CrossEncoder based on cross-encoder/ms-marco-MiniLM-L6-v2

This is a Cross Encoder model finetuned from cross-encoder/ms-marco-MiniLM-L6-v2 using the sentence-transformers library. It computes scores for pairs of texts, which can be used for text reranking and semantic search.

Model Details

Model Description

Model Type: Cross Encoder
Base model: cross-encoder/ms-marco-MiniLM-L6-v2
Maximum Sequence Length: 256 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.

from sentence_transformers import CrossEncoder

# Download from the 🤗 Hub
model = CrossEncoder("cross_encoder_model_id")
# Get scores for pairs of texts
pairs = [
    ['president takes aim at cybersecurity through executive order just a few hours before the state of the union address president obama signed an executive order authorizing new policies to protect u s critical infrastructure cybersecurity the call for additional legislation in this area acknowledges the need for continued vigilance', 'take your child to work day the funniest someecards take your child to work day an excellent idea in all of its incarnations but sometimes goes awry in practice your kids'],
    ['healthy breakfast ideas what health editors eat in the morning when it comes to eating breakfast most experts agree you should do it because health editors spend their days learning', 'eat healthier now simple swaps to make in your next meal there are plenty of ways you may be looking to improve your health and for every goal there are countless diet changes'],
    ['how to kick the feeling that you re not good enough facing impostor feelings can be challenging but speaking with mentors recognizing your personal expertise using strategies to replace current negative habits with more positive beneficial patterns will bring you to a more optimistic experience so you can enjoy your well earned successes', 'how to stick to your resolutions by hacking your brain whether you keep a win journal or just a photo album your story will serve as a constant reminder of what you re capable of and inspire you to shoot even higher in the future the more you reshape your expectations of yourself the more you ll find yourself sticking to your goals'],
    ['voter fraud the gop boogyman we ve put far too much energy into combatting a non existent problem and too little in resolving real issues of enfranchisement remember this when you vote in this general election', 'american dream fraud confession of a stupid idealist while i ve been a true believer and i ve perpetuated the belief in hard work as the great poverty buster now when i talk to some of them i feel that i ve perpetuated a great fraud upon them'],
    ['what soda commercials would look like if they told the truth wet sugar with bubbles yep', 'these photos show how the world might look to a person with autism they don t look at most things in a typical way'],
]
scores = model.predict(pairs)
print(scores.shape)
# (5,)

# Or rank different texts based on similarity to a single text
ranks = model.rank(
    'president takes aim at cybersecurity through executive order just a few hours before the state of the union address president obama signed an executive order authorizing new policies to protect u s critical infrastructure cybersecurity the call for additional legislation in this area acknowledges the need for continued vigilance',
    [
        'take your child to work day the funniest someecards take your child to work day an excellent idea in all of its incarnations but sometimes goes awry in practice your kids',
        'eat healthier now simple swaps to make in your next meal there are plenty of ways you may be looking to improve your health and for every goal there are countless diet changes',
        'how to stick to your resolutions by hacking your brain whether you keep a win journal or just a photo album your story will serve as a constant reminder of what you re capable of and inspire you to shoot even higher in the future the more you reshape your expectations of yourself the more you ll find yourself sticking to your goals',
        'american dream fraud confession of a stupid idealist while i ve been a true believer and i ve perpetuated the belief in hard work as the great poverty buster now when i talk to some of them i feel that i ve perpetuated a great fraud upon them',
        'these photos show how the world might look to a person with autism they don t look at most things in a typical way',
    ]
)
# [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]

Training Details

Training Dataset

Unnamed Dataset

Size: 50,000 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	float
details	min: 45 characters mean: 177.47 characters max: 481 characters	min: 39 characters mean: 177.73 characters max: 1318 characters	min: 0.0 mean: 0.22 max: 1.0

Samples:

sentence_0	sentence_1	label
`president takes aim at cybersecurity through executive order just a few hours before the state of the union address president obama signed an executive order authorizing new policies to protect u s critical infrastructure cybersecurity the call for additional legislation in this area acknowledges the need for continued vigilance`	`take your child to work day the funniest someecards take your child to work day an excellent idea in all of its incarnations but sometimes goes awry in practice your kids`	`0.0`
`healthy breakfast ideas what health editors eat in the morning when it comes to eating breakfast most experts agree you should do it because health editors spend their days learning`	`eat healthier now simple swaps to make in your next meal there are plenty of ways you may be looking to improve your health and for every goal there are countless diet changes`	`1.0`
`how to kick the feeling that you re not good enough facing impostor feelings can be challenging but speaking with mentors recognizing your personal expertise using strategies to replace current negative habits with more positive beneficial patterns will bring you to a more optimistic experience so you can enjoy your well earned successes`	`how to stick to your resolutions by hacking your brain whether you keep a win journal or just a photo album your story will serve as a constant reminder of what you re capable of and inspire you to shoot even higher in the future the more you reshape your expectations of yourself the more you ll find yourself sticking to your goals`	`0.0`

Loss: BinaryCrossEntropyLoss with these parameters:

{
    "activation_fn": "torch.nn.modules.linear.Identity",
    "pos_weight": null
}

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 32
per_device_eval_batch_size: 32
num_train_epochs: 2

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: no
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: 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: 2
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: False
fp16_opt_level: O1
half_precision_backend: auto
bf16_full_eval: False
fp16_full_eval: False
tf32: None
local_rank: 0
ddp_backend: None
tpu_num_cores: None
tpu_metrics_debug: False
debug: []
dataloader_drop_last: False
dataloader_num_workers: 0
dataloader_prefetch_factor: None
past_index: -1
disable_tqdm: False
remove_unused_columns: True
label_names: None
load_best_model_at_end: False
ignore_data_skip: False
fsdp: []
fsdp_min_num_params: 0
fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
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: False
eval_use_gather_object: False
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss
0.6394	500	2.8338
1.2788	1000	0.7721
1.9182	1500	0.6555
0.6394	500	0.4732
1.2788	1000	0.4281
1.9182	1500	0.4102
0.6394	500	0.4003
1.2788	1000	0.3924
1.9182	1500	0.3882
0.6394	500	0.3756
1.2788	1000	0.3735
1.9182	1500	0.3651
0.6394	500	0.3533
1.2788	1000	0.3488
1.9182	1500	0.3489

Framework Versions

Python: 3.12.12
Sentence Transformers: 5.2.3
Transformers: 4.44.2
PyTorch: 2.10.0+cu128
Accelerate: 1.12.0
Datasets: 4.8.3
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",
}