--- base_model: sentence-transformers/all-mpnet-base-v2 language: - en library_name: sentence-transformers license: apache-2.0 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 - dot_accuracy@1 - dot_accuracy@3 - dot_accuracy@5 - dot_accuracy@10 - dot_precision@1 - dot_precision@3 - dot_precision@5 - dot_precision@10 - dot_recall@1 - dot_recall@3 - dot_recall@5 - dot_recall@10 - dot_ndcg@10 - dot_mrr@10 - dot_map@100 pipeline_tag: sentence-similarity tags: - sentence-transformers - sentence-similarity - feature-extraction - generated_from_trainer - dataset_size:5166 - loss:MultipleNegativesRankingLoss widget: - source_sentence: 'Question: Who is the dungeon master in the Knights of the Arcade comedy show, and how are the destinations and battles decided during the performance?' sentences: - 'Event Name: Knights of the Arcade: Epic D&D Adventure Categories: Entertainment, Nightlife Dates: Jun 29, 2024 - Jun 29, 2024 | 9:00 pm - 10:30 pm Location: Arcade Comedy Theater, 943 Liberty Ave, Pittsburgh, PA 15222 Description: “Best Nerd Fantasy Come to Life” by Pittsburgh Magazine“A neo-geek wet dream” – Pittsburgh City PaperA comedy quest awaits! Knights of the Arcade is an award-winning comedy show that takes audiences on a wild, madcap adventure every month. A recurring cast of characters (a dwarf, a monk, a rogue, a sorcerer and a fighter) are joined by special guests and led by their maniacal dungeon master. Where they’re going, who they fight, and if they ultimately succeed is decided upon live dice that are rolled and projected on the theater wall.' - The Pirates are also often referred to as the "Bucs" or the "Buccos" (derived from buccaneer, a synonym for pirate). Since 2001 the team has played its home games at PNC Park, a 39,000-seat stadium along the Allegheny River in Pittsburgh's North Side. The Pirates previously played at Forbes Field from 1909 to 1970 and at Three Rivers Stadium from 1970 to 2000. Since 1948 the Pirates' colors have been black, gold and white, derived from the flag of Pittsburgh and matching the other major professional sports teams in Pittsburgh, the Steelers and the Penguins.The Pittsburgh Pirates are an American professional baseball team based in Pittsburgh. The Pirates compete in Major League Baseball (MLB) as a member club of the National League (NL) Central Division. Founded as part of the American Association in 1881 under the name Pittsburgh Alleghenys, the club joined the National League in 1887 and was a member of the National League East from 1969 through 1993. The Pirates have won five World - "STEELERS IN THE POSTSEASON (36-30)\nYear Record Game Date Opponent Attendance\ \ Steelers Opponent Result\n2015 10-6 AFC Wild Card Game 01/09/2016 at Cincinnati\ \ 63,257 18 16 W\nAFC Divisional Playoff 01/17/2016 at Denver 79,956 16 23 L\n\ 2016# 11-5 AFC Wild Card Game 01/08/2017 Miami 66,726 30 12 W\nAFC Divisional\ \ Playoff 01/15/2017 at Kansas City 75,678 18 16 W\nAFC Championship Game 01/22/2017\ \ at New England 66,829 36 17 L\n2017# 13-3 AFC Divisional Playoff 01/14/2018\ \ Jacksonville 64,524 42 45 L\n2020# 12-4 AFC Wild Card Game 01/03/2021 Cleveland\ \ - 37 48 L\n2021 9-7-1 AFC Wild Card Game 01/16/2022 at Kansas City 73,253 21\ \ 42 L\n2023 10-7 AFC Wild Card Game 01/15/202 4 at Buffalo 70,040 17 31 L\n*AFC\ \ Central Champion\n#AFC North Champion\n+AFC ChampionSTEELERS IN THE POSTSEASON\n\ \ 2023 PITTSBURGH STEELERS\n 421\n STEELERS IN THE POSTSEASON" - source_sentence: 'Question: What is the Local Services Tax and how is it collected?' sentences: - the 1916 Centennial of Pittsburgh's 1816 incorporation as a City. At the March 1916 dedication ceremony, Mayor Joseph Armstrong placed a time capsule into the still under construction building. Two and a half years later in December 1917, he would become the first Mayor to call the City-County Building a second home. The missing time capsule has yet to be discovered. - 'The first City Hall at Market Square. The second City Hall on Smithfield Street. Mayor David Lawrence strikes the first blow for the demolition of the second City Hall.' - "EXEMPT P ERSON – a person who files an