Spaces:
Sleeping
Sleeping
File size: 1,884 Bytes
d70e310 c245a44 d70e310 c245a44 197a245 d70e310 c245a44 d70e310 971836c c245a44 971836c c245a44 971836c d70e310 971836c c245a44 971836c c245a44 b6b3a04 971836c d70e310 b6b3a04 c245a44 971836c d70e310 c245a44 |
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 38 39 40 41 42 43 44 45 46 47 48 |
import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
# Load the tokenizer and model from Hugging Face
model_name = "waterdrops0/mistral-nouns400"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto", torch_dtype=torch.float16)
def generate_text(prompt, max_length=50, temperature=0.7, repetition_penalty=1.2):
# Encode the input prompt
inputs = tokenizer.encode(prompt, return_tensors="pt").to(model.device)
# Generate output based on the prompt with repetition penalty
outputs = model.generate(
inputs,
max_length=max_length + inputs.shape[1], # Ensuring generated text extends beyond the input prompt
temperature=temperature,
repetition_penalty=repetition_penalty, # Add repetition penalty
do_sample=True,
top_p=0.95,
top_k=60
)
# Decode the generated tokens, skipping the input tokens
generated_tokens = outputs[0, inputs.shape[1]:] # Only get the new tokens
generated_text = tokenizer.decode(generated_tokens, skip_special_tokens=True)
return generated_text
# Update the Gradio interface to include repetition penalty slider
iface = gr.Interface(
fn=generate_text,
inputs=[
gr.Textbox(lines=2, placeholder="Enter your prompt here...", label="Prompt"),
gr.Slider(10, 200, step=10, value=50, label="Max Length"),
gr.Slider(0.1, 1.0, step=0.1, value=0.7, label="Temperature"),
gr.Slider(1.0, 2.0, step=0.1, value=1.2, label="Repetition Penalty") # Add a slider for repetition penalty
],
outputs=gr.Textbox(label="Generated Text"),
title="Mistral 7B Nouns Model",
description="Generate text using the fine-tuned Mistral 7B model with repetition penalty."
)
if __name__ == "__main__":
iface.launch()
|