metadata

language: []
library_name: sentence-transformers
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:89218
  - loss:MultipleNegativesRankingLoss
base_model: sentence-transformers/multi-qa-mpnet-base-dot-v1
datasets: []
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
  - dot_accuracy@1
  - dot_accuracy@3
  - dot_accuracy@5
  - dot_accuracy@10
  - dot_precision@1
  - dot_precision@3
  - dot_precision@5
  - dot_precision@10
  - dot_recall@1
  - dot_recall@3
  - dot_recall@5
  - dot_recall@10
  - dot_ndcg@10
  - dot_mrr@10
  - dot_map@100
widget:
  - source_sentence: Pulmonary stenoses, brachytelephalangy, inner ear deafness
    sentences:
      - >-
        This article needs more medical references for verification or relies
        too heavily on primary sources. Please review the contents of the
        article and add the appropriate references if you can. Unsourced or
        poorly sourced material may be challenged and removed.  

        Find sources: "Chondropathy" – news · newspapers · books · scholar ·
        JSTOR (October 2020)  
          
        Chondropathy  

        SpecialtyOrthopedics  
          
        Chondropathy refers to a disease of the cartilage. It is frequently
        divided into 5 grades, with 0-2 defined as normal and 3-4 defined as
        diseased.


        ## Contents

          * 1 Some common diseases affecting/involving the cartilage
          * 2 Repairing articular cartilage damage
          * 3 References
          * 4 External links

        ## Some common diseases affecting/involving the cartilage[edit]
      - >-
        A number sign (#) is used with this entry because of evidence that
        Keutel syndrome (KTLS) is caused by homozygous mutation in the gene
        encoding the human matrix Gla protein (MGP; 154870) on chromosome 12p12.


        Description


        Keutel syndrome is an autosomal recessive disorder characterized by
        multiple peripheral pulmonary stenoses, brachytelephalangy, inner ear
        deafness, and abnormal cartilage ossification or calcification (summary
        by Khosroshahi et al., 2014).


        Clinical Features
      - >-
        ##  Description


        Primary or spontaneous detachment of the retina occurs due to underlying
        ocular disease and often involves the vitreous as well as the retina.
        The precipitating event is formation of a retinal tear or hole, which
        permits fluid to accumulate under the sensory layers of the retina and
        creates an intraretinal cleavage that destroys the neurosensory process
        of visual reception. Vitreoretinal degeneration and tear formation are
        painless phenomena, and in most cases, significant vitreoretinal
        pathology is found only after detachment of the retina starts to cause
        loss of vision or visual field. Without surgical intervention, retinal
        detachment will almost inevitably lead to total blindness (summary by
        McNiel and McPherson, 1971).


        Clinical Features
  - source_sentence: APS, catastrophic, diagnostic criteria, treatment options
    sentences:
      - >-
        A number sign (#) is used with this entry because of evidence that
        myofibrillar myopathy-8 (MFM8) is caused by homozygous or compound
        heterozygous mutation in the PYROXD1 gene (617220) on chromosome 12p12.


        Description


        Myofibrillar myopathy-8 is an autosomal recessive myopathy characterized
        by childhood onset of slowly progressive proximal muscle weakness and
        atrophy resulting in increased falls, gait problems, and difficulty
        running or climbing stairs. Upper and lower limbs are affected, and some
        individuals develop distal muscle weakness and atrophy. Ambulation is
        generally preserved, and patients do not have significant respiratory
        compromise. Muscle biopsy shows a mix of myopathic features, including
        myofibrillar inclusions and sarcomeric disorganization (summary by
        O'Grady et al., 2016).


        For a general phenotypic description and a discussion of genetic
        heterogeneity of myofibrillar myopathy, see MFM1 (601419).


        Clinical Features
      - >-
        Rectal tenesmus  

        SpecialtyGeneral surgery  
          
        Rectal tenesmus is a feeling of incomplete defecation. It is the
        sensation of inability or difficulty to empty the bowel at defecation,
        even if the bowel contents have already been evacuated. Tenesmus
        indicates the feeling of a residue, and is not always correlated with
        the actual presence of residual fecal matter in the rectum. It is
        frequently painful and may be accompanied by involuntary straining and
        other gastrointestinal symptoms. Tenesmus has both a nociceptive and a
        neuropathic component.


        Vesical tenesmus is a similar condition, experienced as a feeling of
        incomplete voiding despite the bladder being empty.


        Often, rectal tenesmus is simply called tenesmus. The term rectal
        tenesmus is a retronym to distinguish defecation-related tenesmus from
        vesical tenesmus.[1]
      - >-
        This article needs additional citations for verification. Please help
        improve this article by adding citations to reliable sources. Unsourced
        material may be challenged and removed.  

