SallySims/AnthroBot_Model_Lora

Model Card for AnthroBot (Llama-3.2-1B-Instruct Fine-tuned)

This model is a fine-tuned version of meta-llama/Llama-3.2-1B-Instruct, adapted for reasoning and generating contextual insights from anthropometric data (e.g., age, sex, weight, height, waist circumference). It can summarise or comment on health-related metrics conversationally.

Model Details

Model Description

This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.

• Developed by: Sally S. Simmons
• Funded by [optional]: NA
• Shared by [optional]: https://huggingface.co/SallySims
• Model type: Causal Language Model (LLM) with Instruction Tuning
• Language(s) (NLP): English
• License: Apache 2.0 (or specify if different)
• Finetuned from model [optional]: meta-llama/Llama-3.2-1B-Instruct

Model Sources [optional]

• Repository: https://huggingface.co/SallySims/AnthroBot_Model_Lora
• Paper [optional]: [More Information Needed]
• Demo [optional]: [More Information Needed]

Uses

Direct Use

The model is intended to analyze structured health-related user inputs and return conversational, personalized feedback.It is designed for educational, wellness, or research purposes.

Downstream Use [optional]

This model can be incorporated into chatbot systems or mobile health platforms that require health-data-aware natural language interaction.

Out-of-Scope Use

*Medical diagnosis or treatment

*Critical healthcare decision-making

*Inputs in languages other than English

Bias, Risks, and Limitations

The model is trained on 20000 observations based on anthropometric data collected during the WHO STEPS survey and 32000 synthetic data not in clinical settings. Outputs may reflect biases present in the training prompts or may misinterpret edge cases.

Recommendations

Seek professional guidance in addition to the outcomes produced by the model

How to Get Started with the Model

Use the code below to get started with the model.

from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline

model_id = "SallySims/AnthroBot_Model_Lora"

tokenizer = AutoTokenizer.from_pretrained(model_id) model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")

pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)

input_text = "Age: 30, Sex: female, Height: 150.5 cm, Weight: 75.3 kg, WC: 68.0 cm" output = pipe(input_text, max_new_tokens=150, do_sample=True) print(output[0]['generated_text'])

Training Details

Training Data

Custom curated structured anthropometric prompts designed to simulate health-focused instruction-following behavior.

Training Procedure

Preprocessing [optional]

Prompts were normalised for consistent numerical formats and tokenization performance.

Training Hyperparameters

• Training regime: [More Information Needed] Epochs: 5

Batch size: 2 (accumulation: 4 steps)

Learning rate: 2e-4

Precision: Mixed precision (fp16 / bf16)

LoRA Parameters:

r=16, alpha=32, dropout=0.05

Quantization 4-bit quantization using BitsAndBytesConfig

Enabled llm_int8_enable_fp32_cpu_offload

Speeds, Sizes, Times [optional]

Evaluation

Testing Data, Factors & Metrics

Testing Data

Evaluation performed on held-out anthropometricindices and recommendations prompts with expected interpretive outputs.

Factors

Metrics

Human-judged relevance, clarity, and accuracy.

Results

Manual inspection shows clear, concise, and useful summaries in the majority of cases. Some rare edge cases may produce vague or overly generic responses.

Summary

Model Examination [optional]

Environmental Impact

Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

• Hardware Type: NVIDIA T4 GPU
• Hours used: ~ 2 hours
• Cloud Provider: Google Colab
• Compute Region: USA
• Carbon Emitted: ~1.2 kg CO₂eq (approx.)

Technical Specifications [optional]

Model Architecture and Objective

Decoder-only transformer based on the LLaMA 3.2B architecture.

Compute Infrastructure

Hardware

Google Colab (A100)

Software

PyTorch, Hugging Face Transformers, PEFT, BitsAndBytes

Citation [optional]

@misc{AnthroBot2025, author = {Sally Sonia Simmons}, title = {AnthroBot: Instruction-Tuned LLaMA-3.2-1B for Anthropometric Reasoning}, year = {2025}, url = {https://huggingface.co/SallySimmons/AnthroBot_Model_Lora} }

BibTeX:

APA:

Glossary [optional]

NA

More Information [optional]

NA

Model Card Authors [optional]

NA

Model Card Contact

simmonssallysonia@gmail.com