You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
So I guess this a general question but I would really appreciate an ELI5 response.
I will provide all the information anybody would need to debug it. I will make my best effort summarising it for other's benefit too.
I have devika installed and runnning with ollama and I have tested it a few times and have run a few taske.
The system I am running is ubuntu linux with 96GB VRAM, 128GB RAM and I have currently Deepseek 67B chat model running.
I asked a simple question:
Create me an app with modern ui where I click on the text "Generate Cat" and a cute cat art is displayed and when I click on "Generate dog" a cute dog art is displayed. source the images for the project. and create the code in python.
And I am really impressed with Devika's reasoning capabilities. I ally am!. Its beautiful how its managing this.
I got:
Here's my step-by-step plan:
Step 1: Source high-quality cat and dog art images from reputable sources such as Shutterstock, Unsplash, or Getty Images. Ensure that these images are free of copyright issues for our use in the application. Save them in separate folders named 'cats' and 'dogs'.
Step 2: Create a Python script (main.py) to handle image generation based on user input. The script will have two functions - generate_cat() and generate_dog(). These functions should read an image from the corresponding folder, apply some random effects or filters (if required), and save it in a new location within the respective folders.
Step 3: Set up a modern UI for the application using a Python framework like Tkinter, PyQt, or Kivy. Create two buttons named "Generate Cat" and "Generate Dog". These buttons should call the appropriate function from step 2 when clicked.
Step 4: Design an appealing layout for the app to enhance user experience. Ensure that the UI is clean, intuitive, and easy to navigate. Include necessary elements like a loading screen or progress bar if image generation takes time.
Step 5: Integrate the Python script (main.py) into the application developed in step 3. Test the app thoroughly to ensure it works as expected without any errors or glitches. Verify that clicking on buttons generates appropriate pet images and displays them correctly within the UI.
Step 6: Package the application for distribution, ensuring compatibility with different operating systems (Windows, macOS, Linux). Include an installation guide in the package for users to follow easily.
Step 7: Document the codebase, providing clear explanations of each component and their interactions. This documentation should assist others who might be working on the project or using it as a reference.
Step 8: Share the application with you upon completion, along with instructions on how to install and run it on your device. Provide support for any issues that may arise during installation or usage of the app.
Now here are a list of issues I am facing and I have pulled up the latest version today so this is fresh off the bakery lol.
The terminal shows that the devika is alive. it searches the web, fetches the codes and the information however I can see that its struggling with code. I get Invalid response from the model, trying again... for all the code thats run in the terminal.
So the model tries the code or the same approach over and over for the code and every time its the same response so the code it saves finally in the project is an extremely simple, dumbed down code wqhen I can see in terminal that the code iterations when it started to work were quite good and intelligent. How do I get Devika to run in my environment and iterate it properly over there?
I for the life of me dont see the point of terminal window when I dont get it doing anything till I see the final code. Can I do what devika does but in the vscode environment?
2.1. Can Devika save the codes that it has iterated over in a text folder within the project?
take a look at this:
Model: deepseek-llm:67b-chat, Enum: OLLAMA
24.04.24 09:51:19: root: INFO : SOCKET tokens MESSAGE: {'token_usage': 5726}
24.04.24 09:51:19: root: DEBUG : Response from the model: File: `main.py`
```python
import random
from PIL import Image, ImageFilter
def generate_cat():
"""Generates a cute cat art image."""
base_image = Image.open('cats/base_image.jpg')
effect = random.choice(['blur', 'contour', 'detail', 'edge_enhance', 'smooth'])
filtered_image = base_image.filter(getattr(ImageFilter, effect))
new_name = f'cat_{effect}.jpg'
filtered_image.save('cats/generated/' + new_name)
def generate_dog():
"""Generates a cute dog art image."""
base_image = Image.open('dogs/base_image.jpg')
effect = random.choice(['blur', 'contour', 'detail', 'edge_enhance', 'smooth'])
filtered_image = base_image.filter(getattr(ImageFilter, effect))
new_name = f'dog_{effect}.jpg'
filtered_image.save('dogs/generated/' + new_name)
Invalid response from the model, trying again...
24.04.24 09:51:19: root: INFO : SOCKET tokens MESSAGE: {'token_usage': 7535}
Model: deepseek-llm:67b-chat, Enum: OLLAMA
24.04.24 09:51:47: root: INFO : SOCKET tokens MESSAGE: {'token_usage': 7241}
24.04.24 09:51:47: root: DEBUG : Response from the model: File: `main.py`:
python
import os
from PIL import Image, ImageFilter, ImageOps
def generate_cat(image_path):
"""Generate a cat image with random effects."""
# Load the original image
img = Image.open(f"{image_path}/original.jpg")
# Apply some random effects or filters (optional)
effect = os.urandom(1).hex() # Generate a random hex string for demo purposes
if 'blur' in effect:
img = img.filter(ImageFilter.BLUR)
elif 'invert' in effect:
img = ImageOps.invert(img)
# Save the image with the applied effects to a new location within the 'cats' folder
save_path = f"{image_path}/{effect}.jpg"
if not os.path.exists(save_path): # Ensure the image is saved only once
img.save(save_path)
def generate_dog(image_path):
"""Generate a dog image with random effects."""
