metadata

language: []
library_name: sentence-transformers
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:1182198
  - loss:CachedMultipleNegativesRankingLoss
  - loss:AnglELoss
base_model: nomic-ai/nomic-embed-text-v1.5
datasets: []
metrics:
  - cosine_accuracy
  - dot_accuracy
  - manhattan_accuracy
  - euclidean_accuracy
  - max_accuracy
  - pearson_cosine
  - spearman_cosine
  - pearson_manhattan
  - spearman_manhattan
  - pearson_euclidean
  - spearman_euclidean
  - pearson_dot
  - spearman_dot
  - pearson_max
  - spearman_max
widget:
  - source_sentence: dog instrument toy
    sentences:
      - >-
        VATOS 25-in-1 Mars Rover Building Kit Outer Space Explorer Educational
        Construction Toy for Kids 556 Pieces Solar Powered STEM Science Building
        Blocks Set, VATOS, White
      - >-
        Prefer Green 7 PCS Portion Control Containers Kit (with COMPLETE GUIDE &
        21 DAY DAILY TRACKER & 21 DAY MEAL PLANNER & RECIPES
        PDFs),Label-Coded,Multi-Color-Coded System,Perfect Size for Lose Weight,
        Prefer Green, 7 PCS
      - >-
        Coolibar UPF 50+ Men's Women's Gannett UV Gloves - Sun Protective
        (Medium- Light Blue), Coolibar, Light Blue
  - source_sentence: flame decal stickers
    sentences:
      - >-
        Tribal Flames Splash Pair - Vinyl Decal Sticker - 12" x 5" - Blue
        Flames, Sticker Pimp, Blue Flames
      - >-
        PC Gaming Headset Headphone Hook Holder Hanger Mount, Headphones Stand
        with Adjustable & Rotating Arm Clamp , Under Desk Design , Universal Fit
        , Built in Cable Clip Organizer EURPMASK, EURPMASK Choose the color of
        europe, Black
      - >-
        Quick Charge 3.0 Wall Charger, 4-Pack 18W QC 3.0 USB Charger Adapter
        Fast Charging Block Compatible Wireless Charger Compatible with Samsung
        Galaxy S10 S9 S8 Plus S7 S6 Edge Note 9, LG, Kindle, Tablet, HONOT,
        Black
  - source_sentence: 'search_query: softies women''s ultra soft marshmallow hooded lounger'
    sentences:
      - >-
        search_document: Red-A Placemats for Dining Table Set of 6
        Heat-Resistant Wipeable Table Mats for Kitchen Table Decoration
        Waterproof Vinyl Placemats Easy to Clean,Black w/Brown, Red-A, Black
      - >-
        search_document: Softies Women's Ultra Soft Marshmallow Hooded Lounger,
        Platinum, L/XL, Softies, Platinum
      - >-
        search_document: Ekouaer Women's Sleepwear Robe with Pockets Plus Size
        Maxi Lounger Zipper Short Sleeve Bathrobe Housecoat (Black,L), Ekouaer,
        Black
  - source_sentence: 'search_query: wine glasses without stem'
    sentences:
      - >-
        search_document: STAUBER Best Bulb Changer with PowerLatch Extension
        Pole (Large Suction, 4 Feet), STAUBER, Large Suction
      - >-
        search_document: Hand Blown Italian Style Crystal Burgundy Wine Glasses
        - Lead-Free Premium Crystal Clear Glass - Set of 2 - 21 Ounce - Gift-Box
        for any Occasion, JBHO, Burgundy
      - >-
        search_document: MyGift Modern Copper Stemless Wine Glasses, Set of 4,
        MyGift, Copper
  - source_sentence: 'search_query: weighted blanket without glass beads'
    sentences:
      - >-
        search_document: Eigso Women Men Spike Punk Rock Black Leather Cuff
        Rivet Bracelet Bangle Adjustable Snap Button, Eigso, Black
      - >-
        search_document: Quility Weighted Blanket with Soft Cover - 20 lbs
        Full/Queen Size Heavy Blanket for Adults - Heating & Cooling, Machine
        Washable - (60" X 80") (Navy), Quility, Navy Cover + Grey Cotton Blanket
      - >-
        search_document: Bedsure Queen Weighted Blanket 15 Pounds - Adult
        Weighted Blanket 60x80 - Soft Heavy Blanket with Breathable TPE Insert
        No Glass Beads, Bedsure, Navy
pipeline_tag: sentence-similarity
model-index:
  - name: SentenceTransformer based on nomic-ai/nomic-embed-text-v1.5
    results:
      - task:
          type: triplet
          name: Triplet
        dataset:
          name: Unknown
          type: unknown
        metrics:
          - type: cosine_accuracy
            value: 0.7236
            name: Cosine Accuracy
          - type: dot_accuracy
            value: 0.282
            name: Dot Accuracy
          - type: manhattan_accuracy
            value: 0.7231
            name: Manhattan Accuracy
          - type: euclidean_accuracy
            value: 0.7227
            name: Euclidean Accuracy
          - type: max_accuracy
            value: 0.7236
            name: Max Accuracy
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: Unknown
          type: unknown
        metrics:
          - type: pearson_cosine
            value: 0.4912162846043421
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.4658522123059972
            name: Spearman Cosine
          - type: pearson_manhattan
            value: 0.4599741171303018
            name: Pearson Manhattan
          - type: spearman_manhattan
            value: 0.4428141949345816
            name: Spearman Manhattan
          - type: pearson_euclidean
            value: 0.46194545823984606
            name: Pearson Euclidean
          - type: spearman_euclidean
            value: 0.44478471500226807
            name: Spearman Euclidean
          - type: pearson_dot
            value: 0.45451995456560107
            name: Pearson Dot
          - type: spearman_dot
            value: 0.43844636325741904
            name: Spearman Dot
          - type: pearson_max
            value: 0.4912162846043421
            name: Pearson Max
          - type: spearman_max
            value: 0.4658522123059972
            name: Spearman Max

