kl3m-doc-pico-001 is a domain-specific masked language model (MLM) based on the RoBERTa architecture, specifically designed for legal and financial document analysis. With approximately 40M parameters, it provides a compact yet effective model for specialized NLP tasks in both fill-mask prediction and feature extraction for document embeddings.

Model Details

Architecture: RoBERTa
Size: 40M parameters
Hidden Size: 256
Layers: 8
Attention Heads: 8
Intermediate Size: 1024
Max Position Embeddings: 512
Max Sequence Length: 509
Tokenizer: alea-institute/kl3m-004-128k-cased
Vector Dimension: 256 (hidden_size)
Pooling Strategy: CLS token or mean pooling

Use Cases

This model is particularly useful for:

Document classification in legal and financial domains
Entity recognition for specialized terminology
Understanding legal citations and references
Filling in missing terms in legal documents
Feature extraction for downstream legal analysis tasks
Document similarity and retrieval tasks
Semantic search across legal and financial corpora

Performance

The model demonstrates strong performance on domain-specific tasks compared to general-purpose models of similar size. It shows particular strength in masked token prediction for legal and regulatory terminology.

Examples of Fill-Mask Performance

Environmental Law:
- "Under the Migratory Bird Treaty Act, the..." → 93.4% confidence
Financial Contracts:
- "This Farm Credit Agreement is hereby..." → 30.0% confidence

Standard Test Examples

Using our standardized test examples for comparing embedding models:

Fill-Mask Results

Contract Clause Heading:
"<|cls|> 8. REPRESENTATIONS AND<|mask|>. Each party hereby represents and warrants to the other party as of the date hereof as follows: <|sep|>"

Top 5 predictions:
1. APPLICATION (0.0384)
2. PROCEDURES (0.0164)
3. HEARING (0.0150)
4. REGULATIONS (0.0119)
5. DEFINITIONS (0.0117)
Note: The contract-specialized models show stronger performance on this task, with kl3m-doc-pico-contracts-001 predicting "WARRANTIES" with higher confidence and kl3m-doc-nano-001 with 0.927 confidence.
Defined Term Example:
"<|cls|> \"Effective<|mask|>\" means the date on which all conditions precedent set forth in Article V are satisfied or waived by the Administrative Agent. <|sep|>"

Top 5 predictions:
1. date (0.3148)
2. Date (0.2616)
3. Time (0.0220)
4. Order (0.0213)
5. Dates (0.0198)
Regulation Example:
"<|cls|> All transactions shall comply with the requirements set forth in the Truth in<|mask|> Act and its implementing Regulation Z. <|sep|>"

Top 5 predictions:
1. Lending (0.5241)
2. the (0.1538)
3. Disabilities (0.0708)
4. America (0.0228)
5. Truth (0.0147)

Document Similarity Results

Using the standardized document examples for embeddings:

Document Pair	Cosine Similarity (CLS token)	Cosine Similarity (Mean pooling)
Court Complaint vs. Consumer Terms	0.711	0.675
Court Complaint vs. Credit Agreement	0.847	0.833
Consumer Terms vs. Credit Agreement	0.828	0.709

Note how different pooling strategies affect similarity measurements. Mean pooling tends to capture more comprehensive document-level similarity, particularly between legal documents like complaints and agreements (0.833), while CLS token embeddings in this model show generally higher similarity values, particularly between Court Complaint and Credit Agreement (0.847).

Usage

Masked Language Modeling

You can use this model for masked language modeling with the simple pipeline approach:

python

from transformers import pipeline

# Load the fill-mask pipeline with the model
fill_mask = pipeline('fill-mask', model="alea-institute/kl3m-doc-pico-001")

# Example: Contract clause heading
# Note the mask token placement - directly adjacent to "AND" without space
text = "<|cls|> 8. REPRESENTATIONS AND<|mask|>. Each party hereby represents and warrants to the other party as of the date hereof as follows: <|sep|>"
results = fill_mask(text)

# Display predictions
print("Top predictions:")
for result in results:
    print(f"- {result['token_str']} (score: {result['score']:.4f})")

