--- base_model: distilbert/distilroberta-base datasets: [] language: [] library_name: sentence-transformers metrics: - cosine_accuracy - dot_accuracy - manhattan_accuracy - euclidean_accuracy - max_accuracy pipeline_tag: sentence-similarity tags: - sentence-transformers - sentence-similarity - feature-extraction - generated_from_trainer - dataset_size:22878 - loss:TripletLoss widget: - source_sentence: 'The image features an advertisement for Euro-Actie Van Sola, a Dutch company offering various kitchenware items. The poster showcases several images of different utensils and dishes in both English (Euro Acti) and French text alongside the logo "SOLA" at the top right corner. Promotional offers are also highlighted on this page such as discounts like 25% off with promo code MAXIMALJAARSVAARDIGHEID or free delivery service when purchasing from their catalogues through Leenkooken En Nettoelenvoorgangtakken Online Shoppe by using specific codes: €10498637 EFT/NETTOEENWOEGANGTAKKEN OR €€...' sentences: - 'The image features a promotional flyer for three dessert recipes, titled "Zo makkelijk kan leker zijn" which translates to "So easy can make it." The text provides information about the ingredients and preparation times required for each recipe: Aardbeienbaken (vanilla frisee vlees-toffees), Brownies with caramelized pineapple topper; Gevulde Ananasen met frietse Vogel Toefjes ...' - The image features a woman holding an open wallet, seemingly in distress or contemplation. The headline "Altid geld ektor? Is het hoop hoepel met simpel-budgetplan" suggests that the article discusses whether it's possible to make ends meet with simple budgeting plans and possibly touches on financial matters related to money management for individuals who may be struggling financially due to lack of funds within their wallets at any given time during certain periods throughout January 2013 as indicated by February hours mentioned above (hourly wages) like those associated with NIBU which could refer either directly towards her situation described earlier but seems more likely indirectly indicating how one might utilize these resources efficiently over different timeslots/days if they are unable to manage finances effectively currently; however this is purely speculative without further context provided from what we see here only showing someone going through something quite personal about themselves while trying out various methods such as seeking help via phone calls made using numbers listed below title line etc... - The image is a magazine cover featuring an advertisement for Libelle. The main focus of the ad depicts someone walking on what appears to be farmland, suggesting that it may promote rural or agricultural products and services associated with farming activities in such areas. - source_sentence: 'The image is an advertisement for a product called "Predictor". It features a woman in black dress and tank top, standing against the backdrop of red color with white text that reads: ''Dat krijg je evan als jezwaanger bent'' which translates to English as ''When you want your wish comes true'', indicating some form of predictive power or magic associated with this brand. The background suggests it might be related to pregnancy-related products given its association with maternity wear like knickers (underwear) from Italian brands such as Mister & Mrs Italy and Sonia Bianca among others listed on their website www.predictorwebster.com/www_predictortorator' sentences: - The image is a magazine page featuring an article titled "uit". The cover of the issue includes various elements such as photos, text in different languages and designs like logos or graphics that contribute to its overall aesthetic appeal. - 'The image is a magazine cover featuring an article titled " Het patron: ontcoog moet niet langer eten het maar waard aan werk" translated to English as "[Patron]: On the table you must not only eat but also value your work.” The author of this piece, Door Lilien MacNackon''s picture and name are prominently displayed at the top left corner in red text against white background with her smiling face capturing attention towards right side center-bottom part of page 15 from Cosmopolitan Magazine dated April/May issue for year 2003 published by Studio Consult Uitgeverij B.V Amsterdam Netherlands.' - The image is a promotional poster for "Van de Straat" featuring an orange and blue background. In the center of this vibrant backdrop, there's a woman carrying out her belongings from inside what appears to be either car or storage unit near water bodies such as docks with boats nearby in various sizes. The text on top reads 'VAN DE STRAAT,' while below it says that she has found something interesting about Viva magazine which features articles like those described by Tweedehandtopstukken voor binnenen (Second-Hand Items Toppicks For Binns) suggesting they are offering unique items at affordable prices within their stores across Europe. - source_sentence: The image depicts a black and white advertisement for Jil Sander's "SUMMER IN YOU" eau de toilette. A