newsroom_embedding_model

SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2

This is a sentence-transformers model finetuned from sentence-transformers/all-MiniLM-L6-v2. It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for retrieval.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: sentence-transformers/all-MiniLM-L6-v2 <!-- at revision c9745ed1d9f207416be6d2e6f8de32d1f16199bf -->
• Maximum Sequence Length: 256 tokens
• Output Dimensionality: 384 dimensions
• Similarity Function: Cosine Similarity
• Supported Modality: Text

<!-- - **Training Dataset:** Unknown --> <!-- - **Language:** Unknown --> <!-- - **License:** Unknown -->

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', 'architecture': 'BertModel'})
  (1): Pooling({'embedding_dimension': 384, 'pooling_mode': 'mean', 'include_prompt': True})
  (2): Normalize({})
)

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("sentence_transformers_model_id")
# Run inference
sentences = [
    'guest post by katherine kelly lutton, principal and global head of litigation, fish & richardson pc at last week ’ s fortune most powerful women dinner in washington, i found myself sandwiched between cnn national political correspondent jessica yellin and jane roberts, former litigator and law firm leader and wife of chief justice john roberts. jessica and jane, meanwhile, were sitting between two mentees from haiti — two courageous women among 33 who are part of fortune ‘ s mentoring partnership with the u. s. state department. i ’ m a mentor. how can i not be? every april, fortune invites women who come to its most powerful women summit to mentor rising women leaders from emerging countries around the world. the program seems right for the time. we have globalization on all fronts including the globalization of “ communities. ” think about it. while social relationships and values and “ mentoring ” used to spring out of physically cohesive groups of people connecting according to commonalities ( beliefs about god, job types, organizations ), connecting this way too often meant a senior caucasian man taking a junior caucasian man “ under his wing. ” mentoring women leaders across the globe turns traditional concepts of community and mentoring on their head. i was inspired when i first saw this in action at the mpwomen summit in 2008. where do i sign up?! i did sign up — but i felt empowered to do more. what if i mentored with one other woman from silicon valley who would showcase her skills, experiences and contacts?. or what if i teamed up with two women or three … or 50? on behalf of all the women i knew, i signed up to “ community ” mentor. fast forward about nine months ( the birth of anything worthwhile takes about that long ), and “ our ” mentee, susan rammekwa, steps off the plane from her home country, south africa. susan is a highly religious woman who runs an ngo called tshepang ( “ have hope ” ) programme and an “ empowerment village ” for children whose parents have died from hiv / aids. for 200 homeless and parentless children ages three to 17, tshepang provides daily care, access to education, social skills and a balanced meal.',
    "Guest Post by Katherine Kelly Lutton, Principal and Global Head of Litigation, Fish & Richardson PC At last week's Fortune Most Powerful Women dinner in Washington, I found myself sandwiched between CNN National Political Correspondent Jessica Yellin\xa0and Jane Roberts, former litigator and law firm leader and wife of Chief Justice John Roberts. Jessica and Jane,…",
    'Massachusetts has its share of defense industry contractors, but none like this. Sterlingwear Boston peacoats for the US Navy -- 40,000 of them this year.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[ 1.0000,  0.6002,  0.0343],
#         [ 0.6002,  1.0000, -0.0063],
#         [ 0.0343, -0.0063,  1.0000]])

<!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use *List how the model may foreseeably be misused and address what users ought not to do with the model.* --> <!-- ## Bias, Risks and Limitations *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.* --> <!-- ### Recommendations *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.* -->

Training Details

Training Dataset

Unnamed Dataset

• Size: 2,082,630 training samples
• Columns: <code>sentence_0</code> and <code>sentence_1</code>
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1
  type	string	string
  details	<ul><li>min: 4 tokens</li><li>mean: 190.49 tokens</li><li>max: 256 tokens</li></ul>	<ul><li>min: 5 tokens</li><li>mean: 50.76 tokens</li><li>max: 256 tokens</li></ul>

