suhwan3/mpnet-step2

SentenceTransformer based on suhwan3/mpnet_step1

This is a sentence-transformers model finetuned from suhwan3/mpnet_step1 on the stage2-dataset 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: suhwan3/mpnet_step1
• Maximum Sequence Length: 384 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • stage2-dataset

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

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:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("sentence_transformers_model_id")
# Run inference
sentences = [
    "The Global X S&P 500 Risk Managed Income ETF seeks to track the Cboe S&P 500 Risk Managed Income Index by investing at least 80% of its assets in index securities. The index's strategy involves holding the underlying stocks of the S&P 500 Index while applying an options collar, specifically selling at-the-money covered call options and buying monthly 5% out-of-the-money put options corresponding to the portfolio's value. This approach aims to generate income, ideally resulting in a net credit from the options premiums, and provide risk management, though selling at-the-money calls inherently caps the fund's potential for upside participation.",
    'Walgreens Boots Alliance, Inc. operates as a pharmacy-led health and beauty retail company. It operates through two segments, the United States and International. The United States segment sells prescription drugs and an assortment of retail products, including health, wellness, beauty, personal care, consumable, and general merchandise products through its retail drugstores. It also provides central specialty pharmacy services and mail services. As of August 31, 2021, this segment operated 8,965 retail stores under the Walgreens and Duane Reade brands in the United States; and five specialty pharmacies. The International segment sells prescription drugs; and health and wellness, beauty, personal care, and other consumer products through its pharmacy-led health and beauty retail stores and optical practices, as well as through boots.com and an integrated mobile application. It also engages in pharmaceutical wholesaling and distribution business in Germany. As of August 31, 2021, this segment operated 4,031 retail stores under the Boots, Benavides, and Ahumada in the United Kingdom, Thailand, Norway, the Republic of Ireland, the Netherlands, Mexico, and Chile; and 548 optical practices, including 160 on a franchise basis. Walgreens Boots Alliance, Inc. was founded in 1901 and is based in Deerfield, Illinois.',
    'Liberty Broadband Corporation engages in the communications businesses. It operates through GCI Holdings and Charter segments. The GCI Holdings segment provides a range of wireless, data, video, voice, and managed services to residential customers, businesses, governmental entities, and educational and medical institutions primarily in Alaska under the GCI brand. The Charter segment offers subscription-based video services comprising video on demand, high-definition television, and digital video recorder service; local and long-distance calling, voicemail, call waiting, caller ID, call forwarding, and other voice services, as well as international calling services; and Spectrum TV. It also provides internet services, including an in-home Wi-Fi product that provides customers with high-performance wireless routers and managed Wi-Fi services; advanced community Wi-Fi; mobile internet; and a security suite that offers protection against computer viruses and spyware. In addition, this segment offers internet access, data networking, fiber connectivity to cellular towers and office buildings, video entertainment, and business telephone services; advertising services on cable television networks and digital outlets; and operates regional sports and news networks. Liberty Broadband Corporation was incorporated in 2014 and is based in Englewood, Colorado.',
]
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]

Training Details

Training Dataset

stage2-dataset

• Dataset: stage2-dataset at cd393c2
• Size: 128,997 training samples
• Columns: anchor and positive
• Approximate statistics based on the first 1000 samples:

  	anchor	positive
  type	string	string
  details	min: 101 tokens mean: 143.15 tokens max: 186 tokens	min: 35 tokens mean: 238.69 tokens max: 384 tokens