# Load the original image
img = Image.open(f"{image_path}/original.jpg")
# Apply some random effects or filters (optional)
effect = os.urandom(1).hex() # Generate a random hex string for demo purposes
if 'blur' in effect:
img = img.filter(ImageFilter.BLUR)
elif 'invert' in effect:
img = ImageOps.invert(img)
# Save the image with the applied effects to a new location within the 'dogs' folder
save_path = f"{image_path}/{effect}.jpg"
if not os.path.exists(save_path): # Ensure the image is saved only once
img.save(save_path)
Invalid response from the model, trying again...
24.04.24 09:51:47: root: INFO : SOCKET tokens MESSAGE: {'token_usage': 8894}
Model: deepseek-llm:67b-chat, Enum: OLLAMA
24.04.24 09:52:13: root: INFO : SOCKET tokens MESSAGE: {'token_usage': 8496}
24.04.24 09:52:13: root: DEBUG : Response from the model: File: `main.py`:
python
import os
from PIL import Image
from random import randint
def generate_cat():
"""Generate a cat image with a random effect."""
# Load the original cat image
cat_image = Image.open(os.path.join('cats', 'original.jpg'))
# Apply a random filter or effect to the image
if randint(0, 1):
cat_image = cat_image.rotate(randint(-30, 30))
else:
cat_image = cat_image.resize((640, 480), Image.ANTIALIAS)
# Save the modified image in a new location within the 'cats' folder
cat_image.save(os.path.join('cats', f"cat_{randint(1,100)}.jpg"))
def generate_dog():
"""Generate a dog image with a random effect."""
# Load the original dog image
dog_image = Image.open(os.path.join('dogs', 'original.jpg'))
# Apply a random filter or effect to the image
if randint(0, 1):
dog_image = dog_image.rotate(randint(-30, 30))
else:
dog_image = dog_image.resize((640, 480), Image.ANTIALIAS)
# Save the modified image in a new location within the 'dogs' folder
dog_image.save(os.path.join('dogs', f"dog_{randint(1,100)}.jpg"))
Invalid response from the model, trying again...
24.04.24 09:52:13: root: INFO : SOCKET tokens MESSAGE: {'token_usage': 10201}
Model: deepseek-llm:67b-chat, Enum: OLLAMA
24.04.24 09:52:39: root: INFO : SOCKET tokens MESSAGE: {'token_usage': 9771}
24.04.24 09:52:39: root: DEBUG : Response from the model: File: `main.py`:
python
import tkinter as tk
from PIL import ImageTk, Image
class PetArtGenApp(tk.Frame):
def __init__(self, master=None):
super().__init__(master)
self.master = master
self.pack()
self.create_widgets()
def create_widgets(self):
# Generate Cat Button
self.generate_cat_button = tk.Button(self, text="Generate Cat", command=self.generate_cat)
self.generate_cat_button.pack(side="top")
# Generate Dog Button
self.generate_dog_button = tk.Button(self, text="Generate Dog", command=self.generate_dog)
self.generate_dog_button.pack(side="top")
def generate_cat(self):
# Load an image file
cat_image = ImageTk.PhotoImage(Image.open("./cats/kitten.jpg"))
# Create a Tkinter label that contains the image
cat_label = tk.Label(self, image=cat_image)
cat_label.pack()
def generate_dog(self):
# Load an image file
dog_image = ImageTk.PhotoImage(Image.open("./dogs/puppy.jpg"))
# Create a Tkinter label that contains the image
dog_label = tk.Label(self, image=dog_image)
dog_label.pack()
root = tk.Tk()
app = PetArtGenApp(master=root)
app.mainloop()
Invalid response from the model, trying again...
and the iterations are endless. I have run it for 30 mins now.
Thats it. Whatever this project is doing, its doing awesome and I would love to donate to the cause.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
So I guess this a general question but I would really appreciate an ELI5 response.
I will provide all the information anybody would need to debug it. I will make my best effort summarising it for other's benefit too.
I have devika installed and runnning with ollama and I have tested it a few times and have run a few taske.
The system I am running is ubuntu linux with 96GB VRAM, 128GB RAM and I have currently Deepseek 67B chat model running.
I asked a simple question:
Create me an app with modern ui where I click on the text "Generate Cat" and a cute cat art is displayed and when I click on "Generate dog" a cute dog art is displayed. source the images for the project. and create the code in python.
And I am really impressed with Devika's reasoning capabilities. I ally am!. Its beautiful how its managing this.
I got:
Now here are a list of issues I am facing and I have pulled up the latest version today so this is fresh off the bakery lol.
Invalid response from the model, trying again...
for all the code thats run in the terminal.So the model tries the code or the same approach over and over for the code and every time its the same response so the code it saves finally in the project is an extremely simple, dumbed down code wqhen I can see in terminal that the code iterations when it started to work were quite good and intelligent. How do I get Devika to run in my environment and iterate it properly over there?
2.1. Can Devika save the codes that it has iterated over in a text folder within the project?
take a look at this:
and the iterations are endless. I have run it for 30 mins now.
Thats it. Whatever this project is doing, its doing awesome and I would love to donate to the cause.
Beta Was this translation helpful? Give feedback.
All reactions