legora_model / checkpoint-98 /README.md

Vinsuka

Upload folder using huggingface_hub

d52ad1b verified 5 months ago

preview code

raw

history blame contribute delete

32.9 kB

metadata

language:
  - en
license: apache-2.0
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:6190
  - loss:MatryoshkaLoss
  - loss:MultipleNegativesRankingLoss
base_model: nomic-ai/modernbert-embed-base
widget:
  - source_sentence: What is the duration of the period mentioned in the text?
    sentences:
      - >-
        . The only excep Ɵon to the requirement that the plainƟﬀ must be a
        lending i nsƟtuƟon in order to invoke the provisions of the Act is
        contained in SecƟon 25, in terms of which a person who inter alia
        knowingly draws a cheque which is subsequently dishonoured by the bank
        for want of funds is guilty of an oﬀence under the Act, and proceedings
        can be insƟtuted against such person in the Magistrate’s
      - >-
        ? The 1st question of law is formulated on the basis that , the 1st
        Defendant is the licensee of the 2nd Defendant and therefore, the 1st
        Defendant cannot claim prescriptive title to the subject matter
      - >-
        .50,000/ - (that is , a period of 36 months) but such “Facility” is
        subject to review on 30 /09/2000”, (that is, a period of about only 5
        months from the date of P4)
  - source_sentence: >-
      What is the purpose of the disposition of the property by Lanka Tractors
      Limited as mentioned in the text?
    sentences:
      - >-
        . (3) is whether the said disposition of the property by Lanka Tractors
        Limited was done with the sole object of defrauding its creditors.
        Section 348 of the Companies Act which describes about Fraudulent
        reference would be relevant in this regard
      - >-
        . In the arbitration process, the Government is not involved; the court
        system is not involved (except as provided for in the Act); the parties
        do not have to rely on any Government institution for resolution of
        their dispute. Process of conducting the arbitration, venue, time, mode
        of adducing evidence are all decided by agreement of parties
      - >-
        . This is broadly similar to the provision in the summary procedure on
        liquid claims. The amendment in clause 8 of the Bill, repeals the deﬁni
        Ɵon of the term ‘debt’ in sec Ɵon 30. The subs Ɵtuted deﬁni Ɵon excludes
        the words referred to above which limit its applicability to money owed
        under a promise or agreement which is in wri Ɵng
  - source_sentence: What is one of the topics covered in the training program?
    sentences:
      - >-
        . The resul Ɵng posiƟon is that the court would not have any wri Ʃen
        evidence of the commitment on the part of the debtor when it issues
        decree nisi in the ﬁrst instance
      - >-
        ? Before this C ourt, there is no dispute on the manner in which the
        appellant obtained the title of the land in question
      - >-
        . Detail reporting procedures to government of Sri Lanka’s contact
        points. - 4 Weeks Phase 3 Training of Port Facility Security Officers
        SATHSINDU/BAGNOLD undertakes to design a training program and conducted
        aid program for up to ten persons. • Understanding the reasons for the
        ISPS code • ISPS Code content and requirements. • Understanding the ISPS
        Code
  - source_sentence: What type of action was taken by the Divisional Secretary?
    sentences:
      - >-
        .2020 was also sent by the Divisional Secretary of Th amankaduwa
        imposing similar restrictions as by the Polonnaruwa Pradeshiya Sabha
      - >-
        . When Seylan Bank published the resolution of its board of directors
        which exercised its powers of Parate Execution in the newspaper on 10th
        March 2006-, HNB had made the application dated 21st March [SC Appeal
        No. 85A /2009 ] Page 6 of 25 2006 to the District Court of Colombo in
        terms of Sections 260, 261, 348, 359 and 352 of the Companies Act No
      - >-
        . Having regard to the above -mentioned stipulated circumstances , I
        consider the facts put forward for the appellant , seeking a reduction
        of sentence. The offence was committed in 2004. The appellant had been
        in remand custody for more than three years and the appell ant did not
        have any previous convictions
  - source_sentence: What is described in Section 25 of the Arbitration Act?
    sentences:
      - >-
        . But where a matter is within the plenary jurisdiction of the Court if
        no objection is taken, the Court will then have jurisdiction to proceed
        on with the matter and make a valid order.” 14 31. Further , in the case
        of Don Tilakaratne v
      - >-
        . (3) The provision of subsections (1) and (2) shall apply only to the
        extent agreed to by the parties. (4) The arbitral tribunal shall decide
        according to considerations of general justice and fairness or trade
        usages only if the parties have expressly authorised it to do so.
        Section 25 of the Arbitration Act describes the form and content of the
        arbitral award as follows: 25
      - >-
        . 9 and 10 based on the objection taken to them by the Counsel for HNB,
        despite the fact that they did not arise from the pleadings, and were
        altogether inconsistent with them, answered the afore-stated question of
        law (in respect of which this Court had granted Leave to Appeal in that
        case) in the affirmative and in favour of HNB, and stated as follows:
        “In conclusion, it needs to be emphasised
pipeline_tag: sentence-similarity
library_name: sentence-transformers
metrics:
  - cosine_accuracy@1
  - cosine_accuracy@3
  - cosine_accuracy@5
  - cosine_accuracy@10
  - cosine_precision@1
  - cosine_precision@3
  - cosine_precision@5
  - cosine_precision@10
  - cosine_recall@1
  - cosine_recall@3
  - cosine_recall@5
  - cosine_recall@10
  - cosine_ndcg@10
  - cosine_mrr@10
  - cosine_map@100
model-index:
  - name: Fine-tuned with [QuicKB](https://github.com/ALucek/QuicKB)
    results:
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 768
          type: dim_768
        metrics:
          - type: cosine_accuracy@1
            value: 0.5784883720930233
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.7601744186046512
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.8197674418604651
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.8880813953488372
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.5784883720930233
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.253391472868217
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.163953488372093
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.08880813953488371
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.5784883720930233
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.7601744186046512
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.8197674418604651
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.8880813953488372
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.733110755438693
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.6835271317829454
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.6874162730700494
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 512
          type: dim_512
        metrics:
          - type: cosine_accuracy@1
            value: 0.5770348837209303
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.7616279069767442
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.8197674418604651
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.8822674418604651
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.5770348837209303
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.25387596899224807
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.163953488372093
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.0882267441860465
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.5770348837209303
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.7616279069767442
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.8197674418604651
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.8822674418604651
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.7303694393079266
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.6816127953119231
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.685728403908298
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 256
          type: dim_256
        metrics:
          - type: cosine_accuracy@1
            value: 0.5552325581395349
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.7354651162790697
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.7950581395348837
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.8604651162790697
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.5552325581395349
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.24515503875968994
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.15901162790697673
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.08604651162790695
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.5552325581395349
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.7354651162790697
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.7950581395348837
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.8604651162790697
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.7073914263638542
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.6584688768918417
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.6631022921972395
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 128
          type: dim_128
        metrics:
          - type: cosine_accuracy@1
            value: 0.498546511627907
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.6700581395348837
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.7369186046511628
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.8226744186046512
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.498546511627907
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.22335271317829455
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.14738372093023255
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.08226744186046511
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.498546511627907
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.6700581395348837
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.7369186046511628
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.8226744186046512
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.6569451636973174
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.6045300387596899
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.6099407824679366
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 64
          type: dim_64
        metrics:
          - type: cosine_accuracy@1
            value: 0.39680232558139533
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.5523255813953488
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.626453488372093
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.7209302325581395
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.39680232558139533
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.18410852713178294
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.12529069767441858
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.07209302325581395
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.39680232558139533
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.5523255813953488
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.626453488372093
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.7209302325581395
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.5476937013127858
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.4936317598744924
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.5014802548028123
            name: Cosine Map@100