SentenceTransformer based on nomic-ai/nomic-embed-text-v1.5

This is a sentence-transformers model finetuned from nomic-ai/nomic-embed-text-v1.5 on the triplets and pairs datasets. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

Model Type: Sentence Transformer
Base model: nomic-ai/nomic-embed-text-v1.5
Maximum Sequence Length: 8192 tokens
Output Dimensionality: 768 tokens
Similarity Function: Cosine Similarity
Training Datasets:
- triplets
- pairs

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}) with Transformer model: NomicBertModel 
  (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("lv12/esci-nomic-embed-text-v1_5_4")
# Run inference
sentences = [
    'search_query: weighted blanket without glass beads',
    'search_document: Bedsure Queen Weighted Blanket 15 Pounds - Adult Weighted Blanket 60x80 - Soft Heavy Blanket with Breathable TPE Insert No Glass Beads, Bedsure, Navy',
    'search_document: Quility Weighted Blanket with Soft Cover - 20 lbs Full/Queen Size Heavy Blanket for Adults - Heating & Cooling, Machine Washable - (60" X 80") (Navy), Quility, Navy Cover + Grey Cotton Blanket',
]
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

Triplet

Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.7236
dot_accuracy	0.282
manhattan_accuracy	0.7231
euclidean_accuracy	0.7227
max_accuracy	0.7236

Semantic Similarity

Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.4912
spearman_cosine	0.4659
pearson_manhattan	0.46
spearman_manhattan	0.4428
pearson_euclidean	0.4619
spearman_euclidean	0.4448
pearson_dot	0.4545
spearman_dot	0.4384
pearson_max	0.4912
spearman_max	0.4659

Training Details

Training Datasets

triplets

Dataset: triplets
Size: 684,084 training samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 7 tokens mean: 11.1 tokens max: 22 tokens	min: 17 tokens mean: 42.75 tokens max: 95 tokens	min: 15 tokens mean: 43.8 tokens max: 127 tokens

Samples:

anchor	positive	negative
`search_query: tarps heavy duty waterproof 8x10`	`search_document: 8' x 10' Super Heavy Duty 16 Mil Brown Poly Tarp Cover - Thick Waterproof, UV Resistant, Rip and Tear Proof Tarpaulin with Grommets and Reinforced Edges - by Xpose Safety, Xpose Safety, Brown`	`search_document: Grillkid 6'X8' 4.5 Mil Thick General Purpose Waterproof Poly Tarp, Grillkid, All Purpose`
`search_query: wireless keyboard without number pad`	`search_document: Macally 2.4G Small Wireless Keyboard - Ergonomic & Comfortable Computer Keyboard - Compact Keyboard for Laptop or Windows PC Desktop, Tablet, Smart TV - Plug & Play Mini Keyboard with 12 Hot Keys, Macally, Black`	`search_document: Wireless Keyboard - iClever GKA22S Rechargeable Keyboard with Number Pad, Full-Size Stainless Steel Ultra Slim Keyboard, 2.4G Stable Connection Wireless Keyboard for iMac, Mackbook, PC, Laptop, iClever, Silver`
`search_query: geometry earrings`	`search_document: Simple Stud Earrings for Women, Geometric Minimalist Stud Earring Set Tiny Circle Triangle Square Bar Stud Earrings Mini Cartilage Tragus Earrings, choice of all, B:Circle Sliver`	`search_document: BONALUNA Bohemian Wood And Marble Effect Oblong Shaped Drop Statement Earrings (VIVID TURQUOISE), BONALUNA, VIVID TURQUOISE`

