README.md

metadata

tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:33870508
  - loss:MultipleNegativesRankingLoss
widget:
  - source_sentence: >-
      Physical Behavior Profiles Among Older Adults and Their Associations With
      Physical Capacity and Life-Space Mobility.
    sentences:
      - >-
        Injectable hydrogel-based materials have emerged as promising
        alendronate (ALN) delivery systems for the treatment of osteoporosis.
        However, their intrinsic permeability limits the sustained delivery of
        small-molecule drugs. In response to this challenge, we present the
        multifunctional hybrids composed of mesoporous silica particles
        decorated with hydroxyapatite and loaded with alendronate
        (MSP-NH2-HAp-ALN), which are immobilized in collagen/chitosan/hyaluronic
        acid-based hydrogel. We have mainly focused on the biological in
        vitro/ex vivo evaluation of developed composites. It was found that the
        extracts released from tested systems do not exhibit hemolytic
        properties and are safe for blood elements and the human liver cell
        model. The resulting materials create an environment conducive to
        differentiating human bone marrow mesenchymal stem cells and reduce the
        viability of osteoclast precursors (RAW 264.7). Importantly, even the
        system with the lowest concentration of ALN caused a substantial
        cytotoxic effect on RAW 264.7 cells; their viability decreased to 20 %
        and 10 % of control on 3 and 7 day of culture. Additionally, prolonged
        ALN release (up to 20 days) with minimized burst release was observed,
        while material features (wettability, swellability, degradation,
        mechanical properties) depended on MSP-NH2-HAp-ALN content. The obtained
        data indicate that developed composites establish a high-potential
        formulation for safe and effective osteoporosis therapy.
      - >-
        We identified data-driven multidimensional physical activity (PA)
        profiles using several novel accelerometer-derived metrics. Participants
        aged 75, 80, and 85 (n = 441) wore triaxial accelerometers for 3-7 days.
        PA profiles were formed with k-means cluster analysis based on PA
        minutes, intensity, fragmentation, sit-to-stand transitions, and gait
        bouts for men and women. Associations with physical capacity and
        life-space mobility were examined using age-adjusted general linear
        models. Three profiles emerged: "Exercisers" and "actives" accumulated
        relatively high PA minutes, with actives engaging in lighter intensity
        PA. "Inactives" had the highest activity fragmentation and lowest PA
        volume, intensity, and gait bouts. Inactives showed lower scores in
        physical capacity and life-space mobility compared with exercisers and
        actives. Exercisers and actives had similar physical capacity and
        life-space mobility, except female exercisers had higher walking speed
        in the 6-min walk test. Our findings demonstrate the importance of
        assessing PA as multidimensional behavior rather than focusing on a
        single metric.
      - >-
        Existing exoskeletons for pediatric gait assistance have limitations in
        anthropometric design, structure weight, cost, user safety features, and
        adaptability to diverse users. Additionally, creating precise models for
        pediatric rehabilitation is difficult because the rapid anthropometric
        changes in children result in unknown model parameters. Furthermore,
        external disruptions, like unpredictable movements and involuntary
        muscle contractions, add complexity to the control schemes that need to
        be managed. To overcome these limitations, this study aims to develop an
        affordable stand-aided lower-limb exoskeleton specifically for pediatric
        subjects (8-12 years, 25-40 kg, 128-132 cm) in passive-assist mode. The
        authors modified a previously developed model (LLESv1) for improved
        rigidity, reduced mass, simplified motor arrangement, variable waist
        size, and enhanced mobility. A computer-aided design of the new
        exoskeleton system (LLESv2) is presented. The developed prototype of the
        exoskeleton appended with a pediatric subject (age: 12 years old, body
        mass: 40 kg, body height: 132 cm) is presented with real-time hardware
        architecture. Thereafter, an improved fast non-singular terminal sliding
        mode (IFNSTSM) control scheme is proposed, incorporating a double
        exponential reaching law for expedited error convergence and enhanced
        stability. The Lyapunov stability warrants the control system's
        performance despite uncertainties and disturbances. In contrast to fast
        non-singular terminal sliding mode (FNSTSM) control and time-scaling
        sliding mode (TSSM) control, experimental validation demonstrates the
        effectiveness of IFNSTSM control by a respective average of 5.39% and
        42.1% in tracking desired joint trajectories with minimal and rapid
        finite time converging errors. Moreover, the exoskeleton with the
        proposed IFNSTSM control requires significantly lesser control efforts
        than the exoskeleton using contrast FNSTSM control. The Bland-Altman
        analysis indicates that although there is a minimal mean difference in
        variables when employing FNSTSM and IFNSTSM controllers, the latter
        exhibits significant performance variations as the mean of variables
        changes. This research contributes to affordable and effective pediatric
        gait assistance, improving rehabilitation outcomes and enhancing
        mobility support.
  - source_sentence: >-
      Anatomo-functional basis of emotional and motor resonance elicited by
      facial expressions.
    sentences:
      - >-
        Simulation theories predict that the observation of other's expressions
        modulates neural activity in the same centers controlling their
        production. This hypothesis has been developed by two models,
        postulating that the visual input is directly projected either to the
        motor system for action recognition (motor resonance) or to
        emotional/interoceptive regions for emotional contagion and social
        synchronization (emotional resonance). Here we investigated the role of
        frontal/insular regions in the processing of observed emotional
        expressions by combining intracranial recording, electrical stimulation
        and effective connectivity. First, we intracranially recorded from
        prefrontal, premotor or anterior insular regions of 44 patients during
        the passive observation of emotional expressions, finding widespread
        modulations in prefrontal/insular regions (anterior cingulate cortex,
        anterior insula, orbitofrontal cortex and inferior frontal gyrus) and
        motor territories (rolandic operculum and inferior frontal junction).
        Subsequently, we electrically stimulated the activated sites, finding
        that (a) in the anterior cingulate cortex and anterior insula, the
        stimulation elicited emotional/interoceptive responses, as predicted by
        the 'emotional resonance model', (b) in the rolandic operculum it evoked
        face/mouth sensorimotor responses, in line with the 'motor resonance'
        model, and (c) all other regions were unresponsive or revealed functions
        unrelated to the processing of facial expressions. Finally, we traced
        the effective connectivity to sketch a network-level description of
        these regions, finding that the anterior cingulate cortex and the
        anterior insula are reciprocally interconnected while the rolandic
        operculum is part of the parieto-frontal circuits and poorly connected
        with the formers. These results support the hypothesis that the pathways
        hypothesized by the 'emotional resonance' and the 'motor resonance'
        models work in parallel, differing in terms of spatio-temporal
        fingerprints, reactivity to electrical stimulation and connectivity
        patterns.
      - >-
        STAC3-related myopathy, or Native American myopathy, and myopathic
        facies. Since the first description of NAM, more cases have been
        described worldwide, with three cases reported from the Middle East.
        This study presents a cohort of seven Saudi NAM patients belonging to