exemption certificate with his employer\ \ affirming \nthat he reasonably expects to receive earned income and net profits\ \ from all sources within the \nCity of less than twelve thousand dollars ($12,000)\ \ in the calendar year for wh ich the exemption \ncertificate is filed. See Section\ \ 301(h) below, and Section 2 of the Local Tax Enabling Act, 53 P.S. § \n6924.301.1,\ \ for other exemptions. \nINCOME – all earned income and net profits from whatever\ \ source derived, including but not \nlimited to salaries, wages, bonuses, commissions\ \ and income from self -employment earned in \nPittsburgh. \nLOCAL SERVICES TAX\ \ (LST) – a tax on individuals for the privilege of engaging in an \noccupation.\ \ The Local Services Tax may be levied, assessed and collected by the political\ \ \nsubdivision of the taxpayer’s primary place of employment. \nOCCUPATION –\ \ any livelihood, job, trade, profession, business or enterprise of any kind for" - source_sentence: '"What is the nature of the incident being investigated by Zone Five Officers in Homewood on April 23, 2024?"' sentences: - 'Event Name: Saturday Night Improv @ BGC! Date: Saturdays, 7:30-9:30 p.m. Location: BGC Community Activity Center: 113 N. Pacific Ave., Pittsburgh | Garfield Price Information: GET TICKETS: 10 Categories: Comedy, Theater Description: It''s time to Love, Laugh and Enjoy. Join us at the BGC Activity Center Saturday evenings for an evening of improv with performances by Narsh and Penny Pressed! Shows start promptly at 7:30 PM so don''t be late! 412-441-6950 Event Name: Swing City Date: Saturdays, 8 p.m. Location: Wightman School: 5604 Solway, Pittsburgh | Squirrel Hill Categories: Other Stuff Description: Learn & practice swing dancing skills w/ the Jim Adler Band. 412-759-1569' - 'Police Investigate Stabbing Incident in Beltzhoover - 04.23.2024 Zone Five Officers Investigate Homewood Shooting Incident - 04.23.2024 Violent Crimes Division VCU Detectives Make Firearms Arrest in Spring Garden - 04.19.2024 UPDATE: Detectives Seek Assistance in Search for Missing 12-Year-Old Girl - 04.19.2024 UPDATE: Police Investigate Aggravated Assault on Riverwalk in Point State Park - 04.19.2024 Police Investigate Homicide Inside Larimer Residence - 04.19.2024 UPDATE: Police Seek the Public''s Help in Locating Missing Juvenile Male - 04.19.2024 UPDATE: Pittsburgh Police Ask for Public''s Help to Find Missing Woman - 04.15.2024 Police Investigate Shooting Incident in Allegheny Center - 04.13.2024 UPDATE: Pittsburgh Public Safety Responds to Barge Emergency on Ohio River - 04.12.2024 Police Make Ethnic Intimidation and Criminal Mischief Arrest in Squirrel Hill - 04.12.2024 UPDATE: Police Seek the Public''s Assistance in Locating Missing Boy - 04.11.2024' - "24\n \n$ (Millions)Select Major Expenditures, 2018-2022\n2018 2019 2020\n2021\ \ 2022Health Insurance\nWorkers' CompensationPension and OPEBDebt Service050,000,000100,000,000150,000,000\n\ Health Insurance\nThese expenditures are categorized within the Personnel – Employment\ \ Benefits subclass. Prior to 2016 these \nexpenditures were budgeted centrally\ \ in the Department of Human Resources and Civil Service. Except for retiree \n\ health insurance, these expenditures are budgeted across all divisions based on\ \ staffing levels and plan \nelections.\n Health Insurance\n52101 Health Insurance\n\ 52111 Other Insurance and Benefits\n52121 Retiree Health Insurance\nWorkers’\ \ Compensation\nThese expenditures are categorized within the Personnel – Employment\ \ Benefits subclass. Most medical, \nindemnity, and fees are budgeted across divisions\ \ with outstanding claims. Legal and settlement expenses \nremain budgeted in\ \ the Department of Human Resources and Civil Service with accounts organized\ \ as follows:" - source_sentence: 'Answer: The passage does not provide information about the longest reception for the Steelers in the Wild Card Game against Cincinnati.' sentences: - '09/08 Lions RESERVE/LEAGUE SUSP. T 27-27 + 09/15 at Ravens RESERVE/LEAGUE SUSP. L 17-23 09/22 Panthers RESERVE/LEAGUE SUSP. L 20-38 09/29 Seahawks RESERVE/LEAGUE SUSP. L 10-27 10/06 at