Loss: CachedMultipleNegativesRankingLoss with these parameters:

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

pairs

Dataset: pairs
Size: 498,114 training samples
Columns: sentence1, sentence2, and score
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 score
type string string float
details
min: 3 tokens
mean: 6.73 tokens
max: 33 tokens

min: 10 tokens
mean: 40.14 tokens
max: 98 tokens

min: 0.0
mean: 0.81
max: 1.0

	sentence1	sentence2	score
type	string	string	float
details	min: 3 tokens mean: 6.73 tokens max: 33 tokens	min: 10 tokens mean: 40.14 tokens max: 98 tokens	min: 0.0 mean: 0.81 max: 1.0

Samples:

sentence1	sentence2	score
`I would choose a medium weight waterproof fabric, hip length jacket or longer, long sleeves, zip front, with a hood and deep pockets with zips`	`ZSHOW Men's Winter Hooded Packable Down Jacket(Blue, XX-Large), ZSHOW, Blue`	`1.0`
`sequin dance costume girls`	`Yeahdor Big Girls' Lyrical Latin Ballet Dance Costumes Dresses Halter Sequins Irregular Tutu Skirted Leotard Dancewear Pink 12-14, Yeahdor, Pink`	`1.0`
`paint easel bulk`	`Artecho Artist Easel Display Easel Stand, 2 Pack Metal Tripod Stand Easel for Painting, Hold Canvas from 21" to 66", Floor and Tabletop Displaying, Painting with Portable Bag, Artecho, Black`	`1.0`

Loss: AnglELoss with these parameters:

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

Evaluation Datasets

triplets

Dataset: triplets
Size: 10,000 evaluation samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 7 tokens mean: 11.13 tokens max: 23 tokens	min: 15 tokens mean: 43.11 tokens max: 107 tokens	min: 15 tokens mean: 43.56 tokens max: 99 tokens

Samples:

anchor	positive	negative
`search_query: hitch fifth wheel`	`search_document: ENIXWILL 5th Wheel Trailer Hitch Lifting Device Bracket Pin Fit for Hitch Companion and Patriot Series Hitch, ENIXWILL, Black`	`search_document: ECOTRIC Fifth 5th Wheel Trailer Hitch Mount Rails and Installation Kits for Full-Size Trucks, ECOTRIC, black`
`search_query: dek pro`	`search_document: Cubiker Computer Desk 47 inch Home Office Writing Study Desk, Modern Simple Style Laptop Table with Storage Bag, Brown, Cubiker, Brown`	`search_document: FEZIBO Dual Motor L Shaped Electric Standing Desk, 48 Inches Stand Up Corner Desk, Home Office Sit Stand Desk with Rustic Brown Top and Black Frame, FEZIBO, Rustic Brown`
`search_query: 1 year baby mouth without teeth cleaner`	`search_document: Baby Toothbrush,Infant Toothbrush,Baby Tongue Cleaner,Infant Toothbrush,Baby Tongue Cleaner Newborn,Toothbrush Tongue Cleaner Dental Care for 0-36 Month Baby,36 Pcs + Free 4 Pcs, Babycolor, Blue`	`search_document: Slotic Baby Toothbrush for 0-2 Years, Safe and Sturdy, Toddler Oral Care Teether Brush, Extra Soft Bristle for Baby Teeth and Infant Gums, Dentist Recommended (4-Pack), Slotic, 4 Pack`

Loss: CachedMultipleNegativesRankingLoss with these parameters:

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

pairs

Dataset: pairs
Size: 10,000 evaluation samples
Columns: sentence1, sentence2, and score
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 score
type string string float
details
min: 3 tokens
mean: 6.8 tokens
max: 34 tokens

min: 9 tokens
mean: 39.7 tokens
max: 101 tokens

min: 0.0
mean: 0.77
max: 1.0

	sentence1	sentence2	score
type	string	string	float
details	min: 3 tokens mean: 6.8 tokens max: 34 tokens	min: 9 tokens mean: 39.7 tokens max: 101 tokens	min: 0.0 mean: 0.77 max: 1.0