• Samples:

  anchor	positive
  The Invesco Financial Preferred ETF (PGF) seeks to track the ICE Exchange-Listed Fixed Rate Financial Preferred Securities Index, primarily by investing at least 90% of its total assets in the securities comprising the index. The underlying index is market capitalization weighted and designed to track the performance of exchange-listed, fixed rate, U.S. dollar denominated preferred securities, including functionally equivalent instruments, issued by U.S. financial companies. PGF provides a concentrated portfolio exclusively focused on financial-sector preferred securities and is considered non-diversified, holding both investment- and non-investment-grade securities within this focus.	JPMorgan Chase & Co. operates as a financial services company worldwide. It operates through four segments: Consumer & Community Banking (CCB), Corporate & Investment Bank (CIB), Commercial Banking (CB), and Asset & Wealth Management (AWM). The CCB segment offers s deposit, investment and lending products, payments, and services to consumers; lending, deposit, and cash management and payment solutions to small businesses; mortgage origination and servicing activities; residential mortgages and home equity loans; and credit card, auto loan, and leasing services. The CIB segment provides investment banking products and services, including corporate strategy and structure advisory, and equity and debt markets capital-raising services, as well as loan origination and syndication; payments and cross-border financing; and cash and derivative instruments, risk management solutions, prime brokerage, and research. This segment also offers securities services, including custody, fund accounting ...
  The Invesco Financial Preferred ETF (PGF) seeks to track the ICE Exchange-Listed Fixed Rate Financial Preferred Securities Index, primarily by investing at least 90% of its total assets in the securities comprising the index. The underlying index is market capitalization weighted and designed to track the performance of exchange-listed, fixed rate, U.S. dollar denominated preferred securities, including functionally equivalent instruments, issued by U.S. financial companies. PGF provides a concentrated portfolio exclusively focused on financial-sector preferred securities and is considered non-diversified, holding both investment- and non-investment-grade securities within this focus.	JPMorgan Chase & Co. operates as a financial services company worldwide. It operates through four segments: Consumer & Community Banking (CCB), Corporate & Investment Bank (CIB), Commercial Banking (CB), and Asset & Wealth Management (AWM). The CCB segment offers s deposit, investment and lending products, payments, and services to consumers; lending, deposit, and cash management and payment solutions to small businesses; mortgage origination and servicing activities; residential mortgages and home equity loans; and credit card, auto loan, and leasing services. The CIB segment provides investment banking products and services, including corporate strategy and structure advisory, and equity and debt markets capital-raising services, as well as loan origination and syndication; payments and cross-border financing; and cash and derivative instruments, risk management solutions, prime brokerage, and research. This segment also offers securities services, including custody, fund accounting ...
  The Invesco Financial Preferred ETF (PGF) seeks to track the ICE Exchange-Listed Fixed Rate Financial Preferred Securities Index, primarily by investing at least 90% of its total assets in the securities comprising the index. The underlying index is market capitalization weighted and designed to track the performance of exchange-listed, fixed rate, U.S. dollar denominated preferred securities, including functionally equivalent instruments, issued by U.S. financial companies. PGF provides a concentrated portfolio exclusively focused on financial-sector preferred securities and is considered non-diversified, holding both investment- and non-investment-grade securities within this focus.	The Allstate Corporation, together with its subsidiaries, provides property and casualty, and other insurance products in the United States and Canada. The company operates through Allstate Protection; Protection Services; Allstate Health and Benefits; and Run-off Property-Liability segments. The Allstate Protection segment offers private passenger auto and homeowners insurance; other personal lines products; and commercial lines products under the Allstate and Encompass brand names. The Protection Services segment provides consumer product protection plans and related technical support for mobile phones, consumer electronics, furniture, and appliances; finance and insurance products, including vehicle service contracts, guaranteed asset protection waivers, road hazard tire and wheel, and paint and fabric protection; towing, jump-start, lockout, fuel delivery, and tire change services; device and mobile data collection services; data and analytic solutions using automotive telematics i...

