This is a sentence-transformers model finetuned from ltg/norbert4-base on the all-nli-norwegian dataset. It maps sentences & paragraphs to a 640-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: ltg/norbert4-base
Maximum Sequence Length: 128 tokens
Output Dimensionality: 640 dimensions
Similarity Function: Cosine Similarity
Training Dataset:
- all-nli-norwegian
Language: no

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': 128, 'do_lower_case': False, 'architecture': 'GptBertModel'})
  (1): Pooling({'word_embedding_dimension': 640, '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.

python

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("thivy/norbert4-base-nli-norwegian")
# Run inference
sentences = [
    'En mann lager et sandmaleri på gulvet.',
    'En mann lager kunst.',
    'En kvinne ødelegger et sandmaleri.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 640]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.6251, 0.2931],
#         [0.6251, 1.0000, 0.1305],
#         [0.2931, 0.1305, 1.0000]])

Evaluation

Metrics

Triplet

Dataset: eval
Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.9547

Training Details

Training Dataset

all-nli-norwegian

Dataset: all-nli-norwegian at 98cabde
Size: 556,367 training samples
Columns: anchor, positive, and negative
Approximate statistics based on the first 1000 samples:
anchor positive negative
type string string string
details
min: 6 tokens
mean: 9.53 tokens
max: 47 tokens
min: 5 tokens
mean: 12.03 tokens
max: 40 tokens
min: 5 tokens
mean: 12.7 tokens
max: 49 tokens

Samples:

anchor	positive	negative
`En person på en hest hopper over et havarert fly.`	`En person er utendørs, på en hest.`	`En person er på en diner og bestiller en omelett.`
`Barn smiler og vinker til kameraet`	`Det er barn til stede`	`Barna rynker pannen`
`En gutt hopper på skateboard midt på en rød bro.`	`Gutten gjør et skateboardtriks.`	`Gutten skater nedover fortauet.`

Loss: MultipleNegativesRankingLoss with these parameters:
json
1{ 2 "scale": 20.0, 3 "similarity_fct": "cos_sim", 4 "gather_across_devices": false 5}

Evaluation Dataset

all-nli-norwegian

Dataset: all-nli-norwegian at 98cabde
Size: 6,561 evaluation samples
Columns: anchor, positive, and negative
Approximate statistics based on the first 1000 samples:
anchor positive negative
type string string string
details
min: 5 tokens
mean: 17.72 tokens
max: 74 tokens
min: 4 tokens
mean: 8.98 tokens
max: 31 tokens
min: 3 tokens
mean: 9.5 tokens
max: 29 tokens

Samples:

anchor	positive	negative
`To kvinner klemmer mens de holder take-away pakker.`	`To kvinner holder pakker.`	`Mennene slåss utenfor en deli.`
`To små barn i blå drakter, en med nummer 9 og en med nummer 2, står på trinn i et bad og vasker hendene i en vask.`	`To barn i nummererte drakter vasker hendene.`	`To barn i jakker går til skolen.`
`En mann selger donuts til en kunde under et verdensutstillingsarrangement holdt i byen Angeles`	`En mann selger donuts til en kunde.`	`En kvinne drikker kaffen sin på en liten kafé.`

