SentenceTransformer based on jhu-clsp/mmBERT-small

This is a sentence-transformers model finetuned from jhu-clsp/mmBERT-small on the msmarco-co-condenser-margin-mse-sym-mnrl-mean-v1 dataset. It maps sentences & paragraphs to a 384-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: jhu-clsp/mmBERT-small
• Maximum Sequence Length: 8192 tokens
• Output Dimensionality: 384 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • msmarco-co-condenser-margin-mse-sym-mnrl-mean-v1
• Language: en

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}}, 'module_output_name': 'token_embeddings', 'message_format': 'auto', 'architecture': 'ModernBertModel'})
  (1): Pooling({'embedding_dimension': 384, 'pooling_mode': 'mean', '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.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("tomaarsen/mmBERT-small-msmarco-fa2-flattened")
# Run inference
sentences = [
    'do bond funds pay dividends',
    "A bond fund or debt fund is a fund that invests in bonds, or other debt securities. Bond funds can be contrasted with stock funds and money funds. Bond funds typically pay periodic dividends that include interest payments on the fund's underlying securities plus periodic realized capital appreciation. Bond funds typically pay higher dividends than CDs and money market accounts. Most bond funds pay out dividends more frequently than individual bonds.",
    'You would have $71,200 paying out $1,687 in annual dividends. That is about $4.62 for working up in the morning. Interestingly enough, that 2.37% yield is at a low point because The Wellington Fund is a â\x80\x9cbalanced fundâ\x80\x9d meaning that it holds a combination of stocks and bonds.',
    "If a cavity is causing the toothache, your dentist will fill the cavity or possibly extract the tooth, if necessary. A root canal might be needed if the cause of the toothache is determined to be an infection of the tooth's nerve. Bacteria that have worked their way into the inner aspects of the tooth cause such an infection. An antibiotic may be prescribed if there is fever or swelling of the jaw.",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [4, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[0.4402, 0.2069, 0.2170]])

Evaluation

Metrics

Triplet

• Dataset: msmarco-co-condenser-eval-triplet
• Evaluated with TripletEvaluator

Training Details

Training Dataset

msmarco-co-condenser-margin-mse-sym-mnrl-mean-v1

• Dataset: msmarco-co-condenser-margin-mse-sym-mnrl-mean-v1 at 84ed2d3
• Size: 300,000 training samples
• Columns: query, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	query	positive	negative
  type	string	string	string
  details	min: 11 characters mean: 32.46 characters max: 140 characters	min: 67 characters mean: 336.71 characters max: 864 characters	min: 56 characters mean: 339.96 characters max: 900 characters

• Samples:

  query	positive	negative
  what is the meaning of menu planning	Menu planning is the selection of a menu for an event. Such as picking out the dinner for your wedding or even a meal at a Birthday Party. Menu planning is when you are preparing a calendar of meals and you have to sit down and decide what meat and veggies you want to serve on each certain day.	Menu Costs. In economics, a menu cost is the cost to a firm resulting from changing its prices. The name stems from the cost of restaurants literally printing new menus, but economists use it to refer to the costs of changing nominal prices in general.
  how old is brett butler	Brett Butler is 59 years old. To be more precise (and nerdy), the current age as of right now is 21564 days or (even more geeky) 517536 hours. That's a lot of hours!	Passed in: St. John's, Newfoundland and Labrador, Canada. Passed on: 16/07/2016. Published in the St. John's Telegram. Passed away suddenly at the Health Sciences Centre surrounded by his loving family, on July 16, 2016 Robert (Bobby) Joseph Butler, age 52 years. Predeceased by his special aunt Geri Murrin and uncle Mike Mchugh; grandparents Joe and Margaret Murrin and Jack and Theresa Butler.
  when was the last navajo treaty sign?	In Executive Session, Senate of the United States, July 25, 1868. Resolved, (two-thirds of the senators present concurring,) That the Senate advise and consent to the ratification of the treaty between the United States and the Navajo Indians, concluded at Fort Sumner, New Mexico, on the first day of June, 1868.	Share Treaty of Greenville. The Treaty of Greenville was signed August 3, 1795, between the United States, represented by Gen. Anthony Wayne, and chiefs of the Indian tribes located in the Northwest Territory, including the Wyandots, Delawares, Shawnees, Ottawas, Miamis, and others.

