File size: 45,708 Bytes
9b076d1 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 |
---
language:
- en
license: apache-2.0
tags:
- sentence-transformers
- sentence-similarity
- feature-extraction
- dense
- generated_from_trainer
- dataset_size:441985
- loss:CachedMultipleNegativesRankingLoss
base_model: google/embeddinggemma-300m
widget:
- source_sentence: "title: \nSwords of Revealing Light SDPLEN026 Common 1st Edition"
sentences:
- 'description
Structure Deck Powercode Link is a Structure Deck in the YuGiOh Official Card
Game and YuGiOh Trading Card Game It is the 35th Deck in the OCGs Structure Deck
series following Structure Deck Cyberse Link It is the 42nd Deck in the TCGs Structure
Deck series following Lair of Darkness Structure Deck'
- 'description
Features Glowing in the dark effect These handmade luminous marbles are eyecathing
and in line with the most party themes such as black light party neon theme party
growth party Halloween party and so on so you can use them to decorate your party
to create the party vibe you want Play together These doted style glass marbles
are funny toys for boys and girls They can play the related game on the garden
yard lawn playground park and more It is a nice opportunity to develop their social
skill and get away from electronic products Specifications Material glass Diameter
approx 063 inch Color as pictures shown Package included 35 x Luminous marbles Warm
notices Manual measurement please allow slight errors on size The color might
exist slightly difference due to different displayswhat you will get package comes
with 35 pieces handmade glass marbles in 7 colors 5 pieces of each color with
the same size enough quantity to meet your need You can also share them with your
friends
Size and muticolor doted style glass marbles are 063 inch in diameter featuring
7 colors such as transparent pinkishorange yellow purple green sea blue and dark
blue They are glowing at night and look very interesting then you can imagine
yourself in the universe
Quality material these muticolors luminous marbles are made from quality glass
which are firm and sturdy smooth with a nice finish durable and highstrength designed
for longtime use giving you nice play feeling
How to make marbles glow brighter you can make the glow marbles sunbathe or illuminate
them with ultraviolet light to make the marbles shine brighter and longer bringing
you delightful mood Recommended for ages 5 years and up
Multiple uses the luminous glass marbles can effectively improve boys and girls
eye to hand coordination and enhance their fine motor skills They are also nice
stocking stuffers Easter baskets birthday presents and home decoration For example
you can use them to add a splash of color in your fish tanks and vases'
- 'description
Legendary Encounters A Predator Deck Building Game is really two games in one
You can play as humans working together to escape and fight off the Predator Or
your group can play as Predators competing to hunt humans and earn the most Honor
You can mix both Legendary Encounters games in order to play as Predators hunting
the Aliens You would play on the Aliens board pick a location from the Aliens
game choose Aliens Objectives and more See Rulebook for details If youre feeling
extra adventurous there are even more ways to mix the games such as playing as
Predators and hunting a mix of Human Prey and Aliens Or you could mix in Marvel
Legendary cards for even more craziness700 playable cards all with original art
Game mat and rulebook included
15 players
3060 minute play time
Cards included 35 Experience 25 Brute Strength 15 Role Avatars 10 Human 5 Predator
15 Role Character Cards 10 Human 5 Predator 10 Commanders 8 Killer Instincts 224
Character cards 16 characters with 14 cards each
Cards Included 2 Locations 6 Objectives 66 Enemy cards 6 different minidecks 66
Prey cards 6 different minidecks 24 Young Blood cards 24 Mercenary cards 40 Enemy
Strikes 60 Prey Strikes 20 Traps 20 Gear 20 Tests 20 Challenges'
- source_sentence: "title: \nFunko POP Muppets VINYL Snowth"
sentences:
- 'description
From the Manufacturer
This Muppets Snowth Pop Vinyl figure will have you humming Do Dooo Do Doo doo
too From Jim Hensons Muppets comes Snowth the furry pink creature with horns
and round yellow lips Snowths were performed simultaneously by Frank Oz with each
puppet on each arm Snowth is adorable as ever as it stands 4 tall and comes in
a stylized artful display boxCollect them all
Displayable window box
Muppets Classics'
- 'description
On September 20 1983 in Burbank CA the USPS released this setenant in an effort
to break down he barriers and stereotypes which have been created One stamp simply
shows the American Sign Language sign for I Love You while the other pictures
a mother signing I Love You to her childADA
Deaf
Hearing Impaired'
- 'description
From Avatar Ty Lee as a stylized Pop vinyl from Funko Figure stands 3 34 inches
and comes in a window display box Check out the other Avatar figures from Funko
Collect them allImported
Product TypeToys And Games
Item Package Dimension35 L X45 W X625 H
Item Package Weight025 Lbs
