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: 512 tokens
Number of Output Labels: 1 label
Supported Modality: Text

Model Sources

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

Full Model Architecture

CrossEncoder(
  (0): Transformer({'transformer_task': 'sequence-classification', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'logits'}}, 'module_output_name': 'scores', 'architecture': 'BertForSequenceClassification'})
)

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 inputs
pairs = [
    ['skill: manage health care staff', 'skill: manage healthcare staff'],
    ['skill: manage health care staff', 'skill: manage staff'],
    ['skill: manage health care staff', 'skill: manage physiotherapy staff'],
    ['skill: manage health care staff', 'skill: work with nursing staff'],
    ['skill: manage health care staff', 'skill: manage chiropractic staff'],
]
scores = model.predict(pairs)
print(scores)
# [ 6.9497  2.0168 -0.5377 -3.4953 -4.3348]

# Or rank different texts based on similarity to a single text
ranks = model.rank(
    'skill: manage health care staff',
    [
        'skill: manage healthcare staff',
        'skill: manage staff',
        'skill: manage physiotherapy staff',
        'skill: work with nursing staff',
        'skill: manage chiropractic staff',
    ]
)
# [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]

Training Details

Training Dataset

Unnamed Dataset

Size: 93,031 training samples
Columns: sentence_A, sentence_B, and label

Approximate statistics based on the first 100 samples:

	sentence_A	sentence_B	label
type	string	string	float
modality	text	text
details	min: 7 tokens mean: 8.14 tokens max: 11 tokens	min: 6 tokens mean: 7.94 tokens max: 15 tokens	min: 0.0 mean: 0.22 max: 1.0

Samples:

sentence_A	sentence_B	label
`skill: manage security systems`	`skill: manage technical security systems`	`1.0`
`skill: manage security systems`	`skill: Physical access management software`	`0.0`
`skill: manage security systems`	`skill: manage theft prevention`	`0.0`

Loss: BinaryCrossEntropyLoss with these parameters:

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

Evaluation Dataset

Unnamed Dataset

Size: 11,562 evaluation samples
Columns: sentence_A, sentence_B, and label

Approximate statistics based on the first 100 samples:

	sentence_A	sentence_B	label
type	string	string	float
modality	text	text
details	min: 6 tokens mean: 9.23 tokens max: 14 tokens	min: 6 tokens mean: 8.06 tokens max: 16 tokens	min: 0.0 mean: 0.21 max: 1.0

Samples:

sentence_A	sentence_B	label
`skill: manage health care staff`	`skill: manage healthcare staff`	`1.0`
`skill: manage health care staff`	`skill: manage staff`	`0.0`
`skill: manage health care staff`	`skill: manage physiotherapy staff`	`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
num_train_epochs: 2
learning_rate: 2e-05
warmup_steps: 0.1
per_device_eval_batch_size: 64
load_best_model_at_end: True

