metadata
language:
- en
license: apache-2.0
tags:
- sentence-transformers
- sentence-similarity
- feature-extraction
- dense
- generated_from_trainer
- dataset_size:90000
- loss:QuantizationAwareLoss
- loss:MultipleNegativesRankingLoss
base_model: microsoft/mpnet-base
widget:
- source_sentence: what is the difference between trojan virus and worm?
sentences:
- >-
Worms spread from computer to computer, but unlike a virus, it has the
capability to travel without any help from a person. ... A Trojan horse
is not a virus. It is a destructive program that looks as a genuine
application. Unlike viruses, Trojan horses do not replicate themselves
but they can be just as destructive.
- >-
You're usually no longer infectious 24 hours after starting a course of
antibiotics, but this time period can sometimes vary. For example, the
antibiotics may take longer to work if your body takes longer to absorb
them, or if you're taking other medicine that interacts with the
antibiotics.
- >-
Eating salt raises the amount of sodium in your bloodstream and wrecks
the delicate balance, reducing the ability of your kidneys to remove the
water. The result is a higher blood pressure due to the extra fluid and
extra strain on the delicate blood vessels leading to the kidneys.
- source_sentence: which are the neighbouring countries of pakistan?
sentences:
- >-
Pakistan is bordered by India on the east, the Arabian Sea on the south,
Iran on the southwest, and Afghanistan on the west and north; in the
northeast is the disputed territory (with India) of Kashmir, of which
the part occupied by Pakistan borders on China.
- >-
Age is a big factor when it comes to how much sleep a dog needs. Just as
human babies need a lot of sleep, the AKC notes your puppy needs 15-20
hours of sleep a day to help his central nervous system, immune system
and muscles develop properly.
- >-
Step 1: Connect your iPhone to your computer using n USB cable through
any of the USB ports available on your computer. Step 2: Open iTunes,
click the "Files" tab and check the boxes to sync or transfer your
files. Step 3: Select your desired destination folder for the files and
click "Sync" to complete the transfer.
- source_sentence: what can you do with 1gb of data?
sentences:
- >-
You could even contact your email provider, complain that somebody else
is using your email address, and say that you are worried about your
account being compromised. They're very unlikely to do anything, but if
something goes wrong, at least you can prove you forewarned them.
- >-
1) Under Section 80CCD(1), investment in Atal Pension Yojana or NPS up
to ₹ 1.5 lakh qualifies for income tax deduction. But remember that the
total amount of deduction under sections 80C, 80CCC and 80CCD cannot
exceed ₹ 1.5 lakh.
- >-
1GB (or 1024MB) of data lets you send or receive about 1,000 emails and
browse the Internet for about 20 hours every month. (This limit relates
only to your 1GB mobile data allocation; if you are an 'inclusive mobile
broadband customer' you also get 2000 BT Wi-fi wi-fi minutes every
month.)
- source_sentence: how many carbon atoms are in carbon dioxide?
sentences:
- >-
For CO2 there is one atom of carbon and two atoms of oxygen. For H2O,
there is one atom of oxygen and two atoms of hydrogen. A molecule can be
made of only one type of atom.
- >-
Avian influenza refers to the disease caused by infection with avian
(bird) influenza (flu) Type A viruses. These viruses occur naturally
among wild aquatic birds worldwide and can infect domestic poultry and
other bird and animal species. Avian flu viruses do not normally infect
humans.
- >-
At the end of "Inception," Dom Cobb (Leonardo DiCaprio) finally returns
home to his kids after spending a long time in the dream world. Cobb
carries a little top with him. If the top keeps spinning, that means he
is in a dream. ... The final shot shows the top spinning, but it never
reveals whether it falls over.
- source_sentence: is duchenne muscular dystrophy a dominant or recessive trait?
sentences:
- >-
Duchenne muscular dystrophy is inherited in an X-linked recessive
pattern. Males have only one copy of the X chromosome from their mother
and one copy of the Y chromosome from their father. If their X
chromosome has a DMD gene mutation, they will have Duchenne muscular
dystrophy.
