Edit model card

This model is a chatbot that answers users all their road-related queries.

Model Description

Meet roadGPT, your go-to chatbot powered by advanced AI, designed to provide precise answers to a wide range of road-related queries. Leveraging the comprehensive guidelines of the Indian Roads Congress (IRC), roadGPT is equipped to offer expert advice, best practices, and regulatory standards essential for road engineers, planners, and enthusiasts alike.

Capabilities & Guideline Interpretation:

roadGPT meticulously interprets and applies the latest IRC guidelines to ensure accurate and up-to-date information. Wide-Ranging Expertise: From design and construction to maintenance and safety norms, roadGPT covers all aspects of road engineering. Get immediate responses to your queries, saving valuable time and effort in accessing and interpreting complex IRC documents.

Source of Knowledge:

At the heart of roadGPT's responses lie the authoritative IRC guidelines, which are meticulously encoded into our model. This ensures that every piece of advice or information you receive is grounded in recognized standards and practices.

User Interaction:

Engaging with roadGPT is straightforward and intuitive:

Simply pose your question related to road design, construction, maintenance, or any other road-related topic. roadGPT will process your query, referencing the relevant IRC guidelines to provide a clear, concise, and accurate response.

Deployment:

roadGPT is proudly hosted on Huggingface, making it accessible to professionals, students, and road safety advocates worldwide. Our user-friendly interface ensures that you can get the answers you need with just a few clicks, regardless of your technical expertise.

Inference

Quickly get inference running with the following required installation:

pip install transformers peft accelerate

Now, proceed as usual with HuggingFace:

import os
from transformers import (
    AutoModelForCausalLM,
    AutoTokenizer,
    BitsAndBytesConfig,
    pipeline,
)
from transformers import BitsAndBytesConfig

bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype="float16",
    bnb_4bit_use_double_quant=False,
)

model_name = "rukaiyah-indika-ai/rv-chatbot-2"

model = AutoModelForCausalLM.from_pretrained(
    model_name,
    quantization_config=bnb_config
)
    
def generate_response(prompt):
    inst = "You are a very helpful assistant providing solutions to road-related queries. Ensure you provide correct and relevant answers according to the IRC guidelines. If you don't know the answer to a question, please don't share false information."
    prompt = "What are potholes, and how do they appear? "
    pipe = pipeline(task="text-generation", model=model, tokenizer=tokenizer, temperature=0.2, max_new_tokens=256)
    ranked_results = pipe(f"<s>[INST] {inst}{prompt} [/INST]")

    for result in ranked_results:
        response = result['generated_text']
        response = response.split("[/INST]", 1)[-1]
        response = response.replace("<s>", "")
        response = response.replace("</s>", "")
        return response

>> ' Potholes are bowl-shaped cavities in pavements caused by localized disintegration. They vary in size: small (25mm deep, 200mm wide), medium (25-50mm deep, up to 500mm wide), and large (exceeding 50mm deep, more than 500mm wide). Factors like water infiltration, traffic stress, and weathering lead to their development by breaking down pavement layers.'

Dive into the world of road engineering with roadGPT at your side. Whether you're drafting a project, studying for your next exam, or simply curious about road regulations, roadGPT is here to guide you according to the esteemed IRC guidelines. Start your query now and experience the future of road-related assistance.

Downloads last month
0
Safetensors
Model size
6.74B params
Tensor type
FP16
·
Inference Examples
Inference API (serverless) does not yet support adapter-transformers models for this pipeline type.