This is a Cross Encoder model finetuned from microsoft/MiniLM-L12-H384-uncased using the sentence-transformers library. It computes scores for pairs of texts, which can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

Model Type: Cross Encoder
Base model: microsoft/MiniLM-L12-H384-uncased
Maximum Sequence Length: 512 tokens
Number of Output Labels: 1 label

Model Sources

Documentation: Sentence Transformers Documentation
Documentation: Cross Encoder Documentation
Repository: Sentence Transformers on GitHub
Hugging Face: Cross Encoders on Hugging Face

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.

python

from sentence_transformers import CrossEncoder

# Download from the 🤗 Hub
model = CrossEncoder("sentence_transformers_model_id")
# Get scores for pairs of texts
pairs = [
    ['how much should i pay for breast implants', 'Implant Fees. The cost of buying the implants themselves will influence the overall costs of the procedure. Silicone implants cost between $1,800 and $2,500 though this may go up to slightly over $3,000.Saline implants will cost anywhere between $1,200 and $1,600.The cost of implants varies due to size and manufacturer though there is usually not much of a difference.Anesthesia Fees. There are various options for your anesthesia and this will have a direct impact on the fee you will be charged.otal Costs of Brest Implants. With all the above factors taken into consideration, the total cost of breast implants generally are somewhere in the range of $5,000 and $15,000. However, it is best for your safety and peace of mind to avoid the lowest-charging surgeons and the ones charging very high fees.'],
    ['are merrell shoes lifetime warranty', "Best Answer: Regular shoes. If your horse doesn't have bad feet, don't put shoes on him. Shoes actually weaken the hooves due to a lack of circulation and the nail holes (they just make the hooves stronger while the shoes are on, once you take them off the hooves are weaker than they were at first).f your horse doesn't have bad feet, don't put shoes on him. Shoes actually weaken the hooves due to a lack of circulation and the nail holes (they just make the hooves stronger while the shoes are on, once you take them off the hooves are weaker than they were at first)."],
    ['what is the largest capacity dvd disc available', 'Insert a disc that contains files into the drive that is having the problem. Use a type of disc that is not being recognized in the drive. Good discs to use are game or software discs that were purchased from a store. Do not use music CDs. If the DVD drive can read CDs but not DVDs, insert a DVD movie.nsert a software CD (like a game or business software) into the CD/DVD drive and note what happens. If an AutoPlay window opens, the drive is able to read the disc. The data stored on the disc may still be bad, but an AutoPlay window proves that the drive can read data on the disc.'],
    ['weather in dead sea', "The higher above sea level you go, (for example, the tops of mountains,) the more separated and spaced out the molecules become, which causes cold weather. This is the ACCURATE answer to how elevation affects temperature. learned the answer to this in science this year, so don't worry, it is accurate: The higher above sea level/elevation you are, the colder the temperature becomes. The reas â\x80¦ on for this is because there are air molecules in the air bump closer together when you are lower above sea level-that creates warm weather."],
    ['who should not contribute to roth ira', 'You can contribute to a Roth at any age, even past retirement age, as long as youâ\x80\x99re still earning taxable income. A working spouse can also contribute to a Roth IRA on behalf of a nonworking spouse. For a 401(k), the 2014 contribution limit is $17,500, unless youâ\x80\x99re 50 or older, in which case the limit is $23,000.hen you can strategize your distributions to minimize your tax liability. You can also contribute to a traditional IRA even if you participate in an employer-sponsored retirement plan, but in some cases not all of your traditional IRA contributions will be tax deductible.'],
]
scores = model.predict(pairs)
print(scores.shape)
# (5,)