- >-
An automatic transmission will downshift for you when you drive uphill.
However, for moderately steep slopes, it's wise to shift to the gear
range marked D2, 2, or L to ascend and descend the hill. For steep
slopes that you can't ascend at a speed faster than 10 mph (about 15
kph), shift to D1 or 1.
- >-
The dream suggests captivity and it refers to your fear of punishment.
Another interpretation of this dream refers to a need to do what you
feel is right in waking life. Being in jail suggests that your feelings
may be trapped by a limited mind and body. ... Jail also suggests
repressed feelings.
datasets:
- sentence-transformers/gooaq
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
co2_eq_emissions:
emissions: 24.236900898755138
energy_consumed: 0.0905639330800724
source: codecarbon
training_type: fine-tuning
on_cloud: false
cpu_model: 13th Gen Intel(R) Core(TM) i7-13700K
ram_total_size: 31.777088165283203
hours_used: 0.293
hardware_used: 1 x NVIDIA GeForce RTX 3090
model-index:
- name: MPNet base trained on GooAQ using QAT with InfoNCE + GOR
results:
- task:
type: information-retrieval
name: Information Retrieval
dataset:
name: gooaq dev float32
type: gooaq-dev-float32
metrics:
- type: cosine_accuracy@1
value: 0.7419
name: Cosine Accuracy@1
- type: cosine_accuracy@3
value: 0.8825
name: Cosine Accuracy@3
- type: cosine_accuracy@5
value: 0.9237
name: Cosine Accuracy@5
- type: cosine_accuracy@10
value: 0.96
name: Cosine Accuracy@10
- type: cosine_precision@1
value: 0.7419
name: Cosine Precision@1
- type: cosine_precision@3
value: 0.29416666666666663
name: Cosine Precision@3
- type: cosine_precision@5
value: 0.18474000000000002
name: Cosine Precision@5
- type: cosine_precision@10
value: 0.09600000000000002
name: Cosine Precision@10
- type: cosine_recall@1
value: 0.7419
name: Cosine Recall@1
- type: cosine_recall@3
value: 0.8825
name: Cosine Recall@3
- type: cosine_recall@5
value: 0.9237
name: Cosine Recall@5
- type: cosine_recall@10
value: 0.96
name: Cosine Recall@10
- type: cosine_ndcg@10
value: 0.8536715392203515
name: Cosine Ndcg@10
- type: cosine_mrr@10
value: 0.8192423412698366
name: Cosine Mrr@10
- type: cosine_map@100
value: 0.8211527599211433
name: Cosine Map@100
- task:
type: information-retrieval
name: Information Retrieval
dataset:
name: gooaq dev int8
type: gooaq-dev-int8
metrics:
- type: cosine_accuracy@1
value: 0.7336
name: Cosine Accuracy@1
- type: cosine_accuracy@3
value: 0.8753
name: Cosine Accuracy@3
- type: cosine_accuracy@5
value: 0.919
name: Cosine Accuracy@5
- type: cosine_accuracy@10
value: 0.9569
name: Cosine Accuracy@10
- type: cosine_precision@1
value: 0.7336
name: Cosine Precision@1
- type: cosine_precision@3
value: 0.2917666666666666
name: Cosine Precision@3
- type: cosine_precision@5
value: 0.18380000000000002
name: Cosine Precision@5
- type: cosine_precision@10
value: 0.09569000000000001
name: Cosine Precision@10
- type: cosine_recall@1
value: 0.7336
name: Cosine Recall@1
- type: cosine_recall@3
value: 0.8753
name: Cosine Recall@3
- type: cosine_recall@5
value: 0.919
name: Cosine Recall@5
- type: cosine_recall@10
value: 0.9569
name: Cosine Recall@10
- type: cosine_ndcg@10
