README.md

metadata

language:
  - en
license: apache-2.0
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - dense
  - generated_from_trainer
  - dataset_size:10000
  - loss:MarginMSELoss
base_model: answerdotai/ModernBERT-base
widget:
  - source_sentence: driving distance miami to fort lauderdale
    sentences:
      - "This Distance calculator provides both the by air, and by road distance between cities in both miles and kms, along with a map and driving directions â\x80\x93 please scroll down to see it, and a little further for the city to city driving directions. This is the by air, as the crow flies distance between the two cities."
      - >-
        Driving distance from Fort Lauderdale, FL to Miami, FL. The total
        driving distance from Fort Lauderdale, FL to Miami, FL is 27 miles or 43
        kilometers. Your trip begins in Fort Lauderdale, Florida. It ends in
        Miami, Florida. If you are planning a road trip, you might also want to
        calculate the total driving time from Fort Lauderdale, FL to Miami, FL
        so you can see when you'll arrive at your destination.
      - >-
        The total driving distance from Fort Lauderdale to Miami Beach, FL is 32
        miles. It can take 1 hour to 1.5 hours depending on how bad the traffic
        is or in rush hour. Or 11.30 at night it may be 1 hour, but will be less
        1 hour.
      - >-
        Distance Between Cordova Tennessee United States and Memphis
        International Airport (MEM) Tennessee United States, flight and driving
        distances and airport information. Distance calculator.
      - >-
        Distance Map Info. Optimal route map between New Smyrna Beach, FL and
        Melbourne, FL. This route will be about 75 Miles. The driving route
        information(distance, estimated time, directions), flight route, traffic
        information and print the map features are placed on the top right
        corner of the map.
      - >-
        Driving distance from Fort Lauderdale, FL to Miami, FL. The total
        driving distance from Fort Lauderdale, FL to Miami, FL is 27 miles or 43
        kilometers.Your trip begins in Fort Lauderdale, Florida. It ends in
        Miami, Florida.ou can also calculate the cost of driving from Fort
        Lauderdale, FL to Miami, FL based on current local fuel prices and an
        estimate of your car's best gas mileage.
      - >-
        There are 25.15 miles from Miami to Fort Lauderdale in north direction
        and 29.11 miles (46.85 kilometers) by car, following the I-95 route.
        Miami and Fort Lauderdale are 31 minutes far apart, if you drive
        non-stop. This is the fastest route from Miami, FL to Fort Lauderdale,
        FL. The halfway point is Aventura, FL.
      - >-
        Need to know towns or cities within a specific radius of other places
        try this Towns within a Radius of Treasure Island tool. The distance
        between Treasure Island and St Augustine in a straight line is 172 miles
        or 276.75 Kilometers.
      - >-
        Distance calculator helps you to find the distance between cities and
        calculate the flying and driving distance in both kilometers and miles.
        Distance Calculator Distance calculator is a tool for calculating
        distance between cities or places on map.
  - source_sentence: datsun 1200 starter motor diagram
    sentences:
      - >-
        Here you will find scans of the original wiring diagram for the 1971,
        1972, and 1973 Datsun 1200. 1971 Datsun 1200 Wiring Diagram. 1972 Datsun
        1200 Wiring Diagram. 1973 Datsun 1200 Wiring Diagram.
      - >-
        Free Universal VIN decoder to check vehicle data and history. This is a
        universal VIN decoder. Every car has a unique identifier code called a
        VIN. This number contains vital information about the car, such as its
        manufacturer, year of production, the plant it was produced in, type of
        engine, model and more.
      - >-
        This review is from: Powerhorse Portable Generator with Electric Start -
        9000 Surge Watts, 7250 Rated Watts (Misc.) Update Jan. 4, 2012... I
        found out the motor for this generator is manufactured by a company
        called Ducar. I found this out by speaking with a customer Service rep
        at Northern Tool.
      - >-
        Starter Motor. The engine starter is an electrical motor that is bolted
        to the engine crankcase. The starter turns the engine flywheel teeth
        with the teeth on the starter motor plunger which starts the riding
        lawnmower engine. Once the battery, electrical connectors, wiring and
        starter solenoid have been tested and are functioning properly, the
        problem likely lies with the engine starter motor.
      - "Datsun is a Japanese car manufacturer that produced a great number of cars, starting with the first Datsun model in 1931 and on until the brand was discontinued in 1983. The Datsun brand became particularly famous because of one of their sports cars, the Fairlady model. Headquartered in Japan, Datsun was absorbed by another Japanese carmaker, Nissan, and the Datsun models were rebadged as Nissanâ\x80\x99s from 1983 onwards."
      - >-
        Datsun is an automobile brand owned by Nissan. Datsun's original
        production run began in 1931. From 1958 to 1986, only vehicles exported
        by Nissan were identified as Datsun. By 1986 Nissan had phased out the
        Datsun name, but re-launched it in 2013 as the brand for low-cost
        vehicles manufactured for emerging markets.
      - >-
        The dial reading should be 12 volts or more. Work the starter switch,
        and the reading should fall, but not below 10.5 volts. If the reading
        does not fall, there is a fault in the ignition-switch circuit or in the
        solenoid.
      - >-
        echo dot. Controls lights, fans, switches, thermostats, garage doors,
        sprinklers, and more with compatible connected devices. Connects to
        speakers or headphones through Bluetooth or 3.5 mm stereo cable to play
        music.
      - >-
        Acura TSX vs Honda Accord. Compare price, expert/user reviews, mpg,
        engines, safety, cargo capacity and other specs at a glance.
  - source_sentence: what primary plant tissue is involved in transport of minerals
    sentences:
      - >-
        The ground tissue comprises the bulk of the primary plant body.
        Parenchyma, collenchyma, and sclerenchyma cells are common in the ground
        tissue. Vascular tissue transports food, water, hormones and minerals
        within the plant. Vascular tissue includes xylem, phloem, parenchyma,
        and cambium cells. Two views of the structure of the root and root
        meristem.
      - >-
        · just now. Report Abuse. 1.Water transport-in xylem from the roots up
        the plant. A passive process = transpiration 2.Sugar transport-sucrose
        is the main sugar transported in phloem from leaves up and down the
        plant. An active process-needs ATP.Process called translocation Look up
        transpiration and translocation Discuss the cells used in each, the
        direction and the mechanism. Peter S · 6 years ago.ugar i.e. glucose is
        produced in leaves (green parts of the plant) and it is transported to
        other parts of the plants such as reproductive organs, fruits and seeds,
        roots and stem (growing tips etc) through phloem tissue. Xylems are made
        up of dead cells while phloem tissue is made up of living cells.
      - >-
        1 of 6. Plants have tissues to transport water, nutrients and minerals.
        Xylem transports water and mineral salts from the roots up to other
        parts of the plant, while phloem transports sucrose and amino acids
        between the leaves and other parts of the plant. Xylem and phloem in the