# Or rank different texts based on similarity to a single text
ranks = model.rank(
    'how much should i pay for breast implants',
    [
        'Implant Fees. The cost of buying the implants themselves will influence the overall costs of the procedure. Silicone implants cost between $1,800 and $2,500 though this may go up to slightly over $3,000.Saline implants will cost anywhere between $1,200 and $1,600.The cost of implants varies due to size and manufacturer though there is usually not much of a difference.Anesthesia Fees. There are various options for your anesthesia and this will have a direct impact on the fee you will be charged.otal Costs of Brest Implants. With all the above factors taken into consideration, the total cost of breast implants generally are somewhere in the range of $5,000 and $15,000. However, it is best for your safety and peace of mind to avoid the lowest-charging surgeons and the ones charging very high fees.',
        "Best Answer: Regular shoes. If your horse doesn't have bad feet, don't put shoes on him. Shoes actually weaken the hooves due to a lack of circulation and the nail holes (they just make the hooves stronger while the shoes are on, once you take them off the hooves are weaker than they were at first).f your horse doesn't have bad feet, don't put shoes on him. Shoes actually weaken the hooves due to a lack of circulation and the nail holes (they just make the hooves stronger while the shoes are on, once you take them off the hooves are weaker than they were at first).",
        'Insert a disc that contains files into the drive that is having the problem. Use a type of disc that is not being recognized in the drive. Good discs to use are game or software discs that were purchased from a store. Do not use music CDs. If the DVD drive can read CDs but not DVDs, insert a DVD movie.nsert a software CD (like a game or business software) into the CD/DVD drive and note what happens. If an AutoPlay window opens, the drive is able to read the disc. The data stored on the disc may still be bad, but an AutoPlay window proves that the drive can read data on the disc.',
        "The higher above sea level you go, (for example, the tops of mountains,) the more separated and spaced out the molecules become, which causes cold weather. This is the ACCURATE answer to how elevation affects temperature. learned the answer to this in science this year, so don't worry, it is accurate: The higher above sea level/elevation you are, the colder the temperature becomes. The reas â\x80¦ on for this is because there are air molecules in the air bump closer together when you are lower above sea level-that creates warm weather.",
        'You can contribute to a Roth at any age, even past retirement age, as long as youâ\x80\x99re still earning taxable income. A working spouse can also contribute to a Roth IRA on behalf of a nonworking spouse. For a 401(k), the 2014 contribution limit is $17,500, unless youâ\x80\x99re 50 or older, in which case the limit is $23,000.hen you can strategize your distributions to minimize your tax liability. You can also contribute to a traditional IRA even if you participate in an employer-sponsored retirement plan, but in some cases not all of your traditional IRA contributions will be tax deductible.',
    ]
)
# [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]

Evaluation

Metrics

Cross Encoder Reranking

Datasets: train-eval, NanoMSMARCO, NanoNFCorpus and NanoNQ
Evaluated with CERerankingEvaluator

Metric	train-eval	NanoMSMARCO	NanoNFCorpus	NanoNQ
map	0.6511	0.5909 (+0.1013)	0.3364 (+0.0660)	0.6673 (+0.2466)
mrr@10	0.6494	0.5862 (+0.1087)	0.5282 (+0.0284)	0.6862 (+0.2595)
ndcg@10	0.7082	0.6658 (+0.1254)	0.3656 (+0.0405)	0.7191 (+0.2185)

Cross Encoder Nano BEIR

Dataset: NanoBEIR_mean
Evaluated with CENanoBEIREvaluator

Metric	Value
map	0.5315 (+0.1380)
mrr@10	0.6002 (+0.1322)
ndcg@10	0.5835 (+0.1281)

Training Details

Training Dataset

Unnamed Dataset

Size: 2,000,000 training samples
Columns: sentence_0, sentence_1, and label
Approximate statistics based on the first 1000 samples:
sentence_0 sentence_1 label
type string string int
details
min: 7 characters
mean: 34.08 characters
max: 118 characters
min: 83 characters
mean: 342.99 characters
max: 1018 characters
0: ~81.70%
1: ~18.30%

