feat(ImageGen): add stable diffusion support

scripts/commands.py

@@ -9,6 +9,7 @@ import ai_functions as ai
 from file_operations import read_file, write_to_file, append_to_file, delete_file, search_files
 from execute_code import execute_python_file
 from json_parser import fix_and_parse_json
+from image_gen import generate_image
 from duckduckgo_search import ddg
 from googleapiclient.discovery import build
 from googleapiclient.errors import HttpError
@@ -102,6 +103,8 @@ def execute_command(command_name, arguments):
             return ai.write_tests(arguments["code"], arguments.get("focus"))
         elif command_name == "execute_python_file":  # Add this command
             return execute_python_file(arguments["file"])
+        elif command_name == "generate_image":  # Add this command
+            return generate_image(arguments["prompt"])
         elif command_name == "task_complete":
             shutdown()
         else:

scripts/config.py

@@ -53,6 +53,8 @@ class Config(metaclass=Singleton):
         self.pinecone_api_key = os.getenv("PINECONE_API_KEY")
         self.pinecone_region = os.getenv("PINECONE_ENV")
 
+        self.huggingface_api_token = os.getenv("HUGGINGFACE_API_TOKEN")
+
         # User agent headers to use when browsing web
         # Some websites might just completely deny request with an error code if no user agent was found.
         self.user_agent_header = {"User-Agent":"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/83.0.4103.97 Safari/537.36"}

scripts/data/prompt.txt

@@ -23,6 +23,7 @@ COMMANDS:
 17. Write Tests: "write_tests", args: "code": "<full_code_string>", "focus": "<list_of_focus_areas>"
 18. Execute Python File: "execute_python_file", args: "file": "<file>"
 19. Task Complete (Shutdown): "task_complete", args: "reason": "<reason>"
+20. Generate Image: "generate_image", args: "prompt": "<prompt>"
 
 RESOURCES:
 

scripts/image_gen.py

@@ -1,44 +1,28 @@
-from kandinsky2 import get_kandinsky2
+import requests
+import io
+import os.path
+from PIL import Image
 from config import Config
+import uuid
 
 cfg = Config()
 
+working_directory = "auto_gpt_workspace"
+
+API_URL = "https://api-inference.huggingface.co/models/CompVis/stable-diffusion-v1-4"
+headers = {"Authorization": "Bearer " + cfg.huggingface_api_token}
+
 def generate_image(prompt):
-        
+    response = requests.post(API_URL, headers=headers, json={
-    model = get_kandinsky2('cuda', task_type='text2img', model_version='2.1', use_flash_attention=False)
+        "inputs": prompt,
-    images = model.generate_text2img(
+    })
-        "red cat, 4k photo", # prompt
+    image = Image.open(io.BytesIO(response.content))
-        num_steps=100,
+    print("Image Generated for prompt:" + prompt)
-        batch_size=1, 
-        guidance_scale=4,
-        h=768, w=768,
-        sampler='p_sampler', 
-        prior_cf_scale=4,
-        prior_steps="5"
-    )
-    return images
-    
-    # base_url = 'http://export.arxiv.org/api/query?'
-    # query = f'search_query=all:{search_query}&start=0&max_results={max_results}'
-    # url = base_url + query
-    # response = requests.get(url)
 
-    # if response.status_code == 200:
+    filename = str(uuid.uuid4()) + ".jpg"
-    #     soup = BeautifulSoup(response.content, 'xml')
-    #     entries = soup.find_all('entry')
 
-    #     articles = []
+    image.save(os.path.join(working_directory, filename))
-    #     for entry in entries:
-    #         title = entry.title.text.strip()
-    #         url = entry.id.text.strip()
-    #         published = entry.published.text.strip()
 
-    #         articles.append({
+    print("Saved to disk:" + filename)
-    #             'title': title,
-    #             'url': url,
-    #             'published': published
-    #         })
 
-    #     return articles
+    return str("Image " + filename + " saved to disk for prompt: " + prompt)
-    # else:
-    #     return None