Country Of Origin Viet Nam'
- source_sentence: "title: \nFuryu Accel World Kuroyukihime Figure"
sentences:
- 'description
From the Manufacturer
A rescue themed vehicle with a Dalmation RollaRound that both magically come to
life when baby plays Push along the Firetruck to see the siren lights move up
and down Put the Round into the the truck to see the Dogs head turn back and forth
as he rollA rescue themed vehicle with a Dalmation RollaRound that both magically
come to life when baby plays
Push along the Firetruck to see the siren lights move up and down
Put the Round into the the truck to see the Dogs head turn back and forth as he
rolls along
Part of RollARounds Collection
Age Range 6 to 36 Months'
- 'description
Disney Toy Story Buzz Lightyear Little Lights with Hook for Samsung Galaxy S3
Mini i8190'
- 'description
Accel World FuRyu 5 Kuroyukihime Sitting PVC Statue KuroyukihimeBrand New Official
Item
Great for Collectors'
- source_sentence: "title: \nJellycat Pretty Patisserie Tarte Au Citron Food Plush"
sentences:
- 'description
Features Benefits 60pcs New Year latex balloons and foil balloons colored in
gold and black printed with festive words are great atmosphere creator for your
home and party Made of durable latex or foil material wonderful for hanging on
the wall ceiling window and other places you want Specifications Size 12 inches
18 inch Color black gold Material latex foil material Package Includes 60 x New
Year Balloon 4 x Foil BalloonLarge Quantity Are you looking for New Year decorations
Our cute balloons will be a great choice for you 60pcs latex balloons and 4pcs
large foil balloons great combination for your home and party decoration
Delicate Design Taking black and gold as theme colors printed with festive words
or designed with confetti the foil balloons are shaped in star or round classic
and exquisite
Good Quality Made of good quality latex or foil material safe and durable the
foil one is reusable you can fill them with air or helium
Size Each latex balloon measures 12 inch and the foil balloons measure 18 inch
appropriate size for hanging on the wall ceiling window and other places you want
Wide Application Perfect to decorate your New Year party decoration Christmas
school activities and other occasions suitable for both indoor and outdoor bar
stage props etc'
- 'description
Finally A pet bear that wont gnaw on your head or bite your arms off This adorable
bouncing head animal doll is ready for you to give him a great big bear hug Soft
white fur makes this shaking head figure look great on any dashboard bookshelf
desktop or countertop He loves the dashboard adhesive that travels with him because
the sticky pad reminds him of honey Youll love it because it will securely attach
him to your car dashboard but it wont leave any permanent residue With adorable
beady eyes this bobblehead will make you the envy of all your friends Perfect
gift for any bobble head collector a trucker with an empty dashboard or children
Kids love these cute little bobbing head toys One of many ridiculously awesome
nodding head dash board therapists brought to you as part of Batty Bargains Legendary
Bobbleheads collection Order yours todayNo live bears were harmed in the making
of this bobblehead bear Fuzzy black fur made of velvet textured flocking
Big strong paws and great listening skills make for a phenomenal dash board therapist
When your not strong lean on this bobbing head bear Its bouncing head will help
you though the tough times
A beautiful toy dash board doll great for any road trip One of many dashboard
figures available as part of Batty Bargains Legendary Bobbleheads Collection
Adjustable bobble head moves easily and smoothly making it look like it is nodding
or shaking its head Guaranteed to bob at a wide range of angles Just loosen or
tighten the weighted bolt on the back of its bobbing head
Easily mounts on almost any car or vehicle dashboard in seconds with included
adhesive pad Easily removed without leaving any permanent residue'
- 'description
The Jellycat brand was established in London in 1999 to create quirky original
and innovative soft toys for all ages Jellycat offers the best selection of soft
plush stuffed animals and toys in the cutest and most luxurious of fabrics and
textures With unmatched quality find the perfect stuffed animal with the coolest
designs for babies kids and adults alike Plush toys from Jellycat come in a variety
of sizes to cuddle including mini small medium large huge and really big They
also feature a variety of collections and themes to adorn any nursery or childrens
room for both boys and girls making a Jellycat the perfect gift Best of all your
Jellycat stuffed animal will provide countless hours of soft hugs and memories
for years to comePlush measures 4 x 4 x 2 inches
Suitable for all ages
Made of 95 Polyester 5 Spandex
Spot clean only
Designed by Jellycat in London UK'
- source_sentence: "title: \nFairytale and Historic Minifigure Set"
sentences:
- 'description
Features 1 Cartoon Dinosaur Plush Doll is suitable for being Ornament Gift Pillow
