mudasir13cs commited on
Commit
282e0a8
·
verified ·
1 Parent(s): 6794aaa

Update model card with comprehensive information

Browse files
Files changed (1) hide show
  1. README.md +109 -537
README.md CHANGED
@@ -1,580 +1,152 @@
1
  ---
 
 
 
2
  tags:
 
 
3
  - sentence-transformers
4
- - sentence-similarity
5
- - feature-extraction
6
- - dense
7
- - generated_from_trainer
8
- - dataset_size:32000
9
- - loss:MultipleNegativesRankingLoss
10
- base_model: sentence-transformers/all-MiniLM-L6-v2
11
- widget:
12
- - source_sentence: how to present delivery offers creatively
13
- sentences:
14
- - Introducing the "Delivery Offer with Fresh Raw Plums" presentation template, featuring
15
- a vibrant poster adorned with lush plums, some artfully flying through the air.
16
- Perfect for marketing ads, business finance, and fashion style industries, this
17
- template adds a fresh twist to presentations. Ideal for holidays, celebrations,
18
- and city-themed events, it captures the essence of express courier and delivery
19
- services. Engage your audience with this visually captivating and professionally
20
- designed template.
21
- - Elevate your presentations with the "Inclusive Urban Design Contribution Gratitude"
22
- template. This visually engaging design features a certificate motif, perfect
23
- for recognizing achievements in construction and urban development. Ideal for
24
- industries like Marketing Ads, Entertainment Leisure, and Business Finance, this
25
- template is also suited for Holidays Celebration, Fashion Style, and Travels Vacations
26
- contexts. Seamlessly blend Courier, Delivery, and Celebration themes to captivate
27
- your audience and underscore your message with professional flair.
28
- - 'Unleash your creativity with the "Volunteer Work Quote with Animal Skull" presentation
29
- template. Featuring a striking black and white image of a ram, this graphic design
30
- is perfect for industries like Marketing Ads, Entertainment Leisure, and Services.
31
- Ideal for Holidays Celebration, Food & Drinks, and Fashion Style presentations,
32
- this template captivates with its artistic flair. Engage your audience with its
33
- bold visual elements, making it a standout choice for those seeking impactful,
34
- professional presentations. Keywords: graphic design, animal skull, creative presentation.'
35
- - source_sentence: leisure activities presentation style
36
- sentences:
37
- - Elevate your presentations with the "Fashion Quote Businessman Wearing Suit in
38
- Grey" template. Featuring sleek visuals of a confident man in a suit and tie,
39
- this template embodies sophistication and style. Ideal for industries like Marketing
40
- Ads, Entertainment, Leisure, and Business Finance, it seamlessly fits categories
41
- such as Holidays Celebration and Food & Drinks. Perfect for presentations focused
42
- on celebrating, socializing, and enjoying life's finer moments, this template
43
- ensures your message is both impactful and memorable.
44
- - Dive into a lush paradise with our "Amazing Tropical Vegetation" presentation
45
- template. Featuring vibrant green plant imagery against a chic white and pink
46
- striped backdrop, this template exudes a lively yet professional aesthetic. Perfect
47
- for marketing ads, entertainment projects, or business finance presentations,
48
- it's ideal for holiday celebrations, leisure activities, or pet-related content.
49
- Captivate your audience with this versatile, visually stunning template, bringing
50
- a touch of tropical flair to your next project.
51
- - 'Introducing the "Nature Expo Announcement: Blooming Daisy Flower" presentation
52
- template, a vibrant blend of lush green fields and pristine white flowers, perfect
53
- for capturing the essence of nature. Ideal for marketing ads, entertainment, and
54
- service industries, this template is perfect for holidays, celebrations, and food
55
- and drinks events. With elements like music notes and food frames, it''s designed
56
- for those celebrating nature''s beauty. Engage your audience with this captivating,
57
- nature-inspired template. Perfect for creating an unforgettable Nature Expo 2019
58
- presentation.'
59
- - source_sentence: Gaming championship announcement design
60
- sentences:
61
- - Unleash the fun with our Video Games Championship Announcement template! Featuring
62
- a sleek black background and the intriguing phrase, "Is this video game your passion
63
- without challenges?" this template is perfect for marketing ads and entertainment
64
- events. Ideal for holidays, celebrations, and leisure activities, it seamlessly
65
- integrates a food frame for added flair. Engage your audience with its amusing
66
- style, making it perfect for celebrating gaming enthusiasts and their love for
67
- competition.
68
- - Elevate your presentations with the "Teacher Helping Kids" template, featuring
69
- heartwarming visuals of a dedicated woman assisting a young boy in a vibrant classroom
70
- setting. Ideal for marketing ads, entertainment, and service industries, this
71
- template seamlessly fits holiday celebrations and food and drink events. Perfect
72
- for parties, socializing, and cultural gatherings, it brings an engaging, professional
73
- touch to your message. Inspire your audience with these captivating, relatable
74
- scenes that highlight education and community.
75
- - 'Unleash your creativity with the "Do It Yourself Inspirational Banner" presentation
76
- template. Featuring a sleek black and white logo with the phrase ''dot yourself,''
77
- this design is framed by a vibrant yellow-bordered white background, exuding a
78
- modern yet professional aesthetic. Perfect for marketing ads, entertainment, and
79
- business finance sectors, this template is ideal for celebrating holidays, showcasing
80
- food and drinks, or highlighting cities and places. Keywords: courier, delivery,
