metadata
library_name: sentence-transformers
pipeline_tag: sentence-similarity
tags:
- sentence-transformers
- sentence-similarity
- feature-extraction
- generated_from_trainer
- dataset_size:26147930
- loss:MultipleNegativesRankingLoss
widget:
- source_sentence: >-
[YEAR_RANGE] 2020-2024 [TEXT] Vitamin B-6 Prevents Heart Failure with
Preserved Ejection Fraction Through Downstream of Kinase 3 in a Mouse
Model.
sentences:
- >-
[YEAR_RANGE] 2020-2024 [TEXT] Colorectal cancer (CRC) is a complex and
genetically heterogeneous disease presenting a specific metastatic
pattern, with the liver being the most common site of metastasis. Around
20%-25% of patients with CRC will develop exclusively hepatic metastatic
disease throughout their disease history. With its specific
characteristics and therapeutic options, liver-limited disease (LLD)
should be considered as a specific entity. The identification of these
patients is particularly relevant in view of the growing interest in
liver transplantation in selected patients with advanced CRC.
Identifying why some patients will develop only LLD remains a challenge,
mainly because of a lack of a systemic understanding of this complex and
interlinked phenomenon given that cancer has traditionally been
investigated according to distinct physiological compartments. Recently,
multidisciplinary efforts and new diagnostic tools have made it possible
to study some of these complex issues in greater depth and may help
identify targets and specific treatment strategies to benefit these
patients. In this review we analyze the underlying biology and available
tools to help clinicians better understand this increasingly common and
specific disease.
- >-
[YEAR_RANGE] 2020-2024 [TEXT] PURPOSE: Secondary breast cancer is a
frequent late adverse event of mediastinal Hodgkin lymphoma
radiotherapy. Secondary breast cancers overwhelmingly correspond to
ductal carcinoma and develop from the glandular mammary tissue. In
addition, during childhood, radiation overexposure of the glandular
tissue may lead to a late breast hypotrophy at adult age. The aim of
this study was to evaluate the radiation exposure to the glandular
tissue in patients treated for mediastinal Hodgkin lymphoma with
intensity-modulated proton therapy, in order to evaluate the potential
dosimetric usefulness of its delineation for breast sparing. MATERIALS
AND METHODS: Sixteen consecutive intermediate-risk mediastinal female
patients with Hodgkin lymphoma treated with consolidation radiation with
deep inspiration breath hold intensity-modulated proton therapy to the
total dose of 30Gy were included. Breasts were delineated according to
the European Society for Radiotherapy and Oncology guidelines for
treatment optimization ("clinical organ at risk"). The glandular tissue
("glandular organ at risk") was retrospectively contoured on the initial
simulation CT scans based on Hounsfield unit (HU) values, using a range
between -80HU and 500HU. RESULTS: The mean and maximum doses delivered
to the glandular organ at risk were significantly lower than the mean
and maximum doses delivered to the clinical organ at risk, but were
statistically correlated. Glandular organ at risk volumes were
significantly smaller. CONCLUSION: Optimizing the treatment plans on the
clinical breast contours will systematically lead to overestimation of
the dose received to the glandular tissue and, consequently, to an
indistinct and involuntary improved glandular tissue sparing. As such,
our findings do not support the consideration of the glandular tissue as
an additional organ at risk when planning intensity-modulated proton
therapy for mediastinal Hodgkin lymphoma in female patients.
- >-
[YEAR_RANGE] 2020-2024 [TEXT] BACKGROUND: There is an urgent need to
develop an efficient therapeutic strategy for heart failure with
preserved ejection fraction (HFpEF), which is mediated by phenotypic
changes in cardiac macrophages. We previously reported that vitamin B-6
inhibits macrophage-mediated inflammasome activation. OBJECTIVES: We
sought to examine whether the prophylactic use of vitamin B-6 prevents
HFpEF. METHODS: HFpEF model was elicited by a combination of high-fat
diet and Nω-nitro-l-arginine methyl ester supplement in mice. Cardiac
function was assessed using conventional echocardiography and Doppler
imaging. Immunohistochemistry and immunoblotting were used to detect
changes in the macrophage phenotype and myocardial remodeling-related
molecules. RESULTS: Co-administration of vitamin B-6 with HFpEF mice
mitigated HFpEF phenotypes, including diastolic dysfunction, cardiac
macrophage phenotypic shifts, fibrosis, and hypertrophy.
Echocardiographic improvements were observed, with the E/E' ratio
decreasing from 42.0 to 21.6 and the E/A ratio improving from 2.13 to
1.17. The exercise capacity also increased from 295.3 to 657.7 min.
However, these beneficial effects were negated in downstream of kinase
(DOK) 3-deficient mice. Mechanistically, vitamin B-6 increased DOK3
protein concentrations and inhibited macrophage phenotypic changes,
which were abrogated by an AMP-activated protein kinase inhibitor.
CONCLUSIONS: Vitamin B-6 increases DOK3 signaling to lower risk of HFpEF
by inhibiting phenotypic changes in cardiac macrophages.
- source_sentence: >-
[YEAR_RANGE] 2020-2024 [TEXT] Resolving phylogenetic relationships and
taxonomic revision in the Pseudogastromyzon (Cypriniformes,
Gastromyzonidae) genus: molecular and morphological evidence for a new
genus, Labigastromyzon.
sentences:
- >-
[YEAR_RANGE] 2020-2024 [TEXT] Bats contain a diverse spectrum of viral
species in their bodies. The RNA virus family Paramyxoviridae tends to
infect several vertebrate species, which are accountable for a variety
of devastating infections in both humans and animals. Viruses of this
kind include measles, mumps, and Hendra. Some synonymous codons are
favoured over others in mRNAs during gene-to-protein synthesis process.
