AdamLucek/ModernBERT-embed-base-legal-MRL

ModernBERT Embed base Legal Matryoshka

This is a sentence-transformers model finetuned from nomic-ai/modernbert-embed-base on the AdamLucek/legal-rag-positives-synthetic dataset. It maps sentences & paragraphs to a 768-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: nomic-ai/modernbert-embed-base <!-- at revision 92168cbee600b1abbfc10842aba988aa69572291 -->
• —Maximum Sequence Length: 8192 tokens
• —Output Dimensionality: 768 dimensions
• —Similarity Function: Cosine Similarity
• —Training Dataset:
• —AdamLucek/legal-rag-positives-synthetic
• —Language: en
• —License: apache-2.0

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: ModernBertModel 
  (1): Pooling({'word_embedding_dimension': 768, '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})
  (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("AdamLucek/ModernBERT-embed-base-legal-MRL")
# Run inference
sentences = [
    'contracting/contracting-assistance-programs/sba-mentor-protege-program (last visited Apr. 19, \n2023). \n5 \n \nprotégé must demonstrate that the added mentor-protégé relationship will not adversely affect the \ndevelopment of either protégé firm (e.g., the second firm may not be a competitor of the first \nfirm).”  13 C.F.R. § 125.9(b)(3).',
    'What must the protégé demonstrate about the mentor-protégé relationship?',
    'What discretion do district courts have regarding a defendant’s invocation of FOIA exemptions?',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

<!--

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. -->

Evaluation

Metrics

Information Retrieval

• —Datasets: dim_768, dim_512, dim_256, dim_128 and dim_64
• —Evaluated with <code>InformationRetrievalEvaluator</code>

<!--

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

AdamLucek/legal-rag-positives-synthetic

• —Dataset: AdamLucek/legal-rag-positives-synthetic
• —Size: 5,822 training samples
• —Columns: <code>positive</code> and <code>anchor</code>
• —Approximate statistics based on the first 1000 samples: | | positive | anchor | |:--------|:-----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------| | type | string | string | | details | <ul><li>min: 15 tokens</li><li>mean: 97.6 tokens</li><li>max: 153 tokens</li></ul> | <ul><li>min: 8 tokens</li><li>mean: 16.68 tokens</li><li>max: 41 tokens</li></ul> |
• —Samples: | positive | anchor | |:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------| | <code>infrastructure security information,” the information at issue must, “if disclosed . . . reveal vulner-<br>abilities in Department of Defense critical infrastructure.” 10 U.S.C. § 130e(f). The closest the <br>Department comes is asserting that the information “individually or in the aggregate, would enable</code> | <code>What type of information must reveal vulnerabilities if disclosed?</code> | | <code>they have bid.” Oral Arg. Tr. at 42:18–20. Plaintiffs also assert that, should this Court require the <br>Polaris Solicitations to consider price at the IDIQ level, such an adjustment “adds a solicitation <br>requirement that would necessarily change the overall structure of the evaluation” GSA must <br>perform in awarding the IDIQ contracts. Oral Arg. Tr. at 43:3–5; see supra Discussion Section</code> | <code>Where in the document can further discussion about the assertion be found?</code> | | <code>otra parte. Fernández v. San Juan Cement Co., Inc., 118 DPR 713, <br>718-719 (1987). Nuestro más Alto Foro ha dispuesto que, la <br>facultad de imponer honorarios de abogados es la mejor arma que <br> <br>22 Id. <br>23 Andamios de PR v. Newport Bonding, 179 DPR 503, 520 (2010); Pérez Rodríguez <br>v. López Rodríguez, supra; SLG González -Figueroa v. Pacheco Romero, supra;</code> | <code>What case is cited with the reference number 118 DPR 713?</code> |
• —Loss: <code>MatryoshkaLoss</code> with these parameters:

  {
      "loss": "MultipleNegativesRankingLoss",
      "matryoshka_dims": [
          768,
          512,
          256,
          128,
          64
      ],
      "matryoshka_weights": [
          1,
          1,
          1,
          1,
          1
      ],
      "n_dims_per_step": -1
  }

Training Hyperparameters

Non-Default Hyperparameters

• —eval_strategy: epoch
• —per_device_train_batch_size: 32
• —per_device_eval_batch_size: 16
• —gradient_accumulation_steps: 16
• —learning_rate: 2e-05
• —num_train_epochs: 4
• —lr_scheduler_type: cosine
• —warmup_ratio: 0.1
• —bf16: True
• —tf32: True
• —load_best_model_at_end: True
• —optim: adamwtorchfused
• —batch_sampler: no_duplicates

All Hyperparameters

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

• —overwrite_output_dir: False
• —do_predict: False
• —eval_strategy: epoch
• —prediction_loss_only: True
• —per_device_train_batch_size: 32
• —per_device_eval_batch_size: 16
• —per_gpu_train_batch_size: None
• —per_gpu_eval_batch_size: None
• —gradient_accumulation_steps: 16
• —eval_accumulation_steps: None
• —torch_empty_cache_steps: None
• —learning_rate: 2e-05
• —weight_decay: 0.0
• —adam_beta1: 0.9
• —adam_beta2: 0.999
• —adam_epsilon: 1e-08
• —max_grad_norm: 1.0
• —num_train_epochs: 4
• —max_steps: -1
• —lr_scheduler_type: cosine
• —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: True
• —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: {'minnumparams': 0, 'xla': False, 'xlafsdpv2': False, 'xlafsdpgrad_ckpt': False}
• —fsdp_transformer_layer_cls_to_wrap: None
• —accelerator_config: {'splitbatches': False, 'dispatchbatches': None, 'evenbatches': True, 'useseedablesampler': True, 'nonblocking': False, 'gradientaccumulationkwargs': None}
• —deepspeed: None
• —label_smoothing_factor: 0.0
• —optim: adamwtorchfused
• —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
• —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
• —dispatch_batches: None
• —split_batches: 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
• —eval_use_gather_object: False
• —average_tokens_across_devices: False
• —prompts: None
• —batch_sampler: no_duplicates
• —multi_dataset_batch_sampler: proportional

</details>

Training Logs

• —The bold row denotes the saved checkpoint.

Framework Versions

• —Python: 3.11.11
• —Sentence Transformers: 3.3.1
• —Transformers: 4.48.0
• —PyTorch: 2.5.1+cu121
• —Accelerate: 1.2.1
• —Datasets: 3.2.0
• —Tokenizers: 0.21.0

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",
}

MatryoshkaLoss

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning},
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

MultipleNegativesRankingLoss

@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}
}