aaa961/modernbert-embed-base-legal-no_MRL_symmetricMNRL

ModernBERT Embed Base Legal Fine-tuned

This is a sentence-transformers model finetuned from nomic-ai/modernbert-embed-base on the 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 d556a88e332558790b210f7bdbe87da2fa94a8d8 -->
• —Maximum Sequence Length: 8192 tokens
• —Output Dimensionality: 768 dimensions
• —Similarity Function: Cosine Similarity
• —Training Dataset:
• —legal-rag-positives-synthetic
• —Language: en
• —License: apache-2.0

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': 8192, 'do_lower_case': False, 'architecture': '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("aaa961/modernbert-embed-base-legal-no_MRL_symmetricMNRL")
# Run inference
sentences = [
    'Did the CIA agree with the relief sought by the plaintiff?',
    'redacted versions of the documents to him.  See id.  Since plaintiff’s counsel wished to submit \nthe non-classified portions of the two documents to the Court, the plaintiff filed a motion on \nAugust 3, 2012 to compel the CIA to “provid[e] [plaintiff’s counsel] with redacted copies” of the \ntwo documents in question.  See id. at 4.  The CIA opposed the relief sought by the plaintiff, \n23',
    'would be an example of something featuring something.”).  At Oral Argument, Defendant’s \ncounsel further suggested that “predominant” means a majority.  Oral Arg. Tr. at 78:16–17 (“I \nwould say predominant means at least the majority.”).   \nThis Court considers the parties’ interpretations of “feature” to be reasonable, given the',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.4220, 0.0226],
#         [0.4220, 1.0000, 0.0205],
#         [0.0226, 0.0205, 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. -->

Evaluation

Metrics

Information Retrieval

• —Dataset: ir_eval
• —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

Training Dataset

legal-rag-positives-synthetic

• —Dataset: legal-rag-positives-synthetic at f11534a
• —Size: 5,822 training samples
• —Columns: <code>anchor</code> and <code>positive</code>
• —Approximate statistics based on the first 1000 samples: | | anchor | positive | |:--------|:----------------------------------------------------------------------------------|:------------------------------------------------------------------------------------| | type | string | string | | details | <ul><li>min: 8 tokens</li><li>mean: 16.54 tokens</li><li>max: 35 tokens</li></ul> | <ul><li>min: 29 tokens</li><li>mean: 98.01 tokens</li><li>max: 170 tokens</li></ul> |
• —Samples: | anchor | positive | |:----------------------------------------------------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| | <code>What statute exempts the numbers of personnel employed by the Agency from disclosure?</code> | <code>numbers of personnel employed by the Agency,” 50 U.S.C. § 403g, while the NSA Act more <br>broadly exempts from disclosure “the organization or any function of the [NSA],” 73 Stat. at 64. <br>The CIA contends that the CIA Act exempts from disclosure, inter alia, “the ‘functions’ of the <br>CIA,” Second Lutz Decl. ¶ 36, but the NSA Act demonstrates that when Congress intends for a</code> | | <code>What would be violated if a heightened requirement was imposed on the protégés according to the Defendant’s Counsel?</code> | <code>VCH MJAR at 29 (same); Oral Arg. Tr. at 72:5–12 (Court: “[W]ould you agree that it would not <br>be okay to impose a heightened requirement on the protégés?” Defendant’s Counsel: “[I]f that <br>were the case here, then, right, we’d be violating 125.8(e).”). <br> <br>36 <br> <br>submit an individually completed Relevant Experience Project, Defendant contends that GSA has</code> | | <code>Where is the direct quotation from FACA § 10(b) found?</code> | <code>other documents which were made available to or prepared for or by” the Commission, a direct <br>quotation from section 10(b) of FACA. Pl.’s Mot. Exs. at 21. EPIC agrees that its FOIA request <br>“exactly track[s] the language of FACA § 10(b)”—i.e., that its FOIA request is meant to be <br>coterminous with FACA’s parameters. Pl.’s Mem. at 24; Pl.’s Reply at 9. <br>25</code> |
• —Loss: <code>CachedMultipleNegativesSymmetricRankingLoss</code> with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim",
      "mini_batch_size": 32,
      "gather_across_devices": false,
      "directions": [
          "query_to_doc",
          "doc_to_query"
      ],
      "partition_mode": "per_direction",
      "hardness_mode": null,
      "hardness_strength": 0.0
  }

Training Hyperparameters

Non-Default Hyperparameters

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

All Hyperparameters

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

• —per_device_train_batch_size: 32
• —num_train_epochs: 4
• —max_steps: -1
• —learning_rate: 2e-05
• —lr_scheduler_type: cosine
• —lr_scheduler_kwargs: None
• —warmup_steps: 0.1
• —optim: adamwtorchfused
• —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: 16
• —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: True
• —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: epoch
• —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: True
• —ignore_data_skip: False
• —restore_callback_states_from_checkpoint: False
• —full_determinism: False
• —seed: 42
• —data_seed: None
• —use_cpu: False
• —accelerator_config: {'splitbatches': False, 'dispatchbatches': None, 'evenbatches': True, 'useseedablesampler': True, 'nonblocking': False, 'gradientaccumulationkwargs': 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: {'minnumparams': 0, 'xla': False, 'xlafsdpv2': False, 'xlafsdpgrad_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: {}

</details>

Training Logs

• —The bold row denotes the saved checkpoint.

Framework Versions

• —Python: 3.12.11
• —Sentence Transformers: 5.3.0
• —Transformers: 5.3.0
• —PyTorch: 2.5.1+cu121
• —Accelerate: 1.13.0
• —Datasets: 4.8.2
• —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",
}

CachedMultipleNegativesSymmetricRankingLoss

@misc{gao2021scaling,
    title={Scaling Deep Contrastive Learning Batch Size under Memory Limited Setup},
    author={Luyu Gao and Yunyi Zhang and Jiawei Han and Jamie Callan},
    year={2021},
    eprint={2101.06983},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

<!--

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. -->