        centre of the plant root.
      - >-
        Xylem is one of two types of vascular tissues found in plant (the other
        being Phloem). The word Xylem is derived from the Greek for wood. The
        xylem is responsible for the transport (translocation) of water and
        soluble mineral nutrients from the roots throughout the plant, this
        facilitates the replacement of water lost during transpiration and
        photosynthesis.
      - >-
        Various vascular tissues in the root allow for transportation of water
        and nutrients to the rest of theplant.Plant cells have a cell wall to
        provide support, a large vacuole for storage of minerals, food,
        andchloroplasts where photosynthesis takes place.
      - >-
        The function of the phloem tissue is to transport food nutrients such as
        glucose and amino acids from the leaves and to all other cells of the
        plant, this is called translocation.
      - >-
        Plants have tissues to transport water, nutrients and minerals. Xylem
        transports water and mineral salts from the roots up to other parts of
        the plant, while phloem transports sucrose and amino acids between the
        leaves and other parts of the plant. Xylem and phloem in the centre of
        the plant root
      - >-
        Seed plants contain 2 types of vascular tissue (xylem & phloem) to help
        transport water, minerals, & food throughout the root & shoot
        systems.Plant cells have several specialized structures including a
        central vacuole for storage, plastids for storage of pigments, and a
        thick cell wall of cellulose.lants have 3 tissue systems --- ground,
        dermal, and vascular tissues. Plant tissues make up the main organs of a
        plant --- root, stem, leaf, & flower.
      - >-
        This article is about vascular tissue in plants. For transport in
        animals, see Circulatory system. Vascular tissue is a complex conducting
        tissue, formed of more than one cell type, found in vascular plants. The
        primary components of vascular tissue are the xylem and phloem. These
        two tissues transport fluid and nutrients internally. There are also two
        meristems associated with vascular tissue: the vascular cambium and the
        cork cambium. All the vascular tissues within a particular plant
        together constitute the vascular tissue system of that plant. The cells
        in vascular tissue are typically long and slender. Since the xylem and
        phloem function in the conduction of water, minerals, and nutrients
        throughout the plant, it is not surprising that their form should be
        similar to pipes.
  - source_sentence: Julia name meaning
    sentences:
      - >-
        In American the meaning of the name Julie is: Young. French Meaning: The
        name Julie is a French baby name. In French the meaning of the name
        Julie is: Downy. French form of Julia. Also can be a feminine form of
        Julian: Youthful.
      - >-
        Jelisa Name Meaning. You are honest, benevolent, brilliant and often
        inventive, full of high inspirations. You are courageous, honest,
        determined, original and creative. You are a leader, especially for a
        cause.
      - >-
        The name Julius is of Latin and Greek origin. The meaning of Julius is
        downy-bearded, soft haired, implying youthful. Julius is generally used
        as a boy's name. It consists of 6 letters and 3 syllables and is
        pronounced Ju-li-us.
      - >-
        The name Julia is a Greek baby name. In Greek the meaning of the name
        Julia is: Downy. Hairy. Derived from the clan name of Roman dictator
        Gaius Julius Caesar. Latin Meaning: The name Julia is a Latin baby name.
      - >-
        The name Julia is an American baby name. In American the meaning of the
        name Julia is: Youthful. Swedish Meaning: The name Julia is a Swedish
        baby name. In Swedish the meaning of the name Julia is: Youth.Greek
        Meaning: The name Julia is a Greek baby name. In Greek the meaning of
        the name Julia is: Downy. Hairy. Derived from the clan name of Roman
        dictator Gaius Julius Caesar.Latin Meaning: The name Julia is a Latin
        baby name.In Latin the meaning of the name Julia is: Young. The feminine
        form of Julius. A character in Shakespeare's play 'Two Gentlemen of
        Verona'. Shakespearean Meaning: The name Julia is a Shakespearean baby
        name.he name Julia is a Latin baby name. In Latin the meaning of the
        name Julia is: Young. The feminine form of Julius. A character in
        Shakespeare's play 'Two Gentlemen of Verona'.
      - >-
        The meaning of Julius is Downy-bearded youth. Its origin is Variant of
        the Roman name Iulius. Julius is a form of Iulius and is generally
        pronounced like JOO lee es. This name is mostly being used as a boys
        name. Last year it ranked 312th in the U.S. Social Security
        Administration list of most popular baby boy names. Show popularity
        chart.
      - "What does Julianna mean? Julianna [ju-lian-na] as a name for girls is a Latin name, and Julianna means youthful; Jove's child. Julianna is a version of Juliana (Latin): feminine of Julius. Associated with: youthful. Juliannaâ\x96² has 1 variant: Julieanna."
      - >-
        The name Jelena is a Russian baby name. In Russian the meaning of the
        name Jelena is: Shining light. SoulUrge Number: 11. Expression Number:
        2. People with this name have a deep inner desire to inspire others in a
        higher cause, and to share their own strongly held views on spiritual
        matters.
      - >-
        The name Julia is a Latin baby name. In Latin the meaning of the name
        Julia is: Young. The feminine form of Julius. A character in
        Shakespeare's play 'Two Gentlemen of Verona'.he name Julia is a Latin
        baby name. In Latin the meaning of the name Julia is: Young. The
        feminine form of Julius. A character in Shakespeare's play 'Two
        Gentlemen of Verona'.
  - source_sentence: what does ly mean in a blood test
    sentences:
      - >-
        According to the Hormone-Refractory Prostate Cancer Association, LY on a
        blood test stands for lymphocytes. The number in the results represents
        the percentage of lymphocytes in the white blood count. Lymphocytes
        should count for 15 to 46.8 percent of white blood cells. Continue
        Reading.
      - >-
        FROM OUR EXPERTS. Trace lysed blood refers to a finding that is usually
        reported from a urinary dip stick analysis. It implies that there is a
        small quantity of red cells in the urine that have broken open. The
        developer on the dip stick reacts with the hemoglobin that is released
        when the red cells are lysed.
      - >-
        Does anybody know what LM and TFT in blood testing is please? Does
        anybody know what LM and TFT in blood testing is please? TFT is Thyroid
        Function Tests , usually done as a series of tests for TSH
        (Thyroid-stimulating hormone) and for the T3 (Triiodothyronine) and T4 (
        Thyroxine), which are the hormones produced by the thyroid gland.
      - >-
        Uses. A blood lipid test measures the levels of HDL (high-density
        lipoprotein) cholesterol and LDL (low-density lipoprotein) cholesterol
        in the blood as well as triglycerides. These results can help doctors
        determine patients' health status.
      - >-
        It is found in almost all body tissues and can be measured by a simple
        blood test. Normal LDH levels are generally low and range between 140
        IU/liter to 333 IU/ liter. Low LDH levels are usually no cause for
        concern. However, elevated LDH levels may indicate cell damage.
