrigerator whirlpool'
 
49
  sentences:
50
- - 'search_document: Assorted Tootsie Frooties - 2 lbs of Delicious Assorted Bulk
51
- Wrapped Candy with Refrigerator Magnet, Emporium Candy, '
52
- - 'search_document: Whirlpool WHER25 Reverse Osmosis (RO) Filtration System With
53
- Chrome Faucet | Extra Long Life | Easy To Replace UltraEase Filter Cartridges,
54
- White, Whirlpool, White'
55
- - 'search_document: Everydrop by Whirlpool Ice and Water Refrigerator Filter 3,
56
- EDR3RXD1, Single-Pack, White, EveryDrop by Whirlpool, White'
57
- - source_sentence: 'search_query: t towels kitchen'
 
58
  sentences:
59
- - 'search_document: Utopia Kitchen Flour Sack Tea Towels 24 Pack, 28" x 28" Ring
60
- Spun 100% Cotton Dish Cloths - Machine Washable - for Cleaning & Drying - White,
61
- Utopia Kitchen, White'
62
- - 'search_document: Rael Certified Organic Cotton Panty Liners (Regular, 44 Count)
63
- & Extra Long Overnight Pads (12 Count) Bundle, Rael, '
64
- - 'search_document: LSK Bath Hand Towel Holder Standing,T-Shape Towel Rack Countertop
65
- with Round Base for Kitchen Bathroom Toliet ,Rustproof Stainless Steel (Black),
66
- LSK, Black'
67
- - source_sentence: 'search_query: cozy lights bedroom'
68
  sentences:
69
- - 'search_document: Fluorescent Light Covers | Fluorescent Light Covers for Ceiling
70
- Lights, Classroom, Office, or Blue Light Covers Fluorescent Filter- Eliminates
71
- Flicker & Glare - 48" by 24" (4 Pack, Sky Blue Panel), Everyday Educate, Blue'
72
- - 'search_document: 43ft Led Globe String Lights, 100LEDs Outdoor Indoor String
73
- Lights Plug in, 8 Modes Twinkle Lights, Warm White, Fairy String Lights Christmas
74
- Decoration, Yard, Party, Bedroom, Yinuo Candle, White'
75
- - 'search_document: PARMIDA 6 inch Dimmable LED Square Recessed Retrofit Lighting,
76
- Easy Downlight Installation, 12W (100W Eqv.), 950lm, Ceiling Can Lights, Energy
77
- Star & ETL-Listed, 5 Year Warranty, 3000K - 4 Pack, Parmida LED Technologies,
78
- 3000k (Soft White)'
79
- - source_sentence: 'search_query: punxhing bag'
80
  sentences:
81
- - 'search_document: Power Core Bag (EA), Everlast, Black/White'
82
- - 'search_document: Nifeida High Mount 3rd Brake Light for 2004-2008 Ford F150,
83
- 2007-2010 Ford Explorer, 2006-2008 Lincoln Mark LT 22LEDs Waterproof Wedge Center
84
- Tail Light Heavy Duty Third Stop Light Cargo Lamp, nifeida, '
85
- - 'search_document: Everlast 70-Pound MMA Poly Canvas Heavy Bag (Black), Everlast,
86
- Black'
 
87
  pipeline_tag: sentence-similarity
88
  model-index:
89
  - name: SentenceTransformer
@@ -96,19 +92,19 @@ model-index:
96
  type: unknown
97
  metrics:
98
  - type: cosine_accuracy
99
- value: 0.7297333333333333
100
  name: Cosine Accuracy
101
  - type: dot_accuracy
102
- value: 0.2886
103
  name: Dot Accuracy
104
  - type: manhattan_accuracy
105
- value: 0.7276666666666667
106
  name: Manhattan Accuracy
107
  - type: euclidean_accuracy
108
- value: 0.7302
109
  name: Euclidean Accuracy
110
  - type: max_accuracy
111
- value: 0.7302
112
  name: Max Accuracy
113
  - task:
114
  type: semantic-similarity
@@ -118,34 +114,34 @@ model-index:
118
  type: unknown
119
  metrics:
120
  - type: pearson_cosine
121
- value: 0.4148970258272059
122
  name: Pearson Cosine
123
  - type: spearman_cosine
124
- value: 0.398627797786331
125
  name: Spearman Cosine
126
  - type: pearson_manhattan
127
- value: 0.37912086775286635
128
  name: Pearson Manhattan
129
  - type: spearman_manhattan
130
- value: 0.3706070496698137
131
  name: Spearman Manhattan
132
  - type: pearson_euclidean
133
- value: 0.3795867481634979
134
  name: Pearson Euclidean
135
  - type: spearman_euclidean
136
- value: 0.3712739507251931
137
  name: Spearman Euclidean
138
  - type: pearson_dot
139
- value: 0.37803301496759
140
  name: Pearson Dot
141
  - type: spearman_dot
142
- value: 0.37716678508954316
143
  name: Spearman Dot
144
  - type: pearson_max
145
- value: 0.4148970258272059
146
  name: Pearson Max
147
  - type: spearman_max
148
- value: 0.398627797786331
149
  name: Spearman Max
150
  - task:
151
  type: information-retrieval
@@ -155,40 +151,40 @@ model-index:
155
  type: unknown
156
  metrics:
157
  - type: cosine_accuracy@10
158
- value: 0.97
159
  name: Cosine Accuracy@10
160
  - type: cosine_precision@10
161
- value: 0.6959
162
  name: Cosine Precision@10
163
  - type: cosine_recall@10
164
- value: 0.6232158089273498
165
  name: Cosine Recall@10
166
  - type: cosine_ndcg@10
167
- value: 0.8306950477475292
168
  name: Cosine Ndcg@10
169
  - type: cosine_mrr@10
170
- value: 0.91050753968254
171
  name: Cosine Mrr@10
172
  - type: cosine_map@10
173
- value: 0.7765982865646258
174
  name: Cosine Map@10
175
  - type: dot_accuracy@10
176
- value: 0.945
177
  name: Dot Accuracy@10
178
  - type: dot_precision@10
179
- value: 0.6282000000000001
180
  name: Dot Precision@10
181
  - type: dot_recall@10
182
- value: 0.5617500930766831
183
  name: Dot Recall@10
184
  - type: dot_ndcg@10
185
- value: 0.7560665190218724
186
  name: Dot Ndcg@10
187
  - type: dot_mrr@10
188
- value: 0.8675638888888889
189
  name: Dot Mrr@10
190
  - type: dot_map@10
191
- value: 0.682285591931217
192
  name: Dot Map@10
193
  ---
194
 
@@ -242,9 +238,9 @@ from sentence_transformers import SentenceTransformer
242
  model = SentenceTransformer("lv12/esci-nomic-embed-text-v1_5_4")
243
  # Run inference
244
  sentences = [
245
- 'search_query: punxhing bag',
246
- 'search_document: Everlast 70-Pound MMA Poly Canvas Heavy Bag (Black), Everlast, Black',
247
- 'search_document: Power Core Bag (EA), Everlast, Black/White',
248
  ]
249
  embeddings = model.encode(sentences)
250
  print(embeddings.shape)
@@ -290,11 +286,11 @@ You can finetune this model on your own dataset.
290
 
291
  | Metric | Value |
292
  |:--------------------|:-----------|
293
- | **cosine_accuracy** | **0.7297** |
294
- | dot_accuracy | 0.2886 |
295
- | manhattan_accuracy | 0.7277 |
296
- | euclidean_accuracy | 0.7302 |
297
- | max_accuracy | 0.7302 |
298
 
299
  #### Semantic Similarity
300
 
@@ -302,16 +298,16 @@ You can finetune this model on your own dataset.
302
 
303
  | Metric | Value |
304
  |:--------------------|:-----------|
305
- | pearson_cosine | 0.4149 |
306
- | **spearman_cosine** | **0.3986** |
307
- | pearson_manhattan | 0.3791 |
308
- | spearman_manhattan | 0.3706 |
309
- | pearson_euclidean | 0.3796 |
310
- | spearman_euclidean | 0.3713 |
311
- | pearson_dot | 0.378 |
312
- | spearman_dot | 0.3772 |
313
- | pearson_max | 0.4149 |
314
- | spearman_max | 0.3986 |
315
 