and Rag Toy 2 Made of quality soft plush and being fully filled with pp cotton
Not afraid of squeezing 3 Not easy to deform with good resilience It has tightly
threaded side which features not easy to produce cotton 4 Perfect to cuddle with
in bed or to keep you company when you nap on the couch Great gift for kids girls
and anyone you love 5 For the cleaning of such short plush doll clean the surface
with cold water and place it in the sun after washing to ensure that the short
plush is fluffy Notice Actual color may be slightly different from the image due
to different display and light effect Please allow 13cm deviation due to manual
measurement Description Material plush PP cotton Size 3560cm Decoration form hanging
placing ColorStyle As the picture shown'
- 'description
Toysmiths mission is to supply quality toys and gifts while delivering superior
customer service to retailers We offer products in many key categories including
active play science discovery arts crafts impulse novelty toys and nostalgic
retro classics Since our inception in 1981 our owners Bill and Nancy Smith have
worked very diligently to grow from a small family business shipping from their
garage to a large office complex and distribution center that stocks and ships
over 1600 products to more than 6000 accounts However our philosophy remains the
same maintaining a strong family work environment while offering the best selection
and quality of products together with the highest standards of service to all
our customersEducational toys that help children learn
Made using safe and high quality materials
Toys for all age groups'
- 'description
The LEGO Education 779349 or 9349 227piece fairytale and historic minifigure set
includes elements to build 22 multicultural male and female minifigures representing
characters from fairytales and history Children four years and older can explore
roleplaying while creating and reenacting fairytales and stories developing imagination
and imitative skills Themes include pirates a witch and wizard a king and queen
knights mine workers a mermaid and merman and more and accessories such as wands
hats flowers spiders and snakes The set is compatible with any LEGO Education
system supports a group of five students and comes with a decorated box for storage
Since 1980 LEGO Education has delivered handson curriculumbased resources for
teachers and students worldwide LEGO Education believes a handson mindson approach
helps students actively take ownership of the learning process and develop 21stcentury
skills such as creative thinking and problem solving through reallife engaging
experiencesExploring differences to real life makebelieve and historic characters
Storytelling through characters and things they do
Developing fantasy imagination and imitative skills
Featuring a variety of minifigures that enable children to create and act out
their favourite fairytales and stories
Ideal for 4 years old children or above'
datasets:
- guyhadad01/Amazon_2023_items_processed_filtered
pipeline_tag: sentence-similarity
library_name: sentence-transformers
metrics:
- cosine_accuracy@1
- cosine_accuracy@3
- cosine_accuracy@5
- cosine_accuracy@10
- cosine_precision@1
- cosine_precision@3
- cosine_precision@5
- cosine_precision@10
- cosine_recall@1
- cosine_recall@3
- cosine_recall@5
- cosine_recall@10
- cosine_ndcg@10
- cosine_mrr@10
- cosine_map@100
model-index:
- name: EmbeddingGemma-300m fine-tuned on Amazon Toys & Games
results:
- task:
type: information-retrieval
name: Information Retrieval
dataset:
name: dev eval
type: dev-eval
metrics:
- type: cosine_accuracy@1
value: 0.308
name: Cosine Accuracy@1
- type: cosine_accuracy@3
value: 0.437
name: Cosine Accuracy@3
- type: cosine_accuracy@5
value: 0.507
name: Cosine Accuracy@5
- type: cosine_accuracy@10
value: 0.585
name: Cosine Accuracy@10
- type: cosine_precision@1
value: 0.308
name: Cosine Precision@1
- type: cosine_precision@3
value: 0.1456666666666667
name: Cosine Precision@3
- type: cosine_precision@5
value: 0.10139999999999998
name: Cosine Precision@5
- type: cosine_precision@10
value: 0.0585
name: Cosine Precision@10
- type: cosine_recall@1
value: 0.308
name: Cosine Recall@1
- type: cosine_recall@3
value: 0.437
name: Cosine Recall@3
- type: cosine_recall@5
value: 0.507
name: Cosine Recall@5
- type: cosine_recall@10
value: 0.585
name: Cosine Recall@10
- type: cosine_ndcg@10
value: 0.43779247661127096
name: Cosine Ndcg@10
- type: cosine_mrr@10
value: 0.39167420634920597
name: Cosine Mrr@10
- type: cosine_map@100
value: 0.3994706245247452
name: Cosine Map@100
---
# EmbeddingGemma-300m fine-tuned on Amazon Toys & Games
This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [google/embeddinggemma-300m](https://huggingface.co/google/embeddinggemma-300m) on the [amazon_2023_items_processed_filtered](https://huggingface.co/datasets/guyhadad01/Amazon_2023_items_processed_filtered) dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
## Model Details
### Model Description
- **Model Type:** Sentence Transformer
- **Base model:** [google/embeddinggemma-300m](https://huggingface.co/google/embeddinggemma-300m) <!-- at revision 57c266a740f537b4dc058e1b0cda161fd15afa75 -->
