CrossEncoder based on BAAI/bge-reranker-base

This is a Cross Encoder model finetuned from BAAI/bge-reranker-base 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: BAAI/bge-reranker-base
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.

from sentence_transformers import CrossEncoder

# Download from the 🤗 Hub
model = CrossEncoder("cross_encoder_model_id")
# Get scores for pairs of texts
pairs = [
    ['user: I really liked that sensitive skin concentrate I bought before, but I need a body cream for lighter skin tones now.\nassistant: Got it! Are you looking for something with a bit of tint or just moisturizing?\nuser: Just a light BB cream type thing for my body, something simple like my radiance wipes were.\nassistant: Okay, so a gentle body BB cream to even out skin tone?\nuser: Yeah exactly, something for lighter skin tones, 7.5 ounce size maybe.', 'SKIN79 Super Plus Beblesh Balm Pink BB 40g All Day Lasting BB Cream - Made In Korea\nBrand: SKIN79 Category: Beauty & Personal Care > Makeup > Face > BB Creams Features: 3 effects functional beblesh balm NEW 2005 PACKAGE whitening, wrinkles improvement Intercept ultraviolet rays UV A and B excellent cover function by composition of make-up base and foundation can correct your irregularly different skin tones naturally excellent for your skin adhesion and sebum control Details: Brand: SKIN79 Scent: Unscented Item Form: Balm Active Ingredients: Damask Rose Water, Niacinamide, Schisandra Extract Unit Count: 1.00 Count Is Discontinued By Manufacturer: No Package Dimensions: 4.02 x 2.05 x 2.01 inches; 1.41 Ounces Item model number: hot pink UPC: 794437141886 784190681859 Manufacturer: SKIN79'],
    ['user: I really liked that sensitive skin concentrate I bought before, but I need a body cream for lighter skin tones now.\nassistant: Got it! Are you looking for something with a bit of tint or just moisturizing?\nuser: Just a light BB cream type thing for my body, something simple like my radiance wipes were.\nassistant: Okay, so a gentle body BB cream to even out skin tone?\nuser: Yeah exactly, something for lighter skin tones, 7.5 ounce size maybe.', "COVERGIRL Clean Matte BB Cream Light 520 For Oily Skin, (packaging may vary) - 1 Fl Oz (1 Count)\nBrand: COVERGIRL Category: Beauty & Personal Care > Makeup > Face > BB Creams Features: Won't clog pores Lasts all day Oil Free BB Cream Suitable BB Cream for sensitive skin Description: Product Description COVERGIRL Clean Matte BB Cream For Light Skin provides the perfect amount of coverage to even skin tone, hide blemishes and keep you shine free all day! Brand Story By CoverGirl Details: Brand: COVERGIRL Item Form: Capsule Active Ingredients: Chloride,silica,sodium benzoate,sodium chloride Unit Count: 1 Fl Oz Number of Items: 1 Is Discontinued By Manufacturer: No Product Dimensions: 1.62 x 0.75 x 4.33 inches; 0.99 Ounces Item model number: 10022700251121 UPC: 022700251124 Manufacturer: Coty Beauty"],
    ['user: I really liked that sensitive skin concentrate I bought before, but I need a body cream for lighter skin tones now.\nassistant: Got it! Are you looking for something with a bit of tint or just moisturizing?\nuser: Just a light BB cream type thing for my body, something simple like my radiance wipes were.\nassistant: Okay, so a gentle body BB cream to even out skin tone?\nuser: Yeah exactly, something for lighter skin tones, 7.5 ounce size maybe.', "Puritan's Pride Arnica Cream-4 oz Cream\nBrand: Puritan's Pride Category: Beauty & Personal Care > Skin Care > Body > Moisturizers > Creams Details: Brand: Puritan's Pride Item Form: Cream Unit Count: 4 Ounce Number of Items: 1 Use for: Whole Body Is Discontinued By Manufacturer: No Package Dimensions: 2.8 x 2.7 x 2.3 inches; 2.4 Ounces Item model number: SG_B00MCJVFF6_US UPC: 766443549277"],
    ['user: I really liked that sensitive skin concentrate I bought before, but I need a body cream for lighter skin tones now.\nassistant: Got it! Are you looking for something with a bit of tint or just moisturizing?\nuser: Just a light BB cream type thing for my body, something simple like my radiance wipes were.\nassistant: Okay, so a gentle body BB cream to even out skin tone?\nuser: Yeah exactly, something for lighter skin tones, 7.5 ounce size maybe.', 'Rimmel London BB Cream, Light\nCategory: Beauty Description: Broad Spectrum SPF 25 Sunscreen 1 Fl oz 30 ml'],
    ['user: I really liked that sensitive skin concentrate I bought before, but I need a body cream for lighter skin tones now.\nassistant: Got it! Are you looking for something with a bit of tint or just moisturizing?\nuser: Just a light BB cream type thing for my body, something simple like my radiance wipes were.\nassistant: Okay, so a gentle body BB cream to even out skin tone?\nuser: Yeah exactly, something for lighter skin tones, 7.5 ounce size maybe.', "L'Oreal Paris Magic Skin Beautifier BB Cream, Fair, 1.0 Ounces\nCategory: Beauty Description: Our first multi-tasking bare skin beautifying cream enhance skin’s 6 dimensions of flawlessness leaving skin perfectly nude and natural."],
]
scores = model.predict(pairs)
print(scores.shape)
# (5,)

