This is a LoRA (Low‑Rank Adaptation) adapter for the unsloth/tinyllama-bnb-4bit base model.
It has been fine‑tuned on a synthetic question‑answer dataset generated from the textbook "Physics of Black Holes" (2009). The adapter enhances the model's ability to answer domain‑specific questions about general relativity, event horizons, Hawking radiation, and black hole thermodynamics.

Developed by: Varun95
Model type: Causal language model with LoRA adapters (PEFT)
Language(s): English
Finetuned from: unsloth/tinyllama-bnb-4bit
License: Apache 2.0

Intended Uses & Limitations

Direct Use

The adapter is meant to be used on top of the base TinyLlama model for retrieval‑augmented generation (RAG) or direct Q&A about black hole physics.
It works best when provided with context from the original textbook (e.g., via vector search).

Limitations

The model is not a standalone – it requires the base model to be loaded.
Knowledge is confined to the contents of a single 2009 textbook; it may not reflect recent discoveries.
Generated answers may be verbose or occasionally hallucinate; always verify against the source.

How to Use

Loading the Adapter (Recommended)

python

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

# Load the quantized base model
base_model = AutoModelForCausalLM.from_pretrained(
    "unsloth/tinyllama-bnb-4bit",
    load_in_4bit=True,
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained("unsloth/tinyllama-bnb-4bit")
tokenizer.pad_token = tokenizer.eos_token

# Load your LoRA adapter
model = PeftModel.from_pretrained(base_model, "Varun95/finetune-lama-vectordb")

# Inference
prompt = "What is a black hole?"
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=200)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))