Samples:

sentence1	sentence2	score
`outdoor ceiling fans without light`	`44" Plaza Industrial Indoor Outdoor Ceiling Fan with Remote Control Oil Rubbed Bronze Damp Rated for Patio Porch - Casa Vieja, Casa Vieja, No Light Kit - Bronze`	`1.0`
`bathroom cabinet`	`Homfa Bathroom Floor Cabinet Free Standing with Single Door Multifunctional Bathroom Storage Organizer Toiletries(Ivory White), Homfa, White`	`1.0`
`fitbit charge 3`	`TreasureMax Compatible with Fitbit Charge 2 Bands for Women/Men,Silicone Fadeless Pattern Printed Replacement Floral Bands for Fitbit Charge 2 HR Wristbands, TreasureMax, Paw 2`	`0.4`

Loss: AnglELoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 16
per_device_eval_batch_size: 4
gradient_accumulation_steps: 2
learning_rate: 1e-06
lr_scheduler_type: cosine_with_restarts
lr_scheduler_kwargs: {'num_cycles': 1}
warmup_ratio: 0.01
dataloader_drop_last: True
dataloader_num_workers: 4
dataloader_prefetch_factor: 4
load_best_model_at_end: True
gradient_checkpointing: True
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
prediction_loss_only: True
per_device_train_batch_size: 16
per_device_eval_batch_size: 4
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 2
eval_accumulation_steps: None
learning_rate: 1e-06
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1.0
num_train_epochs: 3
max_steps: -1
lr_scheduler_type: cosine_with_restarts
lr_scheduler_kwargs: {'num_cycles': 1}
warmup_ratio: 0.01
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
no_cuda: False
use_cpu: False
use_mps_device: False
seed: 42
data_seed: None
jit_mode_eval: False
use_ipex: False
bf16: False
fp16: False
fp16_opt_level: O1
half_precision_backend: auto
bf16_full_eval: False
fp16_full_eval: False
tf32: None
local_rank: 0
ddp_backend: None
tpu_num_cores: None
tpu_metrics_debug: False
debug: []
dataloader_drop_last: True
dataloader_num_workers: 4
dataloader_prefetch_factor: 4
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}
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: True
gradient_checkpointing_kwargs: None
include_inputs_for_metrics: False
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
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	pairs loss	triplets loss	cosine_accuracy	spearman_cosine
0.0027	100	2.4909	-	-	-	-
0.0054	200	2.6666	-	-	-	-
0.0081	300	2.76	-	-	-	-
0.0108	400	2.6945	-	-	-	-
0.0135	500	2.9113	-	-	-	-
0.0162	600	2.3476	-	-	-	-
0.0189	700	2.2818	-	-	-	-
0.0217	800	2.4241	-	-	-	-
0.0244	900	2.5126	-	-	-	-
0.0271	1000	2.4106	4.7376	0.8087	0.6993	0.3844
0.0298	1100	2.2369	-	-	-	-
0.0325	1200	2.0614	-	-	-	-
0.0352	1300	2.2178	-	-	-	-
0.0379	1400	1.974	-	-	-	-
0.0406	1500	1.9364	-	-	-	-
0.0433	1600	2.0906	-	-	-	-
0.0460	1700	1.8783	-	-	-	-
0.0487	1800	2.1149	-	-	-	-
0.0514	1900	1.7162	-	-	-	-