# Or rank different texts based on similarity to a single text
ranks = model.rank(
    'user: I really liked that sensitive skin concentrate I bought before, but I need a body cream for lighter skin tones now.\nassistant: Got it! Are you looking for something with a bit of tint or just moisturizing?\nuser: Just a light BB cream type thing for my body, something simple like my radiance wipes were.\nassistant: Okay, so a gentle body BB cream to even out skin tone?\nuser: Yeah exactly, something for lighter skin tones, 7.5 ounce size maybe.',
    [
        'SKIN79 Super Plus Beblesh Balm Pink BB 40g All Day Lasting BB Cream - Made In Korea\nBrand: SKIN79 Category: Beauty & Personal Care > Makeup > Face > BB Creams Features: 3 effects functional beblesh balm NEW 2005 PACKAGE whitening, wrinkles improvement Intercept ultraviolet rays UV A and B excellent cover function by composition of make-up base and foundation can correct your irregularly different skin tones naturally excellent for your skin adhesion and sebum control Details: Brand: SKIN79 Scent: Unscented Item Form: Balm Active Ingredients: Damask Rose Water, Niacinamide, Schisandra Extract Unit Count: 1.00 Count Is Discontinued By Manufacturer: No Package Dimensions: 4.02 x 2.05 x 2.01 inches; 1.41 Ounces Item model number: hot pink UPC: 794437141886 784190681859 Manufacturer: SKIN79',
        "COVERGIRL Clean Matte BB Cream Light 520 For Oily Skin, (packaging may vary) - 1 Fl Oz (1 Count)\nBrand: COVERGIRL Category: Beauty & Personal Care > Makeup > Face > BB Creams Features: Won't clog pores Lasts all day Oil Free BB Cream Suitable BB Cream for sensitive skin Description: Product Description COVERGIRL Clean Matte BB Cream For Light Skin provides the perfect amount of coverage to even skin tone, hide blemishes and keep you shine free all day! Brand Story By CoverGirl Details: Brand: COVERGIRL Item Form: Capsule Active Ingredients: Chloride,silica,sodium benzoate,sodium chloride Unit Count: 1 Fl Oz Number of Items: 1 Is Discontinued By Manufacturer: No Product Dimensions: 1.62 x 0.75 x 4.33 inches; 0.99 Ounces Item model number: 10022700251121 UPC: 022700251124 Manufacturer: Coty Beauty",
        "Puritan's Pride Arnica Cream-4 oz Cream\nBrand: Puritan's Pride Category: Beauty & Personal Care > Skin Care > Body > Moisturizers > Creams Details: Brand: Puritan's Pride Item Form: Cream Unit Count: 4 Ounce Number of Items: 1 Use for: Whole Body Is Discontinued By Manufacturer: No Package Dimensions: 2.8 x 2.7 x 2.3 inches; 2.4 Ounces Item model number: SG_B00MCJVFF6_US UPC: 766443549277",
        'Rimmel London BB Cream, Light\nCategory: Beauty Description: Broad Spectrum SPF 25 Sunscreen 1 Fl oz 30 ml',
        "L'Oreal Paris Magic Skin Beautifier BB Cream, Fair, 1.0 Ounces\nCategory: Beauty Description: Our first multi-tasking bare skin beautifying cream enhance skin’s 6 dimensions of flawlessness leaving skin perfectly nude and natural.",
    ]
)
# [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]

