metadata

language: []
library_name: sentence-transformers
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - dataset_size:1K<n<10K
  - loss:CachedGISTEmbedLoss
base_model: nomic-ai/nomic-embed-text-v1.5
metrics:
  - cosine_accuracy
  - dot_accuracy
  - manhattan_accuracy
  - euclidean_accuracy
  - max_accuracy
widget:
  - source_sentence: Pilot
    sentences:
      - Episode Two
      - dog dinosaur bone
      - 10' x 12' gazebo
  - source_sentence: skull
    sentences:
      - cool head s
      - trunk bike rack 4
      - bread without gluten
  - source_sentence: pipes
    sentences:
      - chillum pipe
      - Deckle Edge Ruler
      - dog collar for boxer
  - source_sentence: ddj400
    sentences:
      - lc27h711qenxza
      - bed frame for full
      - chicago bears gifts
  - source_sentence: primes
    sentences:
      - Newton
      - big boys sneakers
      - large dog clothes
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: esci dev
          type: esci-dev
        metrics:
          - type: cosine_accuracy
            value: 0.6414052697616061
            name: Cosine Accuracy
          - type: dot_accuracy
            value: 0.36637390213299875
            name: Dot Accuracy
          - type: manhattan_accuracy
            value: 0.6404015056461732
            name: Manhattan Accuracy
          - type: euclidean_accuracy
            value: 0.6406524466750314
            name: Euclidean Accuracy
          - type: max_accuracy
            value: 0.6414052697616061
            name: Max Accuracy

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. 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

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("sentence_transformers_model_id")
# Run inference
sentences = [
    'primes',
    'Newton',
    'big boys sneakers',
]
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

Dataset: esci-dev
Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.6414
dot_accuracy	0.3664
manhattan_accuracy	0.6404
euclidean_accuracy	0.6407
max_accuracy	0.6414

Training Details

Training Dataset

Unnamed Dataset

Size: 9,090 training samples
Columns: query, pos, and neg
Approximate statistics based on the first 1000 samples:
query pos neg
type string string string
details
min: 3 tokens
mean: 7.42 tokens
max: 30 tokens

min: 3 tokens
mean: 29.27 tokens
max: 87 tokens

min: 4 tokens
mean: 29.8 tokens
max: 82 tokens

Samples:

query	pos	neg
`1 3/4 inch tooled belt strap without belt buckle`	`BS3501 Solid Brass Leaf Belt Buckle Fits 1-3/4"(45mm) Wide Belt`	`Nocona Men's Hired Brown Floral Eagle, 40`
`7edge phone case peacock`	`Galaxy S7 Edge Case for Girls Women Clear with Flowers Design Shockproof Protective Cell Phone Cases for Samsung Galaxy S7 Edge 5.5 Inch Cute Floral Pattern Print Flexible Slim Fit Bumper Rubber Cover`	`Galaxy S7 Case, Galaxy S7 Phone Case with HD Screen Protector for Girls Women, Gritup Cute Clear Gradient Glitter Liquid TPU Slim Phone Case for Samsung Galaxy S7 Teal/Purple`
`girls white shoes`	`adidas Women's Coast Star Shoes, ftwr White/Silver Met./ core Black, 6 M US`	`Converse Optical White M7650 - HI TOP Size 6 M US Women / 4 M US Men`

Loss: CachedGISTEmbedLoss with these parameters:

{'guide': SentenceTransformer(
  (0): Transformer({'max_seq_length': 256, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (2): Normalize()
), 'temperature': 0.01}

Evaluation Dataset

Unnamed Dataset

Size: 3,985 evaluation samples
Columns: query, pos, and neg

Approximate statistics based on the first 1000 samples:

	query	pos	neg
type	string	string	string
details	min: 3 tokens mean: 7.28 tokens max: 25 tokens	min: 3 tokens mean: 28.58 tokens max: 116 tokens	min: 3 tokens mean: 29.26 tokens max: 79 tokens