man is standing next to the box, which prominently displays his name in bold red letters against an off-white background that contrasts with him. The slogan on top of this message reads "JIL SANDRER," suggesting it may be related or endorsed by them as well. sentences: - 'The image is a magazine cover featuring an article titled " Het patron: ontcoog moet niet langer eten het maar waard aan werk" translated to English as "[Patron]: On the table you must not only eat but also value your work.” The author of this piece, Door Lilien MacNackon''s picture and name are prominently displayed at the top left corner in red text against white background with her smiling face capturing attention towards right side center-bottom part of page 15 from Cosmopolitan Magazine dated April/May issue for year 2003 published by Studio Consult Uitgeverij B.V Amsterdam Netherlands.' - The image depicts a news article about the city of Goudda, featuring various elements from daily life. The headline "Stammenstrid in gouda" suggests that it's an issue related to urban planning or development within this area. There are several people visible throughout the scene - some walking on sidewalks and others standing near buildings with street signs nearby. A stroller is also present among these individuals, indicating families might be involved too. - The image is a page from an advertisement or brochure, featuring various items related to audio and electronics. In the top left corner of the photo are two headphones with Sony branding on them in different colors - silver for one pair (Bioscoop oren) and black for another set which seems like they're part of some product display setup at Mannenzaken store. The center section shows David sitting comfortably outdoors while wearing these same headsets as he listens intently into his ears during wintertime days when temperatures drop below zero degrees Celsius. Towards the bottom right side there appears to be information about MP417 van David by Capa Coffee Shop & Playboy magazine online shop website www.capacoffeeshopw waterproof tablecloth w/ clip designed specifically for this model laptop case that comes along with its own carrying bag made up out of recycled materials including plastic bottles such as those used before their recycling process starts! - source_sentence: The image features an advertisement for Philips Flat TV, showcasing a man relaxing in bed with his head resting on the armrest of their television. The background includes elements like blue skies and waves as part of its visual appeal to create relaxation or serenity feelings that align well with leisure activities such as watching movies at home during downtime moments. sentences: - The image is a screenshot of an article from the E-Business section. It features two men, likely business figures or executives based on their attire and positions in front of computers with IBM logos visible behind them. The headline indicates that they are discussing enterprise strategies for social media marketing within businesses like Ibm using platforms such as LinkedIn (IBM's name). There appears to be some text related to "Welvaartstaat" which could suggest travel services associated with these companies but it may not fully translate into English due to character limitations here. - The image is a vibrant advertisement for the Pakhuis Amsterdams, showcasing its architectural design and inviting visitors to explore. The building features various elements like red bicycles parked outside on wheels that resemble scooters or skateboards in different colors such as black with orange accents (red). There's also an eye-catching display of bookshelves filled with numerous volumes arranged neatly against one wall. A prominent figure wearing glasses stands near these shelves under a glass roof structure inside the museum entrance area where people can admire works from renowned artists while enjoying refreshments at nearby cafes if they choose so. - The image features a magazine cover page with the title "De goed-gesprek gek" prominently displayed. The background of this colorful and engaging design includes elements such as text, images like people's faces or objects related to fashion, possibly suggesting an article about personal style in magazines that target young women interested in these topics. - source_sentence: The image features a silver hatchback car parked in the middle of an expansive, dense forest. The trees are tall and dark green with visible trunks that reach up into what appears to be early morning light filtering through them due to their proximity at sunset or sunrise. There is no direct sunlight casting shadows on the ground from above; instead, there's soft lighting coming downwards as if it were just after dawn or dusk when natural daylight has not yet reached its peak intensity. sentences: - The image is a magazine cover with various advertisements and articles. The main focus of the advertisement in question appears to be about tomatoes, featuring an attractive woman holding wine glasses next to them on top of what seems like food or kitchen