81
- express.'
82
- - source_sentence: holiday celebration presentation design
83
- sentences:
84
- - Elevate your presentations with the "Digital Photography Tips with Camera" template,
85
- featuring a sleek dark blue background accented by vibrant lights and a striking
86
- black and white clock image. Perfect for professionals in marketing ads, entertainment,
87
- and leisure services, this template is ideal for holidays, food and drinks, and
88
- fashion style presentations. Capture your audience's attention and boost your
89
- Instagram presence with this visually stunning, professional template designed
90
- for engaging storytelling.
91
- - Elevate your marketing campaigns with the "Antique Furniture Ad with Luxury Armchair"
92
- presentation template. Featuring a sophisticated visual style, this template seamlessly
93
- integrates a furniture store logo with an elegant armchair centerpiece. Perfect
94
- for marketing ads in the entertainment and leisure industry, business finance
95
- presentations, and holiday celebrations, this template captures attention effortlessly.
96
- Ideal for use in courier, delivery, and travel sectors, it ensures your message
97
- is delivered with timeless elegance and professional flair.
98
- - 'The "Happy Children at Kids Camp" presentation template features vibrant visuals
99
- of joyful children sitting on lush grass, set against a lively green banner backdrop.
100
- Perfect for marketing ads, entertainment, leisure, and business finance sectors,
101
- this template is ideal for holiday celebrations, food and drinks promotions, and
102
- fashion style events. With its engaging food frame and informational infographics,
103
- it''s designed to captivate audiences while effectively conveying your message.
104
- Keywords: kids camp, celebrations, engaging visuals, marketing.'
105
- - source_sentence: engaging slides for food and drink theme
106
- sentences:
107
- - Elevate your presentations with the "Android Robot Hand" template, featuring a
108
- sleek, futuristic design. This template includes a captivating white robot set
109
- against a dynamic blue background, complemented by a blue and white bar chart
110
- emphasizing the number 1. Perfect for industries like Marketing Ads, Business
111
- Finance, and Fashion Style, it's ideal for discussions on trends and innovations.
112
- Engage your audience with this modern, trend-focused template, suitable for Courier,
113
- Delivery, and Express services.
114
- - Discover the "Cycling Club Tips" presentation template, featuring a minimalistic
115
- design with a striking black and white diagonal striped background and an image
116
- of a woman cycling against a crisp white backdrop. Perfect for marketing ads,
117
- entertainment, leisure, and business finance sectors, this template is ideal for
118
- creating engaging holiday celebrations or food and drinks presentations. Seamlessly
119
- integrate keywords like courier, delivery, and express to captivate your audience
120
- and elevate your message with style and impact.
121
- - Elevate your presentations with the "Smartphone Review Man Scrolling Phone" template,
122
- featuring a sleek design with dynamic visuals of a person holding a smartphone
123
- against a modern white background adorned with triangles. Perfect for marketing
124
- ads, business finance, and fashion style industries, this template is ideal for
125
- holidays, food and drinks, and leisure entertainment themes. Enhance your courier
126
- and delivery presentations or celebration pitches with this versatile, professional
127
- template that captivates and engages your audience effortlessly.
128
- pipeline_tag: sentence-similarity
129
- library_name: sentence-transformers
130
- metrics:
131
- - cosine_accuracy@1
132
- - cosine_accuracy@3
133
- - cosine_accuracy@5
134
- - cosine_accuracy@10
135
- - cosine_precision@1
136
- - cosine_precision@3
137
- - cosine_precision@5
138
- - cosine_precision@10
139
- - cosine_recall@1
140
- - cosine_recall@3
141
- - cosine_recall@5
142
- - cosine_recall@10
143
- - cosine_ndcg@10
144
- - cosine_mrr@10
145
- - cosine_map@1
146
- - cosine_map@5
147
- - cosine_map@10
148
- model-index:
149
- - name: SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2
150
- results:
151
- - task:
152
- type: information-retrieval
153
- name: Information Retrieval
154
- dataset:
155
- name: validation
156
- type: validation
157
- metrics:
158
- - type: cosine_accuracy@1
159
- value: 0.45
160
- name: Cosine Accuracy@1
161
- - type: cosine_accuracy@3
162
- value: 0.62
163
- name: Cosine Accuracy@3
164
- - type: cosine_accuracy@5
165
- value: 0.685625
166
- name: Cosine Accuracy@5
167
- - type: cosine_accuracy@10
168
- value: 0.765625
169
- name: Cosine Accuracy@10
170
- - type: cosine_precision@1
171
- value: 0.45
172
- name: Cosine Precision@1
173
- - type: cosine_precision@3
174
- value: 0.20666666666666664
175
- name: Cosine Precision@3
176
- - type: cosine_precision@5
177
- value: 0.137125
178
- name: Cosine Precision@5
179
- - type: cosine_precision@10
180
- value: 0.0765625
181
- name: Cosine Precision@10
182
- - type: cosine_recall@1
183
- value: 0.45
184
- name: Cosine Recall@1
185
- - type: cosine_recall@3
186
- value: 0.62
187
- name: Cosine Recall@3
188
- - type: cosine_recall@5
189
- value: 0.685625
190
- name: Cosine Recall@5
191
- - type: cosine_recall@10
192
- value: 0.765625
193
- name: Cosine Recall@10
194
- - type: cosine_ndcg@10
195
- value: 0.6029385769142473
196
- name: Cosine Ndcg@10
197
- - type: cosine_mrr@10
198
- value: 0.5515214533730152
199
- name: Cosine Mrr@10
200
- - type: cosine_map@1
201
- value: 0.45
202
- name: Cosine Map@1
203
- - type: cosine_map@5
204
- value: 0.5409479166666666
205
- name: Cosine Map@5
206
- - type: cosine_map@10
207
- value: 0.5515214533730158
208
- name: Cosine Map@10
209
  ---
210
 
