CoolFace
Modelpublic

abirmaheshwari/abirmed_emergency_slm

sourceHugging Facemitupdated 7mo agoView on Hugging Face
0likes5downloads
Model Card

abirmedemergencyslm — Emergency and Triage Response Transformer

Part of the A.B.I.R Ecosystem

abirmedemergencyslm is a specialized medical language model developed as part of the A.B.I.R Ecosystem and the ABIRMED Modular Medical Specialist Transformer System, a distributed artificial intelligence architecture designed to replicate real-world medical specialization using modular transformer models.

This model functions as the Emergency and Triage Specialist, designed to recognize urgent medical conditions, understand emergency symptoms, and provide emergency awareness and triage-level reasoning.

This is Version 1.0, with future versions planned to improve emergency recognition accuracy, incorporate expanded emergency medical datasets, and enhance real-time triage reasoning capabilities.

ABIRMED — Modular Medical Specialist Transformer System

ABIRMED is a modular medical AI ecosystem consisting of multiple specialist Small Language Models (SLMs), each trained for a specific medical domain. Instead of using a single large monolithic model, ABIRMED uses a distributed specialist architecture inspired by real-world clinical specialization.

Each model acts as an independent medical specialist while collectively forming a unified medical reasoning system.

This modular approach provides:

  • —Higher accuracy within specialized domains
  • —Lower computational requirements
  • —CPU-efficient inference capability
  • —Scalable and extensible medical intelligence architecture

Developed by: Abir Maheshwari Architecture: Modular Decoder-only Transformer System Framework: PyTorch + HuggingFace Transformers Training Platform: Google Colab T4 GPU License: MIT

Role of abirmedemergencyslm in the ABIRMED System

abirmedemergencyslm functions as the Emergency Response and Triage Specialist, equivalent to an emergency physician or triage nurse in real-world healthcare systems.

Its primary role is to provide emergency medical reasoning capabilities including:

  • —Emergency symptom recognition
  • —Urgent medical condition identification
  • —Triage-level medical reasoning
  • —Emergency awareness support
  • —Immediate risk assessment explanation

This model operates alongside other ABIRMED specialists such as diagnosis, pharmacology, pathology, psychiatry, cardiology, pediatrics, dermatology, and veterinary models.

Model Details

Model Name: abirmedemergencyslm Version: 1.0 Developer: Abir Maheshwari Organization: A.B.I.R Ecosystem Model Type: Causal Language Model (Decoder-only Transformer) Base Model: None (trained from scratch) License: MIT

Technical Specifications

Architecture: Decoder-only Transformer

Parameters: ~38 Million

Transformer Layers: 8

Attention Heads: 8

Hidden Size: 512

Intermediate Size: 2048

Context Length: 256 tokens

Tokenizer: GPT-2 tokenizer with custom PAD token

Weight Sharing: Embedding and LM Head tied

Training Objective: Causal Language Modeling

Precision: FP16 mixed precision

Framework: PyTorch

Export Formats:

  • —safetensors
  • —PyTorch (.pt)

Checkpoint Support:

  • —Full training state resume capability

Training Details

Training Dataset

Primary datasets used:

medalpaca/medicalmeadowhealth_advice lavita/MedQuAD (emergency-related subsets)

These datasets contain structured emergency medical knowledge including:

  • —Emergency symptom explanations
  • —Urgent care conditions
  • —First-response medical awareness
  • —Emergency clinical reasoning

This enables the model to understand relationships between symptoms and emergency conditions.

Training Procedure

Optimizer: AdamW

Learning Rate: 5e-4

Batch Size: 8

Gradient Accumulation Steps: 2

Training Platform:

  • —Google Colab
  • —NVIDIA T4 GPU

Training Objective:

  • —Predict next token in emergency reasoning sequences

Training Format:

Instruction → Output

Converted to:

Question → Answer format

Identity training lines were included to ensure proper integration into the ABIRMED ecosystem.

Capabilities

abirmedemergencyslm is capable of:

  • —Recognizing emergency medical symptoms
  • —Explaining urgent medical conditions
  • —Supporting triage-level reasoning
  • —Providing emergency awareness explanations
  • —Supporting emergency medical education

Example:

Input: "Chest pain and shortness of breath"

Output: "These symptoms may indicate a serious medical condition such as a heart attack and require immediate medical evaluation."

Intended Use

This model is intended for:

  • —Emergency medical education
  • —Medical AI research
  • —Emergency awareness systems
  • —Healthcare training simulations
  • —Medical chatbot development

Out-of-Scope Use

This model is not intended for:

  • —Real-world emergency decision making
  • —Clinical emergency treatment decisions
  • —Medical treatment recommendations
  • —Replacement of emergency medical professionals

This is a research model only.

Limitations

abirmedemergencyslm:

  • —Is not a licensed emergency medical system
  • —May produce incomplete emergency assessments
  • —Should not replace trained emergency professionals
  • —May lack full clinical emergency accuracy

Design Philosophy

The ABIRMED ecosystem follows a modular specialist architecture inspired by real-world medical systems.

Each model specializes in a specific domain.

abirmedemergencyslm serves as the emergency medical intelligence specialist.

This architecture improves:

  • —Emergency reasoning accuracy
  • —Specialist-level domain understanding
  • —Computational efficiency
  • —Modular scalability

A.B.I.R Ecosystem Integration

abirmedemergencyslm is part of the A.B.I.R Ecosystem, which includes:

  • —Modular transformer intelligence systems
  • —Language models
  • —Domain-specialized AI systems
  • —Medical AI infrastructure

ABIRMED represents the medical intelligence division of the A.B.I.R Ecosystem.

Version

Version: 1.0

Future versions will include:

  • —Expanded emergency medical datasets
  • —Improved emergency reasoning accuracy
  • —Larger training datasets
  • —Enhanced triage intelligence

Author

Abir Maheshwari Independent AI Researcher Founder, A.B.I.R Ecosystem

Hugging Face: https://huggingface.co/abirmaheshwari

License

MIT License