        Find sources: "Catastrophic antiphospholipid syndrome" – news ·
        newspapers · books · scholar · JSTOR (February 2018) (Learn how and when
        to remove this template message)  
          
        Catastrophic antiphospholipid syndrome  

        Other namesCatastrophic APS
  - source_sentence: Excess cholesterol, foam cells, gallbladder wall changes
    sentences:
      - >-
        Cholesterolosis of gallbladder  

        Micrograph of cholesterolosis of the gallbladder, with an annotated foam
        cell. H&E stain.  

        SpecialtyGastroenterology  
          
        In surgical pathology, strawberry gallbladder, more formally
        cholesterolosis of the gallbladder and gallbladder cholesterolosis, is a
        change in the gallbladder wall due to excess cholesterol.[1]


        The name strawberry gallbladder comes from the typically stippled
        appearance of the mucosal surface on gross examination, which resembles
        a strawberry. Cholesterolosis results from abnormal deposits of
        cholesterol esters in macrophages within the lamina propria (foam cells)
        and in mucosal epithelium. The gallbladder may be affected in a patchy
        localized form or in a diffuse form. The diffuse form macroscopically
        appears as a bright red mucosa with yellow mottling (due to lipid),
        hence the term strawberry gallbladder. It is not tied to cholelithiasis
        (gallstones) or cholecystitis (inflammation of the gallbladder).[2]


        ## Contents
      - >-
        Meningococcal meningitis is an acute bacterial disease caused by
        Neisseria meningitides that presents usually, but not always, with a
        rash (non blanching petechial or purpuric rash), progressively
        developing signs of meningitis (fever, vomiting, headache, photophobia,
        and neck stiffness) and later leading to confusion, delirium and
        drowsiness. Neck stiffness and photophobia are often absent in infants
        and young children who may manifest nonspecific signs such as
        irritability, inconsolable crying, poor feeding, and a bulging fontanel.
        Meningococcal meningitis may also present as part of early or late onset
        sepsis in neonates. The disease is potentially fatal. Surviving patients
        may develop neurological sequelae that include sensorineural hearing
        loss, seizures, spasticity, attention deficits and intellectual
        disability.
      - >-
        Retiform parapsoriasis  

        SpecialtyDermatology  
          
        Retiform parapsoriasis is a cutaneous condition, considered to be a type
        of large-plaque parapsoriasis.[1] It is characterized by widespread,
        ill-defined plaques on the skin, that have a net-like or zebra-striped
        pattern.[2] Skin atrophy, a wasting away of the cutaneous tissue,
        usually occurs within the area of these plaques.[1]


        ## See also[edit]

          * Parapsoriasis
          * Poikiloderma vasculare atrophicans
          * List of cutaneous conditions

        ## References[edit]

          1. ^ a b Lambert WC, Everett MA (Oct 1981). "The nosology of parapsoriasis". J. Am. Acad. Dermatol. 5 (4): 373–95. doi:10.1016/S0190-9622(81)70100-2. PMID 7026622.
          2. ^ Rapini, Ronald P.; Bolognia, Jean L.; Jorizzo, Joseph L. (2007). Dermatology: 2-Volume Set. St. Louis: Mosby. ISBN 1-4160-2999-0.

        ## External links[edit]


        Classification


        D

          * ICD-10: L41.5
          * ICD-9-CM: 696.2

          
          
          * v
          * t
          * e

        Papulosquamous disorders  
          
        Psoriasis


        Pustular
  - source_sentence: >-
      Pulmonary hypoplasia, respiratory insufficiency, megaureter,
      hydronephrosis
    sentences:
      - >-
        A rare fetal lower urinary tract obstruction (LUTO) characterized by
        closure or failure to develop an opening in the urethra and resulting in
        obstructive uropathy presenting in utero as megacystis, oligohydramnios
        or anhydramnios, and potter sequence.


        ## Epidemiology


        Prevalence is unknown, but is higher in males than females.


        ## Clinical description


        Atresia of urethra often presents on routine antenatal ultrasound with
        megacystis, oligohydramnios or anhydramnios and sometimes urinary
        ascites. It may cause fetal death. In cases that survive to birth,
        additional symptoms include respiratory insufficiency due to pulmonary
        hypoplasia, megaureter, hydronephrosis and enlarged often cystic and
        functionally impaired/non-functional dysplastic kidneys as well as
        abdominal distention. Furthermore, a Potter sequence can be found due to
        oligo- or anhydramnios. Patients may present with patent urachus or
        vesicocutaneous fistula.


        ## Etiology
      - >-
        X-linked distal spinal muscular atrophy type 3 is a rare distal
        hereditary motor neuropathy characterized by slowly progressive atrophy
        and weakness of distal muscles of hands and feet with normal deep tendon
        reflexes or absent ankle reflexes and minimal or no sensory loss,
        sometimes mild proximal weakness in the legs and feet and hand
        deformities in males.
      - >-
        A number sign (#) is used with this entry because Chudley-McCullough
        syndrome (CMCS) is caused by homozygous or compound heterozygous
        mutation in the GPSM2 gene (609245) on chromosome 1p13.