0.0541	2000	1.6761	3.8862	0.7490	0.7097	0.4082
0.0568	2100	2.1701	-	-	-	-
0.0596	2200	2.1306	-	-	-	-
0.0623	2300	1.6543	-	-	-	-
0.0650	2400	1.8157	-	-	-	-
0.0677	2500	1.7779	-	-	-	-
0.0704	2600	1.9434	-	-	-	-
0.0731	2700	1.7776	-	-	-	-
0.0758	2800	1.8197	-	-	-	-
0.0785	2900	1.9886	-	-	-	-
0.0812	3000	2.0699	3.8031	0.7298	0.7147	0.4282
0.0839	3100	1.9496	-	-	-	-
0.0866	3200	1.8349	-	-	-	-
0.0893	3300	2.111	-	-	-	-
0.0920	3400	1.9956	-	-	-	-
0.0947	3500	2.0379	-	-	-	-
0.0974	3600	1.8975	-	-	-	-
0.1002	3700	1.8552	-	-	-	-
0.1029	3800	1.9566	-	-	-	-
0.1056	3900	2.011	-	-	-	-
0.1083	4000	2.1263	3.7799	0.7221	0.7176	0.4393
0.1110	4100	1.8217	-	-	-	-
0.1137	4200	1.8638	-	-	-	-
0.1164	4300	1.7699	-	-	-	-
0.1191	4400	1.8248	-	-	-	-
0.1218	4500	1.835	-	-	-	-
0.1245	4600	1.9294	-	-	-	-
0.1272	4700	1.9817	-	-	-	-
0.1299	4800	1.877	-	-	-	-
0.1326	4900	1.5824	-	-	-	-
0.1353	5000	1.7429	3.7728	0.7163	0.7196	0.4496
0.1380	5100	1.8552	-	-	-	-
0.1408	5200	1.6888	-	-	-	-
0.1435	5300	1.9409	-	-	-	-
0.1462	5400	1.9389	-	-	-	-
0.1489	5500	1.82	-	-	-	-
0.1516	5600	1.9763	-	-	-	-
0.1543	5700	1.8122	-	-	-	-
0.1570	5800	1.7204	-	-	-	-
0.1597	5900	1.6901	-	-	-	-
0.1624	6000	1.7785	3.7514	0.7124	0.7195	0.4516
0.1651	6100	1.8559	-	-	-	-
0.1678	6200	1.7646	-	-	-	-
0.1705	6300	1.9068	-	-	-	-
0.1732	6400	1.8848	-	-	-	-
0.1759	6500	1.9384	-	-	-	-
0.1787	6600	1.7692	-	-	-	-
0.1814	6700	1.7093	-	-	-	-
0.1841	6800	1.8759	-	-	-	-
0.1868	6900	1.7319	-	-	-	-
0.1895	7000	1.9428	3.7487	0.7076	0.7256	0.4496
0.1922	7100	1.5733	-	-	-	-
0.1949	7200	1.8487	-	-	-	-
0.1976	7300	1.8361	-	-	-	-
0.2003	7400	1.9911	-	-	-	-
0.2030	7500	1.784	-	-	-	-
0.2057	7600	1.8518	-	-	-	-
0.2084	7700	1.6232	-	-	-	-
0.2111	7800	1.6239	-	-	-	-
0.2138	7900	1.7589	-	-	-	-
0.2165	8000	1.8644	3.7387	0.7040	0.7241	0.4552
0.2193	8100	1.7903	-	-	-	-
0.2220	8200	1.7197	-	-	-	-
0.2247	8300	1.9099	-	-	-	-
0.2274	8400	1.6778	-	-	-	-
0.2301	8500	1.9249	-	-	-	-
0.2328	8600	1.8483	-	-	-	-
0.2355	8700	1.6849	-	-	-	-
0.2382	8800	1.8647	-	-	-	-
0.2409	8900	1.8826	-	-	-	-
0.2436	9000	1.7632	3.7403	0.7033	0.7225	0.4545
0.2463	9100	1.8142	-	-	-	-
0.2490	9200	1.7374	-	-	-	-
0.2517	9300	1.8646	-	-	-	-
0.2544	9400	1.7623	-	-	-	-
0.2571	9500	1.7802	-	-	-	-
0.2599	9600	1.843	-	-	-	-
0.2626	9700	1.9797	-	-	-	-
0.2653	9800	1.7748	-	-	-	-
0.2680	9900	1.7031	-	-	-	-
0.2707	10000	1.5536	3.7613	0.7016	0.7259	0.4548
0.2734	10100	1.7663	-	-	-	-
0.2761	10200	1.8218	-	-	-	-
0.2788	10300	1.6327	-	-	-	-
0.2815	10400	1.8802	-	-	-	-
0.2842	10500	1.6294	-	-	-	-
0.2869	10600	1.9001	-	-	-	-
0.2896	10700	1.7873	-	-	-	-
0.2923	10800	1.8121	-	-	-	-
0.2950	10900	2.0197	-	-	-	-
0.2978	11000	1.7006	3.7559	0.7004	0.727	0.4613
0.3005	11100	1.6404	-	-	-	-
0.3032	11200	1.9422	-	-	-	-
0.3059	11300	1.5917	-	-	-	-
0.3086	11400	1.7236	-	-	-	-
0.3113	11500	1.8977	-	-	-	-