Bengals RESERVE/LEAGUE SUSP. W 26-23 10/13 Falcons RESERVE/LEAGUE SUSP. W 34-33 10/20 at Giants S 7701.0 13.0 0 0 1 0 0 0 0 0 1 0 0 0 0000 000 W 27-21 10/27 at Saints S 6510.0 0.0 0 0 0 1 0 0 0 1 0 0 0 0 0000 000 L 9-31 10/31 49ers S 3210.0 0.0 0 0 0 0 0 0 0 0 0 0 0 0 0000 000 L 25-28 11/10 at Buccaneers S 3300.0 0.0 0 0 0 0 0 0 0 0 0 0 0 0 0000 000 L 27-30 11/17 at 49ers S 4400.0 0.0 0 0 0 0 0 0 0 1 0 0 0 0 0000 000 L 26-36 12/01 Rams S 8530.0 0.0 1 10 0 0 0 0 0 0 0 0 0 0 0000 000 L 7-34 12/08 Steelers S 5410.0 0.0 0 0 0 0 0 0 0 0 0 0 0 0 0000 000 L 17-23 12/15 Browns S 7700.0 0.0 0 0 0 1 0 0 0 3 0 0 0 0 0000 000 W 38-24 12/22 at Seahawks S 3300.0 0.0 1 18 0 0 0 0 0 0 0 0 0 0 0000 000 W 27-13 12/29 at Rams S 7610.0 0.0 1 1 0 0 0 0 0 2 0 0 0 0 0000 000 L 24-31' - "Program \n• Clinical field education to emergency medicine physician residents\ \ in the University of Pittsburgh \nEmergency Medicine Residency program \n \n\ 2023 Accomplishments\n \n• Financial Accomplishments:\n◦ Income from transports\ \ increased by $1.8M from same time period last year\n◦ Bureau slated to bring\ \ in an additional $5M in revenue for 2023\n• Personnel Accomplishments:\n◦ 6\ \ new River Rescue Divers went through intensive training and all successfully\ \ completed the \nclass\n◦ Increase in promotions to upper administration\n• Employee\ \ Safety Initiatives: \n◦ Implementation of Cordico App for employee wellness\n\ ◦ Access control security system installed in all EMS facilities \n• Equipment\ \ Initiatives:\n◦ Bureau was approved to receive state of the art mannequins to\ \ simulate real life patients during \nemergencies\n◦ Billing company to purchase\ \ equipment/medication dispensary machines to be located in 5 areas" - "Pittsburgh 31\nCincinnati 17\nCINCINNATI — Pittsburgh scored 24 unanswered points\ \ to turn a 17-7 deficit into a \n31-17 victory over Cincinnati in the AFC Wild\ \ Card Game at Paul Brown Stadium. \nThe Pittsburgh offense compiled 346 total\ \ yards led by QB Ben Roethlisberger, who \ntossed three touchdowns and finished\ \ with a QB rating of 148.7. RB Jerome Bettis ran for 52 \nyards on 10 carries\ \ (5.2 avg.) and one touchdown. WR Cedrick Wilson caught three passes \nfor 104\ \ yards (34.7 avg.), with one touchdown. \nThe Steelers defense recorded four\ \ sacks and two interceptions while holding the \nBengals to just 84 yards rushing.\ \ \nCincinnati was dealt an early blow when starting QB Carson Palmer suffered\ \ a torn \nACL on the first offensive play of the game. The Bengals jumped out\ \ to a 10-0 lead with a \n23-yard field goal by K Shayne Graham and a 20-yard\ \ touchdown run by RB Rudi Johnson.\nPittsburgh got on the board when RB Willie\ \ Parker took a screen pass 19 yards for a" - source_sentence: '"What cultural celebration will be honored at the 2024 Greater Pittsburgh Lunar New Year Gala, and what is the significance of this event in the community?"' sentences: - 'This page informs City of Pittsburgh residents about the city''s Snow Angels program. This page is also where volunteers can sign up, and recipients can submit a request. City Collection Equity Audit The City of Pittsburgh is conducting an audit to identify inequity and bias in the City’s collection of public art and memorials. Davis Avenue Bridge Design and construction for the new Davis Avenue Bridge between Brighton Heights and Riverview Park. South Side Park Public Art A new public art project is being planned in South Side Park. This is being done in coordination with the park’s Phase 1 renovations and funded by the Percent For Art. Projects that are no longer accepting feedback, but are now in the construction or development phase. PHAD Projects Current Projects – find out about ongoing projects underway throughout the city and learn how to apply for new projects each year. Emerald View Phase I Trails & Trailheads' - of Pittsburgh and greater southwestern Pennsylvania. Justin is employed within the Cultural Resources practice of Michael Baker International. He is Director Emeritus of Preservation Pittsburgh and a past president of the East Liberty Valley Historical Society. Justin