        Description


        Chudley-McCullough syndrome is an autosomal recessive neurologic
        disorder characterized by early-onset sensorineural deafness and
        specific brain anomalies on MRI, including hypoplasia of the corpus
        callosum, enlarged cysterna magna with mild focal cerebellar dysplasia,
        and nodular heterotopia. Some patients have hydrocephalus. Psychomotor
        development is normal (summary by Alrashdi et al., 2011).


        Clinical Features
  - source_sentence: >-
      Thyroid-stimulating hormone receptor gene, chromosome 14q31, homozygous
      mutation
    sentences:
      - >-
        A number sign (#) is used with this entry because dermatofibrosarcoma
        protuberans is caused in most cases by a specific fusion of the COL1A1
        gene (120150) with the PDGFB gene (190040); see 190040.0002.


        Description


        Dermatofibrosarcoma protuberans (DFSP) is an uncommon, locally
        aggressive, but rarely metastasizing tumor of the deep dermis and
        subcutaneous tissue. It typically presents during early or middle adult
        life and is most frequently located on the trunk and proximal
        extremities (Sandberg et al., 2003).


        Clinical Features


        DFSP was first described by Taylor (1890). Sirvent et al. (2003) stated
        that, because DFSP is relatively rare, grows slowly, and has a low level
        of aggressiveness, its clinical significance has been underestimated. In
        particular, they noted that the existence of pediatric cases has been
        overlooked.


        Gardner et al. (1998) described a father and son with
        dermatofibrosarcoma protuberans. The tumors arose at ages 43 and 14
        years, respectively.
      - >-
        Visuospatial dysgnosia is a loss of the sense of "whereness" in the
        relation of oneself to one's environment and in the relation of objects
        to each other.[1] Visuospatial dysgnosia is often linked with
        topographical disorientation.


        ## Contents

          * 1 Symptoms
          * 2 Lesion areas
          * 3 Case studies
          * 4 Therapies
          * 5 References

        ## Symptoms[edit]


        The syndrome rarely presents itself the same way in every patient. Some
        symptoms that occur may be:
      - >-
        A number sign (#) is used with this entry because of evidence that
        congenital nongoitrous hypothyroidism-1 (CHNG1) is caused by homozygous
        or compound heterozygous mutation in the gene encoding the
        thyroid-stimulating hormone receptor (TSHR; 603372) on chromosome 14q31.


        Description


        Resistance to thyroid-stimulating hormone (TSH; see 188540), a hallmark
        of congenital nongoitrous hypothyroidism, causes increased levels of
        plasma TSH and low levels of thyroid hormone. Only a subset of patients
        develop frank hypothyroidism; the remainder are euthyroid and
        asymptomatic (so-called compensated hypothyroidism) and are usually
        detected by neonatal screening programs (Paschke and Ludgate, 1997).


        ### Genetic Heterogeneity of Congenital Nongoitrous Hypothyroidism
pipeline_tag: sentence-similarity
model-index:
  - name: >-
      SentenceTransformer based on
      sentence-transformers/multi-qa-mpnet-base-dot-v1
    results:
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: Unknown
          type: unknown
        metrics:
          - type: cosine_accuracy@1
            value: 0.1900990099009901
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.5756875687568757
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.7932893289328933
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.8704070407040704
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.1900990099009901
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.19189585625229189
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.15865786578657867
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.08704070407040705
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.1900990099009901
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.5756875687568757
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.7932893289328933
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.8704070407040704
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.526584144074431
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.41522683220700946
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.4194005014371134
            name: Cosine Map@100
          - type: dot_accuracy@1
            value: 0.188998899889989
            name: Dot Accuracy@1
          - type: dot_accuracy@3
            value: 0.5761826182618262
            name: Dot Accuracy@3
          - type: dot_accuracy@5
            value: 0.7954895489548955
            name: Dot Accuracy@5
          - type: dot_accuracy@10
            value: 0.8710671067106711
            name: Dot Accuracy@10
          - type: dot_precision@1
            value: 0.188998899889989
            name: Dot Precision@1
          - type: dot_precision@3
            value: 0.19206087275394204
            name: Dot Precision@3
          - type: dot_precision@5
            value: 0.15909790979097907
            name: Dot Precision@5
          - type: dot_precision@10
            value: 0.08710671067106711
            name: Dot Precision@10
          - type: dot_recall@1
            value: 0.188998899889989
            name: Dot Recall@1
          - type: dot_recall@3
            value: 0.5761826182618262
            name: Dot Recall@3
          - type: dot_recall@5
            value: 0.7954895489548955
            name: Dot Recall@5
          - type: dot_recall@10
            value: 0.8710671067106711
            name: Dot Recall@10
          - type: dot_ndcg@10
            value: 0.5265923432373186
            name: Dot Ndcg@10
          - type: dot_mrr@10
            value: 0.4149802896956161
            name: Dot Mrr@10
          - type: dot_map@100
            value: 0.41904239679820193
            name: Dot Map@100