211
- # SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2
212
-
213
- This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [sentence-transformers/all-MiniLM-L6-v2](https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2). It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
214
 
215
  ## Model Details
216
 
217
  ### Model Description
218
- - **Model Type:** Sentence Transformer
219
- - **Base model:** [sentence-transformers/all-MiniLM-L6-v2](https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2) <!-- at revision c9745ed1d9f207416be6d2e6f8de32d1f16199bf -->
220
- - **Maximum Sequence Length:** 256 tokens
221
- - **Output Dimensionality:** 384 dimensions
222
- - **Similarity Function:** Cosine Similarity
223
- <!-- - **Training Dataset:** Unknown -->
224
- <!-- - **Language:** Unknown -->
225
- <!-- - **License:** Unknown -->
226
 
227
- ### Model Sources
228
-
229
- - **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
230
- - **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)
231
- - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
232
 
233
- ### Full Model Architecture
234
 
235
- ```
236
- SentenceTransformer(
237
- (0): Transformer({'max_seq_length': 256, 'do_lower_case': False, 'architecture': 'BertModel'})
238
- (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
239
- (2): Normalize()
240
- )
241
- ```
242
 
243
- ## Usage
244
 
245
- ### Direct Usage (Sentence Transformers)
246
-
247
- First install the Sentence Transformers library:
248
-
249
- ```bash
250
- pip install -U sentence-transformers
251
- ```
252
-
253
- Then you can load this model and run inference.
254
- ```python
255
- from sentence_transformers import SentenceTransformer
256
-
257
- # Download from the 🤗 Hub
258
- model = SentenceTransformer("sentence_transformers_model_id")
259
- # Run inference
260
- sentences = [
261
- 'engaging slides for food and drink theme',
262
- 'Elevate your presentations with the "Smartphone Review Man Scrolling Phone" template, featuring a sleek design with dynamic visuals of a person holding a smartphone against a modern white background adorned with triangles. Perfect for marketing ads, business finance, and fashion style industries, this template is ideal for holidays, food and drinks, and leisure entertainment themes. Enhance your courier and delivery presentations or celebration pitches with this versatile, professional template that captivates and engages your audience effortlessly.',
263
- 'Discover the "Cycling Club Tips" presentation template, featuring a minimalistic design with a striking black and white diagonal striped background and an image of a woman cycling against a crisp white backdrop. Perfect for marketing ads, entertainment, leisure, and business finance sectors, this template is ideal for creating engaging holiday celebrations or food and drinks presentations. Seamlessly integrate keywords like courier, delivery, and express to captivate your audience and elevate your message with style and impact.',
264
- ]
265
- embeddings = model.encode(sentences)
266
- print(embeddings.shape)
267
- # [3, 384]
268
-
269
- # Get the similarity scores for the embeddings
270
- similarities = model.similarity(embeddings, embeddings)
271
- print(similarities)
272
- # tensor([[1.0000, 0.1884, 0.1604],
273
- # [0.1884, 1.0000, 0.2392],
274
- # [0.1604, 0.2392, 1.0000]])
275
- ```
276
 
277
- <!--
278
- ### Direct Usage (Transformers)
279
 
280
- <details><summary>Click to see the direct usage in Transformers</summary>
281
 
282
- </details>
283
- -->
284
 
285
- <!--
286
- ### Downstream Usage (Sentence Transformers)
 
 
 
287
 
288
- You can finetune this model on your own dataset.
 
 
 
 
289
 
290
- <details><summary>Click to expand</summary>
 
 
 
291
 
292
- </details>
293
- -->
294
 
295
- <!--
296
- ### Out-of-Scope Use
 
297
 
298
- *List how the model may foreseeably be misused and address what users ought not to do with the model.*
299
- -->
300
 
301
- ## Evaluation
 
 
302
 
303
- ### Metrics
304
-
305
- #### Information Retrieval
306
-
307
- * Dataset: `validation`
308
- * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator)
309
-
310
- | Metric | Value |
311
- |:--------------------|:-----------|
312
- | cosine_accuracy@1 | 0.45 |
313
- | cosine_accuracy@3 | 0.62 |
314
- | cosine_accuracy@5 | 0.6856 |
315
- | cosine_accuracy@10 | 0.7656 |
316
- | cosine_precision@1 | 0.45 |
317
- | cosine_precision@3 | 0.2067 |
318
- | cosine_precision@5 | 0.1371 |
319
- | cosine_precision@10 | 0.0766 |
320
- | cosine_recall@1 | 0.45 |
321
- | cosine_recall@3 | 0.62 |
322
- | cosine_recall@5 | 0.6856 |
323
- | cosine_recall@10 | 0.7656 |
324
- | **cosine_ndcg@10** | **0.6029** |
325
- | cosine_mrr@10 | 0.5515 |
326
- | cosine_map@1 | 0.45 |
327
- | cosine_map@5 | 0.5409 |
328
- | cosine_map@10 | 0.5515 |
329
-
330
- <!--
331
- ## Bias, Risks and Limitations
332
-
333
- *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
334
- -->
335
-
336
- <!--
337
- ### Recommendations
338
-
339
- *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
340
- -->
341
 