is a graduate of the University of Pittsburgh (B.A. Architectural Studies, 2008) and Columbia University (M.S. Historic Preservation, 2010).Todd Wilson, MBA, PE, is an award-winning transportation engineer, named one of Pittsburgh Business Times’ 20 Engineers to Know in 2022. He has co-authored two books on Pittsburgh’s bridges,Images of America Pittsburgh’s Bridges and Engineering Pittsburgh a History of Roads, Rails, Canals, Bridges, and More.An engineering graduate of Carnegie Mellon, Todd has extensive knowledge on bridges, having photographed them in all 50 states and 25 countries, and he has presented at many conferences. Check out his Pittsburgh bridge photography on Instagram @pghbridges.TOUR STARTS/ENDS:Gateway - 'Event Name: 2024 Greater Pittsburgh Lunar New Year Gala Categories: Arts + Culture, Community, Holidays, Nightlife Dates: Feb 3, 2024 - Feb 3, 2024 | 4:00 pm - 9:00 pm Location: PNC Theater, 350 Forbes Avenue, Pittsburgh, PA 15222' model-index: - name: MPNet base trained on synthetic Pittsburgh data results: - task: type: information-retrieval name: Information Retrieval dataset: name: pittsburgh type: pittsburgh metrics: - type: cosine_accuracy@1 value: 0.7375145180023229 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.9037940379403794 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.9368950832365467 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.9628339140534262 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.7375145180023229 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.30126467931345985 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.1873790166473093 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09628339140534262 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.7375145180023229 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.9037940379403794 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.9368950832365467 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.9628339140534262 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.8590408201907759 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.824762258110111 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.8263189855192845 name: Cosine Map@100 - type: dot_accuracy@1 value: 0.7375145180023229 name: Dot Accuracy@1 - type: dot_accuracy@3 value: 0.9037940379403794 name: Dot Accuracy@3 - type: dot_accuracy@5 value: 0.9368950832365467 name: Dot Accuracy@5 - type: dot_accuracy@10 value: 0.9628339140534262 name: Dot Accuracy@10 - type: dot_precision@1 value: 0.7375145180023229 name: Dot Precision@1 - type: dot_precision@3 value: 0.30126467931345985 name: Dot Precision@3 - type: dot_precision@5 value: 0.1873790166473093 name: Dot Precision@5 - type: dot_precision@10 value: 0.09628339140534262 name: Dot Precision@10 - type: dot_recall@1 value: 0.7375145180023229 name: Dot Recall@1 - type: dot_recall@3 value: 0.9037940379403794 name: Dot Recall@3 - type: dot_recall@5 value: 0.9368950832365467 name: Dot Recall@5 - type: dot_recall@10 value: 0.9628339140534262 name: Dot Recall@10 - type: dot_ndcg@10 value: 0.8590408201907759 name: Dot Ndcg@10 - type: dot_mrr@10 value: 0.824762258110111 name: Dot Mrr@10 - type: dot_map@100 value: 0.8263189855192845 name: Dot Map@100 --- # MPNet base trained on synthetic Pittsburgh data This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [sentence-transformers/all-mpnet-base-v2](https://huggingface.co/sentence-transformers/all-mpnet-base-v2). 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:** [sentence-transformers/all-mpnet-base-v2](https://huggingface.co/sentence-transformers/all-mpnet-base-v2) - **Maximum Sequence Length:** 384 tokens - **Output Dimensionality:** 768 tokens - **Similarity Function:** Cosine Similarity - **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': 384, 'do_lower_case': False}) with Transformer model: MPNetModel (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): 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("lizchu414/mpnet-base-all-pittsburgh-squad") # Run inference sentences = [ '"What cultural celebration will be honored at the 2024 Greater Pittsburgh Lunar New Year Gala, and what is the significance of this event in the community?"', 'Event Name: 2024 Greater Pittsburgh Lunar New Year Gala\nCategories: Arts + Culture, Community, Holidays, Nightlife\nDates: Feb 3, 2024 - Feb 3, 2024 | 4:00 pm - 9:00 pm\nLocation: PNC Theater, 350 Forbes Avenue, Pittsburgh, PA 15222', "This page informs City of Pittsburgh residents about the city's Snow Angels program. This page is also where volunteers can sign up, and recipients can submit a request.