        three families. To our knowledge, this cohort is the largest to be
        reported in the Arabian Peninsula and the Middle Eastern region. We will
        also highlight the importance of considering this MH-causing disease
        preoperatively in myopathic children with cleft palate in areas where
        NAM has been described.
      - >-
        The Tibetan Plateau supplies water to nearly 2 billion people in Asia,
        but climate change poses threats to its aquatic microbial resources.
        Here, we construct the Tibetan Plateau Microbial Catalog by sequencing
        498 metagenomes from six water ecosystems (saline lakes, freshwater
        lakes, rivers, hot springs, wetlands and glaciers). Our catalog expands
        knowledge of regional genomic diversity by presenting 32,355
        metagenome-assembled genomes that de-replicated into 10,723
        representative genome-based species, of which 88% were unannotated. The
        catalog contains nearly 300 million non-redundant gene clusters, of
        which 15% novel, and 73,864 biosynthetic gene clusters, of which 50%
        novel, thus expanding known functional diversity. Using these data, we
        investigate the Tibetan Plateau aquatic microbiome's biogeography along
        a distance of 2,500 km and >5 km in altitude. Microbial compositional
        similarity and the shared gene count with the Tibetan Plateau microbiome
        decline along with distance and altitude difference, suggesting a
        dispersal pattern. The Tibetan Plateau Microbial Catalog stands as a
        substantial repository for high-altitude aquatic microbiome resources,
        providing potential for discovering novel lineages and functions, and
        bridging knowledge gaps in microbiome biogeography.
  - source_sentence: >-
      Effect of verbal cues on the coupling and stability of anti-phase bimanual
      coordination pattern in children with probable developmental coordination
      disorder.
    sentences:
      - >-
        BACKGROUND: Tobacco smoking remains a key cause of preventable illness
        and death globally. In response, many countries provide extensive
        services to help people to stop smoking by offering a variety of
        effective behavioural and pharmacological therapies. However, many
        people who wish to stop smoking do not have access to or use stop
        smoking supports, and new modes of support, including the use of
        financial incentives, are needed to address this issue. A realist review
        of published international literature was undertaken to understand how,
        why, for whom, and in which circumstances financial incentives
        contribute to success in stopping smoking for general population groups
        and among pregnant women. METHODS: Systematic searches were undertaken
        from inception to February 2022 of five academic databases: MEDLINE
        (ovid), Embase.com, CIHAHL, Scopus and PsycINFO. Study selection was
        inclusive of all study designs. Twenty-two studies were included. Using
        Pawson and Tilley's iterative realist review approach, data collected
        were screened, selected, coded, analysed, and synthesised into a set of
        explanatory theoretical findings. RESULTS: Data were synthesised into
        six Context-Mechanism-Outcome Configurations and one overarching
        programme theory after iterative rounds of analysis, team discussion,
        and expert panel feedback. Our programme theory shows that financial
        incentives are particularly useful to help people stop smoking if they
        have a financial need, are pregnant or recently post-partum, have a high
        threshold for behaviour change, and/or respond well to external rewards.
        The incentives work through a number of mechanisms including the role
        their direct monetary value can play in a person's life and through a
        process of reinforcement where they can help build confidence and
        self-esteem. CONCLUSION: This is the first realist review to synthesise
        how, why, and for whom financial incentives work among those attempting
        to stop smoking, adding to the existing evidence demonstrating their
        efficacy. The findings will support the implementation of current
        knowledge into effective programmes which can enhance the impact of stop
        smoking care. PROSPERO REGISTRATION NUMBER: CRD42022298941.
      - >-
        We developed a synthetic method for obtaining 4,5-disubstituted
        2-(pyridin-2-yl)oxazoles from picolinamide and aldehydes by employing
        Pd(TFA)2 as the catalyst in n-octane. This cascade reaction involves the
        condensation of picolinamide and two aldehyde molecules promoted by
        trifluoroacetic acid (TFA) generated in situ from Pd(TFA)2. This one-pot
        protocol provides rapid access to synthetically valuable triaryloxazoles
        from readily available starting materials under mild conditions. An 18O
        labeling study revealed that this tandem reaction proceeded via a
        different reaction mechanism compared to the Robinson-Gabriel oxazole
        synthesis.
      - >-
        The study of the emergence and stability of bimanual and interlimb
        coordination patterns in children with Developmental Coordination
        Disorder (DCD) has shown that they encounter greater difficulties in
        coupling their limbs compared to typically developing (TD) children.
        Verbal cues have been identified as strategies to direct children's
        attention to more relevant task information, thus potentially improving
        motor performance. Consequently, this study investigated the effect of
        providing verbal cues on the execution of bimanual tasks in children
        with and without probable DCD. Twenty-eight children aged 9-10, matched
        by age and gender, were divided into two groups: pDCD and TD. The
        children performed bilateral trajectory movements with both hands
        (horizontal back-and-forth), holding a pen on a tablet, in anti-phase
        (180°) coordination pattern, in two conditions: No cues and Verbal cues.
        In the last condition, children received verbal cues to maintain the
        anti-phase pattern even with an increase in hand oscillation frequency.
        Relative phase and variability of relative phase between the hands were
        calculated for analysis of pattern coupling and stability. Hand cycles,
        movement amplitude, and tablet pressure force were calculated to analyze
        pattern control parameters. All these variables were compared between
        groups and conditions. The results indicated that despite the pDCD group
        showing greater variability in the anti-phase coordination pattern
        compared to the TD group, both groups performed better in the Verbal
        cues than the No cues condition. Furthermore, the pDCD group exhibited
        more hand movement cycles and applied greater pressure force compared to
        the TD group, suggesting different motor control strategies during the
        bimanual task. It is suggested that the use of verbal cues during
        bimanual task execution improves children's performance, potentially by
        promoting interaction between attention, as a cognitive function, and
        intrinsic coordination dynamics, thereby reducing variability in the
        perceptual-motor system.
  - source_sentence: >-
      Frailty efficacy as a predictor of clinical and cognitive complications in
      patients undergoing coronary artery bypass grafting: a prospective cohort
      study.
    sentences:
      - >-
        BACKGROUND: Frailty is proposed as a predictor of outcomes in patients
        undergoing major surgeries, although data on the association of frailty
        and coronary artery bypass grafting, cognitive function by Montreal
        Cognitive Assessment (MoCA), and depression by the Geriatric Depression
        Scale (GDS) were obtained. The incidence of adverse outcomes was
        investigated at the three-month follow-up. Outcomes between frail and
        non-frail groups were compared utilizing T-tests and Mann-Whitney U
        tests, as appropriate. RESULTS: We included 170 patients with a median
        age of 66 ± 4 years (75.3% male). Of these, 58 cases were classified as
        frail, and 112 individuals were non-frail, preoperatively. Frail