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim"
  }
  

Evaluation Dataset

stage2-dataset

• Dataset: stage2-dataset at cd393c2
• Size: 16,944 evaluation samples
• Columns: anchor and positive
• Approximate statistics based on the first 1000 samples:

  	anchor	positive
  type	string	string
  details	min: 135 tokens mean: 149.21 tokens max: 214 tokens	min: 42 tokens mean: 252.75 tokens max: 384 tokens

• Samples:

  anchor	positive
  The Global X S&P 500 Risk Managed Income ETF seeks to track the Cboe S&P 500 Risk Managed Income Index by investing at least 80% of its assets in index securities. The index's strategy involves holding the underlying stocks of the S&P 500 Index while applying an options collar, specifically selling at-the-money covered call options and buying monthly 5% out-of-the-money put options corresponding to the portfolio's value. This approach aims to generate income, ideally resulting in a net credit from the options premiums, and provide risk management, though selling at-the-money calls inherently caps the fund's potential for upside participation.	Apple Inc. designs, manufactures, and markets smartphones, personal computers, tablets, wearables, and accessories worldwide. The company offers iPhone, a line of smartphones; Mac, a line of personal computers; iPad, a line of multi-purpose tablets; and wearables, home, and accessories comprising AirPods, Apple TV, Apple Watch, Beats products, and HomePod. It also provides AppleCare support and cloud services; and operates various platforms, including the App Store that allow customers to discover and download applications and digital content, such as books, music, video, games, and podcasts, as well as advertising services include third-party licensing arrangements and its own advertising platforms. In addition, the company offers various subscription-based services, such as Apple Arcade, a game subscription service; Apple Fitness+, a personalized fitness service; Apple Music, which offers users a curated listening experience with on-demand radio stations; Apple News+, a subscription ...
  The Global X S&P 500 Risk Managed Income ETF seeks to track the Cboe S&P 500 Risk Managed Income Index by investing at least 80% of its assets in index securities. The index's strategy involves holding the underlying stocks of the S&P 500 Index while applying an options collar, specifically selling at-the-money covered call options and buying monthly 5% out-of-the-money put options corresponding to the portfolio's value. This approach aims to generate income, ideally resulting in a net credit from the options premiums, and provide risk management, though selling at-the-money calls inherently caps the fund's potential for upside participation.	Microsoft Corporation develops, licenses, and supports software, services, devices, and solutions worldwide. The company operates in three segments: Productivity and Business Processes, Intelligent Cloud, and More Personal Computing. The Productivity and Business Processes segment offers Office, Exchange, SharePoint, Microsoft Teams, Office 365 Security and Compliance, Microsoft Viva, and Skype for Business; Skype, Outlook.com, OneDrive, and LinkedIn; and Dynamics 365, a set of cloud-based and on-premises business solutions for organizations and enterprise divisions. The Intelligent Cloud segment licenses SQL, Windows Servers, Visual Studio, System Center, and related Client Access Licenses; GitHub that provides a collaboration platform and code hosting service for developers; Nuance provides healthcare and enterprise AI solutions; and Azure, a cloud platform. It also offers enterprise support, Microsoft consulting, and nuance professional services to assist customers in developing, de...
  The Global X S&P 500 Risk Managed Income ETF seeks to track the Cboe S&P 500 Risk Managed Income Index by investing at least 80% of its assets in index securities. The index's strategy involves holding the underlying stocks of the S&P 500 Index while applying an options collar, specifically selling at-the-money covered call options and buying monthly 5% out-of-the-money put options corresponding to the portfolio's value. This approach aims to generate income, ideally resulting in a net credit from the options premiums, and provide risk management, though selling at-the-money calls inherently caps the fund's potential for upside participation.	NVIDIA Corporation provides graphics, and compute and networking solutions in the United States, Taiwan, China, and internationally. The company's Graphics segment offers GeForce GPUs for gaming and PCs, the GeForce NOW game streaming service and related infrastructure, and solutions for gaming platforms; Quadro/NVIDIA RTX GPUs for enterprise workstation graphics; vGPU software for cloud-based visual and virtual computing; automotive platforms for infotainment systems; and Omniverse software for building 3D designs and virtual worlds. Its Compute & Networking segment provides Data Center platforms and systems for AI, HPC, and accelerated computing; Mellanox networking and interconnect solutions; automotive AI Cockpit, autonomous driving development agreements, and autonomous vehicle solutions; cryptocurrency mining processors; Jetson for robotics and other embedded platforms; and NVIDIA AI Enterprise and other software. The company's products are used in gaming, professional visualizat...