- **Maximum Sequence Length:** 2048 tokens
- **Output Dimensionality:** 768 dimensions
- **Similarity Function:** Cosine Similarity
- **Training Dataset:**
- [amazon_2023_items_processed_filtered](https://huggingface.co/datasets/guyhadad01/Amazon_2023_items_processed_filtered)
- **Language:** en
- **License:** apache-2.0
### Model Sources
- **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
- **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)
- **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
### Full Model Architecture
```
SentenceTransformer(
(0): Transformer({'max_seq_length': 2048, 'do_lower_case': False, 'architecture': 'Gemma3TextModel'})
(1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
(2): Dense({'in_features': 768, 'out_features': 3072, 'bias': False, 'activation_function': 'torch.nn.modules.linear.Identity'})
(3): Dense({'in_features': 3072, 'out_features': 768, 'bias': False, 'activation_function': 'torch.nn.modules.linear.Identity'})
(4): Normalize()
)
```
## Usage
### Direct Usage (Sentence Transformers)
First install the Sentence Transformers library:
```bash
pip install -U sentence-transformers
```
Then you can load this model and run inference.
```python
from sentence_transformers import SentenceTransformer
# Download from the 🤗 Hub
model = SentenceTransformer("rabaevn/EncodeRec")
# Run inference
queries = [
"title: \nFairytale and Historic Minifigure Set",
]
documents = [
'description\nThe LEGO Education 779349 or 9349 227piece fairytale and historic minifigure set includes elements to build 22 multicultural male and female minifigures representing characters from fairytales and history Children four years and older can explore roleplaying while creating and reenacting fairytales and stories developing imagination and imitative skills Themes include pirates a witch and wizard a king and queen knights mine workers a mermaid and merman and more and accessories such as wands hats flowers spiders and snakes The set is compatible with any LEGO Education system supports a group of five students and comes with a decorated box for storage\nSince 1980 LEGO Education has delivered handson curriculumbased resources for teachers and students worldwide LEGO Education believes a handson mindson approach helps students actively take ownership of the learning process and develop 21stcentury skills such as creative thinking and problem solving through reallife engaging experiencesExploring differences to real life makebelieve and historic characters\nStorytelling through characters and things they do\nDeveloping fantasy imagination and imitative skills\nFeaturing a variety of minifigures that enable children to create and act out their favourite fairytales and stories\nIdeal for 4 years old children or above',
'description\nToysmiths mission is to supply quality toys and gifts while delivering superior customer service to retailers We offer products in many key categories including active play science discovery arts crafts impulse novelty toys and nostalgic retro classics Since our inception in 1981 our owners Bill and Nancy Smith have worked very diligently to grow from a small family business shipping from their garage to a large office complex and distribution center that stocks and ships over 1600 products to more than 6000 accounts However our philosophy remains the same maintaining a strong family work environment while offering the best selection and quality of products together with the highest standards of service to all our customersEducational toys that help children learn\nMade using safe and high quality materials\nToys for all age groups',
'description\nFeatures 1 Cartoon Dinosaur Plush Doll is suitable for being Ornament Gift Pillow and Rag Toy 2 Made of quality soft plush and being fully filled with pp cotton Not afraid of squeezing 3 Not easy to deform with good resilience It has tightly threaded side which features not easy to produce cotton 4 Perfect to cuddle with in bed or to keep you company when you nap on the couch Great gift for kids girls and anyone you love 5 For the cleaning of such short plush doll clean the surface with cold water and place it in the sun after washing to ensure that the short plush is fluffy Notice Actual color may be slightly different from the image due to different display and light effect Please allow 13cm deviation due to manual measurement Description Material plush PP cotton Size 3560cm Decoration form hanging placing ColorStyle As the picture shown',
]
query_embeddings = model.encode_query(queries)
document_embeddings = model.encode_document(documents)
print(query_embeddings.shape, document_embeddings.shape)
# [1, 768] [3, 768]
# Get the similarity scores for the embeddings
similarities = model.similarity(query_embeddings, document_embeddings)
print(similarities)
# tensor([[0.4024, 0.3496, 0.2501]])
```
<!--
### Direct Usage (Transformers)
<details><summary>Click to see the direct usage in Transformers</summary>
</details>
-->
<!--
### Downstream Usage (Sentence Transformers)
You can finetune this model on your own dataset.