Samples:

query	pos	neg
`colors for dining room`	`AOOS CUSTOM Dimmable LED Neon Signs for Home Bedroom Salon Dining Room Wall Decor (Customization: Texts, Designs, Logos, Languages, Colors, Sizes, Fonts, Color-Changing) (24" / 1 Line Text)`	`Jetec 5 Pieces EAT Sign Kitchen Wood Rustic Sign Arrow Wall Decor EAT Farmhouse Decoration Hanging Arrow Wooden Sign for Kitchen Wall Home Dining Room (Charming Color)`
`mix no 6 heels for women`	`DREAM PAIRS Women's Hi-Chunk Gold Glitter High Heel Pump Sandals - 6 M US`	`Fashare Womens High Heels Pointed Toe Bowtie Back Ankle Buckle Strap Wedding Evening Party Dress Pumps Shoes`
`goxlrmini`	`Singing Machine SMM-205 Unidirectional Dynamic Microphone with 10 Ft. Cord,Black, one size`	`Behringer U-Phoria Studio Pro Complete Recording Bundle with UMC202HD USB Audio Interface - With 20' 6mm Rubber XLR Microphone Cable, On-Stage MBS5000 Broadcast/Webcast Boom Arm with XLR Cable`

Loss: CachedGISTEmbedLoss with these parameters:

{'guide': SentenceTransformer(
  (0): Transformer({'max_seq_length': 256, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (2): Normalize()
), 'temperature': 0.01}

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 16
per_device_eval_batch_size: 16
num_train_epochs: 10
warmup_ratio: 0.1
fp16: 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: 16
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 1
eval_accumulation_steps: None
learning_rate: 5e-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: 10
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
no_cuda: False
use_cpu: False
use_mps_device: False
seed: 42
data_seed: None
jit_mode_eval: False
use_ipex: False
bf16: False
fp16: True
fp16_opt_level: O1
half_precision_backend: auto
bf16_full_eval: False
fp16_full_eval: False
tf32: None
local_rank: 0
ddp_backend: None
tpu_num_cores: None
tpu_metrics_debug: False
debug: []
dataloader_drop_last: False
dataloader_num_workers: 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}
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: False
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

Epoch	Step	Training Loss	esci-dev_max_accuracy
0	0	-	0.6414
0.1757	100	0.8875	-
0.3515	200	0.5281	-
0.5272	300	0.4621	-
0.7030	400	0.4669	-
0.8787	500	0.4501	-
1.0545	600	0.5379	-
1.2302	700	0.4288	-
1.4060	800	0.2112	-
1.5817	900	0.1508	-
1.7575	1000	0.1133	-
1.9332	1100	0.1312	-
2.1090	1200	0.0784	-
2.2847	1300	0.0983	-
2.4605	1400	0.106	-
2.6362	1500	0.1058	-
2.8120	1600	0.0673	-
2.9877	1700	0.0355	-
3.1634	1800	0.0175	-
3.3392	1900	0.0366	-
3.5149	2000	0.0332	-
3.6907	2100	0.0682	-
3.8664	2200	0.0378	-
4.0422	2300	0.0239	-
4.2179	2400	0.0282	-
4.3937	2500	0.0401	-
4.5694	2600	0.0268	-
4.7452	2700	0.0208	-
4.9209	2800	0.0117	-
5.0967	2900	0.0045	-
5.2724	3000	0.0145	-
5.4482	3100	0.029	-
5.6239	3200	0.0009	-
5.7996	3300	0.0033	-
5.9754	3400	0.0088	-
6.1511	3500	0.0014	-
6.3269	3600	0.0027	-
6.5026	3700	0.0021	-
6.6784	3800	0.0001	-
6.8541	3900	0.0025	-
7.0299	4000	0.0059	-
7.2056	4100	0.0025	-
7.3814	4200	0.0029	-
7.5571	4300	0.0007	-
7.7329	4400	0.0018	-
7.9086	4500	0.0032	-
8.0844	4600	0.0007	-
8.2601	4700	0.0027	-
8.4359	4800	0.0027	-
8.6116	4900	0.0	-
8.7873	5000	0.0025	-
8.9631	5100	0.0025	-
9.1388	5200	0.0014	-
9.3146	5300	0.0027	-
9.4903	5400	0.0021	-
9.6661	5500	0.0	-
9.8418	5600	0.0025	-

Framework Versions

Python: 3.10.12
Sentence Transformers: 3.0.0
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",
}