Training Details

Training Dataset

Unnamed Dataset

Size: 4,830 training samples
Columns: query, docs, and labels
Approximate statistics based on the first 1000 samples:
query docs labels
type string list list
details
min: 177 characters
mean: 437.5 characters
max: 881 characters

size: 16 elements

size: 16 elements

	query	docs	labels
type	string	list	list
details	min: 177 characters mean: 437.5 characters max: 881 characters	size: 16 elements	size: 16 elements

Samples:

query	docs	labels
`user: I really liked that sensitive skin concentrate I bought before, but I need a body cream for lighter skin tones now. assistant: Got it! Are you looking for something with a bit of tint or just moisturizing? user: Just a light BB cream type thing for my body, something simple like my radiance wipes were. assistant: Okay, so a gentle body BB cream to even out skin tone? user: Yeah exactly, something for lighter skin tones, 7.5 ounce size maybe.`	['SKIN79 Super Plus Beblesh Balm Pink BB 40g All Day Lasting BB Cream - Made In Korea\nBrand: SKIN79 Category: Beauty & Personal Care > Makeup > Face > BB Creams Features: 3 effects functional beblesh balm NEW 2005 PACKAGE whitening, wrinkles improvement Intercept ultraviolet rays UV A and B excellent cover function by composition of make-up base and foundation can correct your irregularly different skin tones naturally excellent for your skin adhesion and sebum control Details: Brand: SKIN79 Scent: Unscented Item Form: Balm Active Ingredients: Damask Rose Water, Niacinamide, Schisandra Extract Unit Count: 1.00 Count Is Discontinued By Manufacturer: No Package Dimensions: 4.02 x 2.05 x 2.01 inches; 1.41 Ounces Item model number: hot pink UPC: 794437141886 784190681859 Manufacturer: SKIN79', "COVERGIRL Clean Matte BB Cream Light 520 For Oily Skin, (packaging may vary) - 1 Fl Oz (1 Count)\nBrand: COVERGIRL Category: Beauty & Personal Care > Makeup > Face > BB Creams Features: Won't clog ...	`[0.0, 0.0, 0.0, 0.0, 0.0, ...]`
user: I've been happy with that silk protein hair treatment and my foaming face scrub, but my hair's been feeling kinda weak lately. assistant: Got it! Since you liked those, maybe we can find something to boost your scalp health. What have you tried for that? user: I'm not convinced another hair product will do much—most just coat the hair and feel fake. assistant: Some intensive lotions actually target the scalp with natural extracts rather than just coating strands. user: Yeah but I bet any lotion claims to revive fine hair and never does. Hard to trust that.	['Mizani Scalp Care Calming Scalp Lotion \|Dandruff Treatment \| Nourishes & Soothes Scalp \| For Curly Hair \| 4 Fl Oz\nBrand: MIZANI Category: Beauty & Personal Care > Salon & Spa > Hair Care > Hair & Scalp Treatments Details: Brand: MIZANI Hair Type: Curly Product Benefits: A nourishing lotion to soothe and treat dry scalp. Primes hair with a healthy no,dandruff shine. Item Form: Lotion,Oil Scent: Eucalyptus Is Discontinued By Manufacturer: No Product Dimensions: 1.25 x 2.19 x 6.06 inches; 4.66 Ounces Item model number: 955600 UPC: 884486493859 884486307361 Manufacturer: Mizani Country of Origin: USA', 'Renpure Plant-Based Beauty Tea Tree Lemon Sage Refreshing Moisture Scalp Serum, 4 Fluid Ounces\nBrand: RENPURE Category: Beauty & Personal Care > Hair Care Features: Scalp Serum: The perfect partner to our scalp scrub, this scalp serum uses plant based conditioning nutrients to give your scalp a fresh finish; Tea tree oil and ginger mint cool and invigorate Invigorating Nutrients: Our Te...	`[0.0, 0.0, 1.0, 0.0, 0.0, ...]`
`user: I want a collectible figure from a movie I like. Something small to put on my shelf would be nice. assistant: Which movie or character are you interested in, and do you prefer a stylized look or a realistic one? user: Guardians of the Galaxy 2, and I like stylized figures better than realistic ones.`	["Funko Pop Marvel: Marvel Studios 10 - Gamora (Gold Chrome) Collectible Figure, Multicolor\nBrand: Funko Category: Toys & Games > Toy Figures & Playsets > Action Figures Features: From Marvel Studios 10, Gamora (Gold Chrome), as a stylized POP vinyl from Funko! Stylized collectable stands 3 ¾ inches tall, perfect for any Marvel Studios 10 fan! Collect and display all Marvel Studios 10 figures from Funko! Funko pop! is the 2017 toy of the year and people's choice award winner Description: From Marvel Studios 10, Gamora (gold chrome), as a stylized pop vinyl from Funko! figure stands 3 3/4 inches and comes in a window display box. Check out the other Marvel Studios 10 figures from Funko! collect them all! Details: Product Dimensions: 3.54 x 2.95 x 3.54 inches Item Weight: 3.53 ounces Item model number: 33519 Manufacturer recommended age: 6 years and up Best Sellers Rank: {'Toys & Games': 119252, 'Action Figures': 7062} Is Discontinued By Manufacturer: No Release date: November 27, 2018 ...	`[0.0, 0.0, 0.0, 0.0, 0.0, ...]`