        patients demonstrated significantly worse baseline MOCA scores (21.08
        versus 22.41, P = 0.045), GDS (2.00 versus 1.00, P = 0.009), and Lawton
        IADL (8.00 versus 6.00, P < 0.001) compared to non-frail. According to
        3-month follow-up data, postoperative MOCA and GDS scores were
        comparable between the two groups, while Lawton IADL (8.00 versus 6.00,
        P < 0.001) was significantly lower in frail cases. A significantly
        higher rate of readmission (1.8% versus 12.1%), sepsis (7.1% versus
        19.0%), as well as a higher Euroscore (1.5 versus 1.9), was observed in
        the frail group. A mildly significantly more extended ICU stay (6.00
        versus 5.00, p = 0.051) was shown in the frail patient. CONCLUSION:
        Frailty showed a significant association with a worse preoperative
        independence level, cognitive function, and depression status, as well
        as increased postoperative complications.
      - >-
        OBJECTIVE: To assess presentation of neurosyphilis with a focus on the
        psychiatric aspects. METHOD: File review of the cases with a positive
        cerebrospinal fluid venereal disease research laboratory test between
        1999 to 2020. RESULTS: Medical records of 143 neurosyphilis patients
        were analysed. Hallucinations, delusions, and catatonia were the
        commonest psychiatric symptoms. Brain atrophy was the commonest
        neuroimaging finding. The number of neurosyphilis patients and the
        proportion with delirium or catatonia declined during the second decade.
        CONCLUSION: Atypical presentation of psychiatric symptoms around the
        fifth decade, with associated neurological symptoms or brain imaging
        changes, should prompt evaluation for neurosyphilis.
      - >-
        INTRODUCTION: Bibliometrics evaluates the quality of biomedical
        journals. The aim of this study was to compare the main bibliometric
        indexes of the official journals of scientific societies of Internal
        Medicine in Europe. MATERIAL AND METHODS: Bibliometric information was
        obtained from the Web of Science European Journal of Internal Medicine,
        which ranked in the first quartile (Q1) for JIF, CiteScore and JCI
        metrics, exceeding values of 1 in Normalized Eigenfactor and SNIP
        metrics; 2) Internal and Emergency Medicine, Q1 for CiteScore and JCI
        metrics, and with values >1 in Normalized EigenFactor and SNIP metrics;
        3) Polish Archives of Internal Medicine, Q1 for JCI metrics; 4) Revista
        Clínica Española, Q2 for JIF, CiteScore and JCI metrics; and 5) Acta
        Medica Belgica, Q2 for CiteScore and JCI metrics. These journals
        increased their impact metrics in the last 3 years, in parallel with the
        COVID pandemic. CONCLUSIONS: Five official journals of European Internal
        Medicine societies, including Revista Clínica Española, meet high
        quality standards.
  - source_sentence: >-
      De Garengeot Hernia, an acute appendicitis in the right femoral hernia
      canal, and successful management with transabdominal closure and
      appendectomy: a case Report.
    sentences:
      - >-
        With the increasing population worldwide more wastewater is created by
        human activities and discharged into the waterbodies. This is causing
        the contamination of aquatic bodies, thus disturbing the marine
        ecosystems. The rising population is also posing a challenge to meet the
        demands of fresh drinking water in the water-scarce regions of the
        world, where drinking water is made available to people by desalination
        process. The fouling of composite membranes remains a major challenge in
        water desalination. In this innovative study, we present a novel
        probabilistic approach to analyse and anticipate the predominant fouling
        mechanisms in the filtration process. Our establishment of a robust
        theoretical framework hinges upon the utilization of both the geometric
        law and the Hermia model, elucidating the concept of resistance in
        series (RIS). By manipulating the transmembrane pressure, we demonstrate
        effective management of permeate flux rate and overall product quality.
        Our investigations reveal a decrease in permeate flux in three distinct
        phases over time, with the final stage marked by a significant reduction
        due to the accumulation of a denser cake layer. Additionally, an
        increase in transmembrane pressure leads to a correlative rise in
        permeate flux, while also exerting negative effects such as membrane
        ruptures. Our study highlights the minimal immediate impact of the
        intermediate blocking mechanism (n = 1) on permeate flux, necessitating
        continuous monitoring for potential long-term effects. Additionally, we
        note a reduced membrane selectivity across all three fouling types (n =
        0, n = 1.5, n = 2). Ultimately, our findings indicate that the membrane
        undergoes complete fouling with a probability of P = 0.9 in the presence
        of all three fouling mechanisms. This situation renders the membrane
        unable to produce water at its previous flow rate, resulting in a
        significant reduction in the desalination plant's productivity. I have
        demonstrated that higher pressure values notably correlate with
        increased permeate flux across all four membrane types. This correlation
        highlights the significant role of TMP in enhancing the production rate
        of purified water or desired substances through membrane filtration
        systems. Our innovative approach opens new perspectives for water
        desalination management and optimization, providing crucial insights
        into fouling mechanisms and proposing potential strategies to address
        associated challenges.
      - >-
        Incarceration of the appendix within a femoral hernia is a rare
        condition of abdominal wall hernia about 0.1 to 0.5% in reported femoral
        hernia. We report a case of a 56-year-old female whose appendix was
        trapped in the right femoral canal. There are few reports in the
        literature on entrapment of the appendix within a femoral hernia. The
        management of this condition includes antibiotics, drainage
        appendectomy, hernioplasty and mesh repair.
      - >-
        INTRODUCTION: Globally, the prevalence of obesity tripled from 1975 to
        2016. There is evidence that air pollution may contribute to the obesity
        epidemic through an increase in oxidative stress and inflammation of
        adipose tissue. However, the impact of air pollution on body weight at a
        population level remains inconclusive. This systematic review and
        meta-analysis will estimate the association of ambient air pollution
        with obesity, distribution of ectopic adipose tissue, and the incidence
        and prevalence of non-alcoholic fatty liver disease among adults.
        METHODS AND ANALYSIS: The study will follow the Preferred Reporting
        Items for Systematic Reviews and Meta-Analyses guidelines for conduct
        and reporting. The search will include the following databases: Ovid
        Medline, Embase, PubMed, Web of Science and Latin America and the
        Caribbean Literature on Health Sciences, and will be supplemented by a
        grey literature search. Each article will be independently screened by
        two reviewers, and relevant data will be extracted independently and in
        duplicate. Study-specific estimates of associations and their 95%
        Confidence Intervals will be pooled using a DerSimonian and Laird
        random-effects model, implemented using the RevMan software. The I2
        statistic will be used to assess interstudy heterogeneity. The
        confidence in the body of evidence will be assessed using the Grading of
        Recommendations Assessment, Development and Evaluation (GRADE) approach.
        ETHICS AND DISSEMINATION: As per institutional policy, ethical approval
        is not required for secondary data analysis. In addition to being
        published in a peer-reviewed journal and presented at conferences, the
        results of the meta-analysis will be shared with key stakeholders,
        health policymakers and healthcare professionals. PROSPERO REGISTRATION
        NUMBER: CRD42023423955.
pipeline_tag: sentence-similarity
library_name: sentence-transformers