316
  #### Information Retrieval
317
 
@@ -319,18 +315,18 @@ You can finetune this model on your own dataset.
319
 
320
  | Metric | Value |
321
  |:--------------------|:-----------|
322
- | cosine_accuracy@10 | 0.97 |
323
- | cosine_precision@10 | 0.6959 |
324
- | cosine_recall@10 | 0.6232 |
325
- | cosine_ndcg@10 | 0.8307 |
326
- | cosine_mrr@10 | 0.9105 |
327
- | **cosine_map@10** | **0.7766** |
328
- | dot_accuracy@10 | 0.945 |
329
- | dot_precision@10 | 0.6282 |
330
- | dot_recall@10 | 0.5618 |
331
- | dot_ndcg@10 | 0.7561 |
332
- | dot_mrr@10 | 0.8676 |
333
- | dot_map@10 | 0.6823 |
334
 
335
  <!--
336
  ## Bias, Risks and Limitations
@@ -351,24 +347,24 @@ You can finetune this model on your own dataset.
351
  #### triplets
352
 
353
  * Dataset: triplets
354
- * Size: 1,500,000 training samples
355
  * Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
356
  * Approximate statistics based on the first 1000 samples:
357
- | | anchor | positive | negative |
358
- |:--------|:---------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|
359
- | type | string | string | string |
360
- | details | <ul><li>min: 7 tokens</li><li>mean: 11.0 tokens</li><li>max: 27 tokens</li></ul> | <ul><li>min: 9 tokens</li><li>mean: 38.87 tokens</li><li>max: 91 tokens</li></ul> | <ul><li>min: 9 tokens</li><li>mean: 39.17 tokens</li><li>max: 96 tokens</li></ul> |
361
  * Samples:
362
- | anchor | positive | negative |
363
- |:--------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
364
- | <code>search_query: chiffon ivory fabric</code> | <code>search_document: DJBM 59’’ Solid Color Sheer Chiffon Fabric Yards Continuous All Colors for DIY Decoration Valance Ivory/1 Yard, DJBM, Ivory</code> | <code>search_document: 58" Blush Solid Color Sheer Chiffon Fabric by The Bolt - 25 Yards, Stylish FABRIC, </code> |
365
- | <code>search_query: tummy control lingerie for women for sex</code> | <code>search_document: Women's Plus Size Chemise Floral Lace Lingerie Sexy Bodysuit Mesh Babydoll Sleepwear (R007,White, XXXX-Large), Chic Lover, White</code> | <code>search_document: Avidlove Lingerie for Women Teddy One Piece Lace Babydoll Bodysuit Black Medium, Avidlove, Black</code> |
366
- | <code>search_query: photo metal prints</code> | <code>search_document: Smile Art Design Personalized Photo Print Desktop Photo Panel Print with Kickstand Upload Your Tabletop Berlin Shape Frame Custom Photo Print Personalized Gift for Man Woman 5" x7", Smile Art Design, Berlin 5" X 7"</code> | <code>search_document: They Whispered to Her You Cannot Withstand The Storm - Positive Motivational Uplifting Encouragement Gifts for Women Teens - Inspirational Quote Wall Art - Boho Decoration Print - Dragonfly Wall Decor, Yellowbird Art & Design, </code> |
367
  * Loss: [<code>TripletLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#tripletloss) with these parameters:
368
  ```json
369
  {
370
  "distance_metric": "TripletDistanceMetric.COSINE",
371
- "triplet_margin": 0.7
372
  }
373
  ```
374
 
@@ -377,37 +373,39 @@ You can finetune this model on your own dataset.
377
  #### triplets
378
 
379
  * Dataset: triplets
380
- * Size: 15,000 evaluation samples
381
  * Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
382
  * Approximate statistics based on the first 1000 samples:
383
- | | anchor | positive | negative |
384
- |:--------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
385
- | type | string | string | string |
386
- | details | <ul><li>min: 7 tokens</li><li>mean: 11.14 tokens</li><li>max: 31 tokens</li></ul> | <ul><li>min: 9 tokens</li><li>mean: 38.83 tokens</li><li>max: 105 tokens</li></ul> | <ul><li>min: 9 tokens</li><li>mean: 39.12 tokens</li><li>max: 113 tokens</li></ul> |
387
  * Samples:
388
- | anchor | positive | negative |
389
- |:------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
390
- | <code>search_query: a cold dark place</code> | <code>search_document: A Cold Dark Place (An Emily Kenyon Thriller Book 1), , </code> | <code>search_document: The Great Alone, , </code> |
391
- | <code>search_query: quay sunglasses</code> | <code>search_document: Quay Women's Hindsight Sunglasses, Matte Black/Rainbow Mirror, Quay, Multicolor</code> | <code>search_document: SORVINO Aviator Sunglasses for Women Classic Oversized Sun Glasses UV400 Protection (A-Black Frame/Faded Lens, 60), SORVINO, Black Frame/Faded Lens</code> |
392
- | <code>search_query: baby girl jean fress</code> | <code>search_document: Toddler Baby Girl Play Wear Floral Princess Summer Dresses with Bow Ruffle Denim Skirt Set for Summer 9-12months Light Blue, NZRVAWS, </code> | <code>search_document: Slowera Baby Toddler Girls Denim Ruffled Bodysuit Blue Soft One-Piece Romper (Blue, 6-12 Months), Slowera, Blue</code> |
393
  * Loss: [<code>TripletLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#tripletloss) with these parameters:
394
  ```json
395
  {
396
  "distance_metric": "TripletDistanceMetric.COSINE",
397
- "triplet_margin": 0.7
398
  }
399
  ```
400
 
401
  ### Training Hyperparameters
402
  #### Non-Default Hyperparameters
403
 
404
- - `per_device_train_batch_size`: 32
405
  - `per_device_eval_batch_size`: 16
406
  - `gradient_accumulation_steps`: 2
407
  - `learning_rate`: 1e-07
 
408
  - `lr_scheduler_type`: polynomial
409
  - `lr_scheduler_kwargs`: {'lr_end': 1e-08, 'power': 2.0}
410
  - `warmup_ratio`: 0.05
 
411
  - `dataloader_num_workers`: 4
412
  - `dataloader_prefetch_factor`: 4
413
  - `load_best_model_at_end`: True
@@ -421,7 +419,7 @@ You can finetune this model on your own dataset.
421
  - `overwrite_output_dir`: False
422
  - `do_predict`: False
423
  - `prediction_loss_only`: True
424
- - `per_device_train_batch_size`: 32
425
  - `per_device_eval_batch_size`: 16
426
  - `per_gpu_train_batch_size`: None
427
  - `per_gpu_eval_batch_size`: None
@@ -433,7 +431,7 @@ You can finetune this model on your own dataset.
433
  - `adam_beta2`: 0.999
434
  - `adam_epsilon`: 1e-08
435
  - `max_grad_norm`: 1.0
436
- - `num_train_epochs`: 3
437
  - `max_steps`: -1
438
  - `lr_scheduler_type`: polynomial
439
  - `lr_scheduler_kwargs`: {'lr_end': 1e-08, 'power': 2.0}
@@ -529,206 +527,206 @@ You can finetune this model on your own dataset.
529
 