value: 0.8480672791448349
name: Cosine Ndcg@10
- type: cosine_mrr@10
value: 0.8128262301587247
name: Cosine Mrr@10
- type: cosine_map@100
value: 0.8148631237973415
name: Cosine Map@100
- task:
type: information-retrieval
name: Information Retrieval
dataset:
name: gooaq dev binary
type: gooaq-dev-binary
metrics:
- type: cosine_accuracy@1
value: 0.7171
name: Cosine Accuracy@1
- type: cosine_accuracy@3
value: 0.8612
name: Cosine Accuracy@3
- type: cosine_accuracy@5
value: 0.907
name: Cosine Accuracy@5
- type: cosine_accuracy@10
value: 0.9488
name: Cosine Accuracy@10
- type: cosine_precision@1
value: 0.7171
name: Cosine Precision@1
- type: cosine_precision@3
value: 0.28706666666666664
name: Cosine Precision@3
- type: cosine_precision@5
value: 0.18140000000000003
name: Cosine Precision@5
- type: cosine_precision@10
value: 0.09488000000000002
name: Cosine Precision@10
- type: cosine_recall@1
value: 0.7171
name: Cosine Recall@1
- type: cosine_recall@3
value: 0.8612
name: Cosine Recall@3
- type: cosine_recall@5
value: 0.907
name: Cosine Recall@5
- type: cosine_recall@10
value: 0.9488
name: Cosine Recall@10
- type: cosine_ndcg@10
value: 0.8346138412124019
name: Cosine Ndcg@10
- type: cosine_mrr@10
value: 0.7977967063492009
name: Cosine Mrr@10
- type: cosine_map@100
value: 0.8001907977756205
name: Cosine Map@100
MPNet base trained on GooAQ using QAT with InfoNCE + GOR
This is a sentence-transformers model finetuned from microsoft/mpnet-base on the gooaq 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: microsoft/mpnet-base
- Maximum Sequence Length: 512 tokens
- Output Dimensionality: 768 dimensions
- Similarity Function: Cosine Similarity
- Training Dataset:
- Language: en
- License: apache-2.0
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': 512, 'do_lower_case': False, 'architecture': '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})
)
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("tomaarsen/mpnet-base-gooaq-qat")
# Run inference
queries = [
"is duchenne muscular dystrophy a dominant or recessive trait?",
]
documents = [
'Duchenne muscular dystrophy is inherited in an X-linked recessive pattern. Males have only one copy of the X chromosome from their mother and one copy of the Y chromosome from their father. If their X chromosome has a DMD gene mutation, they will have Duchenne muscular dystrophy.',
'The dream suggests captivity and it refers to your fear of punishment. Another interpretation of this dream refers to a need to do what you feel is right in waking life. Being in jail suggests that your feelings may be trapped by a limited mind and body. ... Jail also suggests repressed feelings.',
"An automatic transmission will downshift for you when you drive uphill. However, for moderately steep slopes, it's wise to shift to the gear range marked D2, 2, or L to ascend and descend the hill. For steep slopes that you can't ascend at a speed faster than 10 mph (about 15 kph), shift to D1 or 1.",
]
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([[0.8103, 0.1611, 0.2026]])
Evaluation
Metrics
Information Retrieval
- Datasets:
gooaq-dev-float32,gooaq-dev-int8andgooaq-dev-binary - Evaluated with
InformationRetrievalEvaluator