Fine-tuned with QuicKB

This is a sentence-transformers model finetuned from nomic-ai/modernbert-embed-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: nomic-ai/modernbert-embed-base
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 dimensions
Similarity Function: Cosine Similarity
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}) with Transformer model: ModernBertModel 
  (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 = [
    'What is described in Section 25 of the Arbitration Act?',
    '. (3) The provision of subsections (1) and (2) shall apply only to the extent agreed to by the parties. (4) The arbitral tribunal shall decide according to considerations of general justice and fairness or trade usages only if the parties have expressly authorised it to do so. Section 25 of the Arbitration Act describes the form and content of the arbitral award as follows: 25',
    '. 9 and 10 based on the objection taken to them by the Counsel for HNB, despite the fact that they did not arise from the pleadings, and were altogether inconsistent with them, answered the afore-stated question of law (in respect of which this Court had granted Leave to Appeal in that case) in the affirmative and in favour of HNB, and stated as follows: “In conclusion, it needs to be emphasised',
]
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

Datasets: dim_768, dim_512, dim_256, dim_128 and dim_64
Evaluated with InformationRetrievalEvaluator

Metric	dim_768	dim_512	dim_256	dim_128	dim_64
cosine_accuracy@1	0.5785	0.577	0.5552	0.4985	0.3968
cosine_accuracy@3	0.7602	0.7616	0.7355	0.6701	0.5523
cosine_accuracy@5	0.8198	0.8198	0.7951	0.7369	0.6265
cosine_accuracy@10	0.8881	0.8823	0.8605	0.8227	0.7209
cosine_precision@1	0.5785	0.577	0.5552	0.4985	0.3968
cosine_precision@3	0.2534	0.2539	0.2452	0.2234	0.1841
cosine_precision@5	0.164	0.164	0.159	0.1474	0.1253
cosine_precision@10	0.0888	0.0882	0.086	0.0823	0.0721
cosine_recall@1	0.5785	0.577	0.5552	0.4985	0.3968
cosine_recall@3	0.7602	0.7616	0.7355	0.6701	0.5523
cosine_recall@5	0.8198	0.8198	0.7951	0.7369	0.6265
cosine_recall@10	0.8881	0.8823	0.8605	0.8227	0.7209
cosine_ndcg@10	0.7331	0.7304	0.7074	0.6569	0.5477
cosine_mrr@10	0.6835	0.6816	0.6585	0.6045	0.4936
cosine_map@100	0.6874	0.6857	0.6631	0.6099	0.5015