SentenceTransformer

This is a sentence-transformers model trained on the parquet dataset. It maps sentences & paragraphs to a 384-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: 1024 tokens
• Output Dimensionality: 384 dimensions
• 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': 1024, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, '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-16L-UMLS-Pubmed_PMC-Forward_TCE-Epoch-3")
# Run inference
sentences = [
    'De Garengeot Hernia, an acute appendicitis in the right femoral hernia canal, and successful management with transabdominal closure and appendectomy: a case Report.',
    'Incarceration of the appendix within a femoral hernia is a rare condition of abdominal wall hernia about 0.1 to 0.5% in reported femoral hernia. We report a case of a 56-year-old female whose appendix was trapped in the right femoral canal. There are few reports in the literature on entrapment of the appendix within a femoral hernia. The management of this condition includes antibiotics, drainage appendectomy, hernioplasty and mesh repair.',
    "With the increasing population worldwide more wastewater is created by human activities and discharged into the waterbodies. This is causing the contamination of aquatic bodies, thus disturbing the marine ecosystems. The rising population is also posing a challenge to meet the demands of fresh drinking water in the water-scarce regions of the world, where drinking water is made available to people by desalination process. The fouling of composite membranes remains a major challenge in water desalination. In this innovative study, we present a novel probabilistic approach to analyse and anticipate the predominant fouling mechanisms in the filtration process. Our establishment of a robust theoretical framework hinges upon the utilization of both the geometric law and the Hermia model, elucidating the concept of resistance in series (RIS). By manipulating the transmembrane pressure, we demonstrate effective management of permeate flux rate and overall product quality. Our investigations reveal a decrease in permeate flux in three distinct phases over time, with the final stage marked by a significant reduction due to the accumulation of a denser cake layer. Additionally, an increase in transmembrane pressure leads to a correlative rise in permeate flux, while also exerting negative effects such as membrane ruptures. Our study highlights the minimal immediate impact of the intermediate blocking mechanism (n = 1) on permeate flux, necessitating continuous monitoring for potential long-term effects. Additionally, we note a reduced membrane selectivity across all three fouling types (n = 0, n = 1.5, n = 2). Ultimately, our findings indicate that the membrane undergoes complete fouling with a probability of P = 0.9 in the presence of all three fouling mechanisms. This situation renders the membrane unable to produce water at its previous flow rate, resulting in a significant reduction in the desalination plant's productivity. I have demonstrated that higher pressure values notably correlate with increased permeate flux across all four membrane types. This correlation highlights the significant role of TMP in enhancing the production rate of purified water or desired substances through membrane filtration systems. Our innovative approach opens new perspectives for water desalination management and optimization, providing crucial insights into fouling mechanisms and proposing potential strategies to address associated challenges.",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# 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: 33,870,508 training samples
• Columns: anchor and positive
• Approximate statistics based on the first 1000 samples:

  	anchor	positive
  type	string	string
  details	min: 5 tokens mean: 36.24 tokens max: 106 tokens	min: 30 tokens mean: 328.76 tokens max: 1024 tokens