342
  ## Training Details
343
 
344
- ### Training Dataset
345
-
346
- #### Unnamed Dataset
347
-
348
- * Size: 32,000 training samples
349
- * Columns: <code>sentence_0</code> and <code>sentence_1</code>
350
- * Approximate statistics based on the first 1000 samples:
351
- | | sentence_0 | sentence_1 |
352
- |:--------|:---------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------|
353
- | type | string | string |
354
- | details | <ul><li>min: 5 tokens</li><li>mean: 7.11 tokens</li><li>max: 12 tokens</li></ul> | <ul><li>min: 82 tokens</li><li>mean: 100.09 tokens</li><li>max: 133 tokens</li></ul> |
355
- * Samples:
356
- | sentence_0 | sentence_1 |
357
- |:------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
358
- | <code>entertainment-themed presentation slides</code> | <code>Ignite creativity with our "Reading Inspiration Books on Shelves" presentation template. Featuring a charming visual style with piles of books on a table, this template is perfect for industries like Marketing, Entertainment, and Services. Ideal for Holidays, Celebrations, and Leisure topics, it seamlessly integrates themes of Food, Drinks, and Games. Captivate your audience with a professional yet engaging backdrop that celebrates creativity and leisure. Perfect for marketers looking to inspire and entertain.</code> |
359
- | <code>fashion-forward slides for holidays and celebrations</code> | <code>Elevate your presentations with the "Insurance Company Successful Business Team" template, featuring a sleek design showcasing an insurance logo and a dynamic duo seated on a couch. A pink shield on a pristine white background adds a touch of elegance. Perfect for marketing ads, entertainment, and fashion, it's ideal for holidays, celebrations, or pet-related themes. Keywords like "courier," "beauty," and "celebration" seamlessly blend, making it a captivating choice for professionals seeking style and impact.</code> |
360
- | <code>How to promote a decor event template</code> | <code>Unveil your event with our "Interior Decoration Event Announcement Sofa in Grey" template, featuring a chic living room setting with a stylish grey couch and vibrant green plant. Perfect for marketing ads in the entertainment and leisure industries, this template is ideal for holiday celebrations or home decor events. Capture attention with its modern aesthetic and versatile design, seamlessly integrating keywords like courier, delivery, and express to boost your promotional efforts.</code> |
361
- * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:
362
- ```json
363
- {
364
- "scale": 20.0,
365
- "similarity_fct": "cos_sim"
366
- }
367
- ```
368
 