| Metric | gooaq-dev-float32 | gooaq-dev-int8 | gooaq-dev-binary |
|---|---|---|---|
| cosine_accuracy@1 | 0.7419 | 0.7336 | 0.7171 |
| cosine_accuracy@3 | 0.8825 | 0.8753 | 0.8612 |
| cosine_accuracy@5 | 0.9237 | 0.919 | 0.907 |
| cosine_accuracy@10 | 0.96 | 0.9569 | 0.9488 |
| cosine_precision@1 | 0.7419 | 0.7336 | 0.7171 |
| cosine_precision@3 | 0.2942 | 0.2918 | 0.2871 |
| cosine_precision@5 | 0.1847 | 0.1838 | 0.1814 |
| cosine_precision@10 | 0.096 | 0.0957 | 0.0949 |
| cosine_recall@1 | 0.7419 | 0.7336 | 0.7171 |
| cosine_recall@3 | 0.8825 | 0.8753 | 0.8612 |
| cosine_recall@5 | 0.9237 | 0.919 | 0.907 |
| cosine_recall@10 | 0.96 | 0.9569 | 0.9488 |
| cosine_ndcg@10 | 0.8537 | 0.8481 | 0.8346 |
| cosine_mrr@10 | 0.8192 | 0.8128 | 0.7978 |
| cosine_map@100 | 0.8212 | 0.8149 | 0.8002 |
Training Details
Training Dataset
gooaq
- Dataset: gooaq at b089f72
- Size: 90,000 training samples
- Columns:
questionandanswer - Approximate statistics based on the first 1000 samples:
question answer type string string details - min: 8 tokens
- mean: 11.83 tokens
- max: 20 tokens
- min: 15 tokens
- mean: 60.45 tokens
- max: 180 tokens
- Samples:
question answer how long does halifax take to transfer mortgage funds?Bear in mind that the speed of application will vary depending on your own personal circumstances and the lender's present day-to-day performance. In some cases, applications can be approved by the lender within 24 hours, while some can take weeks or even months.can you get a false pregnancy test?In very rare cases, you can have a false-positive result. This means you're not pregnant but the test says you are. You could have a false-positive result if you have blood or protein in your pee. Certain drugs, such as tranquilizers, anticonvulsants, hypnotics, and fertility drugs, could cause false-positive results.are ahead of its time?Definition of ahead of one's/its time : too advanced or modern to be understood or appreciated during the time when one lives or works As a director, he was ahead of his time. - Loss:
QuantizationAwareLosswith these parameters:{ "loss": "MultipleNegativesRankingLoss", "quantization_precisions": [ "float32", "int8", "binary" ], "quantization_weights": [ 1.0, 1.0, 1.0 ], "n_precisions_per_step": -1 }
Evaluation Dataset
gooaq
- Dataset: gooaq at b089f72
- Size: 10,000 evaluation samples
- Columns:
questionandanswer - Approximate statistics based on the first 1000 samples:
question answer type string string details - min: 8 tokens
- mean: 11.93 tokens
- max: 25 tokens
- min: 14 tokens
- mean: 60.84 tokens
- max: 127 tokens
- Samples:
question answer should you take ibuprofen with high blood pressure?In general, people with high blood pressure should use acetaminophen or possibly aspirin for over-the-counter pain relief. Unless your health care provider has said it's OK, you should not use ibuprofen, ketoprofen, or naproxen sodium. If aspirin or acetaminophen doesn't help with your pain, call your doctor.how old do you have to be to work in sc?The general minimum age of employment for South Carolina youth is 14, although the state allows younger children who are performers to work in show business. If their families are agricultural workers, children younger than age 14 may also participate in farm labor.how to write a topic proposal for a research paper?