Samples:

sentence_0	sentence_1	label
`how much should i pay for breast implants`	Implant Fees. The cost of buying the implants themselves will influence the overall costs of the procedure. Silicone implants cost between $1,800 and $2,500 though this may go up to slightly over $3,000.Saline implants will cost anywhere between $1,200 and $1,600.The cost of implants varies due to size and manufacturer though there is usually not much of a difference.Anesthesia Fees. There are various options for your anesthesia and this will have a direct impact on the fee you will be charged.otal Costs of Brest Implants. With all the above factors taken into consideration, the total cost of breast implants generally are somewhere in the range of $5,000 and $15,000. However, it is best for your safety and peace of mind to avoid the lowest-charging surgeons and the ones charging very high fees.	`1`
`are merrell shoes lifetime warranty`	Best Answer: Regular shoes. If your horse doesn't have bad feet, don't put shoes on him. Shoes actually weaken the hooves due to a lack of circulation and the nail holes (they just make the hooves stronger while the shoes are on, once you take them off the hooves are weaker than they were at first).f your horse doesn't have bad feet, don't put shoes on him. Shoes actually weaken the hooves due to a lack of circulation and the nail holes (they just make the hooves stronger while the shoes are on, once you take them off the hooves are weaker than they were at first).	`0`
`what is the largest capacity dvd disc available`	Insert a disc that contains files into the drive that is having the problem. Use a type of disc that is not being recognized in the drive. Good discs to use are game or software discs that were purchased from a store. Do not use music CDs. If the DVD drive can read CDs but not DVDs, insert a DVD movie.nsert a software CD (like a game or business software) into the CD/DVD drive and note what happens. If an AutoPlay window opens, the drive is able to read the disc. The data stored on the disc may still be bad, but an AutoPlay window proves that the drive can read data on the disc.	`0`

