Closed KalvinThien closed 5 months ago
MertKalkanci
commented
5 months ago Hey there, I think the problem you had happens because you use the 1.0.0 release, I've fixed some problems in newer commits but not made a release.
I made a new release 1.1.0, and updated readme.md.
I hope you won't have the same errors in the new release. I am not closing this issue to incase you face another problem. If you don't face any problem in new release please close the issue. Have a great day.
KalvinThien
commented
5 months ago
C:\Users\KalvinThien\AppData\Local\Programs\Python\Python311\Lib\site-packages\requests\__init__.py:102: RequestsDependencyWarning: urllib3 (1.26.18) or chardet (5.2.0)/charset_normalizer (2.0.12) doesn't match a supported version!
warnings.warn("urllib3 ({}) or chardet ({})/charset_normalizer ({}) doesn't match a supported "
Traceback (most recent call last):
File "d:\Ai_AUto_crop\Highlights-Maker-main\Highlights-Maker-main\main.py", line 12, in <module>
file_path = gr.FileExplorer(label="Video",file_count="single", glob="*/*.mp4")
^^^^^^^^^^^^^^^
AttributeError: module 'gradio' has no attribute 'FileExplorer'```
I'm always in trouble exactly, haha. I asked ChatGPT to fix the error for me.
however, I would like to ask more, in addition to using OpenAI, can Gemini be used instead of it?
Since the 3.5 model is also not really as good as the Gemini-Pro-001actually I'm not sure how to fix gradio errors, try to update packages in the virual environment, it's possible to add gemini support by changing ai.py and main.py to add selections to UI.
Could you try to update packages ? Also how did you fix it in the other issue you made ? Maybe same thing can be a solution to this problem again ?
KalvinThien
commented
5 months ago main.py :
import gradio as gr
from highlight import highlight
from youtube import download
languages = ["EN","VI"]
download_status = 0
generation_status = 0
with gr.Blocks() as interface:
with gr.Tab(label="Highlight"):
file_path = gr.File(label="Video", file_types=[".mp4"], file_count="single")
temperature = gr.Slider(minimum=0, maximum=2,value=0.35, step=0.05, label="AI Temperature")
length = gr.Slider(minimum=18, maximum=60, step=3, label="Dialogue Length", value=30, info="How many sentences AI will process at once (higher = faster generation & less chance of getting rate limited by openai)")
language = gr.Dropdown(languages, label="Language")
keywords = gr.Textbox(label="Keywords", placeholder="viral, funny, highlights", info="write diffrent keywords comma separated")
ai = gr.Dropdown(["OPENAI","LOCAL GGUF"], label="AI to interpret the video script")
ai_path = gr.File(label="Local AI Model", file_types=[".gguf"], file_count="single")
generate_button = gr.Button(value="Generate")
is_generated = gr.Textbox(label="Is Generated?", interactive=False)
with gr.Tab(label="Download"):
link = gr.Textbox(label="Link")
download_button = gr.Button(value="Download")
out_video = gr.Video(label="Video")
download_button.click(fn=download, inputs=link, outputs=out_video)
generate_button.click(fn=highlight, inputs=[file_path, temperature, length, language, keywords, ai, ai_path], outputs=is_generated)
is_shared = len(sys.argv) > 1 and sys.argv[1] == "--share"
interface.launch(share=is_shared)```
. but im get a new trouble
The GUI has said complete, but there are no files in the output directory
this is ai.py :
```import whisper
import openai
from openai import OpenAI
from llama_cpp import Llama
# Load the Whisper model
model_audio = whisper.load_model("base")
# Define the llm_manager class
class llm_manager():
def __init__(self, llm="OPENAI", path=""):
if llm == "OPENAI":
# Set the OpenAI API key (use your own key)
api_key = "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"
self.client = OpenAI(api_key=api_key)
self.type = "OPENAI"
elif llm == "LOCAL GGUF":
# Load the local GGUF model
self.llm = Llama(path)
self.type = "LOCAL GGUF"
def generate(self, system, chat, temperature):
if self.type == "OPENAI":
return self.gpt(system, chat, temperature)
elif self.type == "LOCAL GGUF":
return self.llama(system, chat)
def gpt(self, system, chat, temperature):
# Create the GPT chat completions using the OpenAI API
response = self.client.chat.completions.create(
model="gpt-3.5-turbo-1106",
temperature=temperature,
messages=[
{"role": "system", "content": system},
{"role": "user", "content": chat},
]
)
return response.choices[0].message.content
def llama(self, system, chat):
# Generate a response using the local GGUF llm
return self.llm(f"System: \n{system}\nUser: \n{chat}")["choices"][0]["text"]
# Define the transcribe function
def trasncribe(video_path):
return model_audio.transcribe(video_path)```
In another repositorys issue I've found this:
This is usually caused by gradio not being installed correctly in the python environment.
Can you try
pip uninstall gradioand then rerunpip install gradioand let us know if you are still seeing this?
Also which python version you are using ?
KalvinThien
commented
5 months ago I have fixed it, now I edit it, because mainly the trend is Video Shorts (1080px * 1920 px) also known as 9:16, so I want to add 1 option when ticking to crop the video and the right part needs to be highlighted?
MertKalkanci
commented
5 months ago yes you can add it, create a pull request I will accept it.
KalvinThien
commented
5 months ago Sorry if you misunderstood, I mean it's just an idea. Here I have a code that will crop the video and crop the face to crop
Can you take a look at it and integrate it into the app? https://github.com/NisaarAgharia/AI-Shorts-Creator/blob/main/autocropper.py
MertKalkanci
commented
5 months ago Ok I will have a look onto it.
I ran main.py and had the Running on local URL interface: http://127.0.0.1:7860
But when selecting a video and filling in keywords, then the error