All Hyperparameters

Click to expand

per_device_train_batch_size: 32
num_train_epochs: 2
max_steps: -1
learning_rate: 2e-05
lr_scheduler_type: linear
lr_scheduler_kwargs: None
warmup_steps: 0.1
optim: adamw_torch_fused
optim_args: None
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
optim_target_modules: None
gradient_accumulation_steps: 1
average_tokens_across_devices: True
max_grad_norm: 1.0
label_smoothing_factor: 0.0
bf16: False
fp16: False
bf16_full_eval: False
fp16_full_eval: False
tf32: None
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
torch_compile: False
torch_compile_backend: None
torch_compile_mode: None
use_liger_kernel: False
liger_kernel_config: None
use_cache: False
neftune_noise_alpha: None
torch_empty_cache_steps: None
auto_find_batch_size: False
log_on_each_node: True
logging_nan_inf_filter: True
include_num_input_tokens_seen: no
log_level: passive
log_level_replica: warning
disable_tqdm: False
project: huggingface
trackio_space_id: None
trackio_bucket_id: None
trackio_static_space_id: None
per_device_eval_batch_size: 64
prediction_loss_only: True
eval_on_start: False
eval_do_concat_batches: True
eval_use_gather_object: False
eval_accumulation_steps: None
include_for_metrics: []
batch_eval_metrics: False
save_only_model: False
save_on_each_node: False
enable_jit_checkpoint: False
push_to_hub: False
hub_private_repo: None
hub_model_id: None
hub_strategy: every_save
hub_always_push: False
hub_revision: None
load_best_model_at_end: True
ignore_data_skip: False
restore_callback_states_from_checkpoint: False
full_determinism: False
seed: 42
data_seed: None
use_cpu: False
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
dataloader_drop_last: False
dataloader_num_workers: 0
dataloader_pin_memory: True
dataloader_persistent_workers: False
dataloader_prefetch_factor: None
remove_unused_columns: True
label_names: None
train_sampling_strategy: random
length_column_name: length
ddp_find_unused_parameters: None
ddp_bucket_cap_mb: None
ddp_broadcast_buffers: False
ddp_static_graph: None
ddp_backend: None
ddp_timeout: 1800
fsdp: []
fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
deepspeed: None
debug: []
skip_memory_metrics: True
do_predict: False
resume_from_checkpoint: None
warmup_ratio: None
local_rank: -1
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss
0.0069	20	0.5247	-
0.0138	40	0.5615	-
0.0206	60	0.5603	-
0.0275	80	0.5108	-
0.0344	100	0.5295	-
0.0413	120	0.4418	-
0.0481	140	0.4539	-
0.0550	160	0.3991	-
0.0619	180	0.3650	-
0.0688	200	0.3651	-
0.0757	220	0.3455	-
0.0825	240	0.3742	-
0.0894	260	0.3396	-
0.0963	280	0.3533	-
0.1032	300	0.3732	-
0.1100	320	0.3413	-
0.1169	340	0.3560	-
0.1238	360	0.3473	-
0.1307	380	0.3159	-
0.1376	400	0.3627	-
0.1444	420	0.3376	-
0.1513	440	0.3022	-
0.1582	460	0.3330	-
0.1651	480	0.3373	-
0.1719	500	0.2894	-
0.1788	520	0.3176	-
0.1857	540	0.3271	-
0.1926	560	0.3519	-
0.1994	580	0.3560	-
0.2063	600	0.2581	-
0.2132	620	0.3213	-
0.2201	640	0.2766	-
0.2270	660	0.2782	-
0.2338	680	0.2782	-
0.2407	700	0.3027	-
0.2476	720	0.2784	-