530
  | Epoch | Step | Training Loss | triplets loss | cosine_accuracy | cosine_map@10 | spearman_cosine |
531
  |:------:|:----:|:-------------:|:-------------:|:---------------:|:-------------:|:---------------:|
532
- | 0.0004 | 10 | 0.6484 | - | - | - | - |
533
- | 0.0009 | 20 | 0.6538 | - | - | - | - |
534
- | 0.0013 | 30 | 0.651 | - | - | - | - |
535
- | 0.0017 | 40 | 0.6491 | - | - | - | - |
536
- | 0.0021 | 50 | 0.6433 | - | - | - | - |
537
- | 0.0026 | 60 | 0.6545 | - | - | - | - |
538
- | 0.0030 | 70 | 0.6459 | - | - | - | - |
539
- | 0.0034 | 80 | 0.6517 | - | - | - | - |
540
- | 0.0038 | 90 | 0.6505 | - | - | - | - |
541
- | 0.0043 | 100 | 0.6477 | 0.6560 | 0.7255 | 0.7684 | 0.3931 |
542
- | 0.0047 | 110 | 0.6493 | - | - | - | - |
543
- | 0.0051 | 120 | 0.6514 | - | - | - | - |
544
- | 0.0055 | 130 | 0.6482 | - | - | - | - |
545
- | 0.0060 | 140 | 0.6566 | - | - | - | - |
546
- | 0.0064 | 150 | 0.6476 | - | - | - | - |
547
- | 0.0068 | 160 | 0.6523 | - | - | - | - |
548
- | 0.0073 | 170 | 0.6503 | - | - | - | - |
549
- | 0.0077 | 180 | 0.6568 | - | - | - | - |
550
- | 0.0081 | 190 | 0.6496 | - | - | - | - |
551
- | 0.0085 | 200 | 0.6512 | 0.6558 | 0.725 | 0.7687 | 0.3932 |
552
- | 0.0090 | 210 | 0.642 | - | - | - | - |
553
- | 0.0094 | 220 | 0.644 | - | - | - | - |
554
- | 0.0098 | 230 | 0.6512 | - | - | - | - |
555
- | 0.0102 | 240 | 0.6512 | - | - | - | - |
556
- | 0.0107 | 250 | 0.6477 | - | - | - | - |
557
- | 0.0111 | 260 | 0.6521 | - | - | - | - |
558
- | 0.0115 | 270 | 0.6487 | - | - | - | - |
559
- | 0.0119 | 280 | 0.6541 | - | - | - | - |
560
- | 0.0124 | 290 | 0.6481 | - | - | - | - |
561
- | 0.0128 | 300 | 0.6495 | 0.6555 | 0.7247 | 0.7690 | 0.3933 |
562
- | 0.0132 | 310 | 0.6536 | - | - | - | - |
563
- | 0.0137 | 320 | 0.6515 | - | - | - | - |
564
- | 0.0141 | 330 | 0.6526 | - | - | - | - |
565
- | 0.0145 | 340 | 0.6509 | - | - | - | - |
566
- | 0.0149 | 350 | 0.6506 | - | - | - | - |
567
- | 0.0154 | 360 | 0.6451 | - | - | - | - |
568
- | 0.0158 | 370 | 0.6505 | - | - | - | - |
569
- | 0.0162 | 380 | 0.6529 | - | - | - | - |
570
- | 0.0166 | 390 | 0.6553 | - | - | - | - |
571
- | 0.0171 | 400 | 0.6512 | 0.6550 | 0.7245 | 0.7693 | 0.3934 |
572
- | 0.0175 | 410 | 0.6478 | - | - | - | - |
573
- | 0.0179 | 420 | 0.6497 | - | - | - | - |
574
- | 0.0183 | 430 | 0.6498 | - | - | - | - |
575
- | 0.0188 | 440 | 0.6499 | - | - | - | - |
576
- | 0.0192 | 450 | 0.6462 | - | - | - | - |
577
- | 0.0196 | 460 | 0.6497 | - | - | - | - |
578
- | 0.0201 | 470 | 0.6408 | - | - | - | - |
579
- | 0.0205 | 480 | 0.6507 | - | - | - | - |
580
- | 0.0209 | 490 | 0.6505 | - | - | - | - |
581
- | 0.0213 | 500 | 0.6481 | 0.6545 | 0.7249 | 0.7694 | 0.3936 |
582
- | 0.0218 | 510 | 0.6436 | - | - | - | - |
583
- | 0.0222 | 520 | 0.653 | - | - | - | - |
584
- | 0.0226 | 530 | 0.652 | - | - | - | - |
585
- | 0.0230 | 540 | 0.6509 | - | - | - | - |
586
- | 0.0235 | 550 | 0.6469 | - | - | - | - |
587
- | 0.0239 | 560 | 0.6515 | - | - | - | - |
588
- | 0.0243 | 570 | 0.6485 | - | - | - | - |
589
- | 0.0247 | 580 | 0.6499 | - | - | - | - |
590
- | 0.0252 | 590 | 0.6472 | - | - | - | - |
591
- | 0.0256 | 600 | 0.6569 | 0.6538 | 0.7255 | 0.7702 | 0.3939 |
592
- | 0.0260 | 610 | 0.6457 | - | - | - | - |
593
- | 0.0265 | 620 | 0.6485 | - | - | - | - |
594
- | 0.0269 | 630 | 0.6467 | - | - | - | - |
595
- | 0.0273 | 640 | 0.6494 | - | - | - | - |
596
- | 0.0277 | 650 | 0.6492 | - | - | - | - |
597
- | 0.0282 | 660 | 0.6439 | - | - | - | - |
598
- | 0.0286 | 670 | 0.6468 | - | - | - | - |
599
- | 0.0290 | 680 | 0.6493 | - | - | - | - |
600
- | 0.0294 | 690 | 0.6483 | - | - | - | - |
601
- | 0.0299 | 700 | 0.6474 | 0.6530 | 0.7259 | 0.7704 | 0.3942 |
602
- | 0.0303 | 710 | 0.6512 | - | - | - | - |
603
- | 0.0307 | 720 | 0.6431 | - | - | - | - |
604
- | 0.0311 | 730 | 0.6484 | - | - | - | - |
605
- | 0.0316 | 740 | 0.6488 | - | - | - | - |
606
- | 0.032 | 750 | 0.6525 | - | - | - | - |
607
- | 0.0324 | 760 | 0.6464 | - | - | - | - |
608
- | 0.0329 | 770 | 0.6453 | - | - | - | - |
609
- | 0.0333 | 780 | 0.6471 | - | - | - | - |
610
- | 0.0337 | 790 | 0.6478 | - | - | - | - |
611
- | 0.0341 | 800 | 0.6456 | 0.6520 | 0.7262 | 0.7704 | 0.3945 |
612
- | 0.0346 | 810 | 0.6494 | - | - | - | - |
613
- | 0.0350 | 820 | 0.6374 | - | - | - | - |
614
- | 0.0354 | 830 | 0.6441 | - | - | - | - |
615
- | 0.0358 | 840 | 0.643 | - | - | - | - |
616
- | 0.0363 | 850 | 0.6471 | - | - | - | - |
617
- | 0.0367 | 860 | 0.6401 | - | - | - | - |
618
- | 0.0371 | 870 | 0.6381 | - | - | - | - |
619
- | 0.0375 | 880 | 0.6491 | - | - | - | - |
620
- | 0.0380 | 890 | 0.6428 | - | - | - | - |
621
- | 0.0384 | 900 | 0.6393 | 0.6509 | 0.7262 | 0.7712 | 0.3948 |
622
- | 0.0388 | 910 | 0.6453 | - | - | - | - |
623
- | 0.0393 | 920 | 0.6426 | - | - | - | - |
624
- | 0.0397 | 930 | 0.644 | - | - | - | - |
625
- | 0.0401 | 940 | 0.6469 | - | - | - | - |
626
- | 0.0405 | 950 | 0.6372 | - | - | - | - |
627
- | 0.0410 | 960 | 0.6491 | - | - | - | - |
628
- | 0.0414 | 970 | 0.6441 | - | - | - | - |
629
- | 0.0418 | 980 | 0.6394 | - | - | - | - |
630
- | 0.0422 | 990 | 0.6416 | - | - | - | - |
631