369
  ### Training Hyperparameters
370
- #### Non-Default Hyperparameters
371
-
372
- - `eval_strategy`: steps
373
- - `per_device_train_batch_size`: 64
374
- - `per_device_eval_batch_size`: 64
375
- - `num_train_epochs`: 20
376
- - `multi_dataset_batch_sampler`: round_robin
377
-
378
- #### All Hyperparameters
379
- <details><summary>Click to expand</summary>
380
-
381
- - `overwrite_output_dir`: False
382
- - `do_predict`: False
383
- - `eval_strategy`: steps
384
- - `prediction_loss_only`: True
385
- - `per_device_train_batch_size`: 64
386
- - `per_device_eval_batch_size`: 64
387
- - `per_gpu_train_batch_size`: None
388
- - `per_gpu_eval_batch_size`: None
389
- - `gradient_accumulation_steps`: 1
390
- - `eval_accumulation_steps`: None
391
- - `torch_empty_cache_steps`: None
392
- - `learning_rate`: 5e-05
393
- - `weight_decay`: 0.0
394
- - `adam_beta1`: 0.9
395
- - `adam_beta2`: 0.999
396
- - `adam_epsilon`: 1e-08
397
- - `max_grad_norm`: 1
398
- - `num_train_epochs`: 20
399
- - `max_steps`: -1
400
- - `lr_scheduler_type`: linear
401
- - `lr_scheduler_kwargs`: {}
402
- - `warmup_ratio`: 0.0
403
- - `warmup_steps`: 0
404
- - `log_level`: passive
405
- - `log_level_replica`: warning
406
- - `log_on_each_node`: True
407
- - `logging_nan_inf_filter`: True
408
- - `save_safetensors`: True
409
- - `save_on_each_node`: False
410
- - `save_only_model`: False
411
- - `restore_callback_states_from_checkpoint`: False
412
- - `no_cuda`: False
413
- - `use_cpu`: False
414
- - `use_mps_device`: False
415
- - `seed`: 42
416
- - `data_seed`: None
417
- - `jit_mode_eval`: False
418
- - `use_ipex`: False
419
- - `bf16`: False
420
- - `fp16`: False
421
- - `fp16_opt_level`: O1
422
- - `half_precision_backend`: auto
423
- - `bf16_full_eval`: False
424
- - `fp16_full_eval`: False
425
- - `tf32`: None
426
- - `local_rank`: 0
427
- - `ddp_backend`: None
428
- - `tpu_num_cores`: None
429
- - `tpu_metrics_debug`: False
430
- - `debug`: []
431
- - `dataloader_drop_last`: False
432
- - `dataloader_num_workers`: 0
433
- - `dataloader_prefetch_factor`: None
434
- - `past_index`: -1
435
- - `disable_tqdm`: False
436
- - `remove_unused_columns`: True
437
- - `label_names`: None
438
- - `load_best_model_at_end`: False
439
- - `ignore_data_skip`: False
440
- - `fsdp`: []
441
- - `fsdp_min_num_params`: 0
442
- - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
443
- - `fsdp_transformer_layer_cls_to_wrap`: None
444
- - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
445
- - `deepspeed`: None
446
- - `label_smoothing_factor`: 0.0
447
- - `optim`: adamw_torch
448
- - `optim_args`: None
449
- - `adafactor`: False
450
- - `group_by_length`: False
451
- - `length_column_name`: length
452
- - `ddp_find_unused_parameters`: None
453
- - `ddp_bucket_cap_mb`: None
454