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 64
• per_device_eval_batch_size: 32
• learning_rate: 3e-05
• num_train_epochs: 1
• warmup_ratio: 0.1
• bf16: True
• dataloader_drop_last: True
• load_best_model_at_end: 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: 64
• per_device_eval_batch_size: 32
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 1
• eval_accumulation_steps: None
• torch_empty_cache_steps: None
• learning_rate: 3e-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: True
• 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: True
• 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: True
• 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}
• tp_size: 0
• 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: None
• hub_always_push: False
• gradient_checkpointing: False
• 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
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss
0.0050	10	4.6656	-
0.0099	20	4.4733	-
0.0149	30	4.0093	-
0.0199	40	3.9259	-
0.0248	50	3.8315	-
0.0298	60	3.673	-
0.0347	70	3.5076	-
0.0397	80	3.4416	-
0.0447	90	3.4362	-
0.0496	100	3.3934	-
0.0546	110	3.3343	-
0.0596	120	3.3018	-
0.0645	130	3.2882	-
0.0695	140	3.3027	-
0.0744	150	3.2177	-
0.0794	160	3.2708	-
0.0844	170	3.2645	-
0.0893	180	3.1939	-
0.0943	190	3.0575	-
0.0993	200	3.0799	-
0.1042	210	3.0824	-
0.1092	220	3.0693	-
0.1141	230	3.1014	-
0.1191	240	3.0458	-
0.1241	250	3.04	-
0.1290	260	3.0311	-
0.1340	270	2.9778	-
0.1390	280	3.0701	-
0.1439	290	2.9039	-
0.1489	300	3.0449	2.5685
0.1538	310	2.8896	-
0.1588	320	3.0527	-
0.1638	330	3.0153	-
0.1687	340	2.869	-
0.1737	350	2.9678	-
0.1787	360	2.9756	-
0.1836	370	2.9348	-
0.1886	380	2.9967	-
0.1935	390	2.8953	-
0.1985	400	2.9546	-
0.2035	410	2.9919	-
0.2084	420	2.8487	-
0.2134	430	2.7609	-
0.2184	440	2.9126	-
0.2233	450	2.8991	-
0.2283	460	2.9272	-
0.2333	470	2.9084	-
0.2382	480	2.7963	-
0.2432	490	2.822	-
0.2481	500	2.9376	-
0.2531	510	2.8969	-
0.2581	520	2.7745	-
0.2630	530	2.8103	-
0.2680	540	2.8189	-
0.2730	550	2.8322	-
0.2779	560	2.7627	-
0.2829	570	2.7796	-
0.2878	580	2.8515	-
0.2928	590	2.8758	-
0.2978	600	2.7963	2.4142
0.3027	610	2.8259	-
0.3077	620	2.829	-
0.3127	630	2.7699	-
0.3176	640	2.7311	-
0.3226	650	2.735	-
0.3275	660	2.7306	-
0.3325	670	2.7467	-
0.3375	680	2.7494	-
0.3424	690	2.7386	-
0.3474	700	2.8513	-
0.3524	710	2.673	-
0.3573	720	2.8101	-
0.3623	730	2.7527	-
0.3672	740	2.7213	-
0.3722	750	2.753	-
0.3772	760	2.8034	-
0.3821	770	2.8288	-
