completion / app.py
xlr8harder
add provider require_parameters option to exclude providers that do not support completions api to avoid random errors
1ce16f9
import gradio as gr
import requests
import json
import os
default_prompt = "The best thing about being a cat is"
def generate_completion(api_key, model, prompt, temperature, repetition_penalty, stop_phrase, max_tokens):
if not api_key:
return "Please enter your OpenRouter API key."
try:
# Prepare the request payload
payload = {
"model": model,
"prompt": prompt,
"temperature": temperature,
"frequency_penalty": repetition_penalty,
"max_tokens": max_tokens,
"provider": {
"require_parameters": True
}
}
# Add stop phrase if provided
if stop_phrase:
payload["stop"] = [stop_phrase]
# Prepare headers
headers = {
"Authorization": f"Bearer {api_key}",
"Content-Type": "application/json"
}
# Make the API request
response = requests.post(
"https://openrouter.ai/api/v1/completions",
headers=headers,
json=payload,
timeout=60
)
# Check if request was successful
response.raise_for_status()
# Parse the response
result = response.json()
# Extract the completion text
if "choices" in result and len(result["choices"]) > 0:
return result["choices"][0]["text"].strip()
else:
return "No completion generated."
except requests.exceptions.RequestException as e:
return f"Request error: {str(e)}"
except json.JSONDecodeError as e:
return f"JSON decode error: {str(e)}"
except Exception as e:
return f"An error occurred: {str(e)}"
def append_completion(prompt, completion):
new_prompt = f"{prompt}{completion}".strip()
return new_prompt, "" # Return new prompt and empty completion
def clear_fields():
return "", ""
with gr.Blocks(theme=gr.themes.Soft()) as iface:
gr.Markdown("# Base Model Completion Interface via OpenRouter")
# API Key input at the top
api_key_input = gr.Textbox(
label="OpenRouter API Key",
type="password",
placeholder="Enter your OpenRouter API key"
)
with gr.Row():
with gr.Column(scale=2):
prompt_input = gr.Textbox(label="Prompt", lines=6, value="The best thing about being a cat is")
with gr.Column(scale=1):
model_input = gr.Textbox(
label="Model",
value="meta-llama/llama-3.1-405b",
placeholder="Enter model name"
)
temperature_slider = gr.Slider(minimum=0, maximum=1, value=0.7, step=0.1, label="Temperature")
repetition_penalty_slider = gr.Slider(minimum=0, maximum=2, value=0.1, step=0.1, label="Repetition Penalty")
max_tokens_slider = gr.Slider(minimum=1, maximum=4000, value=250, step=1, label="Max Tokens")
stop_phrase_input = gr.Textbox(label="Stop Phrase", placeholder="Enter stop phrase (optional)")
with gr.Row():
generate_button = gr.Button("Generate Completion")
append_button = gr.Button("Append Completion to Prompt")
clear_button = gr.Button("Clear All Fields")
output_text = gr.Textbox(label="Generated Completion", lines=10)
generate_button.click(
generate_completion,
inputs=[api_key_input, model_input, prompt_input, temperature_slider, repetition_penalty_slider, stop_phrase_input, max_tokens_slider],
outputs=output_text
)
append_button.click(
append_completion,
inputs=[prompt_input, output_text],
outputs=[prompt_input, output_text]
)
clear_button.click(
clear_fields,
outputs=[prompt_input, output_text]
)
iface.launch(share=True)