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim",
      "gather_across_devices": false,
      "directions": [
          "query_to_doc"
      ],
      "partition_mode": "joint",
      "hardness_mode": null,
      "hardness_strength": 0.0
  }
  

Evaluation Dataset

msmarco-co-condenser-margin-mse-sym-mnrl-mean-v1

• Dataset: msmarco-co-condenser-margin-mse-sym-mnrl-mean-v1 at 84ed2d3
• Size: 1,000 evaluation samples
• Columns: query, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	query	positive	negative
  type	string	string	string
  details	min: 10 characters mean: 32.61 characters max: 110 characters	min: 81 characters mean: 344.28 characters max: 908 characters	min: 97 characters mean: 342.31 characters max: 963 characters

• Samples:

  query	positive	negative
  what county is holly springs nc in	Holly Springs, North Carolina. Holly Springs is a town in Wake County, North Carolina, United States. As of the 2010 census, the town population was 24,661, over 2½ times its population in 2000. Contents.	The Mt. Holly Springs Park & Resort. One of the numerous trolley routes that carried people around the county at the turn of the century was the Carlisle & Mt. Holly Railway Company. The “Holly Trolley” as it came to be known was put into service by Patricio Russo and made its first run on May 14, 1901.
  how long does nyquil stay in your system	In order to understand exactly how long Nyquil lasts, it is absolutely vital to learn about the various ingredients in the drug. One of the ingredients found in Nyquil is Doxylamine, which is an antihistamine. This specific medication has a biological half-life or 6 to 12 hours. With this in mind, it is possible for the drug to remain in the system for a period of 12 to 24 hours. It should be known that the specifics will depend on a wide variety of different factors, including your age and metabolism.	I confirmed that NyQuil is about 10% alcohol, a higher content than most domestic beers. When I asked about the relatively high proof, I was told that the alcohol dilutes the active ingredients. The alcohol free version is there for customers with addiction issues.. also found that in that version there is twice the amount of DXM. When I asked if I could speak to a chemist or scientist, I was told they didn't have anyone who fit that description there. It’s been eight years since I kicked NyQuil. I've been sober from alcohol for four years.
  what are mineral water	1 Mineral water – water from a mineral spring that contains various minerals, such as salts and sulfur compounds. 2 It comes from a source tapped at one or more bore holes or spring, and originates from a geologically and physically protected underground water source. Mineral water – water from a mineral spring that contains various minerals, such as salts and sulfur compounds. 2 It comes from a source tapped at one or more bore holes or spring, and originates from a geologically and physically protected underground water source.	Minerals for Your Body. Drinking mineral water is beneficial to health and well-being. But it is not only the amount of water you drink that is important-what the water contains is even more essential.inerals for Your Body. Drinking mineral water is beneficial to health and well-being. But it is not only the amount of water you drink that is important-what the water contains is even more essential.