      - >-
        Usually, an LFT blood test measures the amount of bilirubin in the
        blood. Bilirubin is released when red blood cells breakdown, and it is
        the liver that detoxifies the bilirubin and helps to eliminate it from
        the body. Bilirubin is a part of the digestive juice, bile, which the
        liver produces.
      - >-
        Medical Definition of LYE. 1. : a strong alkaline liquor rich in
        potassium carbonate leached from wood ashes and used especially in
        making soap and washing; broadly: a strong alkaline solution (as of
        sodium hydroxide or potassium hydroxide). 2.: a solid caustic (as sodium
        hydroxide).edical Definition of LYE. 1. : a strong alkaline liquor rich
        in potassium carbonate leached from wood ashes and used especially in
        making soap and washing; broadly: a strong alkaline solution (as of
        sodium hydroxide or potassium hydroxide). 2. : a solid caustic (as
        sodium hydroxide).
      - >-
        FROM OUR COMMUNITY. Hi Terry, The LY (Lymphocytes) in your blood test
        is; the type of white blood cell found in the blood and lymph systems;
        part of the immune system. BUN/CREAT - Bun and Creatinine are tests done
        to monitor kidney function. I'm sorry, but I've never heard of the other
        2.
      - >-
        The Lactic Acid Plasma Test, or Lactate Test, measures the amount of
        lactate in the blood to determine if a patient has lactic acidosis.he
        test may be ordered at prescribed intervals to monitor lactate levels.
        The Lactate Test is also known as Lactic Acid, Plasma Lactate, and
        L-Lactate. Patients will be instructed to fast prior to this blood test
        and be in a resting state.
datasets:
  - tomaarsen/msmarco-Qwen3-Reranker-0.6B
pipeline_tag: sentence-similarity
library_name: sentence-transformers
metrics:
  - cosine_accuracy@1
  - cosine_accuracy@3
  - cosine_accuracy@5
  - cosine_accuracy@10
  - cosine_precision@1
  - cosine_precision@3
  - cosine_precision@5
  - cosine_precision@10
  - cosine_recall@1
  - cosine_recall@3
  - cosine_recall@5
  - cosine_recall@10
  - cosine_ndcg@10
  - cosine_mrr@10
  - cosine_map@100
model-index:
  - name: ModernBERT-base finetuned on MSMARCO via distillation
    results:
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: msmarco eval 1kq 1kd
          type: msmarco-eval-1kq-1kd
        metrics:
          - type: cosine_accuracy@1
            value: 0.781
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.887
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.917
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.943
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.781
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.29566666666666663
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.18340000000000004
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.09430000000000001
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.781
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.887
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.917
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.943
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.8655823808772907
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.8404488095238101
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.8423984809497023
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: NanoMSMARCO
          type: NanoMSMARCO
        metrics:
          - type: cosine_accuracy@1
            value: 0.1
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.28
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.36
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.56
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.1
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.09333333333333332
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.07200000000000001
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.05600000000000001
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.1
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.28
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.36
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.56
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.3024724428473199
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.22331746031746025
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.23964016128375398
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: NanoNFCorpus
          type: NanoNFCorpus
        metrics:
          - type: cosine_accuracy@1
            value: 0.12
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.16
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.24
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.34
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.12
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.05999999999999999
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.068
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.068
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.002236456626220743
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.002789859912543284
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.010512144339555038
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.03263772196662673
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.07431301845929561
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.16880158730158734
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.021674198565261597
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: NanoNQ
          type: NanoNQ
        metrics:
          - type: cosine_accuracy@1
            value: 0.06
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.1
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.16
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.24
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.06
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.039999999999999994
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.036000000000000004
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.026000000000000006
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.04
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.08
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.13
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.21
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.11989547330112625
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.10407936507936508
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.10392112551414794
            name: Cosine Map@100
      - task:
          type: nano-beir
          name: Nano BEIR
        dataset:
          name: NanoBEIR mean
          type: NanoBEIR_mean
        metrics:
          - type: cosine_accuracy@1
            value: 0.09333333333333334
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.18000000000000002
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.25333333333333335
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.38000000000000006
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.09333333333333334
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.06444444444444443
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.05866666666666667
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.05000000000000001
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.04741215220874025
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.1209299533041811
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.16683738144651836
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.2675459073222089
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.1655603115359139
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.1653994708994709
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.12174516178772117
            name: Cosine Map@100