Loss: LambdaLoss with these parameters:

{
    "weighting_scheme": "sentence_transformers.cross_encoder.losses.LambdaLoss.NDCGLoss2PPScheme",
    "k": null,
    "sigma": 1.0,
    "eps": 1e-10,
    "reduction_log": "binary",
    "activation_fn": "torch.nn.modules.linear.Identity",
    "mini_batch_size": null
}

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 4
gradient_accumulation_steps: 4
learning_rate: 2e-05
weight_decay: 0.01
num_train_epochs: 1.0
warmup_ratio: 0.1
bf16: True

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: no
prediction_loss_only: True
per_device_train_batch_size: 4
per_device_eval_batch_size: 8
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 4
eval_accumulation_steps: None
torch_empty_cache_steps: None
learning_rate: 2e-05
weight_decay: 0.01
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1.0
num_train_epochs: 1.0
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: True
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}
parallelism_config: None
deepspeed: None
label_smoothing_factor: 0.0
optim: adamw_torch_fused
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
hub_revision: None
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
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
liger_kernel_config: None
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss
0.0662	20	0.6081
0.1325	40	0.4911
0.1987	60	0.3882
0.2649	80	0.3833
0.3311	100	0.3534
0.3974	120	0.3549
0.4636	140	0.3329
0.5298	160	0.3258
0.5960	180	0.3598
0.6623	200	0.356
0.7285	220	0.3405
0.7947	240	0.3487
0.8609	260	0.3551
0.9272	280	0.3589
0.9934	300	0.3399

Framework Versions

Python: 3.12.9
Sentence Transformers: 5.1.0
Transformers: 4.56.1
PyTorch: 2.8.0+cu128
Accelerate: 1.15.0
Datasets: 5.0.1
Tokenizers: 0.22.2

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",
}

LambdaLoss

@inproceedings{wang2018lambdaloss,
  title={The LambdaLoss Framework for Ranking Metric Optimization},
  author={Wang, Xuanhui and Li, Cheng and Golbandi, Nadav and Bendersky, Michael and Najork, Marc},
  booktitle={Proceedings of the 27th ACM international conference on information and knowledge management},
  pages={1313--1322},
  year={2018}
}

Downloads last month: 19

Safetensors

Model size

0.3B params

Tensor type

F32

Model tree for RR52/adl2026-hw2-reranker

Base model

BAAI/bge-reranker-base

Finetuned

(30)

this model

Paper for RR52/adl2026-hw2-reranker

Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks

Paper • 1908.10084 • Published Aug 27, 2019 • 19