SentenceTransformer based on sentence-transformers/multi-qa-mpnet-base-dot-v1

This is a sentence-transformers model finetuned from sentence-transformers/multi-qa-mpnet-base-dot-v1. 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: sentence-transformers/multi-qa-mpnet-base-dot-v1
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 tokens
Similarity Function: Dot Product

Model Sources

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

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: MPNetModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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("sentence_transformers_model_id")
# Run inference
sentences = [
    'Thyroid-stimulating hormone receptor gene, chromosome 14q31, homozygous mutation',
    'A number sign (#) is used with this entry because of evidence that congenital nongoitrous hypothyroidism-1 (CHNG1) is caused by homozygous or compound heterozygous mutation in the gene encoding the thyroid-stimulating hormone receptor (TSHR; 603372) on chromosome 14q31.\n\nDescription\n\nResistance to thyroid-stimulating hormone (TSH; see 188540), a hallmark of congenital nongoitrous hypothyroidism, causes increased levels of plasma TSH and low levels of thyroid hormone. Only a subset of patients develop frank hypothyroidism; the remainder are euthyroid and asymptomatic (so-called compensated hypothyroidism) and are usually detected by neonatal screening programs (Paschke and Ludgate, 1997).\n\n### Genetic Heterogeneity of Congenital Nongoitrous Hypothyroidism',
    'Visuospatial dysgnosia is a loss of the sense of "whereness" in the relation of oneself to one\'s environment and in the relation of objects to each other.[1] Visuospatial dysgnosia is often linked with topographical disorientation.\n\n## Contents\n\n  * 1 Symptoms\n  * 2 Lesion areas\n  * 3 Case studies\n  * 4 Therapies\n  * 5 References\n\n## Symptoms[edit]\n\nThe syndrome rarely presents itself the same way in every patient. Some symptoms that occur may be:',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Information Retrieval

Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1901
cosine_accuracy@3	0.5757
cosine_accuracy@5	0.7933
cosine_accuracy@10	0.8704
cosine_precision@1	0.1901
cosine_precision@3	0.1919
cosine_precision@5	0.1587
cosine_precision@10	0.087
cosine_recall@1	0.1901
cosine_recall@3	0.5757
cosine_recall@5	0.7933
cosine_recall@10	0.8704
cosine_ndcg@10	0.5266
cosine_mrr@10	0.4152
cosine_map@100	0.4194
dot_accuracy@1	0.189
dot_accuracy@3	0.5762
dot_accuracy@5	0.7955
dot_accuracy@10	0.8711
dot_precision@1	0.189
dot_precision@3	0.1921
dot_precision@5	0.1591
dot_precision@10	0.0871
dot_recall@1	0.189
dot_recall@3	0.5762
dot_recall@5	0.7955
dot_recall@10	0.8711
dot_ndcg@10	0.5266
dot_mrr@10	0.415
dot_map@100	0.419

Training Details

Training Dataset

Unnamed Dataset

Size: 89,218 training samples
Columns: queries and chunks
Approximate statistics based on the first 1000 samples:
queries chunks
type string string
details
min: 7 tokens
mean: 18.07 tokens
max: 63 tokens

min: 5 tokens
mean: 161.59 tokens
max: 299 tokens

	queries	chunks
type	string	string
details	min: 7 tokens mean: 18.07 tokens max: 63 tokens	min: 5 tokens mean: 161.59 tokens max: 299 tokens

Samples:

queries	chunks
`Polyhydramnios, megalencephaly, symptomatic epilepsy`	`A number sign (#) is used with this entry because of evidence that polyhydramnios, megalencephaly, and symptomatic epilepsy (PMSE) is caused by homozygous mutation in the STRADA gene (608626) on chromosome 17q23. Clinical Features`
`Polyhydramnios, megalencephaly, STRADA gene mutation`	`A number sign (#) is used with this entry because of evidence that polyhydramnios, megalencephaly, and symptomatic epilepsy (PMSE) is caused by homozygous mutation in the STRADA gene (608626) on chromosome 17q23. Clinical Features`
`Megalencephaly, symptomatic epilepsy, chromosome 17q23`	`A number sign (#) is used with this entry because of evidence that polyhydramnios, megalencephaly, and symptomatic epilepsy (PMSE) is caused by homozygous mutation in the STRADA gene (608626) on chromosome 17q23. Clinical Features`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 1,
    "similarity_fct": "dot_score"
}