Such phenomenon is termed as codon usage bias (CUB). Our research
emphasized many aspects that shape the CUB of genes in the
Paramyxoviridae family found in bats. Here, the nitrogenous base A
occurred the most. AT was found to be abundant in the coding sequences
of the Paramyxoviridae family. RSCU data revealed that A or T ending
codons occurred more frequently than predicted. Furthermore, 3
overrepresented codons (CAT, AGA, and GCA) and 7 underrepresented codons
(CCG, TCG, CGC, CGG, CGT, GCG and ACG) were detected in the viral
genomes. Correspondence analysis, neutrality plot, and parity plots
highlight the combined impact of mutational pressure and natural
selection on CUB. The neutrality plot of GC12 against GC3 yielded a
regression coefficient value of 0.366, indicating that natural selection
had a significant (63.4 %) impact. Moreover, RNA editing analysis was
done, which revealed the highest frequency of C to T mutations. The
results of our research revealed the pattern of codon usage and RNA
editing sites in Paramyxoviridae genomes.
- >-
[YEAR_RANGE] 2020-2024 [TEXT] OBJECTIVE: The preoperative inclination
angle of mandibular incisors was crucial for surgical and postoperative
stability while the effect of proclined mandibular incisors on skeletal
stability has not been investigated. This study aimed to evaluate the
effects of differences in presurgical mandibular incisor inclination on
skeletal stability after orthognathic surgery in patients with skeletal
Class III malocclusion. METHODS: A retrospective cohort study of 80
consecutive patients with skeletal Class III malocclusion who underwent
bimaxillary orthognathic surgery was conducted. According to incisor
mandibular plane angle (IMPA), patients were divided into 3 groups:
retroclined inclination (IMPA < 87°), normal inclination (87° ≤ IMPA <
93°) and proclined inclination (IMPA ≥ 93°). Preoperative
characteristics, surgical changes and postoperative stability were
compared based on lateral cephalograms obtained 1 week before surgery
(T0), 1 week after surgery (T1), and at 6 to 12 months postoperatively
(T2). RESULTS: The mandible demonstrated a forward and upward relapse in
all three groups. No significant differences in skeletal relapse were
observed in the 3 groups of patients. However, the proclined inclination
group showed a negative overbite tendency postoperatively compared with
the other two groups and a clinically significant mandibular relapse
pattern. Proclined IMPA both pre- and postoperatively was correlated
with mandibular relapse. CONCLUSION: Sufficient presurgical mandibular
incisor decompensation was of crucial importance for the maintenance of
skeletal stability in patients with skeletal Class III malocclusion who
subsequently underwent orthognathic surgery.
- >-
[YEAR_RANGE] 2020-2024 [TEXT] The Pseudogastromyzon genus, consisting of
species predominantly distributed throughout southeastern China, has
garnered increasing market attention in recent years due to its
ornamental appeal. However, the overlapping diagnostic attributes render
the commonly accepted criteria for interspecific identification
unreliable, leaving the phylogenetic relationships among
Pseudogastromyzon species unexplored. In the present study, we undertake
molecular phylogenetic and morphological examinations of the
Pseudogastromyzon genus. Our phylogenetic analysis of mitochondrial
genes distinctly segregated Pseudogastromyzon species into two clades:
the Pseudogastromyzon clade and the Labigastromyzon clade. A subsequent
morphological assessment revealed that the primary dermal ridge
(specifically, the second ridge) within the labial adhesive apparatus
serves as an effective and precise interspecific diagnostic
characteristic. Moreover, the distributional ranges of Pseudogastromyzon
and Labigastromyzon are markedly distinct, exhibiting only a narrow area
of overlap. Considering the morphological heterogeneity of the labial
adhesive apparatus and the substantial division within the molecular
phylogeny, we advocate for the elevation of the Labigastromyzon subgenus
to the status of a separate genus. Consequently, we have ascertained the
validity of the Pseudogastromyzon and Labigastromyzon species, yielding
a total of six valid species. To facilitate future research, we present
comprehensive descriptions of the redefined species and introduce novel
identification keys.
- source_sentence: >-
[YEAR_RANGE] 2020-2024 [TEXT] PCa-RadHop: A transparent and lightweight
feed-forward method for clinically significant prostate cancer
segmentation.
sentences:
- >-
[YEAR_RANGE] 2020-2024 [TEXT] According to the importance of time in
treatment of thrombosis disorders, faster than current treatments are
required. For the first time, this research discloses a novel strategy
for rapid dissolution of blood clots by encapsulation of a fibrinolytic
(Reteplase) into a Thrombin sensitive shell formed by polymerization of
acrylamide monomers and bisacryloylated peptide as crosslinker.
Degradability of the peptide units in exposure to Thrombin, creates the
Thrombin-sensitive Reteplase nanocapsules (TSRNPs) as a triggered
release system. Accelerated thrombolysis was achieved by combining three
approaches including: deep penetration of TSRNPs into the blood clots,
changing the clot dissolution mechanism by altering the distribution
pattern of TSRNPs to 3D intra-clot distribution (based on the
distributed intra-clot thrombolysis (DIT) model) instead of peripheral
and unidirectional distribution of unencapsulated fibrinolytics and,
enzyme-stimulated release of the fibrinolytic. Ex-vivo study was carried
out by an occluded tube model that mimics in-vivo brain stroke as an
emergency situation where faster treatment in short time is a golden
key. In in vivo, efficacy of the developed formulation was confirmed by
PET scan and laser Doppler flowmetry (LDF). As the most important
achievements, 40.0 ± 0.7 (n = 3) % and 37.0 ± 0.4 (n = 3) % reduction in
the thrombolysis time (faster reperfusion) were observed by ex-vivo and
in-vivo experiments, respectively. Higher blood flow and larger
digestion mass of clot at similar times in comparison to
non-encapsulated Reteplase were observed that means more effective
thrombolysis by the developed strategy.