- | 0.0427 | 1000 | 0.6465 | 0.6497 | 0.7263 | 0.7719 | 0.3951 |
632
- | 0.0431 | 1010 | 0.6485 | - | - | - | - |
633
- | 0.0435 | 1020 | 0.6391 | - | - | - | - |
634
- | 0.0439 | 1030 | 0.6489 | - | - | - | - |
635
- | 0.0444 | 1040 | 0.6417 | - | - | - | - |
636
- | 0.0448 | 1050 | 0.6355 | - | - | - | - |
637
- | 0.0452 | 1060 | 0.6445 | - | - | - | - |
638
- | 0.0457 | 1070 | 0.6427 | - | - | - | - |
639
- | 0.0461 | 1080 | 0.6443 | - | - | - | - |
640
- | 0.0465 | 1090 | 0.6422 | - | - | - | - |
641
- | 0.0469 | 1100 | 0.6406 | 0.6483 | 0.7265 | 0.7721 | 0.3955 |
642
- | 0.0474 | 1110 | 0.6415 | - | - | - | - |
643
- | 0.0478 | 1120 | 0.6389 | - | - | - | - |
644
- | 0.0482 | 1130 | 0.6437 | - | - | - | - |
645
- | 0.0486 | 1140 | 0.6412 | - | - | - | - |
646
- | 0.0491 | 1150 | 0.6457 | - | - | - | - |
647
- | 0.0495 | 1160 | 0.6364 | - | - | - | - |
648
- | 0.0499 | 1170 | 0.6389 | - | - | - | - |
649
- | 0.0503 | 1180 | 0.6394 | - | - | - | - |
650
- | 0.0508 | 1190 | 0.6465 | - | - | - | - |
651
- | 0.0512 | 1200 | 0.6453 | 0.6468 | 0.7269 | 0.7734 | 0.3959 |
652
- | 0.0516 | 1210 | 0.6465 | - | - | - | - |
653
- | 0.0521 | 1220 | 0.6389 | - | - | - | - |
654
- | 0.0525 | 1230 | 0.6448 | - | - | - | - |
655
- | 0.0529 | 1240 | 0.6325 | - | - | - | - |
656
- | 0.0533 | 1250 | 0.6347 | - | - | - | - |
657
- | 0.0538 | 1260 | 0.6363 | - | - | - | - |
658
- | 0.0542 | 1270 | 0.6387 | - | - | - | - |
659
- | 0.0546 | 1280 | 0.641 | - | - | - | - |
660
- | 0.0550 | 1290 | 0.6381 | - | - | - | - |
661
- | 0.0555 | 1300 | 0.6412 | 0.6452 | 0.7273 | 0.7747 | 0.3962 |
662
- | 0.0559 | 1310 | 0.6384 | - | - | - | - |
663
- | 0.0563 | 1320 | 0.6391 | - | - | - | - |
664
- | 0.0567 | 1330 | 0.6327 | - | - | - | - |
665
- | 0.0572 | 1340 | 0.6388 | - | - | - | - |
666
- | 0.0576 | 1350 | 0.6328 | - | - | - | - |
667
- | 0.0580 | 1360 | 0.6327 | - | - | - | - |
668
- | 0.0585 | 1370 | 0.6333 | - | - | - | - |
669
- | 0.0589 | 1380 | 0.6388 | - | - | - | - |
670
- | 0.0593 | 1390 | 0.6373 | - | - | - | - |
671
- | 0.0597 | 1400 | 0.636 | 0.6434 | 0.7278 | 0.7749 | 0.3966 |
672
- | 0.0602 | 1410 | 0.6357 | - | - | - | - |
673
- | 0.0606 | 1420 | 0.6394 | - | - | - | - |
674
- | 0.0610 | 1430 | 0.6339 | - | - | - | - |
675
- | 0.0614 | 1440 | 0.6385 | - | - | - | - |
676
- | 0.0619 | 1450 | 0.6288 | - | - | - | - |
677
- | 0.0623 | 1460 | 0.6393 | - | - | - | - |
678
- | 0.0627 | 1470 | 0.638 | - | - | - | - |
679
- | 0.0631 | 1480 | 0.6353 | - | - | - | - |
680
- | 0.0636 | 1490 | 0.6335 | - | - | - | - |
681
- | 0.064 | 1500 | 0.6329 | 0.6414 | 0.7281 | 0.7749 | 0.3970 |
682
- | 0.0644 | 1510 | 0.6363 | - | - | - | - |
683
- | 0.0649 | 1520 | 0.6311 | - | - | - | - |
684
- | 0.0653 | 1530 | 0.6367 | - | - | - | - |
685
- | 0.0657 | 1540 | 0.636 | - | - | - | - |
686
- | 0.0661 | 1550 | 0.6351 | - | - | - | - |
687
- | 0.0666 | 1560 | 0.637 | - | - | - | - |
688
- | 0.0670 | 1570 | 0.6352 | - | - | - | - |
689
- | 0.0674 | 1580 | 0.6285 | - | - | - | - |
690
- | 0.0678 | 1590 | 0.6311 | - | - | - | - |
691
- | 0.0683 | 1600 | 0.6347 | 0.6393 | 0.7283 | 0.7752 | 0.3974 |
692
- | 0.0687 | 1610 | 0.6393 | - | - | - | - |
693
- | 0.0691 | 1620 | 0.6368 | - | - | - | - |
694
- | 0.0695 | 1630 | 0.6354 | - | - | - | - |
695
- | 0.0700 | 1640 | 0.6283 | - | - | - | - |
696
- | 0.0704 | 1650 | 0.6289 | - | - | - | - |
697
- | 0.0708 | 1660 | 0.6291 | - | - | - | - |
698
- | 0.0713 | 1670 | 0.6274 | - | - | - | - |
699
- | 0.0717 | 1680 | 0.6217 | - | - | - | - |
700
- | 0.0721 | 1690 | 0.6281 | - | - | - | - |
701
- | 0.0725 | 1700 | 0.6341 | 0.6371 | 0.7289 | 0.7757 | 0.3978 |
702
- | 0.0730 | 1710 | 0.6354 | - | - | - | - |
703
- | 0.0734 | 1720 | 0.6265 | - | - | - | - |
704
- | 0.0738 | 1730 | 0.6214 | - | - | - | - |
705
- | 0.0742 | 1740 | 0.6301 | - | - | - | - |
706
- | 0.0747 | 1750 | 0.621 | - | - | - | - |
707
- | 0.0751 | 1760 | 0.6259 | - | - | - | - |
708
- | 0.0755 | 1770 | 0.6261 | - | - | - | - |
709
- | 0.0759 | 1780 | 0.6273 | - | - | - | - |
710
- | 0.0764 | 1790 | 0.6311 | - | - | - | - |
711
- | 0.0768 | 1800 | 0.6232 | 0.6347 | 0.7292 | 0.7755 | 0.3981 |
712
- | 0.0772 | 1810 | 0.6293 | - | - | - | - |
713
- | 0.0777 | 1820 | 0.617 | - | - | - | - |
714
- | 0.0781 | 1830 | 0.6303 | - | - | - | - |
715
- | 0.0785 | 1840 | 0.6225 | - | - | - | - |
716
- | 0.0789 | 1850 | 0.6313 | - | - | - | - |
717
- | 0.0794 | 1860 | 0.6229 | - | - | - | - |
718
- | 0.0798 | 1870 | 0.6236 | - | - | - | - |
719
- | 0.0802 | 1880 | 0.6265 | - | - | - | - |
720
- | 0.0806 | 1890 | 0.6179 | - | - | - | - |
721
- | 0.0811 | 1900 | 0.6277 | 0.6321 | 0.7292 | 0.7760 | 0.3984 |
722
- | 0.0815 | 1910 | 0.6266 | - | - | - | - |
723
- | 0.0819 | 1920 | 0.6209 | - | - | - | - |
724
- | 0.0823 | 1930 | 0.6258 | - | - | - | - |
725
- | 0.0828 | 1940 | 0.6143 | - | - | - | - |
726
- | 0.0832 | 1950 | 0.6176 | - | - | - | - |
727
- | 0.0836 | 1960 | 0.628 | - | - | - | - |
728
- | 0.0841 | 1970 | 0.6147 | - | - | - | - |
729
- | 0.0845 | 1980 | 0.6175 | - | - | - | - |
730
- | 0.0849 | 1990 | 0.6135 | - | - | - | - |
731
- | 0.0853 | 2000 | 0.6214 | 0.6293 | 0.7297 | 0.7766 | 0.3986 |
732
 