0.2545	740	0.2814	-
0.2613	760	0.2240	-
0.2682	780	0.3190	-
0.2751	800	0.2623	-
0.2820	820	0.2815	-
0.2889	840	0.2631	-
0.2957	860	0.2360	-
0.3026	880	0.2930	-
0.3095	900	0.2815	-
0.3164	920	0.3000	-
0.3232	940	0.3035	-
0.3301	960	0.2495	-
0.3370	980	0.2447	-
0.3439	1000	0.2272	-
0.3508	1020	0.2600	-
0.3576	1040	0.2717	-
0.3645	1060	0.2830	-
0.3714	1080	0.2944	-
0.3783	1100	0.3230	-
0.3851	1120	0.2521	-
0.3920	1140	0.3050	-
0.3989	1160	0.2201	-
0.4058	1180	0.2504	-
0.4127	1200	0.2545	-
0.4195	1220	0.2230	-
0.4264	1240	0.2741	-
0.4333	1260	0.2679	-
0.4402	1280	0.2870	-
0.4470	1300	0.2602	-
0.4539	1320	0.2448	-
0.4608	1340	0.2764	-
0.4677	1360	0.2452	-
0.4746	1380	0.2668	-
0.4814	1400	0.2914	-
0.4883	1420	0.2683	-
0.4952	1440	0.2193	-
0.5021	1460	0.2097	-
0.5089	1480	0.2783	-
0.5158	1500	0.2634	-
0.5227	1520	0.2414	-
0.5296	1540	0.2544	-
0.5365	1560	0.2405	-
0.5433	1580	0.2488	-
0.5502	1600	0.1917	-
0.5571	1620	0.2446	-
0.5640	1640	0.2279	-
0.5708	1660	0.1966	-
0.5777	1680	0.2259	-
0.5846	1700	0.2440	-
0.5915	1720	0.2458	-
0.5983	1740	0.2644	-
0.6052	1760	0.2301	-
0.6121	1780	0.2677	-
0.6190	1800	0.2589	-
0.6259	1820	0.2259	-
0.6327	1840	0.2254	-
0.6396	1860	0.2591	-
0.6465	1880	0.2786	-
0.6534	1900	0.2170	-
0.6602	1920	0.2169	-
0.6671	1940	0.2377	-
0.6740	1960	0.2485	-
0.6809	1980	0.2069	-
0.6878	2000	0.2270	-
0.6946	2020	0.2587	-
0.7015	2040	0.2071	-
0.7084	2060	0.2137	-
0.7153	2080	0.2102	-
0.7221	2100	0.2204	-
0.7290	2120	0.2049	-
0.7359	2140	0.2361	-
0.7428	2160	0.2325	-
0.7497	2180	0.2066	-
0.7565	2200	0.2569	-
0.7634	2220	0.2270	-
0.7703	2240	0.2498	-
0.7772	2260	0.2130	-
0.7840	2280	0.2353	-
0.7909	2300	0.2301	-
0.7978	2320	0.2369	-
0.8047	2340	0.2350	-
0.8116	2360	0.2178	-
0.8184	2380	0.2277	-
0.8253	2400	0.2412	-
0.8322	2420	0.2329	-
0.8391	2440	0.2469	-
0.8459	2460	0.2300	-
0.8528	2480	0.1999	-
0.8597	2500	0.2254	-
0.8666	2520	0.2136	-
0.8735	2540	0.2640	-
0.8803	2560	0.2424	-
0.8872	2580	0.2097	-
0.8941	2600	0.2025	-
0.9010	2620	0.1984	-
0.9078	2640	0.2355	-
0.9147	2660	0.1924	-
0.9216	2680	0.2269	-
0.9285	2700	0.2381	-
0.9354	2720	0.2347	-
0.9422	2740	0.1967	-
0.9491	2760	0.2288	-
0.9560	2780	0.2521	-
0.9629	2800	0.2560	-
0.9697	2820	0.1999	-
0.9766	2840	0.2412	-
0.9835	2860	0.2209	-
0.9904	2880	0.2096	-
0.9972	2900	0.2422	-
1.0	2908	-	0.2988
1.0041	2920	0.2426	-
1.0110	2940	0.1935	-
1.0179	2960	0.2526	-
1.0248	2980	0.1720	-
1.0316	3000	0.2255	-
1.0385	3020	0.2009	-
1.0454	3040	0.1999	-
1.0523	3060	0.2123	-
1.0591	3080	0.1861	-
1.0660	3100	0.1810	-
1.0729	3120	0.2397	-
1.0798	3140	0.2054	-
1.0867	3160	0.1856	-
1.0935	3180	0.2092	-
1.1004	3200	0.1599	-
1.1073	3220	0.2012	-
1.1142	3240	0.2093	-
1.1210	3260	0.2028	-
1.1279	3280	0.1608	-
1.1348	3300	0.2049	-
1.1417	3320	0.2233	-
1.1486	3340	0.2247	-
1.1554	3360	0.1906	-
1.1623	3380	0.1881	-
1.1692	3400	0.2060	-
1.1761	3420	0.1893	-
1.1829	3440	0.2335	-
1.1898	3460	0.2169	-
1.1967	3480	0.2061	-
1.2036	3500	0.1862	-
1.2105	3520	0.1943	-
1.2173	3540	0.1754	-
1.2242	3560	0.1826	-
1.2311	3580	0.2380	-
1.2380	3600	0.2279	-
1.2448	3620	0.2203	-