Evaluation Dataset

Unnamed Dataset

Size: 18,180 evaluation samples
Columns: queries and chunks
Approximate statistics based on the first 1000 samples:
queries chunks
type string string
details
min: 6 tokens
mean: 18.35 tokens
max: 82 tokens

min: 4 tokens
mean: 152.55 tokens
max: 312 tokens

	queries	chunks
type	string	string
details	min: 6 tokens mean: 18.35 tokens max: 82 tokens	min: 4 tokens mean: 152.55 tokens max: 312 tokens

Samples:

queries	chunks
`Weight loss, anorexia, fatigue, epigastric pain and discomfort`	`Undifferentiated carcinoma of stomach is a rare epithelial tumour of the stomach that lacks any features of differentiation beyond an epithelial phenotype. The presenting symptoms are usually vague and nonspecific, such as weight loss, anorexia, fatigue, epigastric pain and discomfort, heartburn and nausea, vomiting or hematemesis. Patients may also be asymptomatic. Ascites, jaundice, intestinal obstruction and peripheral lymphadenopathy indicate advanced stages and metastatic spread.`
`Heartburn, nausea, vomiting, hematemesis`	`Undifferentiated carcinoma of stomach is a rare epithelial tumour of the stomach that lacks any features of differentiation beyond an epithelial phenotype. The presenting symptoms are usually vague and nonspecific, such as weight loss, anorexia, fatigue, epigastric pain and discomfort, heartburn and nausea, vomiting or hematemesis. Patients may also be asymptomatic. Ascites, jaundice, intestinal obstruction and peripheral lymphadenopathy indicate advanced stages and metastatic spread.`
`Ascites, jaundice, intestinal obstruction, peripheral lymphadenopathy`	`Undifferentiated carcinoma of stomach is a rare epithelial tumour of the stomach that lacks any features of differentiation beyond an epithelial phenotype. The presenting symptoms are usually vague and nonspecific, such as weight loss, anorexia, fatigue, epigastric pain and discomfort, heartburn and nausea, vomiting or hematemesis. Patients may also be asymptomatic. Ascites, jaundice, intestinal obstruction and peripheral lymphadenopathy indicate advanced stages and metastatic spread.`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 1,
    "similarity_fct": "dot_score"
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 32
per_device_eval_batch_size: 32
learning_rate: 2e-05
num_train_epochs: 50
warmup_ratio: 0.1
fp16: True
load_best_model_at_end: True
eval_on_start: True
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 32
per_device_eval_batch_size: 32
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: 50
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: 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: True
dataloader_num_workers: 0
dataloader_prefetch_factor: None
past_index: -1
disable_tqdm: False
remove_unused_columns: True
label_names: None
load_best_model_at_end: True
ignore_data_skip: False
fsdp: []
fsdp_min_num_params: 0
fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
fsdp_transformer_layer_cls_to_wrap: None
accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
deepspeed: None
label_smoothing_factor: 0.0
optim: adamw_torch
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: False
hub_always_push: False
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_inputs_for_metrics: False