733
  </details>
734
 
 
6
  - sentence-similarity
7
  - feature-extraction
8
  - generated_from_trainer
9
+ - dataset_size:1600000
10
  - loss:TripletLoss
11
  datasets: []
12
  metrics:
 
38
  - dot_mrr@10
39
  - dot_map@10
40
  widget:
41
+ - source_sentence: 'search_query: pokemon card mewtwo'
42
  sentences:
43
+ - 'search_document: Personal AM/FM Pocket Radio Portable VR-robot, Mini Digital
44
+ Tuning Walkman Radio, with Rechargeable Battery, Earphone, Lock Screen for Walk/Jogging/Gym/Camping,
45
+ VR-robot, Electronics'
46
+ - 'search_document: Pokemon Mewtwo & Pikachu XY Evolutions TCG Card Game Decks -
47
+ 60 Cards Each, Pokemon, '
48
+ - 'search_document: Ultra Pro Pokemon: Charizard Album, 2", Ultra Pro, '
49
+ - source_sentence: 'search_query: table runners 108 inches'
50
  sentences:
51
+ - 'search_document: Sambosk Fall Buffalo Pumpkin Table Runner, Autumn Farmhouse
52
+ Table Runners for Kitchen Dining Coffee or Indoor and Outdoor Home Parties Decor
53
+ 13 x 72 Inches SK006, Sambosk, Black White'
54
+ - 'search_document: EYEGUARD Readers 4 Pack of Thin and Elegant Womens Reading Glasses
55
+ with Beautiful Patterns for Ladies 1.00, EYEGUARD, Mix'
56
+ - 'search_document: Sunfiy 4 Pack Red Satin Table Runner 12 x 108 Inch Long Table
57
+ Runners for Wedding Birthday Parties Banquets Graduations Engagements, Sunfiy,
58
+ Red'
59
+ - source_sentence: 'search_query: nursing shoes for women'
60
  sentences:
61
+ - 'search_document: Hawkwell Women''s Lightweight Comfort Slip Resistant Nursing
62
+ Shoes,White PU,10 M US, Hawkwell, 1923/White'
63
+ - 'search_document: REESE''S Peanut Butter Milk Chocolate You''re Amazing Appreciation
64
+ Candy Bars for Christmas and Holiday Season, 4.2 oz Bars, 12 Count, Reese''s, '
65
+ - 'search_document: adidas womens Cloudfoam Pure Running Shoe, Black/Black, 7.5
66
+ US, adidas, Black/Black/White'
67
+ - source_sentence: 'search_query: mens socks black and white'
 
 
68
  sentences:
69
+ - 'search_document: Fruit of the Loom Men''s Essential 6 Pack Casual Crew Socks
70
+ | Arch Support | Black & White, Black, Shoe Size: 6-12, Fruit of the Loom, Black'
71
+ - 'search_document: adidas Originals Men''s Trefoil Crew Socks (6-Pair), White/Black
72
+ Black/White, Large, (Shoe Size 6-12), adidas Originals, White/Black'
73
+ - 'search_document: Fifty Shades of Grey, , '
74
+ - source_sentence: 'search_query: karoke set 2 microphone for adults'
 
 
 
 
 
75
  sentences:
76
+ - 'search_document: EARISE T26 Portable Karaoke Machine Bluetooth Speaker with Wireless
77
+ Microphone, Rechargeable PA System with FM Radio, Audio Recording, Remote Control,
78
+ Supports TF Card/USB, Perfect for Party, EARISE, '
79
+ - 'search_document: FunWorld Men''s Complete 3D Zombie Costume, Grey, One Size,
80
+ Fun World, Multi'
81
+ - 'search_document: Starion KS829-B Bluetooth Karaoke Machine l Pedestal Design
82
+ w/Light Show l Two Karaoke Microphones, Starion, Black'
83
  pipeline_tag: sentence-similarity
84
  model-index:
85
  - name: SentenceTransformer
 
92
  type: unknown
93
  metrics:
94
  - type: cosine_accuracy
95
+ value: 0.7298125
96
  name: Cosine Accuracy
97
  - type: dot_accuracy
98
+ value: 0.2831875
99
  name: Dot Accuracy
100
  - type: manhattan_accuracy
101
+ value: 0.72825
102
  name: Manhattan Accuracy
103
  - type: euclidean_accuracy
104
+ value: 0.729875
105
  name: Euclidean Accuracy
106
  - type: max_accuracy
107
+ value: 0.729875
108
  name: Max Accuracy
109
  - task:
110
  type: semantic-similarity
 
114
  type: unknown
115
  metrics:
116
  - type: pearson_cosine
117
+ value: 0.4148003591706621
118
  name: Pearson Cosine
119
  - type: spearman_cosine
120
+ value: 0.39973675544358156
121
  name: Spearman Cosine
122
  - type: pearson_manhattan
123
+ value: 0.37708819507475255
124
  name: Pearson Manhattan
125
  - type: spearman_manhattan
126
+ value: 0.36992167570513307
127
  name: Spearman Manhattan
128
  - type: pearson_euclidean
129
+ value: 0.3777862291730549
130
  name: Pearson Euclidean
131
  - type: spearman_euclidean
132
+ value: 0.3707889635811508
133
  name: Spearman Euclidean
134
  - type: pearson_dot
135
+ value: 0.3813644395159763
136
  name: Pearson Dot
137
  - type: spearman_dot
138
+ value: 0.3817136551173837
139
  name: Spearman Dot
140
  - type: pearson_max
141
+ value: 0.4148003591706621
142
  name: Pearson Max
143
  - type: spearman_max
144
+ value: 0.39973675544358156
145
  name: Spearman Max
146
  - task:
147
  type: information-retrieval
 
151
  type: unknown
152
  metrics:
153
  - type: cosine_accuracy@10
154
+ value: 0.967
155
  name: Cosine Accuracy@10
156
  - type: cosine_precision@10
157
+ value: 0.6951
158
  name: Cosine Precision@10
159
  - type: cosine_recall@10
160
+ value: 0.6216729831257005
161
  name: Cosine Recall@10
162
  - type: cosine_ndcg@10
163
+ value: 0.8300106033542061
164
  name: Cosine Ndcg@10
165
  - type: cosine_mrr@10
166
+ value: 0.9111154761904765
167
  name: Cosine Mrr@10
168
  - type: cosine_map@10
169
+ value: 0.7758485833963215
170
  name: Cosine Map@10
171
  - type: dot_accuracy@10
172
+ value: 0.946
173
  name: Dot Accuracy@10
174
  - type: dot_precision@10
175
+ value: 0.6369
176
  name: Dot Precision@10
177
  - type: dot_recall@10
178
+ value: 0.5693415261440723
179
  name: Dot Recall@10
180
  - type: dot_ndcg@10
181
+ value: 0.7668657376718138
182
  name: Dot Ndcg@10
183
  - type: dot_mrr@10
184
+ value: 0.8754059523809526
185
  name: Dot Mrr@10
186
  - type: dot_map@10
187
+ value: 0.6962231903502142
188
  name: Dot Map@10
189
  ---
190
 
 
238
  model = SentenceTransformer("lv12/esci-nomic-embed-text-v1_5_4")
239
  # Run inference
240
  sentences = [
241
+ 'search_query: karoke set 2 microphone for adults',
242
+ 'search_document: Starion KS829-B Bluetooth Karaoke Machine l Pedestal Design w/Light Show l Two Karaoke Microphones, Starion, Black',
243
+ 'search_document: EARISE T26 Portable Karaoke Machine Bluetooth Speaker with Wireless Microphone, Rechargeable PA System with FM Radio, Audio Recording, Remote Control, Supports TF Card/USB, Perfect for Party, EARISE, ',
244
  ]
245
  embeddings = model.encode(sentences)
246
  print(embeddings.shape)
 
286
 
287
  | Metric | Value |
288
  |:--------------------|:-----------|
289
+ | **cosine_accuracy** | **0.7298** |
290
+ | dot_accuracy | 0.2832 |
291
+ | manhattan_accuracy | 0.7282 |
292
+ | euclidean_accuracy | 0.7299 |
293
+ | max_accuracy | 0.7299 |
294
 
295
  #### Semantic Similarity
296
 
 
298
 
299
  | Metric | Value |
300
  |:--------------------|:-----------|
301
+ | pearson_cosine | 0.4148 |
302
+ | **spearman_cosine** | **0.3997** |
303
+ | pearson_manhattan | 0.3771 |
304
+ | spearman_manhattan | 0.3699 |
305
+ | pearson_euclidean | 0.3778 |
306
+ | spearman_euclidean | 0.3708 |
307
+ | pearson_dot | 0.3814 |
308
+ | spearman_dot | 0.3817 |
309
+ | pearson_max | 0.4148 |
310
+ | spearman_max | 0.3997 |
311
 
