app.py

import gradio as gr, spaces
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline


# tokenizer = AutoTokenizer.from_pretrained("LingoIITGN/ganga-1b")
# model = AutoModelForCausalLM.from_pretrained("LingoIITGN/ganga-1b")

# @spaces.GPU
# def greet(input_text):
#     input_token = tokenizer.encode(input_text, return_tensors="pt").to("cuda")
#     output = model.generate(input_token, max_new_tokens=100, num_return_sequences=1, do_sample=True, top_k=50, top_p=0.95, temperature=0.7)
#     output_text = tokenizer.batch_decode(output)[0]
#     return output_text

# demo = gr.Interface(fn=greet, inputs=["text"], outputs=["text"],)

@spaces.GPU
def greet(input_text):
    input_token = tokenizer.encode(input_text, return_tensors="pt").to("cpu")

    output = model.generate(input_token, max_new_tokens=100, num_return_sequences=1, do_sample=True, top_k=50, top_p=0.95, temperature=0.7)
    output_text = tokenizer.batch_decode(output)[0]
    return output_text


demo = gr.Interface(fn=greet, inputs=["text"], outputs=["text"],)
demo.launch()