items. There's also text related to "Medicijn" suggesting it might relate to health-related products as well. - 'The image depicts a man standing among an array of white chairs, arranged in rows and clusters. The background is green with the text "en creatie" prominently displayed at the top left corner. There''s also some Dutch writing visible on one side that translates to English as: “Voegd verwijzen en werken met vrij informatietechnologie systems op het eind van de product te kennunnen dat er waarvoor een betere bezit moet wordt gegeven dan ze momentoan kunnen bestaan” - This phrase roughly means “To evaluate whether it makes sense for you or your organization to acquire products based solely upon their immediate availability,” which suggests this might relate to software evaluation processes within Microsoft enterprises seeking better ways to manage resources efficiently while staying current without unnecessary waste.' - The image is a page from the bridge puzzles section of an older edition. The top left corner features "BRIDGE," while at its center, there's text that reads "(Noor Helijkaal) HORIZONTAL" and then continues with more details about various aspects related to this puzzle or game in general. In contrast, below these sections are instructions for solving crosswords called Kettingwoord (the Dutch word meaning 'setting up en prijswinnaars') which includes hints like identifying patterns among clues provided on each square as well as how winning combinations can be found by connecting them across multiple squares without crossing other words already spelled out within those boxes. model-index: - name: SentenceTransformer based on distilbert/distilroberta-base results: - task: type: triplet name: Triplet dataset: name: Magazine type: Magazine metrics: - type: cosine_accuracy value: 0.9041628383631352 name: Cosine Accuracy - type: dot_accuracy value: 0.0963318962470846 name: Dot Accuracy - type: manhattan_accuracy value: 0.8027422432680755 name: Manhattan Accuracy - type: euclidean_accuracy value: 0.9085447734822248 name: Euclidean Accuracy - type: max_accuracy value: 0.9085447734822248 name: Max Accuracy --- # SentenceTransformer based on distilbert/distilroberta-base This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [distilbert/distilroberta-base](https://huggingface.co/distilbert/distilroberta-base). 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:** [distilbert/distilroberta-base](https://huggingface.co/distilbert/distilroberta-base) - **Maximum Sequence Length:** 512 tokens - **Output Dimensionality:** 768 tokens - **Similarity Function:** Cosine Similarity ### 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': 512, 'do_lower_case': False}) with Transformer model: RobertaModel (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}) ) ``` ## 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("sentence_transformers_model_id") # Run inference sentences = [ "The image features a silver hatchback car parked in the middle of an expansive, dense forest. The trees are tall and dark green with visible trunks that reach up into what appears to be early morning light filtering through them due to their proximity at sunset or sunrise. There is no direct sunlight casting shadows on the ground from above; instead, there's soft lighting coming downwards as if it were just after dawn or dusk when natural daylight has not yet reached its peak intensity.", 'The image is a page from the bridge puzzles section of an older edition. The top left corner features "BRIDGE," while at its center, there\'s text that reads "(Noor Helijkaal) HORIZONTAL" and then continues with more details about various aspects related to this puzzle or game in general. In contrast, below these sections are instructions for solving crosswords called Kettingwoord (the Dutch word meaning \'setting up en prijswinnaars\') which includes hints like identifying patterns among clues provided on each square as well as how winning combinations can be found by connecting them across multiple squares without crossing other words already spelled out within those boxes.', 'The image is a magazine cover with various advertisements and articles. The main focus of the advertisement in question appears to be about tomatoes, featuring an attractive woman holding wine glasses next to them on top of what seems like food or kitchen items. There\'s also text related to "Medicijn" suggesting it might relate to health-related products as well.', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 768] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` ## Evaluation ### Metrics #### Triplet * Dataset: `Magazine` * Evaluated with [TripletEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) | Metric | Value | |:-------------------|:-----------| | cosine_accuracy | 0.9042 | | dot_accuracy | 0.0963 | | manhattan_accuracy | 0.8027 | | euclidean_accuracy | 0.9085 | | **max_accuracy** | **0.9085** | ## Training Details ### Training Dataset #### Unnamed Dataset * Size: 22,878 training samples * Columns: sentence_0, sentence_1, and sentence_2 * Approximate statistics