Loss: MultipleNegativesRankingLoss with these parameters:
json
1{ 2 "scale": 20.0, 3 "similarity_fct": "cos_sim", 4 "gather_across_devices": false 5}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 32
per_device_eval_batch_size: 64
learning_rate: 2e-05
weight_decay: 0.01
num_train_epochs: 1
warmup_ratio: 0.1
bf16: True
load_best_model_at_end: 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: 32
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: 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
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
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: True
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
project: huggingface
trackio_space_id: trackio
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: no
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: True
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	eval_cosine_accuracy
0.0058	100	4.0493	-	-
0.0115	200	3.0097	-	-
0.0173	300	1.4324	-	-
0.0230	400	1.0791	-	-
0.0288	500	0.8985	0.7151	0.8682
0.0345	600	0.7899	-	-
0.0403	700	0.7379	-	-
0.0460	800	0.7333	-	-
0.0518	900	0.6676	-	-
0.0575	1000	0.6593	0.4987	0.9137
0.0633	1100	0.6162	-	-
0.0690	1200	0.6153	-	-
0.0748	1300	0.5763	-	-
0.0805	1400	0.6055	-	-
0.0863	1500	0.5504	0.4496	0.9207
0.0920	1600	0.5622	-	-
0.0978	1700	0.5484	-	-
0.1035	1800	0.5263	-	-
0.1093	1900	0.5789	-	-
0.1150	2000	0.5462	0.4225	0.9273
0.1208	2100	0.5521	-	-
0.1265	2200	0.5368	-	-
0.1323	2300	0.5079	-	-
0.1380	2400	0.5437	-	-
0.1438	2500	0.5123	0.4020	0.9346
0.1495	2600	0.4835	-	-
0.1553	2700	0.473	-	-
0.1610	2800	0.4957	-	-
0.1668	2900	0.4935	-	-
0.1725	3000	0.4894	0.3775	0.9383
0.1783	3100	0.4894	-	-
0.1840	3200	0.5203	-	-
0.1898	3300	0.4907	-	-
0.1955	3400	0.464	-	-
0.2013	3500	0.461	0.3808	0.9387
0.2071	3600	0.4486	-	-
0.2128	3700	0.4753	-	-
0.2186	3800	0.4591	-	-
0.2243	3900	0.4496	-	-
0.2301	4000	0.428	0.3680	0.9383
0.2358	4100	0.433	-	-
0.2416	4200	0.4525	-	-
0.2473	4300	0.4119	-	-
0.2531	4400	0.4335	-	-
0.2588	4500	0.4378	0.3586	0.9407
0.2646	4600	0.4073	-	-
0.2703	4700	0.3997	-	-
0.2761	4800	0.381	-	-
0.2818	4900	0.4064	-	-
0.2876	5000	0.4211	0.3577	0.9438
0.2933	5100	0.4338	-	-
0.2991	5200	0.3951	-	-
0.3048	5300	0.3813	-	-
0.3106	5400	0.4165	-	-
0.3163	5500	0.405	0.3464	0.9428
0.3221	5600	0.395	-	-
0.3278	5700	0.3869	-	-
0.3336	5800	0.3758	-	-
0.3393	5900	0.4021	-	-
0.3451	6000	0.374	0.3511	0.9460
0.3508	6100	0.3696	-	-
0.3566	6200	0.377	-	-
0.3623	6300	0.37	-	-
0.3681	6400	0.3584	-	-
0.3738	6500	0.3485	0.3399	0.9470
0.3796	6600	0.3841	-	-
0.3853	6700	0.3674	-	-
0.3911	6800	0.3843	-	-
0.3968	6900	0.3753	-	-
0.4026	7000	0.3533	0.3435	0.9448
0.4084	7100	0.3577	-	-
0.4141	7200	0.3442	-	-
0.4199	7300	0.3539	-	-
0.4256	7400	0.3723	-	-
0.4314	7500	0.3666	0.3383	0.9456
0.4371	7600	0.3644	-	-
0.4429	7700	0.3644	-	-
0.4486	7800	0.3474	-	-
0.4544	7900	0.3538	-	-
0.4601	8000	0.3733	0.3316	0.9508
0.4659	8100	0.3587	-	-