ModernBERT-base finetuned on MSMARCO via distillation

This is a sentence-transformers model finetuned from answerdotai/ModernBERT-base on the msmarco-qwen3-reranker-0.6_b dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: answerdotai/ModernBERT-base
• Maximum Sequence Length: 8192 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • msmarco-qwen3-reranker-0.6_b
• Language: en
• License: apache-2.0

Model Sources

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

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False, 'architecture': 'ModernBertModel'})
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
)

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("tomaarsen/ModernBERT-base-msmarco-margin-mse")
# Run inference
queries = [
    "what does ly mean in a blood test",
]
documents = [
    'According to the Hormone-Refractory Prostate Cancer Association, LY on a blood test stands for lymphocytes. The number in the results represents the percentage of lymphocytes in the white blood count. Lymphocytes should count for 15 to 46.8 percent of white blood cells. Continue Reading.',
    "FROM OUR COMMUNITY. Hi Terry, The LY (Lymphocytes) in your blood test is; the type of white blood cell found in the blood and lymph systems; part of the immune system. BUN/CREAT - Bun and Creatinine are tests done to monitor kidney function. I'm sorry, but I've never heard of the other 2.",
    'FROM OUR EXPERTS. Trace lysed blood refers to a finding that is usually reported from a urinary dip stick analysis. It implies that there is a small quantity of red cells in the urine that have broken open. The developer on the dip stick reacts with the hemoglobin that is released when the red cells are lysed.',
]
query_embeddings = model.encode_query(queries)
document_embeddings = model.encode_document(documents)
print(query_embeddings.shape, document_embeddings.shape)
# [1, 768] [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(query_embeddings, document_embeddings)
print(similarities)
# tensor([[0.9409, 0.9409, 0.9366]])

Evaluation

Metrics

Information Retrieval

• Datasets: msmarco-eval-1kq-1kd, NanoMSMARCO, NanoNFCorpus and NanoNQ
• Evaluated with InformationRetrievalEvaluator

Metric	msmarco-eval-1kq-1kd	NanoMSMARCO	NanoNFCorpus	NanoNQ
cosine_accuracy@1	0.781	0.1	0.12	0.06
cosine_accuracy@3	0.887	0.28	0.16	0.1
cosine_accuracy@5	0.917	0.36	0.24	0.16
cosine_accuracy@10	0.943	0.56	0.34	0.24
cosine_precision@1	0.781	0.1	0.12	0.06
cosine_precision@3	0.2957	0.0933	0.06	0.04
cosine_precision@5	0.1834	0.072	0.068	0.036
cosine_precision@10	0.0943	0.056	0.068	0.026
cosine_recall@1	0.781	0.1	0.0022	0.04
cosine_recall@3	0.887	0.28	0.0028	0.08
cosine_recall@5	0.917	0.36	0.0105	0.13
cosine_recall@10	0.943	0.56	0.0326	0.21
cosine_ndcg@10	0.8656	0.3025	0.0743	0.1199
cosine_mrr@10	0.8404	0.2233	0.1688	0.1041
cosine_map@100	0.8424	0.2396	0.0217	0.1039

Nano BEIR

• Dataset: NanoBEIR_mean
• Evaluated with NanoBEIREvaluator with these parameters:

  {
      "dataset_names": [
          "msmarco",
          "nfcorpus",
          "nq"
      ]
  }
  

Metric	Value
cosine_accuracy@1	0.0933
cosine_accuracy@3	0.18
cosine_accuracy@5	0.2533
cosine_accuracy@10	0.38
cosine_precision@1	0.0933
cosine_precision@3	0.0644
cosine_precision@5	0.0587
cosine_precision@10	0.05
cosine_recall@1	0.0474
cosine_recall@3	0.1209
cosine_recall@5	0.1668
cosine_recall@10	0.2675
cosine_ndcg@10	0.1656
cosine_mrr@10	0.1654
cosine_map@100	0.1217

Training Details

Training Dataset

msmarco-qwen3-reranker-0.6_b

• Dataset: msmarco-qwen3-reranker-0.6_b at 20c25c8
• Size: 10,000 training samples
• Columns: query, positive, negative_1, negative_2, negative_3, negative_4, negative_5, negative_6, negative_7, negative_8, and score
• Approximate statistics based on the first 1000 samples:

  	query	positive	negative_1	negative_2	negative_3	negative_4	negative_5	negative_6	negative_7	negative_8	score
  type	string	string	string	string	string	string	string	string	string	string	list
  details	min: 5 tokens mean: 9.32 tokens max: 40 tokens	min: 17 tokens mean: 81.37 tokens max: 288 tokens	min: 16 tokens mean: 68.92 tokens max: 208 tokens	min: 12 tokens mean: 70.8 tokens max: 204 tokens	min: 15 tokens mean: 70.5 tokens max: 260 tokens	min: 16 tokens mean: 71.64 tokens max: 244 tokens	min: 13 tokens mean: 72.57 tokens max: 226 tokens	min: 11 tokens mean: 72.34 tokens max: 198 tokens	min: 15 tokens mean: 73.57 tokens max: 202 tokens	min: 15 tokens mean: 71.21 tokens max: 200 tokens	size: 9 elements