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: True
eval_use_gather_object: False
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	loss	dot_map@100
0	0	-	1.1605	0.2419
0.1435	100	1.2016	-	-
0.2869	200	0.7627	-	-
0.4304	300	0.5559	-	-
0.5739	400	0.4541	-	-
0.7174	500	0.1451	0.3600	0.3913
0.8608	600	0.3841	-	-
1.0057	700	0.3334	-	-
1.1492	800	0.3898	-	-
1.2927	900	0.3576	-	-
1.4362	1000	0.3563	0.2719	0.4127
1.5796	1100	0.3186	-	-
1.7231	1200	0.098	-	-
1.8666	1300	0.3038	-	-
2.0115	1400	0.2629	-	-
2.1549	1500	0.3221	0.2579	0.4155
2.2984	1600	0.2936	-	-
2.4419	1700	0.2867	-	-
2.5854	1800	0.2614	-	-
2.7288	1900	0.0716	-	-
2.8723	2000	0.2655	0.2546	0.4152
3.0172	2100	0.2187	-	-
3.1607	2200	0.2623	-	-
3.3042	2300	0.2462	-	-
3.4476	2400	0.2363	-	-
3.5911	2500	0.213	0.2866	0.4227
3.7346	2600	0.0487	-	-
3.8780	2700	0.222	-	-
4.0230	2800	0.1851	-	-
4.1664	2900	0.224	-	-
4.3099	3000	0.2111	0.2562	0.4215
4.4534	3100	0.1984	-	-
4.5968	3200	0.1707	-	-
4.7403	3300	0.0331	-	-
4.8838	3400	0.1896	-	-
5.0287	3500	0.1548	0.2643	0.4151
5.1722	3600	0.19	-	-
5.3156	3700	0.1656	-	-
5.4591	3800	0.1626	-	-
5.6026	3900	0.1303	-	-
5.7461	4000	0.0264	0.2952	0.4186
5.8895	4100	0.1563	-	-
6.0344	4200	0.1286	-	-
6.1779	4300	0.1436	-	-
6.3214	4400	0.1352	-	-
6.4648	4500	0.1344	0.2668	0.4218
6.6083	4600	0.1069	-	-
6.7518	4700	0.0171	-	-
6.8953	4800	0.1246	-	-
7.0402	4900	0.1074	-	-
7.1836	5000	0.1192	0.2837	0.4166
7.3271	5100	0.1176	-	-
7.4706	5200	0.111	-	-
7.6141	5300	0.0889	-	-
7.7575	5400	0.0202	-	-
7.9010	5500	0.1059	0.2797	0.4166
8.0459	5600	0.0854	-	-
8.1894	5700	0.0989	-	-
8.3329	5800	0.0963	-	-
8.4763	5900	0.0967	-	-
8.6198	6000	0.0635	0.2974	0.4223
8.7633	6100	0.0215	-	-
8.9067	6200	0.0897	-	-
9.0516	6300	0.0693	-	-
9.1951	6400	0.0913	-	-
9.3386	6500	0.0883	0.2812	0.4171
9.4821	6600	0.0849	-	-
9.6255	6700	0.0525	-	-
9.7690	6800	0.0196	-	-
9.9125	6900	0.0799	-	-
10.0574	7000	0.0603	0.2899	0.4132
10.2009	7100	0.0816	-	-
10.3443	7200	0.0771	-	-
10.4878	7300	0.0746	-	-
10.6313	7400	0.0373	-	-
10.7747	7500	0.0181	0.3148	0.419
10.9182	7600	0.0702	-	-
11.0631	7700	0.0531	-	-
11.2066	7800	0.0671	-	-
11.3501	7900	0.0742	-	-
11.4935	8000	0.0728	0.2878	0.4177
11.6370	8100	0.0331	-	-
11.7805	8200	0.0206	-	-
11.9240	8300	0.0605	-	-
12.0689	8400	0.05	-	-
12.2123	8500	0.06	0.3169	0.4180
12.3558	8600	0.0613	-	-
12.4993	8700	0.0649	-	-
12.6428	8800	0.0257	-	-
12.7862	8900	0.0184	-	-
12.9297	9000	0.055	0.3107	0.4189
13.0746	9100	0.0417	-	-
13.2181	9200	0.0537	-	-
13.3615	9300	0.0558	-	-
13.5050	9400	0.0619	-	-
13.6485	9500	0.0217	0.3140	0.4173
13.7920	9600	0.0257	-	-
13.9354	9700	0.0398	-	-
14.0803	9800	0.041	-	-
14.2238	9900	0.0451	-	-
14.3673	10000	0.0485	0.3085	0.4188
14.5108	10100	0.0565	-	-
14.6542	10200	0.0159	-	-
14.7977	10300	0.0258	-	-
14.9412	10400	0.0364	-	-
15.0861	10500	0.0368	0.3144	0.4163
15.2296	10600	0.0447	-	-
15.3730	10700	0.0479	-	-
15.5165	10800	0.0535	-	-
15.6600	10900	0.0139	-	-
15.8034	11000	0.0257	0.3149	0.4151
15.9469	11100	0.0324	-	-
16.0918	11200	0.0374	-	-
16.2353	11300	0.0339	-	-
16.3788	11400	0.0423	-	-
16.5222	11500	0.0512	0.3209	0.4164
16.6657	11600	0.0121	-	-
16.8092	11700	0.0245	-	-
16.9527	11800	0.0323	-	-
17.0976	11900	0.0321	-	-
17.2410	12000	0.034	0.3211	0.4140