# Output:
# Top predictions:
# - APPLICATION (score: 0.0384)
# - PROCEDURES (score: 0.0164)
# - HEARING (score: 0.0150)
# - REGULATIONS (score: 0.0119)
# - DEFINITIONS (score: 0.0117)

You can also try other examples:

python

# Example: Defined term
text2 = "<|cls|> \"Effective<|mask|>\" means the date on which all conditions precedent set forth in Article V are satisfied or waived by the Administrative Agent. <|sep|>"
results2 = fill_mask(text2)

# Display predictions
print("Top predictions:")
for result in results2[:5]:
    print(f"- {result['token_str']} (score: {result['score']:.4f})")

# Output:
# Top predictions:
# - date (score: 0.3148)
# - Date (score: 0.2616)
# - Time (score: 0.0220)
# - Order (score: 0.0213)
# - Dates (score: 0.0198)

Feature Extraction for Embeddings

python

from transformers import pipeline
import numpy as np
from sklearn.metrics.pairwise import cosine_similarity

# Load the feature-extraction pipeline
extractor = pipeline('feature-extraction', model="alea-institute/kl3m-doc-pico-001", return_tensors=True)

# Example legal documents
texts = [
    # Court Complaint
    "<|cls|> IN THE UNITED STATES DISTRICT COURT FOR THE EASTERN DISTRICT OF PENNSYLVANIA\n\nJOHN DOE,\nPlaintiff,\n\nvs.\n\nACME CORPORATION,\nDefendant.\n\nCIVIL ACTION NO. 21-12345\n\nCOMPLAINT\n\nPlaintiff John Doe, by and through his undersigned counsel, hereby files this Complaint against Defendant Acme Corporation, and in support thereof, alleges as follows: <|sep|>",
    
    # Consumer Terms
    "<|cls|> TERMS AND CONDITIONS\n\nLast Updated: April 10, 2025\n\nThese Terms and Conditions (\"Terms\") govern your access to and use of the Service. By accessing or using the Service, you agree to be bound by these Terms. If you do not agree to these Terms, you may not access or use the Service. These Terms constitute a legally binding agreement between you and the Company. <|sep|>",
    
    # Credit Agreement
    "<|cls|> CREDIT AGREEMENT\n\nDated as of April 10, 2025\n\nAmong\n\nACME BORROWER INC.,\nas the Borrower,\n\nBANK OF FINANCE,\nas Administrative Agent,\n\nand\n\nTHE LENDERS PARTY HERETO\n\nThis CREDIT AGREEMENT (\"Agreement\") is entered into as of April 10, 2025, among ACME BORROWER INC., a Delaware corporation (the \"Borrower\"), each lender from time to time party hereto (collectively, the \"Lenders\"), and BANK OF FINANCE, as Administrative Agent. <|sep|>"
]

# Strategy 1: CLS token embeddings
cls_embeddings = []
for text in texts:
    features = extractor(text)
    # Get the CLS token (first token) embedding
    features_array = features[0].numpy() if hasattr(features[0], 'numpy') else features[0]
    cls_embedding = features_array[0]
    cls_embeddings.append(cls_embedding)

# Calculate similarity between documents using CLS tokens
cls_similarity = cosine_similarity(np.vstack(cls_embeddings))
print("\nDocument similarity (CLS token):")
print(np.round(cls_similarity, 3))
# Output:
# [[1.    0.711 0.847]
#  [0.711 1.    0.828]
#  [0.847 0.828 1.   ]]

# Strategy 2: Mean pooling
mean_embeddings = []
for text in texts:
    features = extractor(text)
    # Average over all tokens
    features_array = features[0].numpy() if hasattr(features[0], 'numpy') else features[0]
    mean_embedding = np.mean(features_array, axis=0)
    mean_embeddings.append(mean_embedding)

# Calculate similarity using mean pooling
mean_similarity = cosine_similarity(np.vstack(mean_embeddings))
print("\nDocument similarity (Mean pooling):")
print(np.round(mean_similarity, 3))
# Output:
# [[1.    0.675 0.833]
#  [0.675 1.    0.709]
#  [0.833 0.709 1.   ]]