<details><summary>Click to expand</summary>
</details>
-->
<!--
### Out-of-Scope Use
*List how the model may foreseeably be misused and address what users ought not to do with the model.*
-->
## Evaluation
### Metrics
#### Information Retrieval
* Dataset: `dev-eval`
* Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator)
| Metric | Value |
|:--------------------|:-----------|
| cosine_accuracy@1 | 0.308 |
| cosine_accuracy@3 | 0.437 |
| cosine_accuracy@5 | 0.507 |
| cosine_accuracy@10 | 0.585 |
| cosine_precision@1 | 0.308 |
| cosine_precision@3 | 0.1457 |
| cosine_precision@5 | 0.1014 |
| cosine_precision@10 | 0.0585 |
| cosine_recall@1 | 0.308 |
| cosine_recall@3 | 0.437 |
| cosine_recall@5 | 0.507 |
| cosine_recall@10 | 0.585 |
| **cosine_ndcg@10** | **0.4378** |
| cosine_mrr@10 | 0.3917 |
| cosine_map@100 | 0.3995 |
<!--
## Bias, Risks and Limitations
*What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
-->
<!--
### Recommendations
*What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
-->
## Training Details
### Training Dataset
#### amazon_2023_items_processed_filtered
* Dataset: [amazon_2023_items_processed_filtered](https://huggingface.co/datasets/guyhadad01/Amazon_2023_items_processed_filtered) at [6b58dd1](https://huggingface.co/datasets/guyhadad01/Amazon_2023_items_processed_filtered/tree/6b58dd18854109aac31652e941c667725f6352f0)
* Size: 441,985 training samples
* Columns: <code>question</code> and <code>passage_text</code>
* Approximate statistics based on the first 1000 samples:
| | question | passage_text |
|:--------|:----------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------|
| type | string | string |
| details | <ul><li>min: 9 tokens</li><li>mean: 25.35 tokens</li><li>max: 78 tokens</li></ul> | <ul><li>min: 17 tokens</li><li>mean: 194.73 tokens</li><li>max: 892 tokens</li></ul> |
* Samples:
| question | passage_text |
|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| <code>title: <br>My Little Pony Twisty Twirly Hairstyles Applejack</code> | <code>description<br>Style Applejack ponys hair for the best night ever Inspired by the My Little Pony Friendship is Magic television series Applejack is going to the Grand Galloping Gala at Canterlot Castle This Applejack figure is poseable with waxinfused yarn hair that can be styled up in barrettes or fancy braids Theres many style possibilities This pony figure comes with 10 accessories for making Applejack ponys beautiful yellow hair look gorgeous for the gala Also look for Twisty Twirly Hairstyles Rarity and Pinkie Pie figures and mix and match accessories for more hairstyling fun Each sold separately Subject to availability My Little Pony and all related characters are trademarks of HasbroInspired by the My Little Pony Friendship is Magic television series<br>Pretend to do Applejack ponys hair with clips and barrettes<br>Poseable waxinfused yarn hair can be styled up or down<br>Figure scale 3 inches<br>Includes pony figure and 10 accessories</code> |
| <code>title: <br>TUANTUAN 4 Pcs 16 Scale Dollhouse Miniature Furniture Folding Chair Foldable Chair Model Folding Doll Chairs Decor Black Foldable Chair for Figure Accessories</code> | <code>description<br>MATERIALPlasticAlloyColour BlackSize Folding Size 173 85CmExpanding Size 8895153Cm GOOD QUALITYThese Mini Folding Chairs Are Made Of Plastic And Alloy Materials Which Are Environmentally Friendly And NonToxic Strong And Durable Stand Firm UNIQUE DESIGNMini Folding Chair Design Looks Like A Real Chair Small And Exquisite Realistic Appearance More Interesting To Match With Dolls WIDELY USEDThese Mini Folding Chairs Are 16 Scene Accessories Not For Real People Use And Are Exclusively For 12Inch Dolls Suitable For Doll House Accessories Miniature Furniture House Model Decoration It Can Also Be Used As A Mobile Phone Holder PACKAGE INCLUDE4 Pcs Dolls Folding ChairsMATERIALPlasticAlloyColour BlackSize Folding Size 173 85CmExpanding Size 8895153Cm<br>GOOD QUALITYThese Mini Folding Chairs Are Made Of Plastic And Alloy Materials Which Are Environmentally Friendly And NonToxic Strong And Durable Stand Firm<br>UNIQUE DESIGNMini Folding Chair Design Looks Like A Real Chair Small And Exquis...</code> |
