README.md

---
library_name: transformers
tags:
- text-generation
- cemp
- assisted-living
- LoRA
- tinyllama
---

# Model Card for `alfboss/cempbot-tiny`

This model is a fine-tuned version of TinyLlama-1.1B-Chat-v1.0, specialized in answering questions related to CEMP (Comprehensive Emergency Management Plans) for Assisted Living Facilities in Florida. It has been trained using LoRA (Low-Rank Adaptation) with domain-specific instructions and responses generated by Evergreen Brain Pvt. Ltd.

## Model Details

### Model Description

- **Developed by:** Jagdish Sharma and team at Evergreen Brain Pvt. Ltd.
- **Shared by:** [@evergreen-brain](https://huggingface.co/evergreen-brain)
- **Model type:** Causal Language Model (AutoModelForCausalLM)
- **Language(s):** English
- **License:** apache-2.0
- **Finetuned from model:** [TinyLlama/TinyLlama-1.1B-Chat-v1.0](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0)

### Model Sources

- **Repository:** https://huggingface.co/alfboss/cempbot-tiny
- **Demo:** Available via Hugging Face Inference Endpoints (e.g., `/generate` endpoint)

## Uses

### Direct Use

This model can be used to:
- Guide assisted living facilities in Florida on CEMP requirements
- Answer domain-specific questions like document submission, emergency contact planning, AHCA compliance, etc.
- Serve as an onboarding chatbot or compliance assistant for ALF administrators

### Downstream Use

This model can be integrated into:
- Facility portals for onboarding staff
- AI-powered chatbots
- CEMP automation tools

### Out-of-Scope Use

- Not intended for generating general-purpose content
- Not suitable for legal or emergency advice without expert review
- Not trained for multi-turn conversation or open-domain chat

## Bias, Risks, and Limitations

- The model was fine-tuned on a narrow domain, so its responses may not generalize well to other topics
- There may be hallucinations if prompted outside the CEMP compliance domain
- Not suitable for critical compliance filings without human verification

### Recommendations

Always review model outputs with a compliance expert before submission. Use the model in supervised environments.

## How to Get Started with the Model

```python
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("alfboss/cempbot-tiny")
model = AutoModelForCausalLM.from_pretrained("alfboss/cempbot-tiny")

prompt = "What is required in a Florida ALF CEMP plan?"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=150)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))