• Samples:

  query	positive	negative_1	negative_2	negative_3	negative_4	negative_5	negative_6	negative_7	negative_8	score
  what is clomiphene	Uses of This Medicine. Clomiphene is used as a fertility medicine in some women who are unable to become pregnant. Clomiphene probably works by changing the hormone balance of the body. In women, this causes ovulation to occur and prepares the body for pregnancy.ses of This Medicine. Clomiphene is used as a fertility medicine in some women who are unable to become pregnant. Clomiphene probably works by changing the hormone balance of the body. In women, this causes ovulation to occur and prepares the body for pregnancy.	Clomiphene citrate, a synthetic hormone commonly used to induce or regulate ovulation, is the most often prescribed fertility pill. Brand names for clomiphene citrate include Clomid and Serophene. Clomiphene works indirectly to stimulate ovulation.	Occasionally, clomiphene can stimulate the ovaries too much, causing multiple eggs to be released, which can result in multiple births, such as twins or triplets (see Clomid and Twins) . Clomiphene is one of the least expensive and easiest-to-use fertility drugs. However, it will not work for all types of infertility. Your healthcare provider needs to try to find your cause of infertility before you try clomiphene.	Clomiphene Citrate offers two benefits to the performance enhancing athlete with one being primary. Most commonly, this SERM is used for post cycle recovery purposes; specifically to stimulate natural testosterone production that has been suppressed due to the use of anabolic steroids.	PCOS and ovulation problems and Clomid treatment. Clomid (clomiphene citrate or Serophene) is an oral medication that is commonly used for the treatment of infertility. It is often given to try to induce ovulation in women that do not develop and release an egg (ovulate) on their own.	Indication: Clomid (clomiphene citrate) is often the first choice for treating infertility, because it's effective and been used for more than 40 years.	Clomid Description. 1 Clomid (clomiphene citrate tablets USP) is an orally administered, nonsteroidal, ovulatory stimulant designated chemically as 2-[p-(2-chloro-1,2-diphenylvinyl)phenoxy] triethylamine citrate (1:1). It has the molecular formula of C26H28ClNO • C6H8O7 and a molecular weight of 598.09.	PCOS and ovulation problems and Clomid treatment. Clomid (clomiphene citrate or Serophene) is an oral medication that is commonly used for the treatment of infertility. 1 It is often given to try to induce ovulation in women that do not develop and release an egg (ovulate) on their own. Clomid is started early in the menstrual cycle and is taken for five days either from cycle days 3 through 7, or from day 5 through 9. 2 Clomid is usually started at a dose of one tablet (50mg) daily-taken any time of day.	Clomid is taken as a pill. This is unlike the stronger fertility drugs, which require injection. Clomid is also very effective, stimulating ovulation 80 percent of the time. Clomid may also be marketed under the name Serophene, or you may see it sold under its generic name, clomiphene citrate. Note: Clomid can also be used as a treatment for male infertility. This article focuses on Clomid treatment in women.	[4.75390625, 6.9375, 3.92578125, 1.0400390625, 5.61328125, ...]
  typical accountant cost for it contractor	In the current market, we’ve seen rates as low as £50 +VAT, and as high as £180 +VAT for dedicated contractor accountants. Interestingly, the average cost of contractor accounting has not risen in line with inflation over the past decade.	So, how much does a contractor cost, anywhere from 5% to 25% of the total project cost, with the average ranging 10-15%.ypically the contractor' s crew will be general carpentry trades people, some who may have more specialized skills. Exactly how a general contractor charges for a project depends on the type of contract you agree to. There are three common types of cost contracts, fixed price, time & materials and cost plus a fee.	1 Accountants charge $150-$400 or more an hour, depending on the type of work, the size of the firm and its location. 2 You'll pay lower rates for routine work done by a less-experienced associate or lesser-trained employee, such as $30-$50 for bookkeeping services. 3 An accountant's total fee depends on the project. For a simple start-up, expect a minimum of 0.5-1.5 hours of consultation ($75-$600) to go over your business structure and basic tax issues.	So, how much does a contractor cost, anywhere from 5% to 25% of the total project cost, with the average ranging 10-15%.xactly how a general contractor charges for a project depends on the type of contract you agree to. There are three common types of cost contracts, fixed price, time & materials and cost plus a fee. Each contract type has pros and cons for both the consumer and for the contractor.	1 Accountants charge $150-$400 or more an hour, depending on the type of work, the size of the firm and its location. 2 You'll pay lower rates for routine work done by a less-experienced associate or lesser-trained employee, such as $30-$50 for bookkeeping services. 3 An accountant's total fee depends on the project.	average data entry keystrokes per hour salaries the average salary for data entry keystrokes per hour jobs is $ 20000	Accounting services are typically $250 to $400 per month, or $350 to $500 per quarter. Sales tax and bank recs included. We do all the processing, filing and tax deposits. 5 employees, bi-weekly payroll, direct deposit, $135 per month.	The less that is outsourced, the cheaper it will be for you. A bookkeeper should be paid between $15 and $18 per hour. An accountant with a undergraduate degree (4-years) should be paid somewhere around $20/hour but that still depends on what you're having them do. An accountant with a graduate degree (masters) should be paid between $25 and $30 per hour.	Pay by Experience Level for Intelligence Analyst. Median of all compensation (including tips, bonus, and overtime) by years of experience. Intelligence Analysts with a lot of experience tend to enjoy higher earnings.	[7.44921875, 3.271484375, 5.859375, 3.234375, 5.421875, ...]
  what is mch on a blood test	What High Levels Mean. MCH levels in blood tests are considered high if they are 35 or higher. A normal hemoglobin level is considered to be in the range between 26 and 33 picograms per red blood cell. High MCH levels can indicate macrocytic anemia, which can be caused by insufficient vitamin B12.acrocytic RBCs are large so tend to have a higher MCH, while microcytic red cells would have a lower value.”. MCH is one of three red blood cell indices (MCHC and MCV are the other two). The measurements are done by machine and can help with diagnosis of medical problems.	MCH stands for mean corpuscular hemoglobin. It estimates the average amount of hemoglobin in each red blood cell, measured in picograms (a trillionth of a gram). Automated cell counters calculate the MCH, which is reported as part of a complete blood count (CBC) test. MCH may be low in iron-deficiency anemia, and may be high in anemia due to vitamin B12 or folate deficiency. Other forms of anemia can also cause MCH to be abnormal. Doctors only use the MCH as supporting information, not to make a diagnosis.	A. MCH stands for mean corpuscular hemoglobin. It estimates the average amount of hemoglobin in each red blood cell, measured in picograms (a trillionth of a gram). Automated cell counters calculate the MCH, which is reported as part of a complete blood count (CBC) test. MCH may be low in iron-deficiency anemia, and may be high in anemia due to vitamin B12 or folate deficiency. Other forms of anemia can also cause MCH to be abnormal.	The test used to determine the quantity of hemoglobin in the blood is known as the MCH blood test. The full form of MCH is Mean Corpuscular Hemoglobin. This test is therefore used to determine the average amount of hemoglobin per red blood cell in the body. The results of the MCH blood test are therefore reported in picograms, a tiny measure of weight.	MCH blood test high indicates that there is a poor supply of oxygen to the blood where as MCH blood test low mean that hemoglobin is too little in the cells indicating a lack of iron. It is important that iron is maintained at a certain level as too much or too little iron can be dangerous to your body.	slide 1 of 7. What Is MCH? MCH is the initialism for Mean Corpuscular Hemoglobin. Taken from Latin, the term refers to the average amount of hemoglobin found in red blood cells. A CBC (complete blood count) blood test can be used to monitor MCH levels in the blood. Lab Tests Online explains that the MCH aspect of a CBC test “is a measurement of the average amount of oxygen-carrying hemoglobin inside a red blood cell. Macrocytic RBCs are large so tend to have a higher MCH, while microcytic red cells would have a lower value..	The test used to determine the quantity of hemoglobin in the blood is known as the MCH blood test. The full form of MCH is Mean Corpuscular Hemoglobin. This test is therefore used to determine the average amount of hemoglobin per red blood cell in the body. The results of the MCH blood test are therefore reported in picograms, a tiny measure of weight. The normal range of the MCH blood test is between 26 and 33 pg per cell.	A MCHC test is a test that is carried out to test a person for anemia. The MCHC in a MCHC test stands for Mean Corpuscular Hemoglobin Concentration. MCHC is the calculation of the average hemoglobin inside a red blood cell. A MCHC test can be performed along with a MCV test (Mean Corpuscular Volume).Both levels are used to test people for anemia.The MCHC test is also known as the MCH blood test which tests the levels of hemoglobin in the blood. The MCHC test can be ordered as part of a complete blood count (CBC) test.CHC is measured in grams per deciliter. Normal readings for MCHC are 31 grams per deciliter to 35 grams per deciliter. A MCHC blood test may be ordered when a person is showing signs of fatigue or weakness, when there is an infection, is bleeding or bruising easily or when there is an inflammation.	The test looks at the average amount of hemoglobin per red cell. So MCHC = the amount of hemoglobin present in each red blood cell. A MCHC blood test could be ordered for someone who has signs of fatigue or weakness, when there is an infection, is bleeding or bruising easily or when there is noticeable inflammation.	[6.44921875, 7.05078125, 7.2109375, 8.40625, 6.53515625, ...]