- >-
[YEAR_RANGE] 2020-2024 [TEXT] Prostate Cancer is one of the most
frequently occurring cancers in men, with a low survival rate if not
early diagnosed. PI-RADS reading has a high false positive rate, thus
increasing the diagnostic incurred costs and patient discomfort. Deep
learning (DL) models achieve a high segmentation performance, although
require a large model size and complexity. Also, DL models lack of
feature interpretability and are perceived as "black-boxes" in the
medical field. PCa-RadHop pipeline is proposed in this work, aiming to
provide a more transparent feature extraction process using a linear
model. It adopts the recently introduced Green Learning (GL) paradigm,
which offers a small model size and low complexity. PCa-RadHop consists
of two stages: Stage-1 extracts data-driven radiomics features from the
bi-parametric Magnetic Resonance Imaging (bp-MRI) input and predicts an
initial heatmap. To reduce the false positive rate, a subsequent stage-2
is introduced to refine the predictions by including more contextual
information and radiomics features from each already detected Region of
Interest (ROI). Experiments on the largest publicly available dataset,
PI-CAI, show a competitive performance standing of the proposed method
among other deep DL models, achieving an area under the curve (AUC) of
0.807 among a cohort of 1,000 patients. Moreover, PCa-RadHop maintains
orders of magnitude smaller model size and complexity.
- >-
[YEAR_RANGE] 2020-2024 [TEXT] OBJECTIVE: To evaluate rates of remission,
recovery, relapse, and recurrence in suicidal youth who participated in
a clinical trial comparing Dialectical Behavior Therapy (DBT) and
Individual and Group Supportive Therapy (IGST). METHOD: Participants
were 173 youth, aged 12 to 18 years, with repetitive self-harm
(including at least 1 prior suicide attempt [SA]) and elevated suicidal
ideation (SI). Participants received 6 months of DBT or IGST and were
followed for 6 months post-treatment. The sample was 95% female, 56.4%
White, and 27.49% Latina. Remission was defined as absence of SA or
nonsuicidal self-injury (NSSI) across one 3-month interval; recovery was
defined across 2 or more consecutive intervals. Relapse and recurrence
were defined as SA or NSSI following remission or recovery.
Cross-tabulation with χ2 was used for between-group contrasts. RESULTS:
Over 70% of the sample reported remission of SA at each treatment and
follow-up interval. There were significantly higher rates of remission
and recovery and lower rates of relapse and recurrence for SA in DBT
than for IGST. Across treatments and time points, SA had higher
remission and recovery rates and lower relapse and recurrence rates than
NSSI. There were no significant differences in NSSI remission between
conditions; however, participants receiving DBT had significantly higher
NSSI recovery rates than those receiving IGST for the 3- to 9-month, 3-
to 12-month, and 6- to 12-month intervals. CONCLUSION: Results showed
higher percentages of SA remission and recovery for DBT as compared to
IGST. NSSI was less likely to remit than SA. PLAIN LANGUAGE SUMMARY:
This study examined rates of remission, recovery, relapse, and
recurrence of suicide attempts (SA) and nonsuicidal self-injury (NSSI)
among the participants in the CARES Study, a randomized clinical trial
of 6 months of Dialectical Behavior Therapy or Individual and Group
Supportive Therapy. 173 youth aged 12 to 18 years participated in the
study and were followed for 6 months post treatment. Over 70% of the
sample reported remission of SA at each treatment and follow-up
interval. There were significantly higher rates of remission and
recovery and lower rates of relapse and recurrence for SA among
participants who received Dialectical Behavioral Therapy. Across both
treatments, remission and recovery rates were lower and relapse and
recurrence rates were higher for NSSI than for SA. These results
underscore the value of Dialectical Behavioral Therapy as a first line
treatment for youth at high risk for suicide. DIVERSITY & INCLUSION
STATEMENT: We worked to ensure race, ethnic, and/or other types of
diversity in the recruitment of human participants. CLINICAL TRIAL
REGISTRATION INFORMATION: Collaborative Adolescent Research on Emotions
and Suicide (CARES); https://www. CLINICALTRIALS: gov/; NCT01528020.
- source_sentence: >-
[YEAR_RANGE] 2020-2024 [TEXT] Predicting Recovery After Concussion in
Pediatric Patients: A Meta-Analysis.
sentences:
- >-
[YEAR_RANGE] 2020-2024 [TEXT] OBJECTIVE: The authors examined licensing
requirements for select children's behavioral health care providers.
METHODS: Statutes and regulations as of October 2021 were reviewed for
licensed clinical social workers, licensed professional counselors, and
licensed marriage and family therapists for all 50 U.S. states and the
District of Columbia. RESULTS: All jurisdictions had laws regarding
postgraduate training and license portability. No jurisdiction included
language about specialized postgraduate training related to serving
children and families or cultural competence. Other policies that
related to the structure, composition, and authority of licensing boards
varied across states and licensure types. CONCLUSIONS: In their efforts
to address barriers to licensure, expand the workforce, and ensure that
children have access to high-quality and culturally responsive care,
states could consider their statutes and regulations.
- >-
[YEAR_RANGE] 2020-2024 [TEXT] Magnetic Resonance Imaging (MRI) plays a
pivotal role in the accurate measurement of brain subcortical structures
in macaques, which is crucial for unraveling the complexities of brain
structure and function, thereby enhancing our understanding of
neurodegenerative diseases and brain development. However, due to
significant differences in brain size, structure, and imaging
characteristics between humans and macaques, computational tools
developed for human neuroimaging studies often encounter obstacles when
applied to macaques. In this context, we propose an Anatomy Attentional
Fusion Network (AAF-Net), which integrates multimodal MRI data with
anatomical constraints in a multi-scale framework to address the
challenges posed by the dynamic development, regional heterogeneity, and
age-related size variations of the juvenile macaque brain, thus
achieving precise subcortical segmentation. Specifically, we generate a
Signed Distance Map (SDM) based on the initial rough segmentation of the
subcortical region by a network as an anatomical constraint, providing
comprehensive information on positions, structures, and morphology. Then
we construct AAF-Net to fully fuse the SDM anatomical constraints and
multimodal images for refined segmentation. To thoroughly evaluate the
performance of our proposed tool, over 700 macaque MRIs from 19 datasets
were used in this study. Specifically, we employed two manually labeled
longitudinal macaque datasets to develop the tool and complete four-fold
cross-validations. Furthermore, we incorporated various external
datasets to demonstrate the proposed tool's generalization capabilities
and promise in brain development research. We have made this tool
available as an open-source resource at
https://github.com/TaoZhong11/Macaque_subcortical_segmentation for
direct application.