# Print pairwise similarities
doc_names = ["Court Complaint", "Consumer Terms", "Credit Agreement"]
print("\nPairwise similarities:")
for i in range(len(doc_names)):
    for j in range(i+1, len(doc_names)):
        print(f"{doc_names[i]} vs. {doc_names[j]}:")
        print(f"  - CLS token: {cls_similarity[i, j]:.4f}")
        print(f"  - Mean pooling: {mean_similarity[i, j]:.4f}")
# Output:
# Pairwise similarities:
# Court Complaint vs. Consumer Terms:
#   - CLS token: 0.7112
#   - Mean pooling: 0.6749
# Court Complaint vs. Credit Agreement:
#   - CLS token: 0.8474
#   - Mean pooling: 0.8331
# Consumer Terms vs. Credit Agreement:
#   - CLS token: 0.8276
#   - Mean pooling: 0.7087

Training

The model was trained on a diverse corpus of legal and financial documents, ensuring high-quality performance in these domains. It leverages the KL3M tokenizer which provides 9-17% more efficient tokenization for domain-specific content than cl100k_base or the LLaMA/Mistral tokenizers.

Training included both masked language modeling (MLM) objectives and attention to dense document representation for retrieval and classification tasks.

Intended Usage

This model is intended for both:

Masked Language Modeling: Filling in missing words/terms in legal and financial documents
Document Embedding: Generating fixed-length vector representations for document similarity and classification

Special Tokens

This model includes the following special tokens:

CLS token: <|cls|> (ID: 5) - Used for the beginning of input text
MASK token: <|mask|> (ID: 6) - Used to mark tokens for prediction
SEP token: <|sep|> (ID: 4) - Used for the end of input text
PAD token: <|pad|> (ID: 2) - Used for padding sequences to a uniform length
BOS token: <|start|> (ID: 0) - Beginning of sequence
EOS token: <|end|> (ID: 1) - End of sequence
UNK token: <|unk|> (ID: 3) - Unknown token

Important usage notes:

When using the MASK token for predictions, be aware that this model uses a space-prefixed BPE tokenizer. The <|mask|> token should be placed IMMEDIATELY after the previous token with NO space, because most tokens in this tokenizer have an initial space encoded within them. For example: "word<|mask|>" rather than "word <|mask|>".

This space-aware placement is crucial for getting accurate predictions, as demonstrated in our test examples.

Limitations

While compact compared to larger language models, this model has some limitations:

Limited parameter count (40M) means it captures less nuance than larger language models
Primarily focused on English legal and financial texts
Best suited for domain-specific rather than general-purpose tasks
Maximum sequence length of 512 tokens may require chunking for lengthy documents
Requires domain expertise to interpret results effectively

References

Citation

If you use this model in your research, please cite:

bibtex

@misc{kl3m-doc-pico-001,
  author = {ALEA Institute},
  title = {kl3m-doc-pico-001: A Domain-Specific Language Model for Legal and Financial Text Analysis},
  year = {2024},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/alea-institute/kl3m-doc-pico-001}}
}

@article{bommarito2025kl3m,
  title={KL3M Tokenizers: A Family of Domain-Specific and Character-Level Tokenizers for Legal, Financial, and Preprocessing Applications},
  author={Bommarito, Michael J and Katz, Daniel Martin and Bommarito, Jillian},
  journal={arXiv preprint arXiv:2503.17247},
  year={2025}
}

@misc{bommarito2025kl3mdata,
  title={The KL3M Data Project: Copyright-Clean Training Resources for Large Language Models},
  author={Bommarito II, Michael J. and Bommarito, Jillian and Katz, Daniel Martin},
  year={2025},
  eprint={2504.07854},
  archivePrefix={arXiv},
  primaryClass={cs.CL}
}

License

This model is licensed under CC-BY 4.0.

Contact

The KL3M model family is maintained by the ALEA Institute. For technical support, collaboration opportunities, or general inquiries:

Email: hello@aleainstitute.ai
Website: https://aleainstitute.ai
GitHub: https://github.com/alea-institute/kl3m-model-research

logo