312
  #### Information Retrieval
313
 
 
315
 
316
  | Metric | Value |
317
  |:--------------------|:-----------|
318
+ | cosine_accuracy@10 | 0.967 |
319
+ | cosine_precision@10 | 0.6951 |
320
+ | cosine_recall@10 | 0.6217 |
321
+ | cosine_ndcg@10 | 0.83 |
322
+ | cosine_mrr@10 | 0.9111 |
323
+ | **cosine_map@10** | **0.7758** |
324
+ | dot_accuracy@10 | 0.946 |
325
+ | dot_precision@10 | 0.6369 |
326
+ | dot_recall@10 | 0.5693 |
327
+ | dot_ndcg@10 | 0.7669 |
328
+ | dot_mrr@10 | 0.8754 |
329
+ | dot_map@10 | 0.6962 |
330
 
331
  <!--
332
  ## Bias, Risks and Limitations
 
347
  #### triplets
348
 
349
  * Dataset: triplets
350
+ * Size: 1,600,000 training samples
351
  * Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
352
  * Approximate statistics based on the first 1000 samples:
353
+ | | anchor | positive | negative |
354
+ |:--------|:----------------------------------------------------------------------------------|:------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
355
+ | type | string | string | string |
356
+ | details | <ul><li>min: 7 tokens</li><li>mean: 11.03 tokens</li><li>max: 39 tokens</li></ul> | <ul><li>min: 10 tokens</li><li>mean: 39.86 tokens</li><li>max: 104 tokens</li></ul> | <ul><li>min: 9 tokens</li><li>mean: 39.73 tokens</li><li>max: 159 tokens</li></ul> |
357
  * Samples:
358
+ | anchor | positive | negative |
359
+ |:--------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
360
+ | <code>search_query: udt hydraulic fluid</code> | <code>search_document: Triax Agra UTTO XL Synthetic Blend Tractor Transmission and Hydraulic Oil, 6,000 Hour Life, 50% Less wear, 36F Pour Point, Replaces All OEM Tractor Fluids (5 Gallon Pail), TRIAX, </code> | <code>search_document: Shell Rotella T5 Synthetic Blend 15W-40 Diesel Engine Oil (1-Gallon, Case of 3), Shell Rotella, </code> |
361
+ | <code>search_query: cheetah print iphone xs case</code> | <code>search_document: iPhone Xs Case, iPhone Xs Case,Doowear Leopard Cheetah Protective Cover Shell For Girls Women,Slim Fit Anti Scratch Shockproof Soft TPU Bumper Flexible Rubber Gel Silicone Case for iPhone Xs / X-1, Ebetterr, 1</code> | <code>search_document: iPhone Xs & iPhone X Case, J.west Luxury Sparkle Bling Translucent Leopard Print Soft Silicone Phone Case Cover for Girls Women Flex Slim Design Pattern Drop Protective Case for iPhone Xs/x 5.8 inch, J.west, Leopard</code> |
362
+ | <code>search_query: platform shoes</code> | <code>search_document: Teva Women's Flatform Universal Platform Sandal, Black, 5 M US, Teva, Black</code> | <code>search_document: Vans Women's Old Skool Platform Trainers, (Black/White Y28), 5 UK 38 EU, Vans, Black/White</code> |
363
  * Loss: [<code>TripletLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#tripletloss) with these parameters:
364
  ```json
365
  {
366
  "distance_metric": "TripletDistanceMetric.COSINE",
367
+ "triplet_margin": 0.8
368
  }
369
  ```
370
 
 
373
  #### triplets
374
 
375
  * Dataset: triplets
376
+ * Size: 16,000 evaluation samples
377
  * Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
378
  * Approximate statistics based on the first 1000 samples:
379
+ | | anchor | positive | negative |
380
+ |:--------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|
381
+ | type | string | string | string |
382
+ | details | <ul><li>min: 7 tokens</li><li>mean: 11.02 tokens</li><li>max: 29 tokens</li></ul> | <ul><li>min: 10 tokens</li><li>mean: 38.78 tokens</li><li>max: 87 tokens</li></ul> | <ul><li>min: 9 tokens</li><li>mean: 38.81 tokens</li><li>max: 91 tokens</li></ul> |
383
  * Samples:
384
+ | anchor | positive | negative |
385
+ |:---------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------|
386
+ | <code>search_query: hogknobz</code> | <code>search_document: Black 2014-2015 HDsmallPARTS/LocEzy Saddlebag Mounting Hardware Knobs are replacement/compatible for Saddlebag Quick Release Pins on Harley Davidson Touring Motorcycles Theft Deterrent, LocEzy, </code> | <code>search_document: HANSWD Saddlebag Support Bars Brackets For SUZUKI YAMAHA KAWASAKI (Black), HANSWD, Black</code> |
387
+ | <code>search_query: tile sticker key finder</code> | <code>search_document: Tile Sticker (2020) 2-pack - Small, Adhesive Bluetooth Tracker, Item Locator and Finder for Remotes, Headphones, Gadgets and More, Tile, </code> | <code>search_document: Tile Pro Combo (2017) - 2 Pack (1 x Sport, 1 x Style) - Discontinued by Manufacturer, Tile, Graphite/Gold</code> |
388
+ | <code>search_query: adobe incense burner</code> | <code>search_document: AM Incense Burner Frankincense Resin - Luxury Globe Charcoal Bakhoor Burners for Office & Home Decor (Brown), AM, Brown</code> | <code>search_document: semli Large Incense Burner Backflow Incense Burner Holder Incense Stick Holder Home Office Decor, Semli, </code> |
389
  * Loss: [<code>TripletLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#tripletloss) with these parameters:
390
  ```json
391
  {
392
  "distance_metric": "TripletDistanceMetric.COSINE",
393
+ "triplet_margin": 0.8
394
  }
395
  ```
396
 
397
  ### Training Hyperparameters
398
  #### Non-Default Hyperparameters
399
 
400
+ - `per_device_train_batch_size`: 64
401
  - `per_device_eval_batch_size`: 16
402
  - `gradient_accumulation_steps`: 2
403
  - `learning_rate`: 1e-07
404
+ - `num_train_epochs`: 5
405
  - `lr_scheduler_type`: polynomial
406
  - `lr_scheduler_kwargs`: {'lr_end': 1e-08, 'power': 2.0}
407
  - `warmup_ratio`: 0.05
408
+ - `dataloader_drop_last`: True
409
  - `dataloader_num_workers`: 4
410
  - `dataloader_prefetch_factor`: 4
411
  - `load_best_model_at_end`: True
 
419
  - `overwrite_output_dir`: False
420
  - `do_predict`: False
421
  - `prediction_loss_only`: True
422
+ - `per_device_train_batch_size`: 64
423
  - `per_device_eval_batch_size`: 16
424
  - `per_gpu_train_batch_size`: None
425
  - `per_gpu_eval_batch_size`: None
 
431
  - `adam_beta2`: 0.999
432
  - `adam_epsilon`: 1e-08
433
  - `max_grad_norm`: 1.0
434
+ - `num_train_epochs`: 5
435
  - `max_steps`: -1
436
  - `lr_scheduler_type`: polynomial
437
  - `lr_scheduler_kwargs`: {'lr_end': 1e-08, 'power': 2.0}
 