- >-
[YEAR_RANGE] 2020-2024 [TEXT] CONTEXT: Prognostic prediction models
(PPMs) can help clinicians predict outcomes. OBJECTIVE: To critically
examine peer-reviewed PPMs predicting delayed recovery among pediatric
patients with concussion. DATA SOURCES: Ovid Medline, Embase, Ovid
PsycInfo, Web of Science Core Collection, Cumulative Index to Nursing
and Allied Health Literature, Cochrane Library, Google Scholar. STUDY
SELECTION: The study had to report a PPM for pediatric patients to be
used within 28 days of injury to estimate risk of delayed recovery at 28
days to 1 year postinjury. Studies had to have at least 30 participants.
DATA EXTRACTION: The Critical Appraisal and Data Extraction for
Systematic Reviews of Prediction Modeling Studies checklist was
completed. RESULTS: Six studies of 13 PPMs were included. These studies
primarily reflected male patients in late childhood or early adolescence
presenting to an emergency department meeting the Concussion in Sport
Group concussion criteria. No study authors used the same outcome
definition nor evaluated the clinical utility of a model. All studies
demonstrated high risk of bias. Quality of evidence was best for the
Predicting and Preventing Postconcussive Problems in Pediatrics (5P)
clinical risk score. LIMITATIONS: No formal PPM Grading of
Recommendations, Assessment, Development, and Evaluations (GRADE)
process exists. CONCLUSIONS: The 5P clinical risk score may be
considered for clinical use. Rigorous external validations, particularly
in other settings, are needed. The remaining PPMs require external
validation. Lack of consensus regarding delayed recovery criteria limits
these PPMs.
- source_sentence: >-
[YEAR_RANGE] 2020-2024 [TEXT] Intraoperative Monitoring of the External
Urethral Sphincter Reflex: A Novel Adjunct to Bulbocavernosus Reflex
Neuromonitoring for Protecting the Sacral Neural Pathways Responsible for
Urination, Defecation and Sexual Function.
sentences:
- >-
[YEAR_RANGE] 2020-2024 [TEXT] Early menarche has been associated with
adverse health outcomes, such as depressive symptoms. Discovering effect
modifiers across these conditions in the pediatric population is a
constant challenge. We tested whether movement behaviours modified the
effect of the association between early menarche and depression symptoms
among adolescents. This cross-sectional study included 2031 females aged
15-19 years across all Brazilian geographic regions. Data were collected
using a self-administered questionnaire; 30.5% (n = 620) reported having
experienced menarche before age 12 years (that is, early menarche). We
used the Patient Health Questionnaire (PHQ-9) to evaluate depressive
symptoms. Accruing any moderate-vigorous physical activity during
leisure time, limited recreational screen time, and having good sleep
quality were the exposures investigated. Adolescents who experienced
early menarche and met one (B: -4.45, 95% CI: (-5.38, -3.51)), two (B:
-6.07 (-7.02, -5.12)), or three (B: -6.49 (-7.76, -5.21)), and
adolescents who experienced not early menarche and met one (B: -5.33
(-6.20; -4.46)), two (B: -6.12 (-6.99; -5.24)), or three (B: -6.27
(-7.30; -5.24)) of the movement behaviour targets had lower PHQ-9 scores
for depression symptoms than adolescents who experienced early menarche
and did not meet any of the movement behaviours. The disparities in
depressive symptoms among the adolescents (early menarche versus not
early menarche) who adhered to all three target behaviours were not
statistically significant (B: 0.41 (-0.19; 1.01)). Adherence to movement
behaviours modified the effect of the association between early menarche
and depression symptoms.
- >-
[YEAR_RANGE] 2020-2024 [TEXT] PURPOSE: Intraoperative bulbocavernosus
reflex neuromonitoring has been utilized to protect bowel, bladder, and
sexual function, providing a continuous functional assessment of the
somatic sacral nervous system during surgeries where it is at risk.
Bulbocavernosus reflex data may also provide additional functional
insight, including an evaluation for spinal shock, distinguishing upper
versus lower motor neuron injury (conus versus cauda syndromes) and
prognosis for postoperative bowel and bladder function. Continuous
intraoperative bulbocavernosus reflex monitoring has been utilized to
provide the surgeon with an ongoing functional assessment of the
anatomical elements involved in the S2-S4 mediated reflex arc including
the conus, cauda equina and pudendal nerves. Intraoperative
bulbocavernosus reflex monitoring typically includes the electrical
activation of the dorsal nerves of the genitals to initiate the afferent
component of the reflex, followed by recording the resulting muscle
response using needle electromyography recordings from the external anal
sphincter. METHODS: Herein we describe a complementary and novel
technique that includes recording electromyography responses from the
external urethral sphincter to monitor the external urethral sphincter
reflex. Specialized foley catheters embedded with recording electrodes
have recently become commercially available that provide the ability to
perform intraoperative external urethral sphincter muscle recordings.
RESULTS: We describe technical details and the potential utility of
incorporating external urethral sphincter reflex recordings into
existing sacral neuromonitoring paradigms to provide redundant yet
complementary data streams. CONCLUSIONS: We present two illustrative
neurosurgical oncology cases to demonstrate the utility of the external
urethral sphincter reflex technique in the setting of the necessary
surgical sacrifice of sacral nerve roots.
- >-
[YEAR_RANGE] 2020-2024 [TEXT] BACKGROUND: Limited data are available on
the appropriate choice of blood pressure management strategy for
patients with acute basilar artery occlusion assessed by the standard
deviation (SD). Multivariate logistic models were used to investigate
the association between BPV, the primary outcome (futile recanalization,
90-day modified Rankin Scale score 3-6), and the secondary outcome
(30-day mortality). Subgroup analysis was performed as a sensitivity
test. RESULTS: Futile recanalization occurred in 60 (56 %) patients,
while 26 (24 %) patients died within 30 days. In the fully adjusted
model, MAP SD was associated with a higher risk of futile recanalization
(OR adj=1.36, per 1 mmHg increase, 95 % CI: 1.09-1.69, P=0.006) and
30-day mortality (OR adj=1.56, per 1 mmHg increase, 95 % CI: 1.20-2.04,
P=0.001). A significant interaction between MAP SD and the lack of
hypertension history on futile recanalization (P<0.05) was observed.