\nCity Collection Equity Audit\nThe City of Pittsburgh is conducting an audit to identify inequity and bias in the City’s collection of public art and memorials.\nDavis Avenue Bridge\nDesign and construction for the new Davis Avenue Bridge between Brighton Heights and Riverview Park.\nSouth Side Park Public Art\nA new public art project is being planned in South Side Park. This is being done in coordination with the park’s Phase 1 renovations and funded by the Percent For Art.\nProjects that are no longer accepting feedback, but are now in the construction or development phase.\nPHAD Projects\nCurrent Projects – find out about ongoing projects underway throughout the city and learn how to apply for new projects each year.\nEmerald View Phase I Trails & Trailheads", ] 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 #### Information Retrieval * Dataset: `pittsburgh` * Evaluated with [InformationRetrievalEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) | Metric | Value | |:--------------------|:-----------| | cosine_accuracy@1 | 0.7375 | | cosine_accuracy@3 | 0.9038 | | cosine_accuracy@5 | 0.9369 | | cosine_accuracy@10 | 0.9628 | | cosine_precision@1 | 0.7375 | | cosine_precision@3 | 0.3013 | | cosine_precision@5 | 0.1874 | | cosine_precision@10 | 0.0963 | | cosine_recall@1 | 0.7375 | | cosine_recall@3 | 0.9038 | | cosine_recall@5 | 0.9369 | | cosine_recall@10 | 0.9628 | | cosine_ndcg@10 | 0.859 | | cosine_mrr@10 | 0.8248 | | cosine_map@100 | 0.8263 | | dot_accuracy@1 | 0.7375 | | dot_accuracy@3 | 0.9038 | | dot_accuracy@5 | 0.9369 | | dot_accuracy@10 | 0.9628 | | dot_precision@1 | 0.7375 | | dot_precision@3 | 0.3013 | | dot_precision@5 | 0.1874 | | dot_precision@10 | 0.0963 | | dot_recall@1 | 0.7375 | | dot_recall@3 | 0.9038 | | dot_recall@5 | 0.9369 | | dot_recall@10 | 0.9628 | | dot_ndcg@10 | 0.859 | | dot_mrr@10 | 0.8248 | | **dot_map@100** | **0.8263** | ## Training Details ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: steps - `per_device_eval_batch_size`: 2 - `eval_accumulation_steps`: 1 - `learning_rate`: 2e-05 - `warmup_ratio`: 0.1 - `fp16`: True - `batch_sampler`: no_duplicates #### All Hyperparameters
Click to expand - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: steps - `prediction_loss_only`: True - `per_device_train_batch_size`: 8 - `per_device_eval_batch_size`: 2 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 1 - `eval_accumulation_steps`: 1 - `torch_empty_cache_steps`: None - `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`: 3 - `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`: 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 - `eval_on_start`: False - `use_liger_kernel`: False - `eval_use_gather_object`: False - `batch_sampler`: no_duplicates - `multi_dataset_batch_sampler`: proportional
### Training Logs | Epoch | Step | Training Loss | Validation Loss | pittsburgh_dot_map@100 | |:-----:|:----:|:-------------:|:---------------:|:----------------------:| | 0 | 0 | - | - | 0.5984 | | 0.8 | 100 | 0.587 | 0.1954 | 0.7780 | | 1.592 | 200 | 0.1828 | 0.1805 | 0.8020 | | 2.384 | 300 | 0.2224 | 0.1605 | 0.8263 | ### Framework Versions - Python: 3.12.7 - Sentence Transformers: 3.2.0 - Transformers: 4.45.2 - PyTorch: 2.2.2+cu121 - Accelerate: 1.0.1 - Datasets: 3.0.1 - Tokenizers: 0.20.1 ## 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", } ``` #### MultipleNegativesRankingLoss ```bibtex @misc{henderson2017efficient, title={Efficient Natural Language Response Suggestion for Smart Reply}, 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}, year={2017}, eprint={1705.00652}, archivePrefix={arXiv}, primaryClass={cs.CL} } ```