0.3871	780	2.613	-
0.3921	790	2.7315	-
0.3970	800	2.8077	-
0.4020	810	2.7442	-
0.4069	820	2.7351	-
0.4119	830	2.7643	-
0.4169	840	2.8984	-
0.4218	850	2.7377	-
0.4268	860	2.7021	-
0.4318	870	2.6756	-
0.4367	880	2.7852	-
0.4417	890	2.7531	-
0.4467	900	2.6636	2.3456
0.4516	910	2.7089	-
0.4566	920	2.8029	-
0.4615	930	2.721	-
0.4665	940	2.5606	-
0.4715	950	2.6397	-
0.4764	960	2.6563	-
0.4814	970	2.7163	-
0.4864	980	2.6225	-
0.4913	990	2.645	-
0.4963	1000	2.6576	-
0.5012	1010	2.7019	-
0.5062	1020	2.7195	-
0.5112	1030	2.7242	-
0.5161	1040	2.6729	-
0.5211	1050	2.7637	-
0.5261	1060	2.677	-
0.5310	1070	2.7018	-
0.5360	1080	2.6469	-
0.5409	1090	2.7186	-
0.5459	1100	2.6728	-
0.5509	1110	2.6694	-
0.5558	1120	2.7839	-
0.5608	1130	2.5834	-
0.5658	1140	2.6905	-
0.5707	1150	2.7223	-
0.5757	1160	2.7235	-
0.5806	1170	2.636	-
0.5856	1180	2.6314	-
0.5906	1190	2.5941	-
0.5955	1200	2.7827	2.2911
0.6005	1210	2.6104	-
0.6055	1220	2.6148	-
0.6104	1230	2.6355	-
0.6154	1240	2.6269	-
0.6203	1250	2.6003	-
0.6253	1260	2.6256	-
0.6303	1270	2.6326	-
0.6352	1280	2.681	-
0.6402	1290	2.5776	-
0.6452	1300	2.7528	-
0.6501	1310	2.6076	-
0.6551	1320	2.5784	-
0.6600	1330	2.6064	-
0.6650	1340	2.5757	-
0.6700	1350	2.5851	-
0.6749	1360	2.6007	-
0.6799	1370	2.5674	-
0.6849	1380	2.6984	-
0.6898	1390	2.6202	-
0.6948	1400	2.6729	-
0.6998	1410	2.6683	-
0.7047	1420	2.6355	-
0.7097	1430	2.6033	-
0.7146	1440	2.6834	-
0.7196	1450	2.6597	-
0.7246	1460	2.6298	-
0.7295	1470	2.6232	-
0.7345	1480	2.5672	-
0.7395	1490	2.5139	-
0.7444	1500	2.6248	2.3090
0.7494	1510	2.6417	-
0.7543	1520	2.6197	-
0.7593	1530	2.6911	-
0.7643	1540	2.5542	-
0.7692	1550	2.6584	-
0.7742	1560	2.6182	-
0.7792	1570	2.6301	-
0.7841	1580	2.5629	-
0.7891	1590	2.5965	-
0.7940	1600	2.5722	-
0.7990	1610	2.5835	-
0.8040	1620	2.5901	-
0.8089	1630	2.6055	-
0.8139	1640	2.6019	-
0.8189	1650	2.6421	-
0.8238	1660	2.6049	-
0.8288	1670	2.5351	-
0.8337	1680	2.6158	-
0.8387	1690	2.5994	-
0.8437	1700	2.5816	-
0.8486	1710	2.5848	-
0.8536	1720	2.6138	-
0.8586	1730	2.5811	-
0.8635	1740	2.5933	-
0.8685	1750	2.5869	-
0.8734	1760	2.5464	-
0.8784	1770	2.6842	-
0.8834	1780	2.6312	-
0.8883	1790	2.5621	-
0.8933	1800	2.6103	2.2858

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 4.1.0
• Transformers: 4.51.3
• PyTorch: 2.1.0+cu118
• Accelerate: 1.6.0
• Datasets: 3.5.0
• Tokenizers: 0.21.1

Citation

BibTeX

Sentence Transformers

@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

@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}
}