17.3845	12100	0.0387	-	-
17.5280	12200	0.0482	-	-
17.6714	12300	0.0096	-	-
17.8149	12400	0.0252	-	-
17.9584	12500	0.0299	0.3169	0.4170
18.1033	12600	0.0351	-	-
18.2468	12700	0.032	-	-
18.3902	12800	0.0348	-	-
18.5337	12900	0.0452	-	-
18.6772	13000	0.0076	0.3273	0.4158
18.8207	13100	0.0241	-	-
18.9641	13200	0.0277	-	-
19.1090	13300	0.0331	-	-
19.2525	13400	0.0264	-	-
19.3960	13500	0.0311	0.3272	0.4151
19.5395	13600	0.0437	-	-
19.6829	13700	0.0049	-	-
19.8264	13800	0.0263	-	-
19.9699	13900	0.0231	-	-
20.1148	14000	0.0303	0.3293	0.4200
20.2582	14100	0.0229	-	-
20.4017	14200	0.032	-	-
20.5452	14300	0.0395	-	-
20.6887	14400	0.0045	-	-
20.8321	14500	0.0244	0.3202	0.4144
20.9756	14600	0.0219	-	-
21.1205	14700	0.0291	-	-
21.2640	14800	0.0212	-	-
21.4075	14900	0.029	-	-
21.5509	15000	0.0357	0.3312	0.4147
21.6944	15100	0.0025	-	-
21.8379	15200	0.0252	-	-
21.9813	15300	0.0229	-	-
22.1263	15400	0.0261	-	-
22.2697	15500	0.0198	0.3392	0.4123
22.4132	15600	0.0259	-	-
22.5567	15700	0.0343	-	-
22.7001	15800	0.0022	-	-
22.8436	15900	0.0237	-	-
22.9871	16000	0.0199	0.3346	0.4146
23.1320	16100	0.0263	-	-
23.2755	16200	0.0173	-	-
23.4189	16300	0.0276	-	-
23.5624	16400	0.03	-	-
23.7059	16500	0.0022	0.3430	0.4195
23.8494	16600	0.0253	-	-
23.9928	16700	0.0182	-	-
24.1377	16800	0.0216	-	-
24.2812	16900	0.0194	-	-
24.4247	17000	0.0242	0.3335	0.4132
24.5681	17100	0.0289	-	-
24.7116	17200	0.0013	-	-
24.8551	17300	0.0253	-	-
24.9986	17400	0.0137	-	-
25.1435	17500	0.0219	0.3481	0.4118
25.2869	17600	0.017	-	-
25.4304	17700	0.0261	-	-
25.5739	17800	0.0298	-	-
25.7174	17900	0.0013	-	-
25.8608	18000	0.0257	0.3407	0.4160
26.0057	18100	0.014	-	-
26.1492	18200	0.0215	-	-
26.2927	18300	0.0161	-	-
26.4362	18400	0.0228	-	-
26.5796	18500	0.0246	0.3404	0.4131
26.7231	18600	0.0017	-	-
26.8666	18700	0.0244	-	-
27.0115	18800	0.0124	-	-
27.1549	18900	0.019	-	-
27.2984	19000	0.0151	0.3451	0.4139
27.4419	19100	0.0216	-	-
27.5854	19200	0.0255	-	-
27.7288	19300	0.0016	-	-
27.8723	19400	0.0251	-	-
28.0172	19500	0.0133	0.3416	0.4109
28.1607	19600	0.016	-	-
28.3042	19700	0.0186	-	-
28.4476	19800	0.0185	-	-
28.5911	19900	0.0225	-	-
28.7346	20000	0.0009	0.3463	0.4144
28.8780	20100	0.0249	-	-
29.0230	20200	0.0132	-	-
29.1664	20300	0.0145	-	-
29.3099	20400	0.0174	-	-
29.4534	20500	0.0172	0.3425	0.4092
29.5968	20600	0.0235	-	-
29.7403	20700	0.0009	-	-
29.8838	20800	0.0242	-	-
30.0287	20900	0.0128	-	-
30.1722	21000	0.0133	0.3482	0.4131
30.3156	21100	0.0158	-	-
30.4591	21200	0.0226	-	-
30.6026	21300	0.0188	-	-
30.7461	21400	0.0009	-	-
30.8895	21500	0.0249	0.3483	0.4132
31.0344	21600	0.0116	-	-
31.1779	21700	0.0117	-	-
31.3214	21800	0.0162	-	-
31.4648	21900	0.0184	-	-
31.6083	22000	0.0178	0.3390	0.4145
31.7518	22100	0.0012	-	-
31.8953	22200	0.0215	-	-
32.0402	22300	0.014	-	-
32.1836	22400	0.0105	-	-
32.3271	22500	0.0131	0.3556	0.4144
32.4706	22600	0.0199	-	-
32.6141	22700	0.0158	-	-
32.7575	22800	0.0018	-	-
32.9010	22900	0.0236	-	-
33.0459	23000	0.0131	0.3480	0.4136
33.1894	23100	0.0121	-	-
33.3329	23200	0.0164	-	-
33.4763	23300	0.0209	-	-
33.6198	23400	0.0119	-	-
33.7633	23500	0.0029	0.3575	0.4180
33.9067	23600	0.0201	-	-
34.0516	23700	0.0121	-	-
34.1951	23800	0.0109	-	-
34.3386	23900	0.0132	-	-
34.4821	24000	0.0203	0.3446	0.4141
34.6255	24100	0.0087	-	-
