Harshil Karia Llama 2 7B Chat Finetune
This is a fine-tuned version of Llama 2 7B trained to respond as Harshil Karia, a thoughtful entrepreneur.
Usage
1from transformers import AutoModelForCausalLM, AutoTokenizer
2
3model_name = "jyanjain/Harshil-karia-Llama-2-7b-chat-finetune"
4tokenizer = AutoTokenizer.from_pretrained(model_name)
5model = AutoModelForCausalLM.from_pretrained(model_name)
6
7prompt = "<s>[INST] You are Harshil Karia. [Your question] [/INST]"
8inputs = tokenizer(prompt, return_tensors="pt")
9outputs = model.generate(**inputs, max_new_tokens=200)
10
Harshil-Karia-LLaMA-2-7B is a fine-tuned version of LLaMA-2-7B-Chat, customized using QLoRA to mimic Harshil Karia’s communication style. It has been trained on curated data from his YouTube podcasts, LinkedIn posts, and Instagram captions, focusing on branding, startup strategy, and personal storytelling. The model answers with empathy, actionable insights, and reflective anecdotes, making it ideal for creative strategy, entrepreneurship advice, and generating social media content in his tone. Fine-tuning was done in 4-bit precision on Google Colab T4 GPU, and the model is hosted on Hugging Face for easy use in chatbots or AI agents.
Model Description
This is a fine-tuned version of LLaMA-2-7B-Chat trained using QLoRA on a curated dataset of Harshil Karia’s interviews, podcasts, LinkedIn posts, Instagram captions, and other publicly available content.
The model has been optimized to mimic Harshil Karia’s communication style, providing answers with:
Empathy
Storytelling
Clear decision-making frameworks
Spiritual grounding blended with business insights
Base model: NousResearch/Llama-2-7b-chat-hf
Fine-tuning method: QLoRA (Rank 16, Alpha 32)
Dataset size: ~218 curated training samples
Language(s): English
- Developed by: Jyan Jain
- Finetuned from model: Llama-2-7b-chat-hf
Model Sources [optional]
Uses
Direct Use
Ask questions about entrepreneurship, branding, and startup strategy in Harshil Karia’s tone & style
Generate LinkedIn/Instagram-style posts with empathy & storytelling
Provide actionable business advice with a reflective approach
Downstream Use [optional]
Can be used in chatbots, personalized AI agents, or virtual advisors for creative strategy, branding, and startup mentorship.
Out-of-Scope Use
This model is not intended for general-purpose factual Q&A outside the domain of Harshil Karia’s expertise.
Not suitable for medical, legal, or sensitive advice.
Bias, Risks, and Limitations
The model reflects Harshil Karia’s personal viewpoints from the training data, which might not be universally applicable.
May hallucinate details outside the dataset.
Responses might sound opinionated rather than neutral.
Recommendations
Always fact-check critical information before applying it in real-world decisions.
Use it as an assistant for inspiration rather than a single source of truth.
How to Get Started with the Model
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline, BitsAndBytesConfig
model_name = "jyanjain/Harshil-karia-Llama-2-7b-chat-finetune"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto")
pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
response = pipe("< s >[INST] What is Schbang? [/INST]", max_new_tokens=100)[0]["generated_text"] (no space in < s >)
print(response)
Training Details
Training Data
Training Data-
YouTube transcripts: most-viewed podcasts/interviews
LinkedIn posts: high-engagement posts/articles
Instagram captions: posts with long captions
Citation
@misc{harshilkaria2025,
title={Harshil-Karia-LLaMA-2-7B Fine-Tuned Model},
author={Jyan Jain},
year={2025},
howpublished={\url{
https://huggingface.co/jyanjain/Harshil-karia-Llama-2-7b-chat-finetune}}
}
Model Card Aurthor
Author: Jyan Jain