based on the first 1000 samples: | | sentence_0 | sentence_1 | sentence_2 | |:--------|:-------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------| | type | string | string | string | | details | | | | * Samples: | sentence_0 | sentence_1 | sentence_2 | |:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| | The image features a model in an advertisement for Nivea body lotion. She is seated on the ground, dressed only in her underwear and smiling at the camera with blonde hair cascading down around her shoulders. The background of the photo consists mainly of blue tones that contrast against the white border surrounding it. In addition to showcasing the product's branding prominently above its logo "NIVEA BODY" which suggests this ad might be promoting skincare products or personal care items associated with beauty routines such as moisturizing creams or sunscreens. | The image is a book page with the headline "Meer dan moeder" written in bold blue letters. The accompanying text provides an introduction to Miranda Landman, who has been described as having her own unique style and approach towards parenting that differs from traditional norms. It mentions how she shares stories of being a mother during challenging times like when one child was diagnosed with cancer or dealing with difficult family dynamics such as divorce. Her perspective emphasizes self-love for herself first before focusing on raising children effectively through love instead of fear tactics which can lead to negative consequences over time. | The image displays a newspaper article from Australia, with the headline "Nieuwe Aussies gezocht" prominently featured. The text in Dutch provides details about an event where 19 million residents have thrown away approximately one billion new pieces of waste every day for several weeks at large dumpsites near towns and cities across New South Wales (NSW) due to high levels of immigration migration resulting in overcrowding conditions like those seen during World War II's displacement camps known as 'Europa', which were part of Europe’s largest refugee settlement system built by German architect Karl Lagerfeld on former Nazi concentration camp grounds after his expulsion following WWII. This is not only causing environmental problems but also social issues within these communities affected disproportionately more than other areas such as Melbourne or Sydney because they are often left behind when economic growth occurs elsewhere without adequate support systems that can accommodate them effectively; this has led many people seeking better opportunities outside their current living arrangements instead turning towards alternative locations abroad including countries located further down under its geographical location situated closer compared to North America while being farther off geographically speaking so it may become easier depending upon individual circumstances regarding travel plans between continents via airplanes offering flights connecting various points along different routes leading ultimately toward worldwide destinations far beyond our immediate vicinity here today! | | The image is a vibrant collage of five polaroid photos, each capturing the joyous moments shared among friends at various social gatherings. The top left photo features two men clinking glasses and smiling broadly; one man holds his glass high in celebration while both are dressed casually with shirts that suggest an informal setting. In contrast to this joyful scene on its own right side stand three more people enjoying their time together: they all have drinks held up near them as if celebrating something or simply sharing laughter over some casual conversation. All these images exude warmth through candid expressions from individuals engaged within lively group interactions during what appears to be late-night get-togethers where good times abound for everyone present. | The image is a collage of four Polaroid-style photos, all featuring people enjoying drinks and having fun. The top left photo shows two individuals smiling while holding up their glasses filled with beer in unison; the bottom right corner has "de barkeepers" written on it as if they are bartenders celebrating or posing for this memorable moment together.
In addition to these main photographs, there's also text that reads: "Nu eerst een Bavaria," which translates from Dutch into English meaning something like 'Now first an experience,' indicating some sort of celebration related event taking place at Bavarian events such as sports matches involving soccer teams named Bayern Munich (Bavaria) ... | The image displays a book cover with text in Dutch. The main focus is on the name "Suzanne," which appears prominently at various points throughout the page, possibly indicating that it's about or by Suzanne Rensers from Columbia Records and 24x7 Music Publishing Bijzondere Muziekgroepen/Columbia Recordings International Nederland BV Voorheemweg Houten Netherlands-based label specializing in classical music albums for adults over age thirteen; also distributes to