0.3140	11600	1.7686	-	-	-	-
0.3167	11700	1.4493	-	-	-	-
0.3194	11800	1.7447	-	-	-	-
0.3221	11900	1.9412	-	-	-	-
0.3248	12000	1.8	3.7308	0.6997	0.7241	0.4618
0.3275	12100	1.8855	-	-	-	-
0.3302	12200	1.5133	-	-	-	-
0.3329	12300	1.7893	-	-	-	-
0.3356	12400	1.7861	-	-	-	-
0.3384	12500	1.7733	-	-	-	-
0.3411	12600	1.5877	-	-	-	-
0.3438	12700	2.03	-	-	-	-
0.3465	12800	1.7071	-	-	-	-
0.3492	12900	1.7848	-	-	-	-
0.3519	13000	1.7508	3.7326	0.7006	0.7247	0.4583
0.3546	13100	1.7667	-	-	-	-
0.3573	13200	1.6415	-	-	-	-
0.3600	13300	1.7501	-	-	-	-
0.3627	13400	1.8451	-	-	-	-
0.3654	13500	1.7146	-	-	-	-
0.3681	13600	1.6837	-	-	-	-
0.3708	13700	1.92	-	-	-	-
0.3735	13800	1.6925	-	-	-	-
0.3763	13900	1.7799	-	-	-	-
0.3790	14000	1.527	3.7260	0.6989	0.727	0.4510
0.3817	14100	1.7222	-	-	-	-
0.3844	14200	1.8278	-	-	-	-
0.3871	14300	1.7669	-	-	-	-
0.3898	14400	1.5856	-	-	-	-
0.3925	14500	1.8234	-	-	-	-
0.3952	14600	1.7151	-	-	-	-
0.3979	14700	1.6432	-	-	-	-
0.4006	14800	1.9005	-	-	-	-
0.4033	14900	1.6946	-	-	-	-
0.4060	15000	1.5543	3.7222	0.6969	0.7275	0.4634
0.4087	15100	1.6736	-	-	-	-
0.4114	15200	1.8898	-	-	-	-
0.4141	15300	1.7224	-	-	-	-
0.4169	15400	1.7909	-	-	-	-
0.4196	15500	1.6555	-	-	-	-
0.4223	15600	1.523	-	-	-	-
0.4250	15700	1.7539	-	-	-	-
0.4277	15800	1.5763	-	-	-	-
0.4304	15900	1.7247	-	-	-	-
0.4331	16000	1.876	3.7105	0.6977	0.7263	0.4636
0.4358	16100	1.772	-	-	-	-
0.4385	16200	1.6774	-	-	-	-
0.4412	16300	1.7602	-	-	-	-
0.4439	16400	1.705	-	-	-	-
0.4466	16500	1.7893	-	-	-	-
0.4493	16600	1.653	-	-	-	-
0.4520	16700	1.8326	-	-	-	-
0.4547	16800	1.5326	-	-	-	-
0.4575	16900	1.8251	-	-	-	-
0.4602	17000	1.766	3.7193	0.6973	0.7257	0.4655
0.4629	17100	1.7162	-	-	-	-
0.4656	17200	1.6969	-	-	-	-
0.4683	17300	1.5172	-	-	-	-
0.4710	17400	1.7102	-	-	-	-
0.4737	17500	1.8369	-	-	-	-
0.4764	17600	1.8069	-	-	-	-
0.4791	17700	1.6299	-	-	-	-
0.4818	17800	1.8474	-	-	-	-
0.4845	17900	1.5864	-	-	-	-
0.4872	18000	1.7455	3.7087	0.6986	0.7249	0.4626
0.4899	18100	1.8263	-	-	-	-
0.4926	18200	1.8548	-	-	-	-
0.4954	18300	1.6442	-	-	-	-
0.4981	18400	1.7467	-	-	-	-
0.5008	18500	1.6174	-	-	-	-
0.5035	18600	1.4465	-	-	-	-
0.5062	18700	1.8866	-	-	-	-
0.5089	18800	1.72	-	-	-	-
0.5116	18900	1.7466	-	-	-	-
0.5143	19000	1.9124	3.7247	0.6979	0.725	0.4602
0.5170	19100	1.5687	-	-	-	-
0.5197	19200	1.6391	-	-	-	-
0.5224	19300	1.8248	-	-	-	-
0.5251	19400	1.6231	-	-	-	-
0.5278	19500	1.6152	-	-	-	-
0.5305	19600	1.639	-	-	-	-
0.5332	19700	1.6098	-	-	-	-
0.5360	19800	1.6619	-	-	-	-
0.5387	19900	1.6997	-	-	-	-
0.5414	20000	1.718	3.7259	0.6989	0.7264	0.4660
0.5441	20100	1.634	-	-	-	-
0.5468	20200	1.7865	-	-	-	-
0.5495	20300	1.8573	-	-	-	-
0.5522	20400	1.5575	-	-	-	-
0.5549	20500	1.6594	-	-	-	-
0.5576	20600	1.8793	-	-	-	-
0.5603	20700	1.7643	-	-	-	-
0.5630	20800	1.538	-	-	-	-
0.5657	20900	1.8634	-	-	-	-