• Samples:

  anchor	positive
  How TO OBTAIN THE BRAIN OF THE CAT.	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.
  ADDRESS OF COL. GARRICK MALLERY, U. S. ARMY.	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 unde...
  DOLBEAR ON THE NATURE AND CONSTITUTION OF MATTER.	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: MultipleNegativesRankingLoss with these parameters:

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

Evaluation Dataset

parquet

• Dataset: parquet
• Size: 33,870,508 evaluation samples
• Columns: anchor and positive
• Approximate statistics based on the first 1000 samples:

  	anchor	positive
  type	string	string
  details	min: 6 tokens mean: 24.64 tokens max: 64 tokens	min: 9 tokens mean: 281.83 tokens max: 894 tokens

• Samples:

  anchor	positive
  Noticing education campaigns or public health messages about vaping among youth in the United States, Canada and England from 2018 to 2022.	Public health campaigns have the potential to correct vaping misperceptions. However, campaigns highlighting vaping harms to youth may increase misperceptions that vaping is equally/more harmful than smoking. Vaping campaigns have been implemented in the United States and Canada since 2018 and in England since 2017 but with differing focus: youth vaping prevention. Over half of youth reported noticing vaping campaigns, and noticing increased from August 2018 to February 2020. Consistent with implementation of youth vaping prevention campaigns in the United States and Canada, most youth reported noticing vaping campaigns/messages, and most were perceived to negatively portray vaping.
  Comprehensive performance evaluation of six bioaerosol samplers based on an aerosol wind tunnel.	Choosing a suitable bioaerosol sampler for atmospheric microbial monitoring has been a challenge to researchers interested in environmental microbiology, especially during a pandemic. However, a comprehensive and integrated evaluation method to fully assess bioaerosol sampler performance is still lacking. Herein, we constructed a customized wind tunnel operated at 2-20 km/h wind speed to systematically and efficiently evaluate the performance of six frequently used samplers, where various aerosols, including Arizona test dust, bacterial spores, gram-positive and gram-negative bacteria, phages, and viruses, were generated. After 10 or 60 min of sampling, the physical and biological sampling efficiency and short or long-term sampling capabilities were determined by performing aerodynamic particle size analysis, live microbial culturing, and a qPCR assay. The results showed that AGI-30 and BioSampler impingers have good physical and biological sampling efficiencies for short-term sampling...
  The occurrence, sources, and health risks of substituted polycyclic aromatic hydrocarbons (SPAHs) cannot be ignored.	Similar to parent polycyclic aromatic hydrocarbons (PPAHs), substituted PAHs (SPAHs) are prevalent in the environment and harmful to humans. However, they have not received much attention. This study investigated the occurrence, distribution, and sources of 10 PPAHs and 15 SPAHs in soil, water, and indoor and outdoor PM2.5 and dust in high-exposure areas (EAH) near industrial parks and low-exposure areas (EAL) far from industrial parks. PAH pollution in all media was more severe in the EAH than in the EAL. All SPAHs were detected in this study, with alkylated and oxygenated PAHs being predominant. Additionally, 3-OH-BaP and 1-OH-Pyr were detected in all dust samples in this study, and 6-N-Chr, a compound with carcinogenicity 10 times higher than that of BaP, was detected at high levels in all tap water samples. According to the indoor-outdoor ratio, PAHs in indoor PM2.5 in the EAH mainly originated from indoor pollution sources; however, those in the EAL were simultaneously affected by...

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 128
• learning_rate: 2e-05
• num_train_epochs: 2
• max_steps: 502764
• log_level: info
• fp16: True
• dataloader_num_workers: 16
• load_best_model_at_end: True
• resume_from_checkpoint: True