0.4716	8200	0.347	-	-
0.4774	8300	0.3809	-	-
0.4831	8400	0.3222	-	-
0.4889	8500	0.3408	0.3281	0.9492
0.4946	8600	0.3345	-	-
0.5004	8700	0.3492	-	-
0.5061	8800	0.3311	-	-
0.5119	8900	0.3576	-	-
0.5176	9000	0.3377	0.3215	0.9488
0.5234	9100	0.3405	-	-
0.5291	9200	0.3243	-	-
0.5349	9300	0.351	-	-
0.5406	9400	0.3547	-	-
0.5464	9500	0.3438	0.3241	0.9500
0.5521	9600	0.3384	-	-
0.5579	9700	0.3306	-	-
0.5636	9800	0.353	-	-
0.5694	9900	0.299	-	-
0.5751	10000	0.3064	0.3173	0.9509
0.5809	10100	0.3292	-	-
0.5866	10200	0.292	-	-
0.5924	10300	0.3599	-	-
0.5981	10400	0.3271	-	-
0.6039	10500	0.3002	0.3225	0.9492
0.6097	10600	0.3455	-	-
0.6154	10700	0.2981	-	-
0.6212	10800	0.3255	-	-
0.6269	10900	0.3	-	-
0.6327	11000	0.304	0.3170	0.9512
0.6384	11100	0.3136	-	-
0.6442	11200	0.3348	-	-
0.6499	11300	0.3255	-	-
0.6557	11400	0.3101	-	-
0.6614	11500	0.314	0.3149	0.9500
0.6672	11600	0.3157	-	-
0.6729	11700	0.3149	-	-
0.6787	11800	0.2966	-	-
0.6844	11900	0.3145	-	-
0.6902	12000	0.2928	0.3075	0.9532
0.6959	12100	0.3035	-	-
0.7017	12200	0.3142	-	-
0.7074	12300	0.3289	-	-
0.7132	12400	0.3046	-	-
0.7189	12500	0.311	0.3103	0.9529
0.7247	12600	0.2942	-	-
0.7304	12700	0.295	-	-
0.7362	12800	0.2802	-	-
0.7419	12900	0.3258	-	-
0.7477	13000	0.28	0.3027	0.9518
0.7534	13100	0.2887	-	-
0.7592	13200	0.2729	-	-
0.7649	13300	0.2936	-	-
0.7707	13400	0.2883	-	-
0.7764	13500	0.2972	0.3048	0.9549
0.7822	13600	0.2806	-	-
0.7879	13700	0.2851	-	-
0.7937	13800	0.3097	-	-
0.7994	13900	0.2663	-	-
0.8052	14000	0.2743	0.3004	0.9529
0.8110	14100	0.2911	-	-
0.8167	14200	0.2955	-	-
0.8225	14300	0.2892	-	-
0.8282	14400	0.2796	-	-
0.8340	14500	0.2674	0.3000	0.9528
0.8397	14600	0.2604	-	-
0.8455	14700	0.2816	-	-
0.8512	14800	0.2711	-	-
0.8570	14900	0.2897	-	-
0.8627	15000	0.2495	0.3008	0.9544
0.8685	15100	0.3126	-	-
0.8742	15200	0.3151	-	-
0.8800	15300	0.2664	-	-
0.8857	15400	0.2884	-	-
0.8915	15500	0.263	0.2984	0.9552
0.8972	15600	0.2733	-	-
0.9030	15700	0.2755	-	-
0.9087	15800	0.2818	-	-
0.9145	15900	0.2853	-	-
0.9202	16000	0.2742	0.2980	0.9544
0.9260	16100	0.269	-	-
0.9317	16200	0.257	-	-
0.9375	16300	0.2637	-	-
0.9432	16400	0.2752	-	-
0.9490	16500	0.2719	0.2971	0.9546
0.9547	16600	0.282	-	-
0.9605	16700	0.2461	-	-
0.9662	16800	0.2673	-	-
0.9720	16900	0.2646	-	-
0.9777	17000	0.2665	0.2960	0.9547
0.9835	17100	0.258	-	-
0.9892	17200	0.2562	-	-
0.9950	17300	0.2511	-	-

The bold row denotes the saved checkpoint.

Framework Versions

Python: 3.12.12
Sentence Transformers: 5.2.0
Transformers: 4.57.3
PyTorch: 2.9.1
Accelerate: 1.12.0
Datasets: 4.4.2
Tokenizers: 0.22.1

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",
}

MultipleNegativesRankingLoss

bibtex

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}

	anchor	positive	negative
type	string	string	string
details	min: 6 tokens mean: 9.53 tokens max: 47 tokens	min: 5 tokens mean: 12.03 tokens max: 40 tokens	min: 5 tokens mean: 12.7 tokens max: 49 tokens