File size: 1,508 Bytes
0cad702 f6a1e93 0cad702 f6a1e93 0cad702 ef06e2c 0cad702 f6a1e93 0cad702 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 |
import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM
# Load GraphCodeBERT model and tokenizer
tokenizer = AutoTokenizer.from_pretrained("microsoft/graphcodebert-base")
model = AutoModelForCausalLM.from_pretrained("microsoft/graphcodebert-base")
# Define input and output interfaces
input = gr.Textbox(lines=5, label="Input")
output = gr.Textbox(label="Output")
# Define function to use GraphCodeBERT
def use_graphcodebert(input):
# Encode input
input_ids = tokenizer.encode(input, return_tensors="pt")
# Generate output
output_ids = model.generate(input_ids, max_length=50)
# Decode output
output = tokenizer.decode(output_ids[0], skip_special_tokens=True)
# Return output
return output
# Create and launch Gradio interface
# Create and launch Gradio interface
iface = gr.Interface.from_pretrained( # Use from_pretrained instead of from_pipeline
fn=use_graphcodebert,
inputs=input,
outputs=output,
model=model, # Pass the model as a parameter
tokenizer=tokenizer, # Pass the tokenizer as a parameter
title="GraphCodeBERT Code Synthesis", # Add a title for the web app
description="Enter a natural language query and get a code snippet generated by GraphCodeBERT.", # Add a description for the web app
examples=[["create a function that returns the sum of two numbers"], ["sort a list of numbers in ascending order"]], # Add some examples for the input
theme="huggingface" # Choose a theme for the web app
)
iface.launch()
|