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 128
• per_device_eval_batch_size: 8
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 1
• eval_accumulation_steps: None
• torch_empty_cache_steps: None
• learning_rate: 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: 2
• max_steps: 502764
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.0
• warmup_steps: 0
• log_level: info
• log_level_replica: warning
• log_on_each_node: True
• logging_nan_inf_filter: True
• save_safetensors: True
• save_on_each_node: False
• save_only_model: False
• restore_callback_states_from_checkpoint: False
• no_cuda: False
• use_cpu: False
• use_mps_device: False
• seed: 42
• data_seed: None
• jit_mode_eval: False
• use_ipex: False
• bf16: False
• fp16: True
• fp16_opt_level: O1
• half_precision_backend: auto
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: None
• local_rank: 0
• ddp_backend: None
• tpu_num_cores: None
• tpu_metrics_debug: False
• debug: []
• dataloader_drop_last: False
• dataloader_num_workers: 16
• 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
• 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: True
• 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: batch_sampler
• multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss
0.0000	1	1.793	-
0.0040	1000	0.3695	-
0.0080	2000	0.0813	-
0.0119	3000	0.0666	-
0.0159	4000	0.0817	-
0.0199	5000	0.0694	-
0.0239	6000	0.0586	-
0.0278	7000	0.0539	-
0.0318	8000	0.0545	-
0.0358	9000	0.0515	-
0.0398	10000	0.0493	-
0.0438	11000	0.0419	-
0.0477	12000	0.0464	-
0.0517	13000	0.0494	-
0.0557	14000	0.0536	-
0.0597	15000	0.0472	-
0.0636	16000	0.0945	-
0.0676	17000	0.0385	-
0.0716	18000	0.068	-
0.0756	19000	0.0362	-
0.0796	20000	0.0865	-
0.0835	21000	0.0403	-
0.0875	22000	0.0798	-
0.0915	23000	0.0421	-
0.0955	24000	0.0428	-
0.0994	25000	0.035	-
0.1034	26000	0.0736	-
0.1074	27000	0.0395	-
0.1114	28000	0.0837	-
0.1154	29000	0.0432	-
0.1193	30000	0.0695	-
0.1233	31000	0.0584	-
0.1273	32000	0.0394	-
0.1313	33000	0.113	-
0.1353	34000	0.0349	-
0.1392	35000	0.044	-
0.1432	36000	0.0712	-
0.1472	37000	0.0322	-
0.1512	38000	0.0628	-
0.1551	39000	0.035	-
0.1591	40000	0.0305	-
0.1631	41000	0.0733	-
0.1671	42000	0.0449	-
0.1711	43000	0.0434	-
0.1750	44000	0.0597	-
0.1790	45000	0.0464	-
0.1830	46000	0.0428	-
0.1870	47000	0.0657	-
0.1909	48000	0.0346	-
0.1949	49000	0.0537	-
0.1989	50000	0.0577	-
0.2029	51000	0.0349	-
0.2069	52000	0.0376	-
0.2108	53000	0.0476	-
0.2148	54000	0.0453	-
0.2188	55000	0.0366	-
0.2228	56000	0.0295	-
0.2267	57000	0.0427	-
0.2307	58000	0.0352	-
0.2347	59000	0.0319	-
0.2387	60000	0.0316	-
0.2427	61000	0.0433	-
0.2466	62000	0.0272	-
0.2506	63000	0.0253	-
0.2546	64000	0.0356	-
0.2586	65000	0.0429	-
0.2625	66000	0.0301	-
0.2665	67000	0.0293	-
0.2705	68000	0.0269	-
0.2745	69000	0.03	-
0.2785	70000	0.0585	-
0.2824	71000	0.05	-
0.2864	72000	0.0455	-
0.2904	73000	0.0212	-
0.2944	74000	0.0296	-
0.2983	75000	0.043	-
0.3023	76000	0.0277	-
0.3063	77000	0.0592	-
0.3103	78000	0.0247	-
0.3143	79000	0.046	-
0.3182	80000	0.0429	-
0.3222	81000	0.0306	-
0.3262	82000	0.0313	-
0.3302	83000	0.0386	-
0.3342	84000	0.0196	-
0.3381	85000	0.0353	-
0.3421	86000	0.0462	-
0.3461	87000	0.0277	-
0.3501	88000	0.0461	-
0.3540	89000	0.0265	-
0.3580	90000	0.0159	-
0.3620	91000	0.0201	-
0.3660	92000	0.031	-
0.3700	93000	0.0337	-
0.3739	94000	0.0369	-
0.3779	95000	0.0504	-
0.3819	96000	0.0254	-
0.3859	97000	0.0265	-
0.3898	98000	0.0205	-
0.3938	99000	0.0181	-
0.3978	100000	0.0242	-
0.4018	101000	0.0317	-
0.4058	102000	0.0248	-
0.4097	103000	0.0171	-
0.4137	104000	0.0183	-
0.4177	105000	0.0156	-
0.4217	106000	0.0217	-
0.4256	107000	0.0282	-
0.4296	108000	0.0381	-
0.4336	109000	0.0271	-
0.4376	110000	0.0165	-
0.4416	111000	0.01	-
0.4455	112000	0.0241	-
0.4495	113000	0.0226	-
0.4535	114000	0.0161	-
0.4575	115000	0.0172	-
0.4614	116000	0.0129	-
0.4654	117000	0.0147	-
0.4694	118000	0.0346	-
0.4734	119000	0.039	-
0.4774	120000	0.0348	-
0.4813	121000	0.0353	-
0.4853	122000	0.0178	-
0.4893	123000	0.0173	-
0.4933	124000	0.0197	-
0.4972	125000	0.0148	-
0.5012	126000	0.014	-
0.5052	127000	0.0186	-
0.5092	128000	0.0129	-
0.5132	129000	0.0116	-
0.5171	130000	0.0186	-
0.5211	131000	0.0332	-
0.5251	132000	0.0195	-
0.5291	133000	0.0163	-
0.5331	134000	0.0145	-
0.5370	135000	0.0236	-
0.5410	136000	0.0169	-
0.5450	137000	0.0327	-