international markets including China via distribution partners such as Chrysalis Group plc UK based distributor of classic rock & roll artists like Pink Floyd , Queen etc) (www...chrysallismusicgroupplcometl), and Universal Japan Inc Tokyo japan( www…universaljapain com | | The image is a promotional advertisement for the Mercedes-Benz SLK Special Edition car. It features an elegant silver sports car against a soft pink background, emphasizing its sleek design and luxury appeal. The text in Dutch at the bottom of the ad emphasizes that this edition offers exclusivity with no compromises on quality or style - "als exklusief nog niet eindelijk" translates to "as exclusive now not necessarily forever," suggesting ongoing value despite potential changes over time. | The image presents an advertisement for the Audi quattro, showcasing a silver SUV on ice. The car's design is sleek and modern with its distinctive four-ring logo prominently displayed in red at the bottom right corner of the ad. In addition to highlighting the vehicle itself, there are words like "quatto" which suggests it might be associated with performance or driving capabilities due to being named after this legendary model from Volkswagen Group history. | The image is a page from an article about Philips Lighting Poland, featured in the Dutch newspaper "Economie". The main focus of the photo appears to be on the building's architecture and its modern design elements such as large windows with reflective glass panels that enhance visibility outside this facility located at 12 Bulvarz street in Polania-Bulwarza city center area. | * Loss: [TripletLoss](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#tripletloss) with these parameters: ```json { "distance_metric": "TripletDistanceMetric.EUCLIDEAN", "triplet_margin": 5 } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `per_device_train_batch_size`: 16 - `per_device_eval_batch_size`: 16 - `num_train_epochs`: 10 - `multi_dataset_batch_sampler`: round_robin #### All Hyperparameters
Click to expand - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: no - `prediction_loss_only`: True - `per_device_train_batch_size`: 16 - `per_device_eval_batch_size`: 16 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 1 - `eval_accumulation_steps`: None - `learning_rate`: 5e-05 - `weight_decay`: 0.0 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1 - `num_train_epochs`: 10 - `max_steps`: -1 - `lr_scheduler_type`: linear - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.0 - `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`: False - `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`: 0 - `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 - `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`: False - `resume_from_checkpoint`: None - `hub_model_id`: None - `hub_strategy`: every_save - `hub_private_repo`: False - `hub_always_push`: False - `gradient_checkpointing`: False - `gradient_checkpointing_kwargs`: None - `include_inputs_for_metrics`: False - `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 - `dispatch_batches`: None - `split_batches`: None - `include_tokens_per_second`: False - `include_num_input_tokens_seen`: False - `neftune_noise_alpha`: None - `optim_target_modules`: None - `batch_eval_metrics`: False - `batch_sampler`: batch_sampler - `multi_dataset_batch_sampler`: round_robin
### Training Logs | Epoch | Step | Training Loss | Magazine_max_accuracy | |:------:|:-----:|:-------------:|:---------------------:| | 0.3497 | 500 | 3.6095 | - | | 0.6993 | 1000 | 2.1101 | - | | 1.0 | 1430 | - | 0.8553 | | 1.0490 | 1500 | 1.5426 | - | | 1.3986 | 2000 | 1.2006 | - | | 1.7483 | 2500 | 0.9514 | - | | 2.0 | 2860 | - | 0.8845 | | 2.0979 | 3000 | 0.6882 | - | | 2.4476 | 3500 | 0.5506 | - | | 2.7972 | 4000 | 0.4571 | - | | 3.0 | 4290 | - | 0.8956 | | 3.1469 | 4500 | 0.3377 | - | | 3.4965 | 5000 | 0.304 | - | | 3.8462 | 5500 | 0.2386 | - | | 4.0 | 5720 | - | 0.8973 | | 4.1958 | 6000 | 0.1906 | - | | 4.5455 | 6500 | 0.1826 | - | | 4.8951 | 7000 | 0.1479 | - | | 5.0 | 7150 | - | 0.9058 | | 5.2448 | 7500 | 0.1178 | - | | 5.5944 | 8000 | 0.1094 | - | | 5.9441 | 8500 | 0.0814 | - | | 6.0 | 8580 | - | 0.9025 | | 6.2937 | 9000 | 0.0877 | - | | 6.6434 | 9500 | 0.0645 | - | | 6.9930 | 10000 | 0.0601 | - | | 7.0 | 10010 | - | 0.9029 | | 7.3427 | 10500 | 0.0482 | - | | 7.6923 | 11000 | 0.0407 | - | | 8.0 | 11440 | - | 0.9099 | | 8.0420 | 11500 | 0.0299 | - | | 8.3916 | 12000 | 0.0309 | - | | 8.7413 | 12500 | 0.0225 | - | | 9.0 | 12870 | - | 0.9085 | | 9.0909 | 13000 | 0.0158 | - | | 9.4406 | 13500 | 0.0176 | - | ### Framework Versions - Python: 3.10.12 - Sentence Transformers: 3.0.1 - Transformers: 4.41.2 - PyTorch: 2.3.0+cu121 - Accelerate: 0.32.1 - Datasets: 2.20.0 - Tokenizers: 0.19.1 ## 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", } ``` #### TripletLoss ```bibtex @misc{hermans2017defense, title={In Defense of the Triplet Loss for Person Re-Identification}, author={Alexander Hermans and Lucas Beyer and Bastian Leibe}, year={2017}, eprint={1703.07737}, archivePrefix={arXiv}, primaryClass={cs.CV} } ```