- - `ddp_broadcast_buffers`: False
455
- - `dataloader_pin_memory`: True
456
- - `dataloader_persistent_workers`: False
457
- - `skip_memory_metrics`: True
458
- - `use_legacy_prediction_loop`: False
459
- - `push_to_hub`: False
460
- - `resume_from_checkpoint`: None
461
- - `hub_model_id`: None
462
- - `hub_strategy`: every_save
463
- - `hub_private_repo`: None
464
- - `hub_always_push`: False
465
- - `hub_revision`: None
466
- - `gradient_checkpointing`: False
467
- - `gradient_checkpointing_kwargs`: None
468
- - `include_inputs_for_metrics`: False
469
- - `include_for_metrics`: []
470
- - `eval_do_concat_batches`: True
471
- - `fp16_backend`: auto
472
- - `push_to_hub_model_id`: None
473
- - `push_to_hub_organization`: None
474
- - `mp_parameters`:
475
- - `auto_find_batch_size`: False
476
- - `full_determinism`: False
477
- - `torchdynamo`: None
478
- - `ray_scope`: last
479
- - `ddp_timeout`: 1800
480
- - `torch_compile`: False
481
- - `torch_compile_backend`: None
482
- - `torch_compile_mode`: None
483
- - `include_tokens_per_second`: False
484
- - `include_num_input_tokens_seen`: False
485
- - `neftune_noise_alpha`: None
486
- - `optim_target_modules`: None
487
- - `batch_eval_metrics`: False
488
- - `eval_on_start`: False
489
- - `use_liger_kernel`: False
490
- - `liger_kernel_config`: None
491
- - `eval_use_gather_object`: False
492
- - `average_tokens_across_devices`: False
493
- - `prompts`: None
494
- - `batch_sampler`: batch_sampler
495
- - `multi_dataset_batch_sampler`: round_robin
496
- - `router_mapping`: {}
497
- - `learning_rate_mapping`: {}
498
-
499
- </details>
500
-
501
- ### Training Logs
502
- | Epoch | Step | Training Loss | validation_cosine_ndcg@10 |
503
- |:-----:|:-----:|:-------------:|:-------------------------:|
504
- | 1.0 | 500 | 2.3734 | 0.4626 |
505
- | 2.0 | 1000 | 1.9095 | 0.4966 |
506
- | 3.0 | 1500 | 1.7464 | 0.5176 |
507
- | 4.0 | 2000 | 1.6538 | 0.5309 |
508
- | 5.0 | 2500 | 1.5949 | 0.5425 |
509
- | 6.0 | 3000 | 1.5507 | 0.5519 |
510
- | 7.0 | 3500 | 1.5173 | 0.5605 |
511
- | 8.0 | 4000 | 1.4871 | 0.5669 |
512
- | 9.0 | 4500 | 1.4587 | 0.5729 |
513
- | 10.0 | 5000 | 1.4309 | 0.5763 |
514
- | 11.0 | 5500 | 1.4214 | 0.5805 |
515
- | 12.0 | 6000 | 1.4028 | 0.5852 |
516
- | 13.0 | 6500 | 1.3867 | 0.5894 |
517
- | 14.0 | 7000 | 1.3745 | 0.5945 |
518
- | 15.0 | 7500 | 1.3625 | 0.5950 |
519
- | 16.0 | 8000 | 1.3516 | 0.5982 |
520
- | 17.0 | 8500 | 1.3453 | 0.6001 |
521
- | 18.0 | 9000 | 1.3448 | 0.6019 |
522
- | 19.0 | 9500 | 1.3327 | 0.6023 |
523
- | 20.0 | 10000 | 1.3323 | 0.6029 |
524
-
525
-
526
- ### Framework Versions
527
- - Python: 3.10.18
528
- - Sentence Transformers: 5.0.0
529
- - Transformers: 4.54.0.dev0
530
- - PyTorch: 2.6.0+cu124
531
- - Accelerate: 1.8.1
532
- - Datasets: 3.6.0
533
- - Tokenizers: 0.21.2
534
 