CONCLUSIONS: Among recanalized acute BAO ischemic patients, higher blood
pressure variability during the first 24 h after MT was associated with
worse outcomes. This association was stronger in patients without a
history of hypertension.
SentenceTransformer
This is a sentence-transformers model trained on the parquet dataset. It maps sentences & paragraphs to a 512-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
- Maximum Sequence Length: 512 tokens
- Output Dimensionality: 512 tokens
- Similarity Function: Cosine Similarity
- Training Dataset:
- parquet
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: BertModel
(1): Pooling({'word_embedding_dimension': 512, '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("pankajrajdeo/Bioformer-8L-UMLS-Pubmed-ST-TCE-Epoch-1")
# Run inference
sentences = [
'[YEAR_RANGE] 2020-2024 [TEXT] Intraoperative Monitoring of the External Urethral Sphincter Reflex: A Novel Adjunct to Bulbocavernosus Reflex Neuromonitoring for Protecting the Sacral Neural Pathways Responsible for Urination, Defecation and Sexual Function.',
'[YEAR_RANGE] 2020-2024 [TEXT] PURPOSE: Intraoperative bulbocavernosus reflex neuromonitoring has been utilized to protect bowel, bladder, and sexual function, providing a continuous functional assessment of the somatic sacral nervous system during surgeries where it is at risk. Bulbocavernosus reflex data may also provide additional functional insight, including an evaluation for spinal shock, distinguishing upper versus lower motor neuron injury (conus versus cauda syndromes) and prognosis for postoperative bowel and bladder function. Continuous intraoperative bulbocavernosus reflex monitoring has been utilized to provide the surgeon with an ongoing functional assessment of the anatomical elements involved in the S2-S4 mediated reflex arc including the conus, cauda equina and pudendal nerves. Intraoperative bulbocavernosus reflex monitoring typically includes the electrical activation of the dorsal nerves of the genitals to initiate the afferent component of the reflex, followed by recording the resulting muscle response using needle electromyography recordings from the external anal sphincter. METHODS: Herein we describe a complementary and novel technique that includes recording electromyography responses from the external urethral sphincter to monitor the external urethral sphincter reflex. Specialized foley catheters embedded with recording electrodes have recently become commercially available that provide the ability to perform intraoperative external urethral sphincter muscle recordings. RESULTS: We describe technical details and the potential utility of incorporating external urethral sphincter reflex recordings into existing sacral neuromonitoring paradigms to provide redundant yet complementary data streams. CONCLUSIONS: We present two illustrative neurosurgical oncology cases to demonstrate the utility of the external urethral sphincter reflex technique in the setting of the necessary surgical sacrifice of sacral nerve roots.',
'[YEAR_RANGE] 2020-2024 [TEXT] Early menarche has been associated with adverse health outcomes, such as depressive symptoms. Discovering effect modifiers across these conditions in the pediatric population is a constant challenge. We tested whether movement behaviours modified the effect of the association between early menarche and depression symptoms among adolescents. This cross-sectional study included 2031 females aged 15-19 years across all Brazilian geographic regions. Data were collected using a self-administered questionnaire; 30.5% (n = 620) reported having experienced menarche before age 12 years (that is, early menarche). We used the Patient Health Questionnaire (PHQ-9) to evaluate depressive symptoms. Accruing any moderate-vigorous physical activity during leisure time, limited recreational screen time, and having good sleep quality were the exposures investigated. Adolescents who experienced early menarche and met one (B: -4.45, 95% CI: (-5.38, -3.51)), two (B: -6.07 (-7.02, -5.12)), or three (B: -6.49 (-7.76, -5.21)), and adolescents who experienced not early menarche and met one (B: -5.33 (-6.20; -4.46)), two (B: -6.12 (-6.99; -5.24)), or three (B: -6.27 (-7.30; -5.24)) of the movement behaviour targets had lower PHQ-9 scores for depression symptoms than adolescents who experienced early menarche and did not meet any of the movement behaviours. The disparities in depressive symptoms among the adolescents (early menarche versus not early menarche) who adhered to all three target behaviours were not statistically significant (B: 0.41 (-0.19; 1.01)). Adherence to movement behaviours modified the effect of the association between early menarche and depression symptoms.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 512]
# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]
Training Details
Training Dataset
parquet
- Dataset: parquet
- Size: 26,147,930 training samples
- Columns:
anchorandpositive - Approximate statistics based on the first 1000 samples:
anchor positive type string string details - min: 16 tokens
- mean: 45.85 tokens
- max: 137 tokens
- min: 31 tokens
- mean: 268.19 tokens
- max: 512 tokens
- Samples:
anchor positive [YEAR_RANGE] 1880-1884 [TEXT] ADDRESS OF COL. GARRICK MALLERY, U. S. ARMY.[YEAR_RANGE] 1880-1884 [TEXT] It may be conceded that after man had all his present faculties, he did not choose between the adoption of voice and gesture, and never with those faculties, was in a state where the one was used, to the absolute exclusion of the other. The epoch, however, to which our speculations relate is that in which he had not reached the present symmetric development of his intellect and of his bodily organs, and the inquiry is: Which mode of communication was earliest adopted to his single wants and informed intelligence? With the voice he could imitate distinictively but few sounds of nature, while with gesture he could exhibit actions, motions, positions, forms, dimensions, directions and distances, with their derivations and analogues. It would seem from this unequal division of capacity that oral speech remained rudimentary long after gesture had become an efficient mode of communication. With due allowance for all purely imitative sounds, and for the spontaneous action of vocal organs under excitement, it appears that the connection between ideas and words is only to be explained by a compact between speaker and hearer which supposes the existence of a prior mode of communication. This was probably by gesture. At least we may accept it as a clew leading out of the labyrinth of philological confusion, and regulating the immemorial quest of man's primitive speech.[YEAR_RANGE] 1880-1884 [TEXT] How TO OBTAIN THE BRAIN OF THE CAT.[YEAR_RANGE] 1880-1884 [TEXT] How to obtain the Brain of the Cat, (Wilder).-Correction: Page 158, second column, line 7, "grains," should be "grams;" page 159, near middle of 2nd column, "successily," should be "successively;" page 161, the number of Flower's paper is 3.[YEAR_RANGE] 1880-1884 [TEXT] DOLBEAR ON THE NATURE AND CONSTITUTION OF MATTER.[YEAR_RANGE] 1880-1884 [TEXT] Mr. Dopp desires to make the following correction in his paper in the last issue: "In my article on page 200 of "Science", the expression and should have been and being the velocity of light. - Loss:
MultipleNegativesRankingLosswith these parameters:{ "scale": 20.0, "similarity_fct": "cos_sim" }
Evaluation Dataset
parquet
- Dataset: parquet
- Size: 26,147,930 evaluation samples
- Columns:
anchorandpositive - Approximate statistics based on the first 1000 samples:
anchor positive type string string details - min: 15 tokens
- mean: 31.78 tokens
- max: 78 tokens
- min: 16 tokens
- mean: 299.97 tokens
- max: 512 tokens
- Samples:
anchor positive [YEAR_RANGE] 2020-2024 [TEXT] Solubility and thermodynamics of mesalazine in aqueous mixtures of poly ethylene glycol 200/600 at 293.2-313.2K.[YEAR_RANGE] 2020-2024 [TEXT] In this study, the solubility of mesalazine was investigated in binary solvent mixtures of poly ethylene glycols 200/600 and water at temperatures ranging from 293.2K to 313.2K. The solubility of mesalazine was determined using a shake-flask method, and its concentrations were measured using a UV-Vis spectrophotometer. The obtained solubility data were analyzed using mathematical models including the van't Hoff, Jouyban-Acree, Jouyban-Acree-van't Hoff, mixture response surface, and modified Wilson models. The experimental data obtained for mesalazine dissolution encompassed various thermodynamic properties, including ΔG°, ΔH°, ΔS°, and TΔS°. These properties offer valuable insights into the energetic aspects of the dissolution process and were calculated based on the van't Hoff equation.[YEAR_RANGE] 2020-2024 [TEXT] Safety and efficacy of remimazolam versus propofol during EUS: a multicenter randomized controlled study.[YEAR_RANGE] 2020-2024 [TEXT] BACKGROUND AND AIMS: Propofol, a widely used sedative in GI endoscopic procedures, is associated with cardiorespiratory suppression. Remimazolam is a novel ultrashort-acting benzodiazepine sedative with rapid onset and minimal cardiorespiratory depression. This study compared the safety and efficacy of remimazolam and propofol during EUS procedures. METHODS: A multicenter randomized controlled study was conducted between October 2022 and March 2023 in patients who underwent EUS procedures. Patients were randomly assigned to receive either remimazolam or propofol as a sedative agent. The primary endpoint was cardiorespiratory adverse events.[YEAR_RANGE] 2020-2024 [TEXT] Ultrasound-Guided Vs Non-Guided Prolotherapy for Internal Derangement of Temporomandibular Joint. A Randomized Clinical Trial.