527
 
528
  | Epoch | Step | Training Loss | triplets loss | cosine_accuracy | cosine_map@10 | spearman_cosine |
529
  |:------:|:----:|:-------------:|:-------------:|:---------------:|:-------------:|:---------------:|
530
+ | 0.0008 | 10 | 0.7505 | - | - | - | - |
531
+ | 0.0016 | 20 | 0.7499 | - | - | - | - |
532
+ | 0.0024 | 30 | 0.7524 | - | - | - | - |
533
+ | 0.0032 | 40 | 0.7486 | - | - | - | - |
534
+ | 0.004 | 50 | 0.7493 | - | - | - | - |
535
+ | 0.0048 | 60 | 0.7476 | - | - | - | - |
536
+ | 0.0056 | 70 | 0.7483 | - | - | - | - |
537
+ | 0.0064 | 80 | 0.7487 | - | - | - | - |
538
+ | 0.0072 | 90 | 0.7496 | - | - | - | - |
539
+ | 0.008 | 100 | 0.7515 | 0.7559 | 0.7263 | 0.7684 | 0.3941 |
540
+ | 0.0088 | 110 | 0.7523 | - | - | - | - |
541
+ | 0.0096 | 120 | 0.7517 | - | - | - | - |
542
+ | 0.0104 | 130 | 0.7534 | - | - | - | - |
543
+ | 0.0112 | 140 | 0.746 | - | - | - | - |
544
+ | 0.012 | 150 | 0.7528 | - | - | - | - |
545
+ | 0.0128 | 160 | 0.7511 | - | - | - | - |
546
+ | 0.0136 | 170 | 0.7491 | - | - | - | - |
547
+ | 0.0144 | 180 | 0.752 | - | - | - | - |
548
+ | 0.0152 | 190 | 0.7512 | - | - | - | - |
549
+ | 0.016 | 200 | 0.7513 | 0.7557 | 0.7259 | 0.7688 | 0.3942 |
550
+ | 0.0168 | 210 | 0.7505 | - | - | - | - |
551
+ | 0.0176 | 220 | 0.7481 | - | - | - | - |
552
+ | 0.0184 | 230 | 0.7516 | - | - | - | - |
553
+ | 0.0192 | 240 | 0.7504 | - | - | - | - |
554
+ | 0.02 | 250 | 0.7498 | - | - | - | - |
555
+ | 0.0208 | 260 | 0.7506 | - | - | - | - |
556
+ | 0.0216 | 270 | 0.7486 | - | - | - | - |
557
+ | 0.0224 | 280 | 0.7471 | - | - | - | - |
558
+ | 0.0232 | 290 | 0.7511 | - | - | - | - |
559
+ | 0.024 | 300 | 0.7506 | 0.7553 | 0.7258 | 0.7692 | 0.3943 |
560
+ | 0.0248 | 310 | 0.7485 | - | - | - | - |
561
+ | 0.0256 | 320 | 0.7504 | - | - | - | - |
562
+ | 0.0264 | 330 | 0.7456 | - | - | - | - |
563
+ | 0.0272 | 340 | 0.7461 | - | - | - | - |
564
+ | 0.028 | 350 | 0.7496 | - | - | - | - |
565
+ | 0.0288 | 360 | 0.7518 | - | - | - | - |
566
+ | 0.0296 | 370 | 0.7514 | - | - | - | - |
567
+ | 0.0304 | 380 | 0.7479 | - | - | - | - |
568
+ | 0.0312 | 390 | 0.7507 | - | - | - | - |
569
+ | 0.032 | 400 | 0.7511 | 0.7547 | 0.7258 | 0.7695 | 0.3945 |
570
+ | 0.0328 | 410 | 0.7491 | - | - | - | - |
571
+ | 0.0336 | 420 | 0.7487 | - | - | - | - |
572
+ | 0.0344 | 430 | 0.7496 | - | - | - | - |
573
+ | 0.0352 | 440 | 0.7464 | - | - | - | - |
574
+ | 0.036 | 450 | 0.7518 | - | - | - | - |
575
+ | 0.0368 | 460 | 0.7481 | - | - | - | - |
576
+ | 0.0376 | 470 | 0.7493 | - | - | - | - |
577
+ | 0.0384 | 480 | 0.753 | - | - | - | - |
578
+ | 0.0392 | 490 | 0.7475 | - | - | - | - |
579
+ | 0.04 | 500 | 0.7498 | 0.7540 | 0.7262 | 0.7700 | 0.3948 |
580
+ | 0.0408 | 510 | 0.7464 | - | - | - | - |
581
+ | 0.0416 | 520 | 0.7506 | - | - | - | - |
582
+ | 0.0424 | 530 | 0.747 | - | - | - | - |
583
+ | 0.0432 | 540 | 0.7462 | - | - | - | - |
584
+ | 0.044 | 550 | 0.75 | - | - | - | - |
585
+ | 0.0448 | 560 | 0.7522 | - | - | - | - |
586
+ | 0.0456 | 570 | 0.7452 | - | - | - | - |
587
+ | 0.0464 | 580 | 0.7475 | - | - | - | - |
588
+ | 0.0472 | 590 | 0.7507 | - | - | - | - |
589
+ | 0.048 | 600 | 0.7494 | 0.7531 | 0.7269 | 0.7707 | 0.3951 |
590
+ | 0.0488 | 610 | 0.7525 | - | - | - | - |
591
+ | 0.0496 | 620 | 0.7446 | - | - | - | - |
592
+ | 0.0504 | 630 | 0.7457 | - | - | - | - |
593
+ | 0.0512 | 640 | 0.7462 | - | - | - | - |
594
+ | 0.052 | 650 | 0.7478 | - | - | - | - |
595
+ | 0.0528 | 660 | 0.7459 | - | - | - | - |
596
+ | 0.0536 | 670 | 0.7465 | - | - | - | - |
597
+ | 0.0544 | 680 | 0.7495 | - | - | - | - |
598
+ | 0.0552 | 690 | 0.7513 | - | - | - | - |
599
+ | 0.056 | 700 | 0.7445 | 0.7520 | 0.7274 | 0.7705 | 0.3954 |
600
+ | 0.0568 | 710 | 0.7446 | - | - | - | - |
601
+ | 0.0576 | 720 | 0.746 | - | - | - | - |
602
+ | 0.0584 | 730 | 0.7452 | - | - | - | - |
603
+ | 0.0592 | 740 | 0.7459 | - | - | - | - |
604
+ | 0.06 | 750 | 0.7419 | - | - | - | - |
605
+ | 0.0608 | 760 | 0.7462 | - | - | - | - |
606
+ | 0.0616 | 770 | 0.7414 | - | - | - | - |
607
+ | 0.0624 | 780 | 0.7444 | - | - | - | - |
608
+ | 0.0632 | 790 | 0.7419 | - | - | - | - |
609
+ | 0.064 | 800 | 0.7438 | 0.7508 | 0.7273 | 0.7712 | 0.3957 |
610
+ | 0.0648 | 810 | 0.7503 | - | - | - | - |
611
+ | 0.0656 | 820 | 0.7402 | - | - | - | - |
612
+ | 0.0664 | 830 | 0.7435 | - | - | - | - |
613
+ | 0.0672 | 840 | 0.741 | - | - | - | - |
614
+ | 0.068 | 850 | 0.7386 | - | - | - | - |
615
+ | 0.0688 | 860 | 0.7416 | - | - | - | - |
616
+ | 0.0696 | 870 | 0.7473 | - | - | - | - |
617
+ | 0.0704 | 880 | 0.7438 | - | - | - | - |
618
+ | 0.0712 | 890 | 0.7458 | - | - | - | - |
619
+ | 0.072 | 900 | 0.7446 | 0.7494 | 0.7279 | 0.7718 | 0.3961 |
620
+ | 0.0728 | 910 | 0.7483 | - | - | - | - |
621
+ | 0.0736 | 920 | 0.7458 | - | - | - | - |
622
+ | 0.0744 | 930 | 0.7473 | - | - | - | - |
623
+ | 0.0752 | 940 | 0.7431 | - | - | - | - |
624
+ | 0.076 | 950 | 0.7428 | - | - | - | - |
625
+ | 0.0768 | 960 | 0.7385 | - | - | - | - |
626
+ | 0.0776 | 970 | 0.7438 | - | - | - | - |
627
+ | 0.0784 | 980 | 0.7406 | - | - | - | - |
628
+ | 0.0792 | 990 | 0.7426 | - | - | - | - |
629
+ | 0.08 | 1000 | 0.7372 | 0.7478 | 0.7282 | 0.7725 | 0.3965 |
630
+ | 0.0808 | 1010 | 0.7396 | - | - | - | - |
631
+ | 0.0816 | 1020 | 0.7398 | - | - | - | - |