• Samples:

  sentence_0	sentence_1
  <code>escalating its competition with other boston - area hospital chains, for - profit company steward health care system has again lured away a major doctors group from a rival, this time grabbing a large south shore practice from partners healthcare. deep - pocketed newcomer steward said yesterday that compass medical will join its network, a move that could shake up patient and provider relationships in communities south of boston. compass medical includes 90 doctors in eight offices between braintree and taunton, and over time, doctors there probably will refer more of their thousands of patients to nearby steward - owned community hospitals for care, including quincy medical center, good samaritan medical center in brockton, and morton hospital in taunton. the change is a loss for partners, a powerful provider network that includes massachusetts general and brigham and women ’ s hospitals, which has been affiliated with compass for 16 years. steward ’ s partnership with compass comes a...</code>	<code>Escalating its competition with other Boston-area hospital chains, for-profit company Steward Health Care System has lured away another major doctors group from a rival, this time grabbing a large South Shore practice from Partners HealthCare. Deep-pocketed newcomer Steward said yesterday that Compass Medical will join its network - a move that could shake up patient and provider relationships in communities south of Boston.</code>
  <code>04 / 09 / 2015 at 07 : 55 pm edt is ( finally ) getting his own reality show – and who better to produce the docuseries than the guys behind frontman ' s move from london to los angeles, according to george, 53, announced the show by proclaiming, " if marge simpson met dolly parton and went dancing with ziggy stardust, it wouldnat come close to what youall see. " and if you had any questions as to why the former pop star has suddenly decided to try his hand at television, he ' s got the answers : " why now – why not? why me – who else? a for anyone who missed out on the ' 80s, boy george had a string of hits with his band, the culture club – including classics like " karma chameleon, " " do you really want to hurt me, " and " i ' ll tumble 4 ya. " the british singer ' s androgynous persona was considered quite shocking at the time. aboy george is a musical and cultural icon, and itas about time someone captured his story, " said gil goldschein, chairman and ceo of bmp. " we are experts...</code>	<code>The show will chronicle the Culture Club frontman's move from London to Los Angeles</code>
  <code>color in my hair because my grandson says i look younger. ’ “ some like a lot of teasing and the old - style final net hairspray. a lot of ladies feel like they are going bald, but we have all kinds of tricks up on our sleeve, like a fiber product that fills in spots. the memory issues can be a challenge ; we need to have a receptionist because often people forget their appointment or come back in after they were just here yesterday. those with alzheimer ’ s are afraid of the water, and we can ’ t lay their</code>	<code>You’re never too old to want to look beautiful. Ask hair salon director Bernice Cunningham at Brooksby Village in Peabody.</code>

• Loss: <code>MultipleNegativesRankingLoss</code> 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: 16
• fp16: True
• per_device_eval_batch_size: 16
• multi_dataset_batch_sampler: round_robin

All Hyperparameters

<details><summary>Click to expand</summary>

• per_device_train_batch_size: 16
• num_train_epochs: 3
• max_steps: -1
• learning_rate: 5e-05
• lr_scheduler_type: linear
• lr_scheduler_kwargs: None
• warmup_steps: 0
• 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
• label_smoothing_factor: 0.0
• bf16: False
• fp16: True
• 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: no
• per_device_eval_batch_size: 16
• 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: batch_sampler
• multi_dataset_batch_sampler: round_robin
• router_mapping: {}
• learning_rate_mapping: {}

</details>

Training Logs

<details><summary>Click to expand</summary> </details>

Training Time

• Training: 8.2 hours

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 5.4.0
• Transformers: 5.5.3
• PyTorch: 2.11.0+cu128
• Accelerate: 1.13.0
• Datasets: 4.8.4
• 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},
}

<!-- ## Glossary *Clearly define terms in order to be accessible across audiences.* --> <!-- ## Model Card Authors *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.* --> <!-- ## Model Card Contact *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.* -->