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim",
      "gather_across_devices": false,
      "directions": [
          "query_to_doc"
      ],
      "partition_mode": "joint",
      "hardness_mode": null,
      "hardness_strength": 0.0
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 128
• num_train_epochs: 1
• learning_rate: 8e-05
• warmup_steps: 0.05
• bf16: True
• eval_strategy: steps
• per_device_eval_batch_size: 128
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• per_device_train_batch_size: 128
• num_train_epochs: 1
• max_steps: -1
• learning_rate: 8e-05
• lr_scheduler_type: linear
• lr_scheduler_kwargs: None
• warmup_steps: 0.05
• 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: True
• 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: trackio
• eval_strategy: steps
• per_device_eval_batch_size: 128
• 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: False
• 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_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: no_duplicates
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss	msmarco-co-condenser-eval-triplet_cosine_accuracy
-1	-1	-	-	0.5340
0.0102	24	5.4808	-	-
0.0205	48	5.2376	-	-
0.0307	72	4.2629	-	-
0.0410	96	3.0983	-	-
0.0512	120	2.1074	-	-
0.0614	144	1.5078	-	-
0.0717	168	1.3295	-	-
0.0819	192	1.0376	-	-
0.0922	216	1.0018	-	-
0.1003	235	-	1.1467	0.7610
0.1024	240	0.7672	-	-
0.1126	264	0.7724	-	-
0.1229	288	0.7220	-	-
0.1331	312	0.6774	-	-
0.1433	336	0.6984	-	-
0.1536	360	0.6782	-	-
0.1638	384	0.6620	-	-
0.1741	408	0.5325	-	-
0.1843	432	0.5893	-	-
0.1945	456	0.4944	-	-
0.2005	470	-	0.7816	0.7590
0.2048	480	0.5062	-	-
0.2150	504	0.4083	-	-
0.2253	528	0.4436	-	-
0.2355	552	0.5312	-	-
0.2457	576	0.4251	-	-
0.2560	600	0.4881	-	-
0.2662	624	0.4719	-	-
0.2765	648	0.4556	-	-
0.2867	672	0.3566	-	-
0.2969	696	0.3735	-	-
0.3008	705	-	0.4058	0.8410
0.3072	720	0.5212	-	-
0.3174	744	0.4097	-	-
0.3276	768	0.3429	-	-
0.3379	792	0.3341	-	-
0.3481	816	0.3187	-	-
0.3584	840	0.3550	-	-
0.3686	864	0.3735	-	-
0.3788	888	0.3620	-	-
0.3891	912	0.2912	-	-
0.3993	936	0.2705	-	-
0.4010	940	-	0.3238	0.8210
0.4096	960	0.2694	-	-
0.4198	984	0.2875	-	-
0.4300	1008	0.3122	-	-
0.4403	1032	0.2973	-	-
0.4505	1056	0.3642	-	-
0.4608	1080	0.2962	-	-
0.4710	1104	0.2805	-	-
0.4812	1128	0.2454	-	-
0.4915	1152	0.2715	-	-
0.5013	1175	-	0.2668	0.8350
0.5017	1176	0.3350	-	-
0.5119	1200	0.3056	-	-
0.5222	1224	0.2710	-	-
0.5324	1248	0.3994	-	-
0.5427	1272	0.3313	-	-
0.5529	1296	0.3144	-	-
0.5631	1320	0.2923	-	-
0.5734	1344	0.2912	-	-
0.5836	1368	0.2360	-	-
0.5939	1392	0.2508	-	-
0.6015	1410	-	0.3140	0.8310
0.6041	1416	0.2394	-	-
0.6143	1440	0.3832	-	-
0.6246	1464	0.2197	-	-
0.6348	1488	0.2956	-	-
0.6451	1512	0.2050	-	-
0.6553	1536	0.3045	-	-
0.6655	1560	0.2322	-	-
0.6758	1584	0.2233	-	-
0.6860	1608	0.2498	-	-
0.6962	1632	0.2225	-	-
0.7018	1645	-	0.3156	0.8520
0.7065	1656	0.3178	-	-
0.7167	1680	0.3205	-	-
0.7270	1704	0.2739	-	-
0.7372	1728	0.2206	-	-
0.7474	1752	0.2258	-	-
0.7577	1776	0.3038	-	-
0.7679	1800	0.2068	-	-
0.7782	1824	0.2092	-	-
0.7884	1848	0.1976	-	-
0.7986	1872	0.2832	-	-
0.8020	1880	-	0.2351	0.8420
0.8089	1896	0.2208	-	-
0.8191	1920	0.2190	-	-
0.8294	1944	0.2076	-	-
0.8396	1968	0.1939	-	-
0.8498	1992	0.2083	-	-
0.8601	2016	0.2059	-	-
0.8703	2040	0.3119	-	-
0.8805	2064	0.1962	-	-
0.8908	2088	0.2442	-	-
0.9010	2112	0.1859	-	-
0.9023	2115	-	0.2889	0.8390
0.9113	2136	0.1995	-	-
0.9215	2160	0.2216	-	-
0.9317	2184	0.2370	-	-
0.9420	2208	0.2155	-	-
0.9522	2232	0.1852	-	-
0.9625	2256	0.2151	-	-
0.9727	2280	0.2082	-	-
0.9829	2304	0.2741	-	-
0.9932	2328	0.2627	-	-
-1	-1	-	-	0.8400

Environmental Impact

Carbon emissions were measured using CodeCarbon.

• Energy Consumed: 0.213 kWh
• Carbon Emitted: 0.057 kg of CO2
• Hours Used: 0.738 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: 5.4.0.dev0
• Transformers: 5.3.0.dev0
• PyTorch: 2.10.0+cu128
• Accelerate: 1.13.0.dev0
• Datasets: 4.3.0
• 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",
}

MultipleNegativesRankingLoss

@misc{oord2019representationlearningcontrastivepredictive,
      title={Representation Learning with Contrastive Predictive Coding},
      author={Aaron van den Oord and Yazhe Li and Oriol Vinyals},
      year={2019},
      eprint={1807.03748},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/1807.03748},
}