| <code>title: <br>Transformers Movie RD27 NEST Sky stalker japan import</code> | <code>description<br>Manufactured by Takara Tomy Product name Transformers Revenge of the Fallen Scout class RD27 NEST Sky stalker TRANSFORMERS REVENGE OF THE FALLEN SCUOT CLASS NEST SKYSTALKERb safety standards b ST Mark<br>b target Gender b boy<br>b Age b from 5 years</code> |
* Loss: [<code>CachedMultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cachedmultiplenegativesrankingloss) with these parameters:
```json
{
"scale": 20.0,
"similarity_fct": "cos_sim",
"mini_batch_size": 32,
"gather_across_devices": false
}
```
### Evaluation Dataset
#### amazon_2023_items_processed_filtered
* Dataset: [amazon_2023_items_processed_filtered](https://huggingface.co/datasets/guyhadad01/Amazon_2023_items_processed_filtered) at [6b58dd1](https://huggingface.co/datasets/guyhadad01/Amazon_2023_items_processed_filtered/tree/6b58dd18854109aac31652e941c667725f6352f0)
* Size: 110,497 evaluation samples
* Columns: <code>question</code> and <code>passage_text</code>
* Approximate statistics based on the first 1000 samples:
| | question | passage_text |
|:--------|:---------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------|
| type | string | string |
| details | <ul><li>min: 7 tokens</li><li>mean: 25.1 tokens</li><li>max: 93 tokens</li></ul> | <ul><li>min: 15 tokens</li><li>mean: 189.31 tokens</li><li>max: 948 tokens</li></ul> |
* Samples:
| question | passage_text |
|:----------------------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| <code>title: <br>Reaper Miniatures Tengu Warrior 03774 Dark Heaven Legends RPG DD Mini Figure</code> | <code>description<br>Tengu WarriorBy Artist Derek SchubertDark Heaven Legends25mm Heroic Scale Fantasy MiniaturesIntegral builtin basesUnpainted metal models that may require assembly If assembly is needed glue or putty is required not includedHuge selection of characters and monsters for fantasy roleplayers miniatures painters and wargamersDark Heaven Legends is the premier 25mm Heroic Scale fantasy line for miniature painters roleplayers and wargamers Over the last thirteen years the Dark Heaven line has produced over 1300 fantasy miniatures designed and crafted by the top miniatures sculptors in the worldFound in Reaper Miniatures Category Dark Heaven LegendsFound in Reaper Miniatures Category Dark Heaven Legends<br>Unpainted Metal Miniature<br>Figure designed by artist Derek Schubert</code> |
| <code>title: <br>Marvel Legends Black Widow Action Figure</code> | <code>description<br>Scale 6 Inch Format Action Figures Packaging Clamshell Manufacturer Toy Biz Natalia Romanova aka Natasha Romanoff is the Black Widow as beautiful as Russian spy as their has ever been and as deadly as her namesake Working with everyone from SHIELD to the Avengers to the Marvel Knights she is widely known as a valuable asset to any crimefighting team In this 6inch Marvel Legends her beauty and power are perfectly captured in the sculpt of her skintight black outfit We receive both US and Canadian Cases You will get either English bilingual or trilingual carded figures based on availability Please understand this before ordering Package condition may vary due to size and weight We do not guarantee mint on cardMarvel Legends 6 Inch Action Figure Man Thing Series Black Widow<br>Marvel<br>Toybiz<br>6 Inch</code> |
