sync

config/config.yaml

@@ -3,7 +3,7 @@ gradio:
   port: 7860
 openai:
   api_url: "https://api.openai.com/v1/chat/completions"
-  content_token: 3200 # tokens per content_main (e.g. transcript). If exceed it will be splitted and iterated
+  content_token: 15000 # tokens per content_main (e.g. transcript). If exceed it will be splitted and iterated
   timeout_sec: 25
   max_retry: 2
   api_key: ""

digester/chatgpt_service.py

@@ -213,7 +213,7 @@ class ChatGPTService:
         return prompts
 
     @staticmethod
-    def single_call_chatgpt_with_handling(source_md, prompt_str: str, prompt_show_user: str, chatbot, api_key, gpt_model="gpt-3.5-turbo", history=[]):
+    def single_call_chatgpt_with_handling(source_md, prompt_str: str, prompt_show_user: str, chatbot, api_key, gpt_model, history=[]):
         """
         Handling
         - token exceeding -> split input
@@ -272,7 +272,7 @@ class ChatGPTService:
         return gpt_response
 
     @staticmethod
-    def single_rest_call_chatgpt(api_key, prompt_str: str, gpt_model="gpt-3.5-turbo", history=[], observe_window=None):
+    def single_rest_call_chatgpt(api_key, prompt_str: str, gpt_model, history=[], observe_window=None):
         """
         Single call chatgpt only. No handling on multiple call (it should be in upper caller multi_call_chatgpt_with_handling())
         - Support stream=True

digester/gradio_method_service.py

@@ -7,7 +7,7 @@ from digester.chatgpt_service import LLMService, ChatGPTService
 from digester.util import Prompt, provide_text_with_css, GradioInputs
 
 WAITING_FOR_TARGET_INPUT = "Waiting for target source input"
-RESPONSE_SUFFIX = "⚡Powered by DigestEverythingGPT (in Github). Generated by AI"
+RESPONSE_SUFFIX = "⚡by DigestEverythingGPT"
 
 
 class GradioMethodService:
@@ -378,7 +378,7 @@ Instructions: (step by step instructions)
 
 
 if __name__ == '__main__':
-    GPT_MODEL = "gpt-3.5-turbo"
+    GPT_MODEL = "gpt-3.5-turbo-16k"
     API_KEY = ""
     input_1 = """Give me 2 ideas for the summer"""
     # input_1 = """Explain more on the first idea"""

digester/gradio_ui_service.py

@@ -8,7 +8,7 @@ title_html = """
 <p align=\"center\">
 DigestEverythingGPT leverages ChatGPT/LLMs to help users quickly understand essential information from various forms of content, such as podcasts, YouTube videos, and PDF documents.<br>
 The prompt engineering is chained and tuned so that is result is of high quality and fast. It is not a simple single query and response tool.<br>
-Version 20230614 (
+Version 20230614_2 (
 <a href="https://github.com/michaelthwan/digest-everything-gpt"><i class="fa fa-github"></i> Github</a>
 ) (
 <a href="https://huggingface.co/spaces/michaelthwan/digest-everything-gpt"> HFSpace</a>
@@ -71,8 +71,8 @@ class GradioUIService:
                     with gr.Accordion("Options", open=True):
                         with gr.Row():
                             gpt_model_textbox = gr.Dropdown(
-                                ["gpt-3.5-turbo", "gpt-4"],
-                                value="gpt-3.5-turbo", label="GPT model", info="gpt-3.5 is cheaper.\nBut if you found that the result is not good, try gpt-4 \nYour API key must support gpt-4"
+                                ["gpt-3.5-turbo-16k", "gpt-4"],
+                                value="gpt-3.5-turbo-16k", label="GPT model", info="gpt-3.5 is cheaper.\nBut if you found that the result is not good, try gpt-4 \nYour API key must support gpt-4"
                             )
                         with gr.Row():
                             language_textbox = gr.Dropdown(

digester/test_chatgpt.py

@@ -102,5 +102,5 @@ Example format:
 8:00 - This is the second part
 9:22 - This is the third part
     """
-    GPT_MODEL = "gpt-3.5-turbo"
+    GPT_MODEL = "gpt-3.5-turbo-16k"
     ChatGPTService.single_rest_call_chatgpt(api_key, prompt_str, GPT_MODEL)

digester/test_youtube_chain.py

@@ -46,7 +46,7 @@ class VideoExample:
 
 
 class YoutubeTestChain:
-    def __init__(self, api_key: str, gpt_model="gpt-3.5-turbo"):
+    def __init__(self, api_key: str, gpt_model):
         self.api_key = api_key
         self.gpt_model = gpt_model
 

digester/util.py

@@ -4,7 +4,7 @@ from pathlib import Path
 import tiktoken
 import yaml
 
-tokenizer = tiktoken.encoding_for_model("gpt-3.5-turbo")
+tokenizer = tiktoken.encoding_for_model("gpt-3.5-turbo-16k")
 
 
 class GradioInputs:

requirements.txt

@@ -4,4 +4,4 @@ tiktoken>=0.3.3
 openai
 Markdown
 latex2mathml
-everything2text4prompt==0.0.11
+everything2text4prompt==0.0.14