0.5490	138000	0.0332	-
0.5529	139000	0.034	-
0.5569	140000	0.0317	-
0.5609	141000	0.0372	-
0.5649	142000	0.0246	-
0.5689	143000	0.0278	-
0.5728	144000	0.0196	-
0.5768	145000	0.0217	-
0.5808	146000	0.0223	-
0.5848	147000	0.0138	-
0.5887	148000	0.0114	-
0.5927	149000	0.0122	-
0.5967	150000	0.0199	-
0.6007	151000	0.0204	-
0.6047	152000	0.0155	-
0.6086	153000	0.015	-
0.6126	154000	0.0196	-
0.6166	155000	0.0183	-
0.6206	156000	0.0225	-
0.6245	157000	0.0232	-
0.6285	158000	0.0389	-
0.6325	159000	0.0267	-
0.6365	160000	0.0264	-
0.6405	161000	0.0123	-
0.6444	162000	0.0144	-
0.6484	163000	0.018	-
0.6524	164000	0.0327	-
0.6564	165000	0.0283	-
0.6603	166000	0.0357	-
0.6643	167000	0.0148	-
0.6683	168000	0.0137	-
0.6723	169000	0.0165	-
0.6763	170000	0.0237	-
0.6802	171000	0.0218	-
0.6842	172000	0.0143	-
0.6882	173000	0.027	-
0.6922	174000	0.025	-
0.6961	175000	0.0211	-
0.7001	176000	0.0191	-
0.7041	177000	0.0213	-
0.7081	178000	0.0177	-
0.7121	179000	0.0178	-
0.7160	180000	0.0263	-
0.7200	181000	0.0263	-
0.7240	182000	0.0265	-
0.7280	183000	0.0236	-
0.7320	184000	0.0183	-
0.7359	185000	0.012	-
0.7399	186000	0.0192	-
0.7439	187000	0.0221	-
0.7479	188000	0.0223	-
0.7518	189000	0.021	-
0.7558	190000	0.0234	-
0.7598	191000	0.0221	-
0.7638	192000	0.0246	-
0.7678	193000	0.0212	-
0.7717	194000	0.0191	-
0.7757	195000	0.0122	-
0.7797	196000	0.0111	-
0.7837	197000	0.0094	-
0.7876	198000	0.0107	-
0.7916	199000	0.0103	-
0.7956	200000	0.0093	-
0.7996	201000	0.0128	-
0.8036	202000	0.0104	-
0.8075	203000	0.0161	-
0.8115	204000	0.0221	-
0.8155	205000	0.0243	-
0.8195	206000	0.0209	-
0.8234	207000	0.0241	-
0.8274	208000	0.0224	-
0.8314	209000	0.0131	-
0.8354	210000	0.0105	-
0.8394	211000	0.0118	-
0.8433	212000	0.0122	-
0.8473	213000	0.0112	-
0.8513	214000	0.0113	-
0.8553	215000	0.0108	-
0.8592	216000	0.0117	-
0.8632	217000	0.0111	-
0.8672	218000	0.0123	-
0.8712	219000	0.0112	-
0.8752	220000	0.0109	-
0.8791	221000	0.011	-
0.8831	222000	0.0122	-
0.8871	223000	0.0287	-
0.8911	224000	0.0234	-
0.8950	225000	0.0234	-
0.8990	226000	0.0222	-
0.9030	227000	0.0193	-
0.9070	228000	0.0166	-
0.9110	229000	0.0113	-
0.9149	230000	0.012	-
0.9189	231000	0.0108	-
0.9229	232000	0.0106	-
0.9269	233000	0.0107	-
0.9309	234000	0.0105	-
0.9348	235000	0.0091	-
0.9388	236000	0.0095	-
0.9428	237000	0.0066	-
0.9468	238000	0.0093	-
0.9507	239000	0.0049	-
0.9547	240000	0.0058	-
0.9587	241000	0.0065	-
0.9627	242000	0.0144	-
0.9667	243000	0.0181	-
0.9706	244000	0.0105	-
0.9746	245000	0.0066	-
0.9786	246000	0.0057	-
0.9826	247000	0.0053	-
0.9865	248000	0.005	-
0.9905	249000	0.006	-
0.9945	250000	0.0047	-
0.9985	251000	0.0055	-
1.0000	251382	-	0.0021
1.0025	252000	0.2602	-
1.0064	253000	0.0967	-
1.0104	254000	0.0643	-
1.0144	255000	0.057	-
1.0184	256000	0.0614	-
1.0223	257000	0.062	-
1.0263	258000	0.0471	-
1.0303	259000	0.0445	-
1.0343	260000	0.0439	-
1.0383	261000	0.0339	-
1.0422	262000	0.0376	-
1.0462	263000	0.0445	-
1.0502	264000	0.0331	-
1.0542	265000	0.0392	-
1.0581	266000	0.0539	-
1.0621	267000	0.0595	-
1.0661	268000	0.0595	-
1.0701	269000	0.0472	-
1.0741	270000	0.0421	-
1.0780	271000	0.0705	-
1.0820	272000	0.0343	-
1.0860	273000	0.0702	-
1.0900	274000	0.0385	-
1.0939	275000	0.0348	-
1.0979	276000	0.0338	-
1.1019	277000	0.065	-
1.1059	278000	0.032	-
1.1099	279000	0.0318	-
1.1138	280000	0.0768	-
1.1178	281000	0.0372	-
1.1218	282000	0.0771	-
1.1258	283000	0.0346	-
1.1298	284000	0.0781	-
1.1337	285000	0.0528	-
1.1377	286000	0.0282	-
1.1417	287000	0.0723	-
1.1457	288000	0.0286	-
1.1496	289000	0.0403	-
1.1536	290000	0.0439	-
1.1576	291000	0.0286	-
1.1616	292000	0.0517	-
1.1656	293000	0.0504	-
1.1695	294000	0.0348	-
1.1735	295000	0.0537	-
1.1775	296000	0.0364	-
1.1815	297000	0.04	-
1.1854	298000	0.0587	-
1.1894	299000	0.0332	-
1.1934	300000	0.0429	-
1.1974	301000	0.0522	-
1.2014	302000	0.0348	-
1.2053	303000	0.0305	-
1.2093	304000	0.0319	-
1.2133	305000	0.0493	-
1.2173	306000	0.0375	-
1.2212	307000	0.024	-
1.2252	308000	0.0327	-
1.2292	309000	0.0356	-
1.2332	310000	0.0296	-
1.2372	311000	0.0259	-
1.2411	312000	0.0358	-
1.2451	313000	0.0263	-
1.2491	314000	0.0252	-
1.2531	315000	0.0251	-
1.2570	316000	0.0298	-
1.2610	317000	0.0393	-
1.2650	318000	0.0261	-
1.2690	319000	0.0198	-
1.2730	320000	0.0271	-
1.2769	321000	0.048	-
1.2809	322000	0.0421	-
1.2849	323000	0.0483	-
1.2889	324000	0.0173	-