0.5684	21000	1.916	3.7223	0.6982	0.7258	0.4650
0.5711	21100	1.5947	-	-	-	-
0.5738	21200	1.5321	-	-	-	-
0.5766	21300	1.7004	-	-	-	-
0.5793	21400	1.6947	-	-	-	-
0.5820	21500	1.5228	-	-	-	-
0.5847	21600	1.7152	-	-	-	-
0.5874	21700	1.6883	-	-	-	-
0.5901	21800	1.6779	-	-	-	-
0.5928	21900	1.7323	-	-	-	-
0.5955	22000	1.9633	3.7266	0.6996	0.7288	0.4635
0.5982	22100	1.7498	-	-	-	-
0.6009	22200	1.7513	-	-	-	-
0.6036	22300	1.7078	-	-	-	-
0.6063	22400	1.6438	-	-	-	-
0.6090	22500	1.6743	-	-	-	-
0.6117	22600	1.6701	-	-	-	-
0.6145	22700	1.7871	-	-	-	-
0.6172	22800	1.6247	-	-	-	-
0.6199	22900	1.7817	-	-	-	-
0.6226	23000	1.6606	3.7321	0.6993	0.7286	0.4614
0.6253	23100	1.8987	-	-	-	-
0.6280	23200	1.6494	-	-	-	-
0.6307	23300	1.6776	-	-	-	-
0.6334	23400	1.75	-	-	-	-
0.6361	23500	1.5131	-	-	-	-
0.6388	23600	1.7946	-	-	-	-
0.6415	23700	1.665	-	-	-	-
0.6442	23800	1.6681	-	-	-	-
0.6469	23900	1.8255	-	-	-	-
0.6496	24000	1.6759	3.7227	0.7017	0.7281	0.4625
0.6523	24100	1.554	-	-	-	-
0.6551	24200	1.6435	-	-	-	-
0.6578	24300	1.8224	-	-	-	-
0.6605	24400	1.6186	-	-	-	-
0.6632	24500	1.7156	-	-	-	-
0.6659	24600	1.5247	-	-	-	-
0.6686	24700	1.6264	-	-	-	-
0.6713	24800	1.7673	-	-	-	-
0.6740	24900	1.8072	-	-	-	-
0.6767	25000	1.765	3.7407	0.7026	0.7283	0.4589
0.6794	25100	1.6422	-	-	-	-
0.6821	25200	1.7846	-	-	-	-
0.6848	25300	1.7366	-	-	-	-
0.6875	25400	1.7839	-	-	-	-
0.6902	25500	1.441	-	-	-	-
0.6930	25600	1.5533	-	-	-	-
0.6957	25700	1.6922	-	-	-	-
0.6984	25800	1.5544	-	-	-	-
0.7011	25900	1.456	-	-	-	-
0.7038	26000	1.6494	3.7274	0.7059	0.7268	0.4661
0.7065	26100	1.6963	-	-	-	-
0.7092	26200	1.7892	-	-	-	-
0.7119	26300	1.6669	-	-	-	-
0.7146	26400	1.6758	-	-	-	-
0.7173	26500	1.6322	-	-	-	-
0.7200	26600	1.5416	-	-	-	-
0.7227	26700	1.681	-	-	-	-
0.7254	26800	1.5159	-	-	-	-
0.7281	26900	1.715	-	-	-	-
0.7308	27000	1.6164	3.7456	0.7061	0.7257	0.4570
0.7336	27100	1.6784	-	-	-	-
0.7363	27200	1.5886	-	-	-	-
0.7390	27300	1.6736	-	-	-	-
0.7417	27400	1.5659	-	-	-	-
0.7444	27500	1.6552	-	-	-	-
0.7471	27600	1.5672	-	-	-	-
0.7498	27700	1.5873	-	-	-	-
0.7525	27800	1.6746	-	-	-	-
0.7552	27900	1.7503	-	-	-	-
0.7579	28000	1.7287	3.7390	0.7076	0.7244	0.4636
0.7606	28100	1.6216	-	-	-	-
0.7633	28200	1.6101	-	-	-	-
0.7660	28300	1.5651	-	-	-	-
0.7687	28400	1.5659	-	-	-	-
0.7714	28500	1.5248	-	-	-	-
0.7742	28600	1.3725	-	-	-	-
0.7769	28700	1.7881	-	-	-	-
0.7796	28800	1.739	-	-	-	-
0.7823	28900	1.6464	-	-	-	-
0.7850	29000	1.6841	3.7212	0.7073	0.7247	0.4626
0.7877	29100	1.6254	-	-	-	-
0.7904	29200	1.6728	-	-	-	-
0.7931	29300	1.5605	-	-	-	-
0.7958	29400	1.687	-	-	-	-
0.7985	29500	1.7799	-	-	-	-
0.8012	29600	1.6792	-	-	-	-
0.8039	29700	1.5241	-	-	-	-
0.8066	29800	1.6341	-	-	-	-
0.8093	29900	1.5571	-	-	-	-
0.8121	30000	1.5228	3.7397	0.7105	0.7234	0.4682
0.8148	30100	1.5988	-	-	-	-
0.8175	30200	1.4222	-	-	-	-