• Loss: MarginMSELoss

Evaluation Dataset

msmarco-qwen3-reranker-0.6_b

• Dataset: msmarco-qwen3-reranker-0.6_b at 20c25c8
• Size: 1,000 evaluation samples
• Columns: query, positive, negative_1, negative_2, negative_3, negative_4, negative_5, negative_6, negative_7, negative_8, and score
• Approximate statistics based on the first 1000 samples:

  	query	positive	negative_1	negative_2	negative_3	negative_4	negative_5	negative_6	negative_7	negative_8	score
  type	string	string	string	string	string	string	string	string	string	string	list
  details	min: 4 tokens mean: 9.27 tokens max: 24 tokens	min: 22 tokens mean: 82.29 tokens max: 236 tokens	min: 11 tokens mean: 69.82 tokens max: 215 tokens	min: 11 tokens mean: 69.4 tokens max: 209 tokens	min: 16 tokens mean: 71.47 tokens max: 223 tokens	min: 14 tokens mean: 70.96 tokens max: 225 tokens	min: 14 tokens mean: 73.09 tokens max: 218 tokens	min: 16 tokens mean: 69.31 tokens max: 215 tokens	min: 14 tokens mean: 72.3 tokens max: 226 tokens	min: 11 tokens mean: 70.66 tokens max: 210 tokens	size: 9 elements

• Samples:

  query	positive	negative_1	negative_2	negative_3	negative_4	negative_5	negative_6	negative_7	negative_8	score
  how many people employed by shell	Shell worldwide. Royal Dutch Shell was formed in 1907, although our history dates back to the early 19th century, to a small shop in London where the Samuel family sold sea shells. Today, Shell is one of the world’s major energy companies, employing an average of 93,000 people and operating in more than 70 countries. Our headquarters are in The Hague, the Netherlands, and our Chief Executive Officer is Ben van Beurden.	Show sources information. This statistic shows the number of employees at SeaWorld Entertainment, Inc. in the United States, by type. As of December 2016, SeaWorld employed 5,000 full-time employees and counted approximately 13,000 seasonal employees during their peak operating season.	Jobs, companies, people, and articles for LinkedIn’s Payroll Specialist - Addus Homecare, Inc. members. Insights about Payroll Specialist - Addus Homecare, Inc. members on LinkedIn. Median salary $31,300.	As of July 2014, there are 139 million people employed in the United States. This number is up by 209,000 employees from June and by 1.47 million from the beginning of 2014.	average data entry keystrokes per hour salaries the average salary for data entry keystrokes per hour jobs is $ 20000	Research and review Plano Synergy jobs. Learn more about a career with Plano Synergy including all recent jobs, hiring trends, salaries, work environment and more. Find Jobs Company Reviews Find Salaries Find Resumes Employers / Post Job Upload your resume Sign in	From millions of real job salary data. 13 Customer Support Specialist salary data. Average Customer Support Specialist salary is $59,032 Detailed Customer Support Specialist starting salary, median salary, pay scale, bonus data report Register & Know how much $ you can earn | Sign In	From millions of real job salary data. 1 Ceo Ally salary data. Average Ceo Ally salary is $55,000 Detailed Ceo Ally starting salary, median salary, pay scale, bonus data report	HelpSystems benefits and perks, including insurance benefits, retirement benefits, and vacation policy. Reported anonymously by HelpSystems employees. Glassdoor uses cookies to improve your site experience.	[6.265625, -1.3671875, -6.91796875, 1.111328125, -7.96875, ...]
  what is a lcsw	LCSW is an acronym for licensed clinical social worker, and people with this title are skilled professionals who meet certain requirements and work in a variety of fields. The term social worker is not always synonymous with licensed clinical social worker.	LISW means the person is a Licensed Independent Social Worker. LCSW means the person is a Licensed Clinical Social Worker. Source(s): Introduction to Social Work 101 at University of Nevada, Las Vega (UNLV) Dorothy K. · 1 decade ago.	An LCSW is a licensed clinical social worker. A LMHC is the newest addition to the field of mental health. They are highly similar and can do most of the same things with few exceptions. One thing to keep in mind is that because the LMHC lincense is so new, there are fewer in number in the field.n LCSW is a licensed clinical social worker. A LMHC is the newest addition to the field of mental health. They are highly similar and can do most of the same things with few exceptions. One thing to keep in mind is that because the LMHC lincense is so new, there are fewer in number in the field.	The Licensed Clinical Social Worker or LCSW, is a sub-sector within the field of Social Work. They work with clients in order to help them deal with issues involving their mental and emotional health. This could be related to substance abuse, past trauma or mental illness.	Licensed Clinical Social Worker | LCSW. The Licensed Clinical Social Worker or LCSW, is a sub-sector within the field of Social Work. LCSW's work with clients in order to help deal with issues involving mental and emotional health. There are a wide variety of specializations the Licensed Clinical Social Worker can focus on.	The LMSW exam is a computer-based test containing 170 multiple-choice questions designed to measure minimum competencies in four categories of social work practice: Human development, diversity, and behavior in the environment. Assessment and intervention planning.	The Licensed Clinical Social Worker, also known as the LCSW, is a branch of social work that specializes in mental health therapy in a counseling format. Becoming an LCSW requires a significant degree of training, including having earned a Master of Social Work (MSW) degree from a Council on Social Work Education (CSWE) accredited program.	a. The examination requirements for licensure as an LCSW include passing the Clinical Examination of the ASWB or the Clinical Social Workers Examination of the State of California. Scope of practice-Limitations. a.To the extent they are prepared through education and training, an LCSW can engage in all acts and practices defined as the practice of clinical social work. Certified Social Work (CSW): CSW means a licensed certified social worker. A CSW must have a master s degree.	The LTCM Client is a way for companies to stay in touch with you, their customers, in a way that is unobtrusive and completely under the users' control. It's an application that runs quietly on the computer. Users can and should customize the client to match their desired preferences.	[7.34375, 6.046875, 7.09765625, 6.46484375, 7.28515625, ...]
  does oolong tea have much caffeine?	At a given weight, tea contains more caffeine than coffee, but this doesn’t mean that a usual portion of tea contains more caffeine than coffee because tea is usually brewed in a weak way. Some kinds of tea, such as oolong and black tea, contain higher level of caffeine than most other teas. Among six basic teas (green, black, yellow, white, oolong, dark), green tea contains less caffeine than black tea and white tea contains less than green tea. But many studies found that the caffeine content varies more among individual teas than it does among broad categories.	Actually, oolong tea has less caffeine than coffee and black tea. A cup of oolong tea only has about 1/3 of caffeine of a cup of coffee. According to a research conducted by HICKS M.B, the caffeine decreases whenever the tea leaves go through the process of brewing.	Oolong tea contains caffeine. Caffeine works by stimulating the central nervous system (CNS), heart, and muscles. Oolong tea also contains theophylline and theobromine, which are chemicals similar to caffeine. Too much oolong tea, more than five cups per day, can cause side effects because of the caffeine.	Oolong tea, made from more mature leaves, usually have less caffeine than green tea. On the flip side, mature leaves contain less theanine, a sweet, natural relaxant that makes a tea much less caffeinated than it actually is. That is the theory, anyway.	Oolong tea is a product made from the leaves, buds, and stems of the Camellia sinensis plant. This is the same plant that is also used to make black tea and green tea. The difference is in the processing.Oolong tea is partially fermented, black tea is fully fermented, and green tea is unfermented. Oolong tea is used to sharpen thinking skills and improve mental alertness. It is also used to prevent cancer, tooth decay, osteoporosis, and heart disease.owever, do not drink more than 2 cups a day of oolong tea. That amount of tea contains about 200 mg of caffeine. Too much caffeine during pregnancy might cause premature delivery, low birth weight, and harm to the baby.	A Department of Nutritional Services report provides the following ranges of caffeine content for a cup of tea made with loose leaves: 1 Black Tea: 23 - 110 mg. 2 Oolong Tea: 12 - 55 mg. Green Tea: 8 - 36 mg.	Oolong tea is a product made from the leaves, buds, and stems of the Camellia sinensis plant. This is the same plant that is also used to make black tea and green tea. The difference is in the processing. Oolong tea is partially fermented, black tea is fully fermented, and green tea is unfermented. Oolong tea is used to sharpen thinking skills and improve mental alertness. It is also used to prevent cancer, tooth decay, osteoporosis, and heart disease.	Health Effects of Tea – Caffeine. In dry form, a kilogram of black tea has twice the caffeine as a kilogram of coffee…. But one kilogram of black tea makes about 450 cups of tea and one kilogram of coffee makes about 100 cups of coffee, so…. There is less caffeine in a cup of tea than in a cup of coffee. Green teas have less caffeine than black teas, and white teas have even less caffeine than green teas. Oolong teas fall between black and green teas. Herbal tea, because it is not made from the same tea plant, is caffeine-free, naturally. Here is a graphical representation of their respective caffeine content.	The average 8-ounce serving of brewed black tea contains 14 to 70 mg of caffeine. This compares to 24 to 45 mg of caffeine found in green tea. An 8-ounce glass of instant iced tea prepared with water contains 11 to 47 mg of caffeine. Most ready-to-drink bottled teas contain 5 to 40 mg of caffeine. Just as with coffee, decaffeinated tea still contains 5 to 10 mg of caffeine per cup.	[7.60546875, 8.78125, 9.109375, 8.609375, 7.984375, ...]