[YEAR_RANGE] 2020-2024 [TEXT] OBJECTIVES: This randomized clinical trial study aims to compare ultrasound-guided versus non-guided Dextrose 10% injections in patients suffering from internal derangement in the temporomandibular joint (TMJ). MATERIAL AND METHODS: The study population included 22 patients and 43 TMJs suffering from unilateral or bilateral TMJ painful clicking, magnetic resonance imaging (MRI) proved disc displacement with reduction (DDWR), refractory to or failed conservative treatment. The patients were divided randomly into two groups (non-guided and ultrasound (US)-guided groups). The procedure involved injection of 2 mL solution of a mixture of 0.75 mL 0.9% normal saline solution, 0.3 mL 2% lidocaine and 0.75 mL dextrose 10% using a 25G needle in the joint and 1 mL intramuscular injection to the masseter muscle at the most tender point. The Visual Analogue Score (VAS) was used to compare joint pain intensity over four different periods, beginning with pre-injection, 1-, 2-, and 6-months postinjection. RESULTS: Twenty-two patients 5 males (n = 5/22, 22.7%) and 17 females (n = 17/22, 77.2%) were included in this study. The mean age was 27.3 ± 7.4 years (30.2 ± 7.0) for the non-guided group and 24.3 ± 6.9 for the US-guided group. The dextrose injection reduced intensity over time in both groups with statistically significant improvement (P value <.05) at 2 and 6 months in both groups. There was no statistically significant difference in VAS assessment between both groups. CONCLUSION: Intra-articular injection of dextrose 10% for patients with painful clicking and DDWR resulted in reduced pain intensity in both US-guided and non-guided groups with significant symptomatic improvement over time in both groups. US guidance allowed accurate anatomical localization and safe procedure with a single joint puncture. - Loss:
MultipleNegativesRankingLosswith these parameters:{ "scale": 20.0, "similarity_fct": "cos_sim" }
Training Hyperparameters
Non-Default Hyperparameters
eval_strategy: stepsper_device_train_batch_size: 128learning_rate: 2e-05num_train_epochs: 1max_steps: 194066log_level: infofp16: Truedataloader_num_workers: 16load_best_model_at_end: Trueresume_from_checkpoint: True
All Hyperparameters
Click to expand
overwrite_output_dir: Falsedo_predict: Falseeval_strategy: stepsprediction_loss_only: Trueper_device_train_batch_size: 128per_device_eval_batch_size: 8per_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: 194066lr_scheduler_type: linearlr_scheduler_kwargs: {}warmup_ratio: 0.0warmup_steps: 0log_level: infolog_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: Falseuse_ipex: Falsebf16: Falsefp16: Truefp16_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: 16dataloader_prefetch_factor: Nonepast_index: -1disable_tqdm: Falseremove_unused_columns: Truelabel_names: Noneload_best_model_at_end: Trueignore_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}deepspeed: Nonelabel_smoothing_factor: 0.0optim: adamw_torchoptim_args: Noneadafactor: Falsegroup_by_length: Falselength_column_name: lengthddp_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: Truehub_model_id: Nonehub_strategy: every_savehub_private_repo: Falsehub_always_push: Falsegradient_checkpointing: Falsegradient_checkpointing_kwargs: Noneinclude_inputs_for_metrics: Falseeval_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: Nonedispatch_batches: Nonesplit_batches: Noneinclude_tokens_per_second: Falseinclude_num_input_tokens_seen: Falseneftune_noise_alpha: Noneoptim_target_modules: Nonebatch_eval_metrics: Falseeval_on_start: Falseeval_use_gather_object: Falsebatch_sampler: batch_samplermulti_dataset_batch_sampler: proportional
Training Logs
Click to expand
| Epoch | Step | Training Loss | Validation Loss |
|---|---|---|---|
| 0.0000 | 1 | 5.6379 | - |
| 0.0052 | 1000 | 0.6085 | - |
| 0.0103 | 2000 | 0.1263 | - |
| 0.0155 | 3000 | 0.1246 | - |
| 0.0206 | 4000 | 0.1225 | - |
| 0.0258 | 5000 | 0.0909 | - |
| 0.0309 | 6000 | 0.1011 | - |
| 0.0361 | 7000 | 0.0879 | - |
| 0.0412 | 8000 | 0.0858 | - |
| 0.0464 | 9000 | 0.0854 | - |
| 0.0515 | 10000 | 0.1016 | - |
| 0.0567 | 11000 | 0.0919 | - |
| 0.0618 | 12000 | 0.066 | - |
| 0.0670 | 13000 | 0.0947 | - |
| 0.0721 | 14000 | 0.1017 | - |
| 0.0773 | 15000 | 0.0622 | - |
| 0.0824 | 16000 | 0.1529 | - |
| 0.0876 | 17000 | 0.065 | - |
| 0.0928 | 18000 | 0.104 | - |
| 0.0979 | 19000 | 0.0626 | - |
| 0.1031 | 20000 | 0.1228 | - |
| 0.1082 | 21000 | 0.0681 | - |
| 0.1134 | 22000 | 0.1211 | - |
| 0.1185 | 23000 | 0.0608 | - |
| 0.1237 | 24000 | 0.0733 | - |
| 0.1288 | 25000 | 0.0587 | - |
| 0.1340 | 26000 | 0.1249 | - |
| 0.1391 | 27000 | 0.0662 | - |
| 0.1443 | 28000 | 0.1328 | - |
| 0.1494 | 29000 | 0.0647 | - |
| 0.1546 | 30000 | 0.1371 | - |
| 0.1597 | 31000 | 0.0603 | - |
| 0.1649 | 32000 | 0.1005 | - |
| 0.1700 | 33000 | 0.1037 | - |
| 0.1752 | 34000 | 0.0688 | - |
| 0.1804 | 35000 | 0.138 | - |
| 0.1855 | 36000 | 0.0648 | - |
| 0.1907 | 37000 | 0.0709 | - |
| 0.1958 | 38000 | 0.1214 | - |
| 0.2010 | 39000 | 0.0728 | - |
| 0.2061 | 40000 | 0.1218 | - |
| 0.2113 | 41000 | 0.0804 | - |
| 0.2164 | 42000 | 0.0682 | - |
| 0.2216 | 43000 | 0.1122 | - |
| 0.2267 | 44000 | 0.0692 | - |
| 0.2319 | 45000 | 0.0777 | - |
| 0.2370 | 46000 | 0.0947 | - |
| 0.2422 | 47000 | 0.0623 | - |