0.8202	30300	1.4629	-	-	-	-
0.8229	30400	1.6381	-	-	-	-
0.8256	30500	1.4585	-	-	-	-
0.8283	30600	1.6774	-	-	-	-
0.8310	30700	1.811	-	-	-	-
0.8337	30800	1.5872	-	-	-	-
0.8364	30900	1.4762	-	-	-	-
0.8391	31000	1.7079	3.7256	0.7128	0.7215	0.4645
0.8418	31100	1.4948	-	-	-	-
0.8445	31200	1.4556	-	-	-	-
0.8472	31300	1.5191	-	-	-	-
0.8499	31400	1.598	-	-	-	-
0.8527	31500	1.6586	-	-	-	-
0.8554	31600	1.6893	-	-	-	-
0.8581	31700	1.7764	-	-	-	-
0.8608	31800	1.3632	-	-	-	-
0.8635	31900	1.6681	-	-	-	-
0.8662	32000	1.6232	3.7358	0.7161	0.7232	0.4651
0.8689	32100	1.4556	-	-	-	-
0.8716	32200	1.8698	-	-	-	-
0.8743	32300	1.7566	-	-	-	-
0.8770	32400	1.6082	-	-	-	-
0.8797	32500	1.6465	-	-	-	-
0.8824	32600	1.5018	-	-	-	-
0.8851	32700	1.8482	-	-	-	-
0.8878	32800	1.5147	-	-	-	-
0.8905	32900	1.699	-	-	-	-
0.8933	33000	1.5738	3.7323	0.7176	0.7246	0.4657
0.8960	33100	1.635	-	-	-	-
0.8987	33200	1.7069	-	-	-	-
0.9014	33300	1.6272	-	-	-	-
0.9041	33400	1.7648	-	-	-	-
0.9068	33500	1.6683	-	-	-	-
0.9095	33600	1.4867	-	-	-	-
0.9122	33700	1.6677	-	-	-	-
0.9149	33800	1.5527	-	-	-	-
0.9176	33900	1.6804	-	-	-	-
0.9203	34000	1.425	3.7477	0.7172	0.7231	0.4596
0.9230	34100	1.771	-	-	-	-
0.9257	34200	1.5767	-	-	-	-
0.9284	34300	1.5424	-	-	-	-
0.9312	34400	1.5985	-	-	-	-
0.9339	34500	1.6763	-	-	-	-
0.9366	34600	1.6608	-	-	-	-
0.9393	34700	1.7736	-	-	-	-
0.9420	34800	1.8955	-	-	-	-
0.9447	34900	1.5688	-	-	-	-
0.9474	35000	1.6123	3.7410	0.7196	0.7226	0.4671
0.9501	35100	1.7264	-	-	-	-
0.9528	35200	1.5511	-	-	-	-
0.9555	35300	1.6409	-	-	-	-
0.9582	35400	1.47	-	-	-	-
0.9609	35500	1.8675	-	-	-	-
0.9636	35600	1.6868	-	-	-	-
0.9663	35700	1.744	-	-	-	-
0.9690	35800	1.6734	-	-	-	-
0.9718	35900	1.4154	-	-	-	-
0.9745	36000	1.4793	3.7393	0.7190	0.7223	0.4677
0.9772	36100	1.7126	-	-	-	-
0.9799	36200	1.7037	-	-	-	-
0.9826	36300	1.6306	-	-	-	-
0.9853	36400	1.7783	-	-	-	-
0.9880	36500	1.5751	-	-	-	-
0.9907	36600	1.6079	-	-	-	-
0.9934	36700	1.7162	-	-	-	-
0.9961	36800	1.447	-	-	-	-
0.9988	36900	1.6155	-	-	-	-
1.0015	37000	1.7294	3.7512	0.7177	0.7236	0.4659

Framework Versions

Python: 3.10.12
Sentence Transformers: 3.0.1
Transformers: 4.38.2
PyTorch: 2.1.2+cu121
Accelerate: 0.27.2
Datasets: 2.19.1
Tokenizers: 0.15.2

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",
}

CachedMultipleNegativesRankingLoss

@misc{gao2021scaling,
    title={Scaling Deep Contrastive Learning Batch Size under Memory Limited Setup}, 
    author={Luyu Gao and Yunyi Zhang and Jiawei Han and Jamie Callan},
    year={2021},
    eprint={2101.06983},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

AnglELoss

@misc{li2023angleoptimized,
    title={AnglE-optimized Text Embeddings}, 
    author={Xianming Li and Jing Li},
    year={2023},
    eprint={2309.12871},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}