34.7690	24200	0.0032	-	-
34.9125	24300	0.0182	-	-
35.0574	24400	0.0116	-	-
35.2009	24500	0.0105	0.3587	0.4117
35.3443	24600	0.018	-	-
35.4878	24700	0.0194	-	-
35.6313	24800	0.0076	-	-
35.7747	24900	0.0029	-	-
35.9182	25000	0.0167	0.3529	0.4156
36.0631	25100	0.0105	-	-
36.2066	25200	0.0097	-	-
36.3501	25300	0.0165	-	-
36.4935	25400	0.0187	-	-
36.6370	25500	0.0062	0.3517	0.4173
36.7805	25600	0.0034	-	-
36.9240	25700	0.0173	-	-
37.0689	25800	0.0091	-	-
37.2123	25900	0.0093	-	-
37.3558	26000	0.0152	0.3605	0.4147
37.4993	26100	0.0193	-	-
37.6428	26200	0.0065	-	-
37.7862	26300	0.0036	-	-
37.9297	26400	0.017	-	-
38.0746	26500	0.009	0.3627	0.4178
38.2181	26600	0.0087	-	-
38.3615	26700	0.0129	-	-
38.5050	26800	0.0199	-	-
38.6485	26900	0.0047	-	-
38.7920	27000	0.0104	0.3535	0.4191
38.9354	27100	0.0106	-	-
39.0803	27200	0.0083	-	-
39.2238	27300	0.0091	-	-
39.3673	27400	0.0143	-	-
39.5108	27500	0.018	0.3586	0.4137
39.6542	27600	0.0055	-	-
39.7977	27700	0.0097	-	-
39.9412	27800	0.0111	-	-
40.0861	27900	0.0091	-	-
40.2296	28000	0.009	0.3540	0.4166
40.3730	28100	0.0145	-	-
40.5165	28200	0.0165	-	-
40.6600	28300	0.0041	-	-
40.8034	28400	0.009	-	-
40.9469	28500	0.0091	0.3541	0.4159
41.0918	28600	0.0106	-	-
41.2353	28700	0.0064	-	-
41.3788	28800	0.0125	-	-
41.5222	28900	0.0172	-	-
41.6657	29000	0.0028	0.3550	0.4151
41.8092	29100	0.0097	-	-
41.9527	29200	0.0086	-	-
42.0976	29300	0.0099	-	-
42.2410	29400	0.0064	-	-
42.3845	29500	0.0127	0.3619	0.4150
42.5280	29600	0.0157	-	-
42.6714	29700	0.0025	-	-
42.8149	29800	0.0095	-	-
42.9584	29900	0.0087	-	-
43.1033	30000	0.0094	0.3591	0.4153
43.2468	30100	0.007	-	-
43.3902	30200	0.0114	-	-
43.5337	30300	0.0166	-	-
43.6772	30400	0.0023	-	-
43.8207	30500	0.01	0.3582	0.4172
43.9641	30600	0.0097	-	-
44.1090	30700	0.01	-	-
44.2525	30800	0.007	-	-
44.3960	30900	0.0106	-	-
44.5395	31000	0.0164	0.3626	0.4151
44.6829	31100	0.0017	-	-
44.8264	31200	0.0113	-	-
44.9699	31300	0.0081	-	-
45.1148	31400	0.0095	-	-
45.2582	31500	0.0061	0.3669	0.4152
45.4017	31600	0.0111	-	-
45.5452	31700	0.0157	-	-
45.6887	31800	0.0015	-	-
45.8321	31900	0.0109	-	-
45.9756	32000	0.0085	0.3595	0.4139
46.1205	32100	0.0096	-	-
46.2640	32200	0.0062	-	-
46.4075	32300	0.0111	-	-
46.5509	32400	0.017	-	-
46.6944	32500	0.0013	0.3631	0.4154
46.8379	32600	0.0123	-	-
46.9813	32700	0.0076	-	-
47.1263	32800	0.0088	-	-
47.2697	32900	0.0065	-	-
47.4132	33000	0.0116	0.3656	0.4148
47.5567	33100	0.0142	-	-
47.7001	33200	0.0009	-	-
47.8436	33300	0.0101	-	-
47.9871	33400	0.0069	-	-
48.1320	33500	0.0087	0.3643	0.4160
48.2755	33600	0.005	-	-
48.4189	33700	0.0118	-	-
48.5624	33800	0.0147	-	-
48.7059	33900	0.0008	-	-
48.8494	34000	0.0115	0.3632	0.4158
48.9928	34100	0.006	-	-
49.1377	34200	0.0089	-	-
49.2812	34300	0.0063	-	-
49.4247	34400	0.0126	-	-
49.5681	34500	0.0142	0.3643	0.4157
49.7116	34600	0.0008	-	-
49.8551	34700	0.0137	-	-
49.9986	34800	0.0044	0.3148	0.4190

The bold row denotes the saved checkpoint.

Framework Versions

Python: 3.11.9
Sentence Transformers: 3.0.1
Transformers: 4.43.3
PyTorch: 2.3.1+cu121
Accelerate: 0.30.1
Datasets: 2.19.2
Tokenizers: 0.19.1

Citation

BibTeX

Sentence Transformers

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply}, 
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}