| <code>title: <br>Cabbage Patch Kids Cutie Dash The Deer 9 Collectible Adoptable Baby Doll Toy Officially Licensed Gift for Girls and Boys</code> | <code>description<br>This Holiday complete your collection of Cabbage Patch Cuties by adopting the Dash Reindeer doll Each Cabbage Patch Cutie features a snuggly onesie with adjustable hood and can really suck its thumb Cutie dolls come with the traditional signature baby powder scent that Cabbage Patch Kid fans know and love Take the Oath of Adoption Cabbage Patch Kid Cuties are numbered for collectibility and make a great toy gift for boys and girls who love Cabbage Patch Kids Officially licensed Cabbage Patch Kids merchandise 3 Pack Each Measures approximately 9 tall Comes in sealed polybag packaging with official Cabbage Patch tagThis Holiday complete your collection of Cabbage Patch Cuties by adopting Dash the Reindeer baby doll<br>Its a Cabbage Christmas Each Cabbage Patch Cutie features a snuggly onesie with adjustable hood and can really suck its thumb<br>Cutie baby dolls come with the traditional signature baby powder scent that Cabbage Patch Kid fans know and love<br>Take the Oath of Adoption ...</code> |
* Loss: [<code>CachedMultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cachedmultiplenegativesrankingloss) with these parameters:
```json
{
"scale": 20.0,
"similarity_fct": "cos_sim",
"mini_batch_size": 32,
"gather_across_devices": false
}
```
### Training Hyperparameters
#### Non-Default Hyperparameters
- `eval_strategy`: steps
- `per_device_train_batch_size`: 2
- `per_device_eval_batch_size`: 1
- `gradient_accumulation_steps`: 32
- `torch_empty_cache_steps`: 50
- `learning_rate`: 2e-05
- `num_train_epochs`: 1
- `warmup_ratio`: 0.1
- `fp16`: True
- `dataloader_num_workers`: 4
- `push_to_hub`: True
- `hub_model_id`: rabaevn/EncodeRec
- `hub_strategy`: checkpoint
- `gradient_checkpointing`: True
- `batch_sampler`: no_duplicates
#### All Hyperparameters
<details><summary>Click to expand</summary>
- `overwrite_output_dir`: False
- `do_predict`: False
- `eval_strategy`: steps
- `prediction_loss_only`: True
- `per_device_train_batch_size`: 2
- `per_device_eval_batch_size`: 1
- `per_gpu_train_batch_size`: None
- `per_gpu_eval_batch_size`: None
- `gradient_accumulation_steps`: 32
- `eval_accumulation_steps`: None
- `torch_empty_cache_steps`: 50
- `learning_rate`: 2e-05
- `weight_decay`: 0.0
- `adam_beta1`: 0.9
- `adam_beta2`: 0.999
- `adam_epsilon`: 1e-08
- `max_grad_norm`: 1.0
- `num_train_epochs`: 1
- `max_steps`: -1
- `lr_scheduler_type`: linear
- `lr_scheduler_kwargs`: {}
- `warmup_ratio`: 0.1
- `warmup_steps`: 0
- `log_level`: passive
- `log_level_replica`: warning
- `log_on_each_node`: True
- `logging_nan_inf_filter`: True
- `save_safetensors`: True
- `save_on_each_node`: False
- `save_only_model`: False
- `restore_callback_states_from_checkpoint`: False
- `no_cuda`: False
- `use_cpu`: False
- `use_mps_device`: False
- `seed`: 42
- `data_seed`: None
- `jit_mode_eval`: False
- `use_ipex`: False
- `bf16`: False
- `fp16`: True
- `fp16_opt_level`: O1
- `half_precision_backend`: auto
- `bf16_full_eval`: False
- `fp16_full_eval`: False
- `tf32`: None
- `local_rank`: 0
- `ddp_backend`: None
- `tpu_num_cores`: None
- `tpu_metrics_debug`: False
- `debug`: []
- `dataloader_drop_last`: False
- `dataloader_num_workers`: 4
- `dataloader_prefetch_factor`: None
- `past_index`: -1
- `disable_tqdm`: False
- `remove_unused_columns`: True
- `label_names`: None
- `load_best_model_at_end`: False