632
+ | 0.0824 | 1030 | 0.7376 | - | - | - | - |
633
+ | 0.0832 | 1040 | 0.7417 | - | - | - | - |
634
+ | 0.084 | 1050 | 0.7408 | - | - | - | - |
635
+ | 0.0848 | 1060 | 0.7415 | - | - | - | - |
636
+ | 0.0856 | 1070 | 0.7468 | - | - | - | - |
637
+ | 0.0864 | 1080 | 0.7427 | - | - | - | - |
638
+ | 0.0872 | 1090 | 0.7371 | - | - | - | - |
639
+ | 0.088 | 1100 | 0.7375 | 0.7460 | 0.7279 | 0.7742 | 0.3970 |
640
+ | 0.0888 | 1110 | 0.7434 | - | - | - | - |
641
+ | 0.0896 | 1120 | 0.7441 | - | - | - | - |
642
+ | 0.0904 | 1130 | 0.7378 | - | - | - | - |
643
+ | 0.0912 | 1140 | 0.735 | - | - | - | - |
644
+ | 0.092 | 1150 | 0.739 | - | - | - | - |
645
+ | 0.0928 | 1160 | 0.7408 | - | - | - | - |
646
+ | 0.0936 | 1170 | 0.7346 | - | - | - | - |
647
+ | 0.0944 | 1180 | 0.7389 | - | - | - | - |
648
+ | 0.0952 | 1190 | 0.7367 | - | - | - | - |
649
+ | 0.096 | 1200 | 0.7358 | 0.7440 | 0.729 | 0.7747 | 0.3975 |
650
+ | 0.0968 | 1210 | 0.7381 | - | - | - | - |
651
+ | 0.0976 | 1220 | 0.7405 | - | - | - | - |
652
+ | 0.0984 | 1230 | 0.7348 | - | - | - | - |
653
+ | 0.0992 | 1240 | 0.737 | - | - | - | - |
654
+ | 0.1 | 1250 | 0.7393 | - | - | - | - |
655
+ | 0.1008 | 1260 | 0.7411 | - | - | - | - |
656
+ | 0.1016 | 1270 | 0.7359 | - | - | - | - |
657
+ | 0.1024 | 1280 | 0.7276 | - | - | - | - |
658
+ | 0.1032 | 1290 | 0.7364 | - | - | - | - |
659
+ | 0.104 | 1300 | 0.7333 | 0.7418 | 0.7293 | 0.7747 | 0.3979 |
660
+ | 0.1048 | 1310 | 0.7367 | - | - | - | - |
661
+ | 0.1056 | 1320 | 0.7352 | - | - | - | - |
662
+ | 0.1064 | 1330 | 0.7333 | - | - | - | - |
663
+ | 0.1072 | 1340 | 0.737 | - | - | - | - |
664
+ | 0.108 | 1350 | 0.7361 | - | - | - | - |
665
+ | 0.1088 | 1360 | 0.7299 | - | - | - | - |
666
+ | 0.1096 | 1370 | 0.7339 | - | - | - | - |
667
+ | 0.1104 | 1380 | 0.7349 | - | - | - | - |
668
+ | 0.1112 | 1390 | 0.7318 | - | - | - | - |
669
+ | 0.112 | 1400 | 0.7336 | 0.7394 | 0.7292 | 0.7749 | 0.3983 |
670
+ | 0.1128 | 1410 | 0.7326 | - | - | - | - |
671
+ | 0.1136 | 1420 | 0.7317 | - | - | - | - |
672
+ | 0.1144 | 1430 | 0.7315 | - | - | - | - |
673
+ | 0.1152 | 1440 | 0.7321 | - | - | - | - |
674
+ | 0.116 | 1450 | 0.7284 | - | - | - | - |
675
+ | 0.1168 | 1460 | 0.7308 | - | - | - | - |
676
+ | 0.1176 | 1470 | 0.7287 | - | - | - | - |
677
+ | 0.1184 | 1480 | 0.727 | - | - | - | - |
678
+ | 0.1192 | 1490 | 0.7298 | - | - | - | - |
679
+ | 0.12 | 1500 | 0.7306 | 0.7368 | 0.7301 | 0.7755 | 0.3988 |
680
+ | 0.1208 | 1510 | 0.7269 | - | - | - | - |
681
+ | 0.1216 | 1520 | 0.7299 | - | - | - | - |
682
+ | 0.1224 | 1530 | 0.7256 | - | - | - | - |
683
+ | 0.1232 | 1540 | 0.721 | - | - | - | - |
684
+ | 0.124 | 1550 | 0.7274 | - | - | - | - |
685
+ | 0.1248 | 1560 | 0.7251 | - | - | - | - |
686
+ | 0.1256 | 1570 | 0.7248 | - | - | - | - |
687
+ | 0.1264 | 1580 | 0.7244 | - | - | - | - |
688
+ | 0.1272 | 1590 | 0.7275 | - | - | - | - |
689
+ | 0.128 | 1600 | 0.7264 | 0.7339 | 0.7298 | 0.7756 | 0.3991 |
690
+ | 0.1288 | 1610 | 0.7252 | - | - | - | - |
691
+ | 0.1296 | 1620 | 0.7287 | - | - | - | - |
692
+ | 0.1304 | 1630 | 0.7263 | - | - | - | - |
693
+ | 0.1312 | 1640 | 0.7216 | - | - | - | - |
694
+ | 0.132 | 1650 | 0.7231 | - | - | - | - |
695
+ | 0.1328 | 1660 | 0.728 | - | - | - | - |
696
+ | 0.1336 | 1670 | 0.7309 | - | - | - | - |
697
+ | 0.1344 | 1680 | 0.7243 | - | - | - | - |
698
+ | 0.1352 | 1690 | 0.7239 | - | - | - | - |
699
+ | 0.136 | 1700 | 0.7219 | 0.7309 | 0.7302 | 0.7768 | 0.3994 |
700
+ | 0.1368 | 1710 | 0.7212 | - | - | - | - |
701
+ | 0.1376 | 1720 | 0.7217 | - | - | - | - |
702
+ | 0.1384 | 1730 | 0.7118 | - | - | - | - |
703
+ | 0.1392 | 1740 | 0.7226 | - | - | - | - |
704
+ | 0.14 | 1750 | 0.7185 | - | - | - | - |
705
+ | 0.1408 | 1760 | 0.7228 | - | - | - | - |
706
+ | 0.1416 | 1770 | 0.7257 | - | - | - | - |
707
+ | 0.1424 | 1780 | 0.7177 | - | - | - | - |
708
+ | 0.1432 | 1790 | 0.722 | - | - | - | - |
709
+ | 0.144 | 1800 | 0.712 | 0.7276 | 0.7307 | 0.7763 | 0.3997 |
710
+ | 0.1448 | 1810 | 0.7193 | - | - | - | - |
711
+ | 0.1456 | 1820 | 0.7138 | - | - | - | - |
712
+ | 0.1464 | 1830 | 0.7171 | - | - | - | - |
713
+ | 0.1472 | 1840 | 0.7191 | - | - | - | - |
714
+ | 0.148 | 1850 | 0.7172 | - | - | - | - |
715
+ | 0.1488 | 1860 | 0.7168 | - | - | - | - |
716
+ | 0.1496 | 1870 | 0.7111 | - | - | - | - |
717
+ | 0.1504 | 1880 | 0.7203 | - | - | - | - |
718
+ | 0.1512 | 1890 | 0.7095 | - | - | - | - |
719
+ | 0.152 | 1900 | 0.7064 | 0.7240 | 0.7301 | 0.7762 | 0.3998 |
720
+ | 0.1528 | 1910 | 0.7147 | - | - | - | - |
721
+ | 0.1536 | 1920 | 0.7098 | - | - | - | - |
722
+ | 0.1544 | 1930 | 0.7193 | - | - | - | - |
723
+ | 0.1552 | 1940 | 0.7096 | - | - | - | - |
724
+ | 0.156 | 1950 | 0.7107 | - | - | - | - |
725
+ | 0.1568 | 1960 | 0.7146 | - | - | - | - |
726
+ | 0.1576 | 1970 | 0.7106 | - | - | - | - |
727
+ | 0.1584 | 1980 | 0.7079 | - | - | - | - |
728
+ | 0.1592 | 1990 | 0.7097 | - | - | - | - |
729
+ | 0.16 | 2000 | 0.71 | 0.7202 | 0.7298 | 0.7758 | 0.3997 |
730
 
731
  </details>
732
 
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3cf0ece0c783f88dbd3b658097e010b54fb1c12a8c464d21cac81fd4625675b8
3
  size 546938168
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4b395b636e2dc670c0f323ade7bd13eaa584aa5371b726d2d29dfa7d10df25e5
3
  size 546938168