1.2517	3640	0.2410	-
1.2586	3660	0.1571	-
1.2655	3680	0.2157	-
1.2724	3700	0.2192	-
1.2792	3720	0.2153	-
1.2861	3740	0.2307	-
1.2930	3760	0.1739	-
1.2999	3780	0.2120	-
1.3067	3800	0.2165	-
1.3136	3820	0.1765	-
1.3205	3840	0.1777	-
1.3274	3860	0.1772	-
1.3343	3880	0.2254	-
1.3411	3900	0.2183	-
1.3480	3920	0.2020	-
1.3549	3940	0.2100	-
1.3618	3960	0.1925	-
1.3686	3980	0.1794	-
1.3755	4000	0.2011	-
1.3824	4020	0.1543	-
1.3893	4040	0.1909	-
1.3961	4060	0.1938	-
1.4030	4080	0.1802	-
1.4099	4100	0.2081	-
1.4168	4120	0.2081	-
1.4237	4140	0.1646	-
1.4305	4160	0.1885	-
1.4374	4180	0.1925	-
1.4443	4200	0.1769	-
1.4512	4220	0.1669	-
1.4580	4240	0.1686	-
1.4649	4260	0.2476	-
1.4718	4280	0.2286	-
1.4787	4300	0.1728	-
1.4856	4320	0.1854	-
1.4924	4340	0.1959	-
1.4993	4360	0.2048	-
1.5062	4380	0.1991	-
1.5131	4400	0.2075	-
1.5199	4420	0.2041	-
1.5268	4440	0.1922	-
1.5337	4460	0.2105	-
1.5406	4480	0.1713	-
1.5475	4500	0.1831	-
1.5543	4520	0.1975	-
1.5612	4540	0.1769	-
1.5681	4560	0.1843	-
1.5750	4580	0.1789	-
1.5818	4600	0.1695	-
1.5887	4620	0.2102	-
1.5956	4640	0.1782	-
1.6025	4660	0.1929	-
1.6094	4680	0.1753	-
1.6162	4700	0.2114	-
1.6231	4720	0.1518	-
1.6300	4740	0.1843	-
1.6369	4760	0.2096	-
1.6437	4780	0.1401	-
1.6506	4800	0.2266	-
1.6575	4820	0.1895	-
1.6644	4840	0.1863	-
1.6713	4860	0.1654	-
1.6781	4880	0.2123	-
1.6850	4900	0.2300	-
1.6919	4920	0.2338	-
1.6988	4940	0.2037	-
1.7056	4960	0.1678	-
1.7125	4980	0.1867	-
1.7194	5000	0.1795	-
1.7263	5020	0.1478	-
1.7331	5040	0.2140	-
1.7400	5060	0.2166	-
1.7469	5080	0.1704	-
1.7538	5100	0.1770	-
1.7607	5120	0.2272	-
1.7675	5140	0.1665	-
1.7744	5160	0.1787	-
1.7813	5180	0.1377	-
1.7882	5200	0.1696	-
1.7950	5220	0.2295	-
1.8019	5240	0.1893	-
1.8088	5260	0.2273	-
1.8157	5280	0.2121	-
1.8226	5300	0.1870	-
1.8294	5320	0.1963	-
1.8363	5340	0.1970	-
1.8432	5360	0.2161	-
1.8501	5380	0.1744	-
1.8569	5400	0.1885	-
1.8638	5420	0.1869	-
1.8707	5440	0.1699	-
1.8776	5460	0.2254	-
1.8845	5480	0.1932	-
1.8913	5500	0.2139	-
1.8982	5520	0.1921	-
1.9051	5540	0.1888	-
1.9120	5560	0.1721	-
1.9188	5580	0.1971	-
1.9257	5600	0.2195	-
1.9326	5620	0.2034	-
1.9395	5640	0.1702	-
1.9464	5660	0.1827	-
1.9532	5680	0.1951	-
1.9601	5700	0.1685	-
1.9670	5720	0.1577	-
1.9739	5740	0.2155	-
1.9807	5760	0.1811	-
1.9876	5780	0.1779	-
1.9945	5800	0.2156	-
2.0	5816	-	0.3412

The bold row denotes the saved checkpoint.

Training Time

Training: 11.7 minutes
Evaluation: 22.5 seconds
Total: 12.1 minutes

Framework Versions

Python: 3.13.13
Sentence Transformers: 5.5.0
Transformers: 5.8.1
PyTorch: 2.12.0
Accelerate: 1.14.0
Datasets: 5.0.0
Tokenizers: 0.22.2

Additional Resources

Training and Finetuning Reranker Models with Sentence Transformers: the end-to-end guide for training or finetuning Cross Encoder (reranker) models.
Multimodal Embedding & Reranker Models with Sentence Transformers: use text, image, audio, and video reranker models through the same API.
Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers: training multimodal Cross Encoders.

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",
}