1.2928	325000	0.0174	-
1.2968	326000	0.0375	-
1.3008	327000	0.0261	-
1.3048	328000	0.0563	-
1.3088	329000	0.0238	-
1.3127	330000	0.02	-
1.3167	331000	0.0495	-
1.3207	332000	0.0218	-
1.3247	333000	0.031	-
1.3286	334000	0.0366	-
1.3326	335000	0.0188	-
1.3366	336000	0.0179	-
1.3406	337000	0.0547	-
1.3446	338000	0.0197	-
1.3485	339000	0.0372	-
1.3525	340000	0.0327	-
1.3565	341000	0.0131	-
1.3605	342000	0.019	-
1.3645	343000	0.0119	-
1.3684	344000	0.038	-
1.3724	345000	0.0324	-
1.3764	346000	0.0495	-
1.3804	347000	0.0196	-
1.3843	348000	0.0256	-
1.3883	349000	0.0176	-
1.3923	350000	0.0195	-
1.3963	351000	0.0157	-
1.4003	352000	0.0267	-
1.4042	353000	0.0285	-
1.4082	354000	0.0145	-
1.4122	355000	0.0183	-
1.4162	356000	0.012	-
1.4201	357000	0.0175	-
1.4241	358000	0.022	-
1.4281	359000	0.028	-
1.4321	360000	0.0319	-
1.4361	361000	0.0157	-
1.4400	362000	0.0107	-
1.4440	363000	0.0158	-
1.4480	364000	0.0209	-
1.4520	365000	0.0168	-
1.4559	366000	0.0125	-
1.4599	367000	0.0151	-
1.4639	368000	0.0106	-
1.4679	369000	0.0232	-
1.4719	370000	0.0318	-
1.4758	371000	0.031	-
1.4798	372000	0.0314	-
1.4838	373000	0.023	-
1.4878	374000	0.0151	-
1.4917	375000	0.0144	-
1.4957	376000	0.0165	-
1.4997	377000	0.011	-
1.5037	378000	0.0138	-
1.5077	379000	0.0149	-
1.5116	380000	0.0087	-
1.5156	381000	0.0154	-
1.5196	382000	0.0245	-
1.5236	383000	0.0199	-
1.5275	384000	0.0174	-
1.5315	385000	0.0103	-
1.5355	386000	0.018	-
1.5395	387000	0.0166	-
1.5435	388000	0.0249	-
1.5474	389000	0.028	-
1.5514	390000	0.0306	-
1.5554	391000	0.0264	-
1.5594	392000	0.0325	-
1.5634	393000	0.0282	-
1.5673	394000	0.0189	-
1.5713	395000	0.0246	-
1.5753	396000	0.0189	-
1.5793	397000	0.0192	-
1.5832	398000	0.0155	-
1.5872	399000	0.0108	-
1.5912	400000	0.0085	-
1.5952	401000	0.0171	-
1.5992	402000	0.0176	-
1.6031	403000	0.0159	-
1.6071	404000	0.0127	-
1.6111	405000	0.016	-
1.6151	406000	0.0169	-
1.6190	407000	0.0199	-
1.6230	408000	0.0149	-
1.6270	409000	0.0364	-
1.6310	410000	0.0259	-
1.6350	411000	0.0294	-
1.6389	412000	0.0109	-
1.6429	413000	0.0132	-
1.6469	414000	0.0109	-
1.6509	415000	0.0269	-
1.6548	416000	0.0259	-
1.6588	417000	0.0304	-
1.6628	418000	0.0216	-
1.6668	419000	0.0133	-
1.6708	420000	0.0125	-
1.6747	421000	0.0197	-
1.6787	422000	0.0211	-
1.6827	423000	0.015	-
1.6867	424000	0.0183	-
1.6906	425000	0.0262	-
1.6946	426000	0.0217	-
1.6986	427000	0.0163	-
1.7026	428000	0.0201	-
1.7066	429000	0.0188	-
1.7105	430000	0.015	-
1.7145	431000	0.019	-
1.7185	432000	0.0271	-
1.7225	433000	0.0236	-
1.7264	434000	0.0239	-
1.7304	435000	0.0173	-
1.7344	436000	0.0159	-
1.7384	437000	0.0143	-
1.7424	438000	0.0176	-
1.7463	439000	0.0183	-
1.7503	440000	0.0204	-
1.7543	441000	0.0216	-
1.7583	442000	0.0196	-
1.7623	443000	0.0215	-
1.7662	444000	0.021	-
1.7702	445000	0.0197	-
1.7742	446000	0.0131	-
1.7782	447000	0.0107	-
1.7821	448000	0.0079	-
1.7861	449000	0.01	-
1.7901	450000	0.0097	-
1.7941	451000	0.0079	-
1.7981	452000	0.0105	-
1.8020	453000	0.01	-
1.8060	454000	0.0103	-
1.8100	455000	0.0217	-
1.8140	456000	0.0204	-
1.8179	457000	0.0206	-
1.8219	458000	0.0218	-
1.8259	459000	0.0207	-
1.8299	460000	0.0187	-
1.8339	461000	0.0083	-
1.8378	462000	0.0104	-
1.8418	463000	0.0119	-
1.8458	464000	0.01	-
1.8498	465000	0.0108	-
1.8537	466000	0.0101	-
1.8577	467000	0.0106	-
1.8617	468000	0.0098	-
1.8657	469000	0.0108	-
1.8697	470000	0.0109	-
1.8736	471000	0.0104	-
1.8776	472000	0.0098	-
1.8816	473000	0.0097	-
1.8856	474000	0.0244	-
1.8895	475000	0.019	-
1.8935	476000	0.0238	-
1.8975	477000	0.0207	-
1.9015	478000	0.0198	-
1.9055	479000	0.0184	-
1.9094	480000	0.0124	-
1.9134	481000	0.0106	-
1.9174	482000	0.0113	-
1.9214	483000	0.0095	-
1.9253	484000	0.0106	-
1.9293	485000	0.0097	-
1.9333	486000	0.0094	-
1.9373	487000	0.0088	-
1.9413	488000	0.0076	-
1.9452	489000	0.0095	-
1.9492	490000	0.005	-
1.9532	491000	0.0048	-
1.9572	492000	0.0063	-
1.9612	493000	0.0088	-
1.9651	494000	0.0191	-
1.9691	495000	0.0137	-
1.9731	496000	0.0067	-
1.9771	497000	0.0062	-
1.9810	498000	0.0056	-
1.9850	499000	0.0049	-
1.9890	500000	0.0064	-
1.9930	501000	0.0047	-
1.9970	502000	0.0051	-
2.0000	502764	-	0.0012

Framework Versions

• Python: 3.11.11
• Sentence Transformers: 3.4.1
• Transformers: 4.48.2
• PyTorch: 2.6.0+cu124
• Accelerate: 1.3.0
• Datasets: 3.2.0
• Tokenizers: 0.21.0

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}
}