Training Details

Training Dataset

Unnamed Dataset

Size: 6,190 training samples
Columns: anchor and positive
Approximate statistics based on the first 1000 samples:
anchor positive
type string string
details
min: 7 tokens
mean: 15.11 tokens
max: 32 tokens

min: 3 tokens
mean: 69.53 tokens
max: 214 tokens

	anchor	positive
type	string	string
details	min: 7 tokens mean: 15.11 tokens max: 32 tokens	min: 3 tokens mean: 69.53 tokens max: 214 tokens

Samples:

anchor	positive
`How must the District Court exercise its discretion?`	`imposition of ‘ a’ term; (5) It is not mandatory to impose security, as evinced by the use of the conjunction “or”; (6) In imposing terms, the District Court must be mindful of the objectives of the Act, and its discretion must be exercised judicially`
`What is the source of the observation made by Christian Appu?`	`. Christian Appu , (1895) 1 NLR 288 observed that , “possession is "disturbed" either by an action intended to remove the possessor from the land, or by acts which prevent the possessor from enjoying the free and full use of 12 the land of which he is in the course of acquiring the dominion, and which convert his continuous user into a disconnected and divided user ”`
`What must the defendant do regarding the plaintiff's claim?`	`. The Court of Appeal in Ramanayake v Sampath Bank Ltd and Others [(1993) 1 Sri LR 145 at page 153] has held that, “The defendant has to deal with the plaintiff’s claim on its merits; it is not competent for the defendant to merely set out technical objections. It is also incumbent on the defendant to reveal his defence, if he has any`

Loss: MatryoshkaLoss with these parameters:

{
    "loss": "MultipleNegativesRankingLoss",
    "matryoshka_dims": [
        768,
        512,
        256,
        128,
        64
    ],
    "matryoshka_weights": [
        1,
        1,
        1,
        1,
        1
    ],
    "n_dims_per_step": -1
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: epoch
per_device_train_batch_size: 16
gradient_accumulation_steps: 8
learning_rate: 2e-05
lr_scheduler_type: cosine
warmup_ratio: 0.1
tf32: True
load_best_model_at_end: True
optim: adamw_torch_fused
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: epoch
prediction_loss_only: True
per_device_train_batch_size: 16
per_device_eval_batch_size: 8
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 8
eval_accumulation_steps: None
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: cosine
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: False
fp16_opt_level: O1
half_precision_backend: auto
bf16_full_eval: False
fp16_full_eval: False
tf32: True
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: 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}
fsdp_transformer_layer_cls_to_wrap: None
accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
deepspeed: None
label_smoothing_factor: 0.0
optim: adamw_torch_fused
optim_args: None
adafactor: False
group_by_length: False
length_column_name: length
ddp_find_unused_parameters: None
ddp_bucket_cap_mb: None
ddp_broadcast_buffers: False
dataloader_pin_memory: True
dataloader_persistent_workers: False
skip_memory_metrics: True
use_legacy_prediction_loop: False
push_to_hub: 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
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
average_tokens_across_devices: False
prompts: None
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	dim_768_cosine_ndcg@10	dim_512_cosine_ndcg@10	dim_256_cosine_ndcg@10	dim_128_cosine_ndcg@10	dim_64_cosine_ndcg@10
0.1034	5	29.8712	-	-	-	-	-
0.2067	10	26.1323	-	-	-	-	-
0.3101	15	17.8585	-	-	-	-	-
0.4134	20	14.0232	-	-	-	-	-
0.5168	25	11.6897	-	-	-	-	-
0.6202	30	10.8431	-	-	-	-	-
0.7235	35	9.264	-	-	-	-	-
0.8269	40	11.2186	-	-	-	-	-
0.9302	45	9.9143	-	-	-	-	-
1.0	49	-	0.7134	0.7110	0.6902	0.6341	0.5282
1.0207	50	7.2581	-	-	-	-	-
1.1240	55	6.066	-	-	-	-	-
1.2274	60	6.3626	-	-	-	-	-
1.3307	65	6.8135	-	-	-	-	-
1.4341	70	5.5556	-	-	-	-	-
1.5375	75	6.0144	-	-	-	-	-
1.6408	80	6.1965	-	-	-	-	-
1.7442	85	5.596	-	-	-	-	-
1.8475	90	6.631	-	-	-	-	-
1.9509	95	6.3319	-	-	-	-	-
2.0	98	-	0.7331	0.7304	0.7074	0.6569	0.5477

Framework Versions

Python: 3.13.3
Sentence Transformers: 3.4.0
Transformers: 4.48.1
PyTorch: 2.6.0+cu126
Accelerate: 1.3.0
Datasets: 3.2.0
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",
}

MatryoshkaLoss

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning},
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

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}
}