placed bot_comms and reversions to the chimera which is now multimodal_and_generation()

app.py

@@ -7,6 +7,10 @@ import accelerate
 import spaces
 from PIL import Image
 import threading
+from openai import OpenAI
+import os
+
+API_KEY = os.getenv('OPEN_AI_API_KEYS')
 
 DESCRIPTION = '''
 <div>
@@ -47,8 +51,7 @@ refiner = DiffusionPipeline.from_pretrained(
 refiner.to('cuda')
 
 # All Installed. Let's instance them in the function
-@spaces.GPU(duration=120)
-def chimera(message, history):
+def multimodal_and_generation(message, history):
     """
     Receives input from gradio from the prompt but also
     if any images were passed that i also placed for formatting
@@ -70,22 +73,17 @@ def chimera(message, history):
     prompt = f"<|start_header_id|>user<|end_header_id|>\n\n<image>\n{message['text']}<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n"
 
     if image_path is None:
-        image = base(
-            prompt=prompt,
-            num_inference_steps=40,
-            denoising_end=0.8,
-            output_type="latent",
-        ).images
-        image = refiner(
-            prompt=prompt,
-            num_inference_steps=40,
-            denoising_start=0.8,
-            image=image
-        ).images[0]
-        return image
+        # GPT Generation
+        client = OpenAI(api_key=API_KEY)
+        stream = client.chat.completions.create(
+            model="gpt-3.5-turbo",
+            messages=[{"role": "system", "content": "You are a helpful assistant called 'chimera'."},
+                      {"role": "user", "content": message}],
+            stream=True,
+        )
+        return stream
 
     else:
-
         # Time to instance the llava
         image = Image.open(image_path)
         inputs = processor(prompt, image, return_tensors='pt').to(0, torch.float16)
@@ -95,12 +93,85 @@ def chimera(message, history):
         thread = threading.Thread(target=llava_model.generate, kwargs=generation_kwargs)
         thread.start()
 
-        buffer = ""
-        for new_text in streamer:
-            # find <|eot_id|> and remove it from the new_text
-            if "<|eot_id|>" in new_text:
-                new_text = new_text.split("<|eot_id|>")[0]
-            buffer += new_text
+        # buffer = ""
+        # for new_text in streamer:
+        #     # find <|eot_id|> and remove it from the new_text
+        #     if "<|eot_id|>" in new_text:
+        #         new_text = new_text.split("<|eot_id|>")[0]
+        #     buffer += new_text
+        #     generated_text_no_prompt = buffer
+        #     yield generated_text_no_prompt
+        return streamer
+    
+def diffusing(prompt):
+    """
+    Uses stable diffusion on the prompt and
+    returns the image.
+    """
+    image = base(
+        prompt=prompt,
+        num_inference_steps=40,
+        denoising_end=0.8,
+        output_type="latent",
+    ).images
+    image = refiner(
+        prompt=prompt,
+        num_inference_steps=40,
+        denoising_start=0.8,
+        image=image
+    ).images[0]
+    return image
+
+def check_cuda_availability():
+    if torch.cuda.is_available():
+        result = f"GPU: {torch.cuda.get_device_name(0)}"
+        return result
+    else:
+        return "No CUDA device found."
+
+mode = ""
+
+@spaces.GPU(duration=120)
+def bot_comms(message,
+              history):
+    """
+    Communication between gradio and the models.
+    """
+    global mode
+
+    if message == "check cuda":
+        result = check_cuda_availability()
+        yield result
+        return
+    
+    if message == "imagery":
+        mode = message
+        yield "Imagery On! Type your prompt to make the image 🖼️"
+        return
+    
+    if message == "chatting":
+        mode = message
+        yield "Imagery Off. Ask me any questions. ☄️"
+        return
+    
+    if mode == "imagery":
+        image = diffusing(
+            message=message,
+            history=history,
+        )
+        return image
+        
+    buffer = ""
+    if mode == "chatting" or mode == "":
+        stream = multimodal_and_generation(
+            message=message,
+            history=history,
+        )
+        for text in stream:
+            # find <|eot_id|> and remove it from the text
+            if "<|eot_id|>" in text:
+                text = text.split("<|eot_id|>")[0]
+            buffer += text
             generated_text_no_prompt = buffer
             yield generated_text_no_prompt
 
@@ -110,7 +181,7 @@ chat_input = gr.MultimodalTextbox(interactive=True, file_types=["images"], place
 with gr.Blocks(fill_height=True) as demo:
     gr.Markdown(DESCRIPTION)
     gr.ChatInterface(
-        fn=chimera,
+        fn=bot_comms,
         chatbot=chatbot,
         fill_height=True,
         multimodal=True,