535
- ## Citation
536
 
537
- ### BibTeX
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
538
 
539
- #### Sentence Transformers
540
- ```bibtex
541
- @inproceedings{reimers-2019-sentence-bert,
542
- title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
543
- author = "Reimers, Nils and Gurevych, Iryna",
544
- booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
545
- month = "11",
546
- year = "2019",
547
- publisher = "Association for Computational Linguistics",
548
- url = "https://arxiv.org/abs/1908.10084",
549
- }
550
- ```
551
 
552
- #### MultipleNegativesRankingLoss
553
  ```bibtex
554
- @misc{henderson2017efficient,
555
- title={Efficient Natural Language Response Suggestion for Smart Reply},
556
- 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},
557
- year={2017},
558
- eprint={1705.00652},
559
- archivePrefix={arXiv},
560
- primaryClass={cs.CL}
561
  }
562
  ```
563
 
564
- <!--
565
- ## Glossary
566
 
567
- *Clearly define terms in order to be accessible across audiences.*
568
- -->
569
-
570
- <!--
571
  ## Model Card Authors
 
572
 
573
- *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
574
- -->
575
-
576
- <!--
577
  ## Model Card Contact
 
 
578
 
579
- *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
580
- -->
 
 
 
1
  ---
2
+ library_name: sentence-transformers
3
+ pipeline_tag: sentence-similarity
4
+ license: apache-2.0
5
  tags:
6
+ - embeddings
7
+ - semantic-search
8
  - sentence-transformers
9
+ - presentation-templates
10
+ - information-retrieval
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
11
  ---
12
 
13
+ # Field-adaptive-bi-encoder
 
 
14
 
15
  ## Model Details
16
 
17
  ### Model Description
18
+ A fine-tuned SentenceTransformers bi-encoder model for semantic similarity and information retrieval. This model is specifically trained for finding relevant presentation templates based on user queries, descriptions, and metadata (industries, categories, tags).
 
 
 
 
 
 
 
19
 
20
+ **Developed by:** Mudasir Syed (mudasir13cs)
 
 
 
 
21
 
22
+ **Model type:** SentenceTransformer (Bi-encoder)
23
 
24
+ **Language(s) (NLP):** English
 
 
 
 
 
 
25
 
26
+ **License:** Apache 2.0
27
 
28
+ **Finetuned from model:** Microsoft/MiniLM-L12-H384-uncased
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
29
 
30
+ ### Model Sources
31
+ **Repository:** https://github.com/mudasir13cs/hybrid-search
32
 
33
+ ## Uses
34
 
35
+ ### Direct Use
36
+ This model is designed for semantic search and information retrieval tasks, specifically for finding relevant presentation templates based on natural language queries.
37
 
38
+ ### Downstream Use
39
+ - Presentation template recommendation systems
40
+ - Content discovery platforms
41
+ - Semantic search engines
42
+ - Information retrieval systems
43
 
44
+ ### Out-of-Scope Use
45
+ - Text generation
46
+ - Question answering
47
+ - Machine translation
48
+ - Any task not related to semantic similarity
49
 
50
+ ## Bias, Risks, and Limitations
51
+ - The model is trained on presentation template data and may not generalize well to other domains
52
+ - Performance may vary based on the quality and diversity of training data
53
+ - The model inherits biases present in the base model and training data
54
 