• Loss: MarginMSELoss

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• learning_rate: 4e-05
• num_train_epochs: 1
• warmup_ratio: 0.1
• bf16: 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: 16
• per_device_eval_batch_size: 16
• 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: 4e-05
• weight_decay: 0.0
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 1
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.1
• warmup_steps: 0
• log_level: passive
• log_level_replica: warning
• log_on_each_node: True
• logging_nan_inf_filter: True
• save_safetensors: True
• save_on_each_node: False
• save_only_model: False
• restore_callback_states_from_checkpoint: False
• no_cuda: False
• use_cpu: False
• use_mps_device: False
• seed: 42
• data_seed: None
• jit_mode_eval: False
• use_ipex: False
• bf16: True
• fp16: False
• fp16_opt_level: O1
• half_precision_backend: auto
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: None
• local_rank: 0
• ddp_backend: None
• tpu_num_cores: None
• tpu_metrics_debug: False
• debug: []
• dataloader_drop_last: False
• dataloader_num_workers: 0
• dataloader_prefetch_factor: None
• past_index: -1
• disable_tqdm: False
• remove_unused_columns: True
• label_names: None
• load_best_model_at_end: False
• ignore_data_skip: False
• fsdp: []
• fsdp_min_num_params: 0
• fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
• tp_size: 0
• fsdp_transformer_layer_cls_to_wrap: None
• accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
• deepspeed: None
• label_smoothing_factor: 0.0
• optim: adamw_torch
• optim_args: None
• adafactor: False
• group_by_length: False
• length_column_name: length
• ddp_find_unused_parameters: None
• ddp_bucket_cap_mb: None
• ddp_broadcast_buffers: False
• dataloader_pin_memory: True
• dataloader_persistent_workers: False
• skip_memory_metrics: True
• use_legacy_prediction_loop: False
• push_to_hub: False
• resume_from_checkpoint: None
• hub_model_id: None
• hub_strategy: every_save
• hub_private_repo: None
• hub_always_push: False
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• include_for_metrics: []
• eval_do_concat_batches: True
• fp16_backend: auto
• push_to_hub_model_id: None
• push_to_hub_organization: None
• mp_parameters:
• auto_find_batch_size: False
• full_determinism: False
• torchdynamo: None
• ray_scope: last
• ddp_timeout: 1800
• torch_compile: False
• torch_compile_backend: None
• torch_compile_mode: None
• include_tokens_per_second: False
• include_num_input_tokens_seen: False
• neftune_noise_alpha: None
• optim_target_modules: None
• batch_eval_metrics: False
• eval_on_start: False
• use_liger_kernel: False
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	Validation Loss	msmarco-eval-1kq-1kd_cosine_ndcg@10	NanoMSMARCO_cosine_ndcg@10	NanoNFCorpus_cosine_ndcg@10	NanoNQ_cosine_ndcg@10	NanoBEIR_mean_cosine_ndcg@10
0.032	20	626.2853	-	-	-	-	-	-
0.064	40	150.4493	-	-	-	-	-	-
0.096	60	67.3182	-	-	-	-	-	-
0.128	80	53.5684	-	-	-	-	-	-
0.16	100	37.7594	37.6225	0.0996	0.0127	0.0160	0.0	0.0096
0.192	120	34.6859	-	-	-	-	-	-
0.224	140	36.1137	-	-	-	-	-	-
0.256	160	30.4027	-	-	-	-	-	-
0.288	180	29.148	-	-	-	-	-	-
0.32	200	33.1368	29.7916	0.2748	0.0432	0.0228	0.0	0.0220
0.352	220	27.5536	-	-	-	-	-	-
0.384	240	27.2182	-	-	-	-	-	-
0.416	260	25.0055	-	-	-	-	-	-
0.448	280	24.3704	-	-	-	-	-	-
0.48	300	23.1422	23.1264	0.5906	0.1270	0.0205	0.0261	0.0579
0.512	320	22.3186	-	-	-	-	-	-
0.544	340	24.1421	-	-	-	-	-	-
0.576	360	20.9064	-	-	-	-	-	-
0.608	380	18.2549	-	-	-	-	-	-
0.64	400	19.5288	19.2762	0.7293	0.2216	0.0404	0.0758	0.1126
0.672	420	19.1762	-	-	-	-	-	-
0.704	440	17.4021	-	-	-	-	-	-
0.736	460	18.0734	-	-	-	-	-	-
0.768	480	17.8102	-	-	-	-	-	-
0.8	500	18.1116	17.3469	0.8224	0.2664	0.0576	0.1053	0.1431
0.832	520	16.9641	-	-	-	-	-	-
0.864	540	17.378	-	-	-	-	-	-
0.896	560	16.3021	-	-	-	-	-	-
0.928	580	14.9917	-	-	-	-	-	-
0.96	600	17.5367	16.2952	0.8594	0.3014	0.0721	0.1187	0.1641
0.992	620	15.6708	-	-	-	-	-	-
-1	-1	-	-	0.8656	0.3025	0.0743	0.1199	0.1656

Framework Versions

• Python: 3.11.10
• Sentence Transformers: 5.1.0.dev0
• Transformers: 4.51.2
• PyTorch: 2.5.1+cu124
• Accelerate: 1.5.2
• Datasets: 3.5.0
• 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",
}

MarginMSELoss

@misc{hofstätter2021improving,
    title={Improving Efficient Neural Ranking Models with Cross-Architecture Knowledge Distillation},
    author={Sebastian Hofstätter and Sophia Althammer and Michael Schröder and Mete Sertkan and Allan Hanbury},
    year={2021},
    eprint={2010.02666},
    archivePrefix={arXiv},
    primaryClass={cs.IR}
}