Merged_Aura_Airavata

This model merges Aura_v3_7B and Airavata using upscaling and weighted blending.

Model Details

Base Models: ResplendentAI/Aura_v3_7B + ai4bharat/Airavata
Merging Method: Upscaling + Weighted Averaging (α=0.5)
Tokenizer: Compatible with Llama tokenizers

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("YOUR_USERNAME/Merged_Aura_Airavata")
tokenizer = AutoTokenizer.from_pretrained("YOUR_USERNAME/Merged_Aura_Airavata")

prompt = "Hello, how are you?"
input_ids = tokenizer(prompt, return_tensors="pt").input_ids
output = model.generate(input_ids)
print(tokenizer.decode(output[0], skip_special_tokens=True))