55
+ ## How to Get Started with the Model
 
56
 
57
+ ```python
58
+ from sentence_transformers import SentenceTransformer
59
+ import torch
60
 
61
+ # Load the model
62
+ model = SentenceTransformer("mudasir13cs/Field-adaptive-bi-encoder")
63
 
64
+ # Encode text for similarity search
65
+ queries = ["business presentation template", "marketing slides for startups"]
66
+ embeddings = model.encode(queries)
67
 
68
+ # Compute similarity
69
+ from sentence_transformers import util
70
+ cosine_scores = util.cos_sim(embeddings[0], embeddings[1])
71
+ print(f"Similarity: {cosine_scores.item():.4f}")
72
+ ```
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
73
 
74
  ## Training Details
75
 
76
+ ### Training Data
77
+ - **Dataset:** Presentation template dataset with descriptions and queries
78
+ - **Size:** Custom dataset of presentation templates with metadata
79
+ - **Source:** Curated presentation template collection
80
+
81
+ ### Training Procedure
82
+ - **Architecture:** SentenceTransformer with triplet loss
83
+ - **Loss Function:** Triplet loss with hard negative mining
84
+ - **Optimizer:** AdamW
85
+ - **Learning Rate:** 2e-5
86
+ - **Batch Size:** 16
87
+ - **Epochs:** 3
 
 
 
 
 
 
 
 
 
 
 
 
88
 
89
  ### Training Hyperparameters
90
+ - **Training regime:** Supervised learning with triplet loss
91
+ - **Hardware:** GPU (NVIDIA)
92
+ - **Training time:** ~2 hours
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
93
 
94
+ ## Evaluation
95
 
96
+ ### Testing Data, Factors & Metrics
97
+ - **Testing Data:** Validation split from presentation template dataset
98
+ - **Factors:** Query-description similarity, template relevance
99
+ - **Metrics:**
100
+ - MAP@K (Mean Average Precision at K)
101
+ - MRR@K (Mean Reciprocal Rank at K)
102
+ - Cosine similarity scores
103
+
104
+ ### Results
105
+ - **MAP@10:** ~0.85
106
+ - **MRR@10:** ~0.90
107
+ - **Performance:** Optimized for presentation template retrieval
108
+
109
+ ## Environmental Impact
110
+ - **Hardware Type:** NVIDIA GPU
111
+ - **Hours used:** ~2 hours
112
+ - **Cloud Provider:** Local/Cloud
113
+ - **Carbon Emitted:** Minimal (local training)
114
+
115
+ ## Technical Specifications
116
+
117
+ ### Model Architecture and Objective
118
+ - **Architecture:** Transformer-based bi-encoder
119
+ - **Objective:** Learn semantic representations for similarity search
120
+ - **Input:** Text sequences (queries and descriptions)
121
+ - **Output:** 384-dimensional embeddings
122
+
123
+ ### Compute Infrastructure
124
+ - **Hardware:** NVIDIA GPU
125
+ - **Software:** PyTorch, SentenceTransformers, Transformers
126
 
127
+ ## Citation
 
 
 
 
 
 
 
 
 
 
 
128
 
129
+ **BibTeX:**
130
  ```bibtex
131
+ @misc{field-adaptive-bi-encoder,
132
+ title={Field-adaptive Bi-encoder for Presentation Template Search},
133
+ author={Mudasir Syed},
134
+ year={2024},
135
+ url={https://huggingface.co/mudasir13cs/Field-adaptive-bi-encoder}
 
 
136
  }
137
  ```
138
 
139
+ **APA:**
140
+ Syed, M. (2024). Field-adaptive Bi-encoder for Presentation Template Search. Hugging Face. https://huggingface.co/mudasir13cs/Field-adaptive-bi-encoder
141
 
 
 
 
 
142
  ## Model Card Authors
143
+ Mudasir Syed (mudasir13cs)
144
 
 
 
 
 
145
  ## Model Card Contact
146
+ - **GitHub:** https://github.com/mudasir13cs
147
+ - **Hugging Face:** https://huggingface.co/mudasir13cs
148
 
149
+ ## Framework versions
150
+ - SentenceTransformers: 2.2.2
151
+ - Transformers: 4.35.0
152
+ - PyTorch: 2.0.0