Loss: BinaryCrossEntropyLoss

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 64
per_device_eval_batch_size: 64
num_train_epochs: 1
fp16: 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: 64
per_device_eval_batch_size: 64
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: 5e-05
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1
num_train_epochs: 1
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.0
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: 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, '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
dispatch_batches: None
split_batches: None
include_tokens_per_second: False
include_num_input_tokens_seen: False
neftune_noise_alpha: None
optim_target_modules: None
batch_eval_metrics: False
eval_on_start: False
use_liger_kernel: False
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	train-eval_ndcg@10	NanoMSMARCO_ndcg@10	NanoNFCorpus_ndcg@10	NanoNQ_ndcg@10	NanoBEIR_mean_ndcg@10
-1	-1	-	0.0312	0.0280 (-0.5124)	0.2260 (-0.0991)	0.0315 (-0.4691)	0.0952 (-0.3602)
0.016	500	0.6271	-	-	-	-	-
0.032	1000	0.4867	-	-	-	-	-
0.048	1500	0.3551	-	-	-	-	-
0.064	2000	0.2768	-	-	-	-	-
0.08	2500	0.2455	-	-	-	-	-
0.096	3000	0.2186	-	-	-	-	-
0.112	3500	0.2151	-	-	-	-	-
0.128	4000	0.2002	-	-	-	-	-
0.144	4500	0.1973	-	-	-	-	-
0.16	5000	0.1928	0.6389	0.6178 (+0.0774)	0.3541 (+0.0291)	0.6869 (+0.1862)	0.5529 (+0.0976)
0.176	5500	0.1841	-	-	-	-	-
0.192	6000	0.1835	-	-	-	-	-
0.208	6500	0.1828	-	-	-	-	-
0.224	7000	0.1777	-	-	-	-	-
0.24	7500	0.1674	-	-	-	-	-
0.256	8000	0.1655	-	-	-	-	-
0.272	8500	0.1706	-	-	-	-	-
0.288	9000	0.1629	-	-	-	-	-
0.304	9500	0.1641	-	-	-	-	-
0.32	10000	0.1631	0.6859	0.6220 (+0.0815)	0.3849 (+0.0598)	0.6951 (+0.1944)	0.5673 (+0.1119)
0.336	10500	0.1616	-	-	-	-	-
0.352	11000	0.1575	-	-	-	-	-
0.368	11500	0.1565	-	-	-	-	-
0.384	12000	0.1523	-	-	-	-	-
0.4	12500	0.1628	-	-	-	-	-
0.416	13000	0.1569	-	-	-	-	-
0.432	13500	0.1581	-	-	-	-	-
0.448	14000	0.1527	-	-	-	-	-
0.464	14500	0.1484	-	-	-	-	-
0.48	15000	0.1531	0.6939	0.6455 (+0.1051)	0.3663 (+0.0413)	0.6977 (+0.1970)	0.5698 (+0.1145)
0.496	15500	0.1482	-	-	-	-	-
0.512	16000	0.1523	-	-	-	-	-
0.528	16500	0.1532	-	-	-	-	-
0.544	17000	0.1513	-	-	-	-	-
0.56	17500	0.1486	-	-	-	-	-
0.576	18000	0.1438	-	-	-	-	-
0.592	18500	0.1496	-	-	-	-	-
0.608	19000	0.1455	-	-	-	-	-
0.624	19500	0.1474	-	-	-	-	-
0.64	20000	0.1484	0.7025	0.6423 (+0.1019)	0.3637 (+0.0387)	0.7162 (+0.2156)	0.5741 (+0.1187)
0.656	20500	0.1436	-	-	-	-	-
0.672	21000	0.1427	-	-	-	-	-
0.688	21500	0.1463	-	-	-	-	-
0.704	22000	0.1475	-	-	-	-	-
0.72	22500	0.1446	-	-	-	-	-
0.736	23000	0.1424	-	-	-	-	-
0.752	23500	0.1397	-	-	-	-	-
0.768	24000	0.1405	-	-	-	-	-
0.784	24500	0.1405	-	-	-	-	-
0.8	25000	0.1397	0.7014	0.6492 (+0.1088)	0.3672 (+0.0422)	0.7229 (+0.2222)	0.5798 (+0.1244)
0.816	25500	0.1378	-	-	-	-	-
0.832	26000	0.1409	-	-	-	-	-
0.848	26500	0.1368	-	-	-	-	-
0.864	27000	0.1389	-	-	-	-	-
0.88	27500	0.1354	-	-	-	-	-
0.896	28000	0.1412	-	-	-	-	-
0.912	28500	0.138	-	-	-	-	-
0.928	29000	0.1369	-	-	-	-	-
0.944	29500	0.1321	-	-	-	-	-
0.96	30000	0.137	0.7150	0.6576 (+0.1172)	0.3655 (+0.0405)	0.7211 (+0.2204)	0.5814 (+0.1260)
0.976	30500	0.1342	-	-	-	-	-
0.992	31000	0.137	-	-	-	-	-
1.0	31250	-	0.7082	0.6658 (+0.1254)	0.3656 (+0.0405)	0.7191 (+0.2185)	0.5835 (+0.1281)

Environmental Impact

Carbon emissions were measured using CodeCarbon.

Energy Consumed: 0.501 kWh
Carbon Emitted: 0.195 kg of CO2
Hours Used: 1.403 hours

Training Hardware

On Cloud: No
GPU Model: 1 x NVIDIA GeForce RTX 3090
CPU Model: 13th Gen Intel(R) Core(TM) i7-13700K
RAM Size: 31.78 GB

Framework Versions

Python: 3.11.6
Sentence Transformers: 3.5.0.dev0
Transformers: 4.48.3
PyTorch: 2.5.0+cu121
Accelerate: 1.3.0
Datasets: 2.20.0
Tokenizers: 0.21.0

Citation

BibTeX

Sentence Transformers

bibtex

@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",
}

	sentence_0	sentence_1	label
type	string	string	int
details	min: 7 characters mean: 34.08 characters max: 118 characters	min: 83 characters mean: 342.99 characters max: 1018 characters	0: ~81.70% 1: ~18.30%

tomaarsen/reranker-MiniLM-L12-msmarco-scratch

CrossEncoder based on microsoft/MiniLM-L12-H384-uncased