| 0.2473 | 48000 | 0.0785 | - |
| 0.2525 | 49000 | 0.0862 | - |
| 0.2576 | 50000 | 0.0703 | - |
| 0.2628 | 51000 | 0.0614 | - |
| 0.2679 | 52000 | 0.063 | - |
| 0.2731 | 53000 | 0.086 | - |
| 0.2783 | 54000 | 0.0763 | - |
| 0.2834 | 55000 | 0.0536 | - |
| 0.2886 | 56000 | 0.0634 | - |
| 0.2937 | 57000 | 0.0598 | - |
| 0.2989 | 58000 | 0.063 | - |
| 0.3040 | 59000 | 0.057 | - |
| 0.3092 | 60000 | 0.0733 | - |
| 0.3143 | 61000 | 0.0447 | - |
| 0.3195 | 62000 | 0.0504 | - |
| 0.3246 | 63000 | 0.0558 | - |
| 0.3298 | 64000 | 0.0669 | - |
| 0.3349 | 65000 | 0.0664 | - |
| 0.3401 | 66000 | 0.0519 | - |
| 0.3452 | 67000 | 0.0396 | - |
| 0.3504 | 68000 | 0.0531 | - |
| 0.3555 | 69000 | 0.069 | - |
| 0.3607 | 70000 | 0.0793 | - |
| 0.3659 | 71000 | 0.0873 | - |
| 0.3710 | 72000 | 0.034 | - |
| 0.3762 | 73000 | 0.0428 | - |
| 0.3813 | 74000 | 0.0536 | - |
| 0.3865 | 75000 | 0.0433 | - |
| 0.3916 | 76000 | 0.1024 | - |
| 0.3968 | 77000 | 0.051 | - |
| 0.4019 | 78000 | 0.0672 | - |
| 0.4071 | 79000 | 0.0343 | - |
| 0.4122 | 80000 | 0.0597 | - |
| 0.4174 | 81000 | 0.074 | - |
| 0.4225 | 82000 | 0.0421 | - |
| 0.4277 | 83000 | 0.0463 | - |
| 0.4328 | 84000 | 0.0651 | - |
| 0.4380 | 85000 | 0.0292 | - |
| 0.4431 | 86000 | 0.0403 | - |
| 0.4483 | 87000 | 0.0459 | - |
| 0.4535 | 88000 | 0.0705 | - |
| 0.4586 | 89000 | 0.0577 | - |
| 0.4638 | 90000 | 0.0462 | - |
| 0.4689 | 91000 | 0.0342 | - |
| 0.4741 | 92000 | 0.0402 | - |
| 0.4792 | 93000 | 0.0472 | - |
| 0.4844 | 94000 | 0.0472 | - |
| 0.4895 | 95000 | 0.0843 | - |
| 0.4947 | 96000 | 0.031 | - |
| 0.4998 | 97000 | 0.0347 | - |
| 0.5050 | 98000 | 0.0326 | - |
| 0.5101 | 99000 | 0.0349 | - |
| 0.5153 | 100000 | 0.0276 | - |
| 0.5204 | 101000 | 0.0412 | - |
| 0.5256 | 102000 | 0.0445 | - |
| 0.5307 | 103000 | 0.0232 | - |
| 0.5359 | 104000 | 0.0301 | - |
| 0.5411 | 105000 | 0.0268 | - |
| 0.5462 | 106000 | 0.0215 | - |
| 0.5514 | 107000 | 0.0538 | - |
| 0.5565 | 108000 | 0.0475 | - |
| 0.5617 | 109000 | 0.0302 | - |
| 0.5668 | 110000 | 0.0397 | - |
| 0.5720 | 111000 | 0.0438 | - |
| 0.5771 | 112000 | 0.0416 | - |
| 0.5823 | 113000 | 0.0289 | - |
| 0.5874 | 114000 | 0.0374 | - |
| 0.5926 | 115000 | 0.0514 | - |
| 0.5977 | 116000 | 0.0321 | - |
| 0.6029 | 117000 | 0.0759 | - |
| 0.6080 | 118000 | 0.0433 | - |
| 0.6132 | 119000 | 0.0362 | - |
| 0.6183 | 120000 | 0.041 | - |
| 0.6235 | 121000 | 0.0306 | - |
| 0.6286 | 122000 | 0.0407 | - |
| 0.6338 | 123000 | 0.0393 | - |
| 0.6390 | 124000 | 0.0478 | - |
| 0.6441 | 125000 | 0.0429 | - |
| 0.6493 | 126000 | 0.0521 | - |
| 0.6544 | 127000 | 0.0442 | - |
| 0.6596 | 128000 | 0.0428 | - |
| 0.6647 | 129000 | 0.0346 | - |
| 0.6699 | 130000 | 0.027 | - |
| 0.6750 | 131000 | 0.0219 | - |
| 0.6802 | 132000 | 0.0417 | - |
| 0.6853 | 133000 | 0.0376 | - |
| 0.6905 | 134000 | 0.0341 | - |
| 0.6956 | 135000 | 0.0304 | - |
| 0.7008 | 136000 | 0.0375 | - |
| 0.7059 | 137000 | 0.0352 | - |
| 0.7111 | 138000 | 0.0413 | - |
| 0.7162 | 139000 | 0.0346 | - |
| 0.7214 | 140000 | 0.0437 | - |
| 0.7266 | 141000 | 0.0447 | - |
| 0.7317 | 142000 | 0.0335 | - |
| 0.7369 | 143000 | 0.0287 | - |
| 0.7420 | 144000 | 0.0271 | - |
| 0.7472 | 145000 | 0.0295 | - |
| 0.7523 | 146000 | 0.0238 | - |
| 0.7575 | 147000 | 0.03 | - |
| 0.7626 | 148000 | 0.0369 | - |
| 0.7678 | 149000 | 0.0351 | - |
| 0.7729 | 150000 | 0.0302 | - |
| 0.7781 | 151000 | 0.0369 | - |
| 0.7832 | 152000 | 0.0374 | - |
| 0.7884 | 153000 | 0.0298 | - |
| 0.7935 | 154000 | 0.0387 | - |
| 0.7987 | 155000 | 0.0344 | - |
| 0.8038 | 156000 | 0.0374 | - |
| 0.8090 | 157000 | 0.0338 | - |
| 0.8142 | 158000 | 0.0317 | - |
| 0.8193 | 159000 | 0.0378 | - |
| 0.8245 | 160000 | 0.0375 | - |
| 0.8296 | 161000 | 0.0276 | - |
| 0.8348 | 162000 | 0.0238 | - |
| 0.8399 | 163000 | 0.0178 | - |
| 0.8451 | 164000 | 0.0226 | - |
| 0.8502 | 165000 | 0.0233 | - |
| 0.8554 | 166000 | 0.0183 | - |
| 0.8605 | 167000 | 0.0253 | - |
| 0.8657 | 168000 | 0.0232 | - |
| 0.8708 | 169000 | 0.0252 | - |
| 0.8760 | 170000 | 0.0258 | - |
| 0.8811 | 171000 | 0.0256 | - |
| 0.8863 | 172000 | 0.0285 | - |
| 0.8914 | 173000 | 0.0279 | - |
| 0.8966 | 174000 | 0.0237 | - |
| 0.9018 | 175000 | 0.0277 | - |
| 0.9069 | 176000 | 0.0245 | - |
| 0.9121 | 177000 | 0.0282 | - |
| 0.9172 | 178000 | 0.0248 | - |
| 0.9224 | 179000 | 0.0212 | - |
| 0.9275 | 180000 | 0.0199 | - |
| 0.9327 | 181000 | 0.0236 | - |
| 0.9378 | 182000 | 0.0246 | - |
| 0.9430 | 183000 | 0.0225 | - |
| 0.9481 | 184000 | 0.0228 | - |
| 0.9533 | 185000 | 0.0235 | - |
| 0.9584 | 186000 | 0.0226 | - |
| 0.9636 | 187000 | 0.0201 | - |
| 0.9687 | 188000 | 0.0201 | - |
| 0.9739 | 189000 | 0.0221 | - |
| 0.9790 | 190000 | 0.022 | - |
| 0.9842 | 191000 | 0.0359 | - |
| 0.9893 | 192000 | 0.0242 | - |
| 0.9945 | 193000 | 0.0192 | - |
| 0.9997 | 194000 | 0.0208 | - |
| 1.0000 | 194066 | - | 0.0007 |
Framework Versions
- Python: 3.12.2
- Sentence Transformers: 3.2.1
- Transformers: 4.44.2
- PyTorch: 2.5.0
- Accelerate: 1.0.1
- Datasets: 3.0.2
- Tokenizers: 0.19.1
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}
}