- `ignore_data_skip`: False
- `fsdp`: []
- `fsdp_min_num_params`: 0
- `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
- `fsdp_transformer_layer_cls_to_wrap`: None
- `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
- `deepspeed`: None
- `label_smoothing_factor`: 0.0
- `optim`: adamw_torch_fused
- `optim_args`: None
- `adafactor`: False
- `group_by_length`: False
- `length_column_name`: length
- `ddp_find_unused_parameters`: None
- `ddp_bucket_cap_mb`: None
- `ddp_broadcast_buffers`: False
- `dataloader_pin_memory`: True
- `dataloader_persistent_workers`: False
- `skip_memory_metrics`: True
- `use_legacy_prediction_loop`: False
- `push_to_hub`: True
- `resume_from_checkpoint`: None
- `hub_model_id`: rabaevn/EncodeRec
- `hub_strategy`: checkpoint
- `hub_private_repo`: None
- `hub_always_push`: False
- `hub_revision`: None
- `gradient_checkpointing`: True
- `gradient_checkpointing_kwargs`: None
- `include_inputs_for_metrics`: False
- `include_for_metrics`: []
- `eval_do_concat_batches`: True
- `fp16_backend`: auto
- `push_to_hub_model_id`: None
- `push_to_hub_organization`: None
- `mp_parameters`:
- `auto_find_batch_size`: False
- `full_determinism`: False
- `torchdynamo`: None
- `ray_scope`: last
- `ddp_timeout`: 1800
- `torch_compile`: False
- `torch_compile_backend`: None
- `torch_compile_mode`: None
- `include_tokens_per_second`: False
- `include_num_input_tokens_seen`: False
- `neftune_noise_alpha`: None
- `optim_target_modules`: None
- `batch_eval_metrics`: False
- `eval_on_start`: False
- `use_liger_kernel`: False
- `liger_kernel_config`: None
- `eval_use_gather_object`: False
- `average_tokens_across_devices`: False
- `prompts`: None
- `batch_sampler`: no_duplicates
- `multi_dataset_batch_sampler`: proportional
- `router_mapping`: {}
- `learning_rate_mapping`: {}
</details>
### Training Logs
| Epoch | Step | Training Loss | Validation Loss | dev-eval_cosine_ndcg@10 |
|:------:|:----:|:-------------:|:---------------:|:-----------------------:|
| 0.0072 | 50 | 0.0428 | - | - |
| 0.0145 | 100 | 0.0118 | - | - |
| 0.0217 | 150 | 0.0087 | - | - |
| 0.0290 | 200 | 0.0064 | - | - |
| 0.0362 | 250 | 0.0069 | - | - |
| 0.0434 | 300 | 0.0088 | - | - |
| 0.0507 | 350 | 0.0055 | - | - |
| 0.0579 | 400 | 0.0067 | - | - |
| 0.0652 | 450 | 0.0098 | - | - |
| 0.0724 | 500 | 0.0096 | - | - |
| 0.0796 | 550 | 0.0104 | - | - |
| 0.0869 | 600 | 0.0155 | - | - |
| 0.0941 | 650 | 0.0109 | - | - |
| 0.1014 | 700 | 0.0144 | - | - |
| 0.1086 | 750 | 0.0109 | - | - |
| 0.1158 | 800 | 0.0107 | 0.0 | 0.4378 |
### Framework Versions
- Python: 3.10.18
- Sentence Transformers: 5.1.0
- Transformers: 4.55.2
- PyTorch: 2.8.0+cu126
- Accelerate: 1.10.0
- Datasets: 4.1.1
- Tokenizers: 0.21.4
## Citation
### BibTeX
#### Sentence Transformers
```bibtex
@inproceedings{reimers-2019-sentence-bert,
title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
author = "Reimers, Nils and Gurevych, Iryna",
booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
month = "11",
year = "2019",
publisher = "Association for Computational Linguistics",
url = "https://arxiv.org/abs/1908.10084",
}
```
#### CachedMultipleNegativesRankingLoss
```bibtex
@misc{gao2021scaling,
title={Scaling Deep Contrastive Learning Batch Size under Memory Limited Setup},
author={Luyu Gao and Yunyi Zhang and Jiawei Han and Jamie Callan},
year={2021},
eprint={2101.06983},
archivePrefix={arXiv},
primaryClass={cs.LG}
}
```
<!--
## Glossary
*Clearly define terms in order to be accessible across audiences.*
-->
<!--
## Model Card Authors
*Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
-->
<!--
## Model Card Contact
*Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
--> |