import gradio as gr import os import openai from diffusers import StableDiffusionPipeline, DPMSolverMultistepScheduler import torch model_id = "stabilityai/stable-diffusion-2-1" pipe = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5") # pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16) # pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config) # pipe = pipe.to("cuda") openai.api_key = os.getenv("OPENAI_API_KEY") def generate_prompt(radio,word1,word2): #prompt = 'Create an analogy for this phrase:\n\n{word1}' # 50/50 in that/because # pluralize singluar words if radio == "normal": prompt_in = f'Create an analogy for this phrase:\n\n{word1} is like {word2} in that:' else: prompt_in = f'Create a {radio} analogy for this phrase:\n\n{word1} is like {word2} in that:' response = openai.Completion.create( model="text-davinci-003", prompt=prompt_in, temperature=0.5, max_tokens=60, top_p=1.0, frequency_penalty=0.0, presence_penalty=0.0 )['choices'][0]['text'] response_txt = response.replace('\n','') diffusion_in = f'a dramatic painting of: {response_txt.split(".")[0]}' image = pipe(diffusion_in).images[0] return response_txt, image demo = gr.Interface( generate_prompt, [ gr.Radio(["normal", "very insulting"],value='normal',label="Flavor"), gr.Textbox(label="Thing 1"), gr.Textbox(label="Thing 2") # gr.Dropdown(team_list, value=[team_list[random.randint(1,30)]], multiselect=True), # gr.Checkbox(label="Is it the morning?"), ], ["text","image"], # "image", allow_flagging="never", title="GPT-3 Analogy Lab 🧪", description="Enter two things you want to connect.", css="footer {visibility: hidden}" ) demo.launch() #openai.api_key = os.getenv("OPENAI_API_KEY") # openai.api_key = "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" # response = openai.Completion.create( # model="text-davinci-003", # prompt="Create an analogy for this phrase:\n\nQuestions are arrows in that:", # temperature=0.5, # max_tokens=60, # top_p=1.0, # frequency_penalty=0.0, # presence_penalty=0.0 # )