['Write down the main topic of your paper. ... ', 'Write two or three short sentences under the main topic that explain why you chose that topic. ... ', 'Write a thesis sentence that states the angle and purpose of your research paper. ... ', 'List the items you will cover in the body of the paper that support your thesis statement.'] - Loss:
QuantizationAwareLosswith these parameters:{ "loss": "MultipleNegativesRankingLoss", "quantization_precisions": [ "float32", "int8", "binary" ], "quantization_weights": [ 1.0, 1.0, 1.0 ], "n_precisions_per_step": -1 }
Training Hyperparameters
Non-Default Hyperparameters
eval_strategy: stepsper_device_train_batch_size: 64per_device_eval_batch_size: 64learning_rate: 2e-05num_train_epochs: 1warmup_ratio: 0.1bf16: Truebatch_sampler: no_duplicates
All Hyperparameters
Click to expand
overwrite_output_dir: Falsedo_predict: Falseeval_strategy: stepsprediction_loss_only: Trueper_device_train_batch_size: 64per_device_eval_batch_size: 64per_gpu_train_batch_size: Noneper_gpu_eval_batch_size: Nonegradient_accumulation_steps: 1eval_accumulation_steps: Nonetorch_empty_cache_steps: Nonelearning_rate: 2e-05weight_decay: 0.0adam_beta1: 0.9adam_beta2: 0.999adam_epsilon: 1e-08max_grad_norm: 1.0num_train_epochs: 1max_steps: -1lr_scheduler_type: linearlr_scheduler_kwargs: Nonewarmup_ratio: 0.1warmup_steps: 0log_level: passivelog_level_replica: warninglog_on_each_node: Truelogging_nan_inf_filter: Truesave_safetensors: Truesave_on_each_node: Falsesave_only_model: Falserestore_callback_states_from_checkpoint: Falseno_cuda: Falseuse_cpu: Falseuse_mps_device: Falseseed: 42data_seed: Nonejit_mode_eval: Falsebf16: Truefp16: Falsefp16_opt_level: O1half_precision_backend: autobf16_full_eval: Falsefp16_full_eval: Falsetf32: Nonelocal_rank: 0ddp_backend: Nonetpu_num_cores: Nonetpu_metrics_debug: Falsedebug: []dataloader_drop_last: Falsedataloader_num_workers: 0dataloader_prefetch_factor: Nonepast_index: -1disable_tqdm: Falseremove_unused_columns: Truelabel_names: Noneload_best_model_at_end: Falseignore_data_skip: Falsefsdp: []fsdp_min_num_params: 0fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}fsdp_transformer_layer_cls_to_wrap: Noneaccelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}parallelism_config: Nonedeepspeed: Nonelabel_smoothing_factor: 0.0optim: adamw_torch_fusedoptim_args: Noneadafactor: Falsegroup_by_length: Falselength_column_name: lengthproject: huggingfacetrackio_space_id: trackioddp_find_unused_parameters: Noneddp_bucket_cap_mb: Noneddp_broadcast_buffers: Falsedataloader_pin_memory: Truedataloader_persistent_workers: Falseskip_memory_metrics: Trueuse_legacy_prediction_loop: Falsepush_to_hub: Falseresume_from_checkpoint: Nonehub_model_id: Nonehub_strategy: every_savehub_private_repo: Nonehub_always_push: Falsehub_revision: Nonegradient_checkpointing: Falsegradient_checkpointing_kwargs: Noneinclude_inputs_for_metrics: Falseinclude_for_metrics: []eval_do_concat_batches: Truefp16_backend: autopush_to_hub_model_id: Nonepush_to_hub_organization: Nonemp_parameters:auto_find_batch_size: Falsefull_determinism: Falsetorchdynamo: Noneray_scope: lastddp_timeout: 1800torch_compile: Falsetorch_compile_backend: Nonetorch_compile_mode: Noneinclude_tokens_per_second: Falseinclude_num_input_tokens_seen: noneftune_noise_alpha: Noneoptim_target_modules: Nonebatch_eval_metrics: Falseeval_on_start: Falseuse_liger_kernel: Falseliger_kernel_config: Noneeval_use_gather_object: Falseaverage_tokens_across_devices: Trueprompts: Nonebatch_sampler: no_duplicatesmulti_dataset_batch_sampler: proportionalrouter_mapping: {}learning_rate_mapping: {}
Training Logs
| Epoch | Step | Training Loss | Validation Loss | gooaq-dev-float32_cosine_ndcg@10 | gooaq-dev-int8_cosine_ndcg@10 | gooaq-dev-binary_cosine_ndcg@10 |
|---|---|---|---|---|---|---|
| -1 | -1 | - | - | 0.2155 | 0.5116 | 0.3432 |
| 0.0007 | 1 | 8.8919 | - | - | - | - |
| 0.0505 | 71 | 4.6028 | - | - | - | - |
| 0.1002 | 141 | - | 0.3973 | 0.7842 | 0.7799 | 0.7606 |
| 0.1009 | 142 | 0.8168 | - | - | - | - |
| 0.1514 | 213 | 0.4967 | - | - | - | - |
| 0.2004 | 282 | - | 0.2611 | 0.8125 | 0.8082 | 0.7879 |
| 0.2018 | 284 | 0.4427 | - | - | - | - |
| 0.2523 | 355 | 0.4156 | - | - | - | - |
| 0.3006 | 423 | - | 0.2213 | 0.8282 | 0.8230 | 0.8047 |
| 0.3028 | 426 | 0.3245 | - | - | - | - |
| 0.3532 | 497 | 0.3354 | - | - | - | - |
| 0.4009 | 564 | - | 0.2026 | 0.8333 | 0.8291 | 0.8129 |
| 0.4037 | 568 | 0.2926 | - | - | - | - |
| 0.4542 | 639 | 0.317 | - | - | - | - |
| 0.5011 | 705 | - | 0.1854 | 0.8384 | 0.8340 | 0.8192 |
| 0.5046 | 710 | 0.2779 | - | - | - | - |
| 0.5551 | 781 | 0.278 | - | - | - | - |
| 0.6013 | 846 | - | 0.1768 | 0.8440 | 0.8398 | 0.8245 |
| 0.6055 | 852 | 0.2696 | - | - | - | - |
| 0.6560 | 923 | 0.2752 | - | - | - | - |
| 0.7015 | 987 | - | 0.1679 | 0.8504 | 0.8449 | 0.8287 |
| 0.7065 | 994 | 0.2318 | - | - | - | - |
| 0.7569 | 1065 | 0.2398 | - | - | - | - |
| 0.8017 | 1128 | - | 0.1621 | 0.8498 | 0.8454 | 0.8317 |
| 0.8074 | 1136 | 0.2274 | - | - | - | - |
| 0.8579 | 1207 | 0.2376 | - | - | - | - |
| 0.9019 | 1269 | - | 0.1572 | 0.8518 | 0.8464 | 0.8305 |
| 0.9083 | 1278 | 0.238 | - | - | - | - |
| 0.9588 | 1349 | 0.2168 | - | - | - | - |
| -1 | -1 | - | - | 0.8537 | 0.8481 | 0.8346 |
Environmental Impact
Carbon emissions were measured using CodeCarbon.
- Energy Consumed: 0.091 kWh
- Carbon Emitted: 0.024 kg of CO2
- Hours Used: 0.293 hours
Training Hardware
- On Cloud: No
- GPU Model: 1 x NVIDIA GeForce RTX 3090
- CPU Model: 13th Gen Intel(R) Core(TM) i7-13700K
- RAM Size: 31.78 GB
Framework Versions
- Python: 3.11.6
- Sentence Transformers: 5.3.0.dev0
- Transformers: 4.57.6
- PyTorch: 2.10.0+cu126
- Accelerate: 1.12.0
- Datasets: 4.3.0
- Tokenizers: 0.22.2
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",
}
QuantizationAwareLoss
@article{jacob2018quantization,
title={Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference},
author={Jacob, Benoit and Kligys, Skirmantas and Chen, Bo and Zhu, Menglong and Tang, Matthew and Howard, Andrew and Adam, Hartwig and Kalenichenko, Dmitry},
journal={arXiv preprint arXiv:1712.05877},
year={2018}
}
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}
}