Update app.py

app.py

@@ -3,143 +3,11 @@ import os
 import json 
 import requests
 
-
-#Chatbot2
-from transformers import BlenderbotTokenizer, BlenderbotForConditionalGeneration
-import torch
-from datasets import load_dataset
-# PersistDataset -----
-import os
-import csv
-from gradio import inputs, outputs
-import huggingface_hub
-#from huggingface_hub import Repository, hub_download, upload_file
-from datetime import datetime
-import fastapi
-from typing import List, Dict
-import httpx
-import pandas as pd
-import datasets as ds
-
-#Chatbot2 constants
-title = """<h1 align="center">💬ChatGPT ChatBack🧠💾</h1>"""
-#description = """Chatbot With persistent memory dataset allowing multiagent system AI to access a shared dataset as memory pool with stored interactions.  """
-UseMemory=True
-
-#ChatGPT info
+#Streaming endpoint 
 API_URL = "https://api.openai.com/v1/chat/completions" #os.getenv("API_URL") + "/generate_stream"
 OPENAI_API_KEY= os.environ["HF_TOKEN"] # Add a token to this space .  Then copy it to the repository secret in this spaces settings panel.  os.environ reads from there.
 # Keys for Open AI ChatGPT API usage are created from here: https://platform.openai.com/account/api-keys
-description = """
-
-Chatbot With persistent memory dataset allowing multiagent system AI to access a shared dataset as memory pool with stored interactions. 
-
-
-## ChatGPT Datasets 📚
-- WebText
-- Common Crawl
-- BooksCorpus
-- English Wikipedia
-- Toronto Books Corpus
-- OpenWebText
-
-## ChatGPT Datasets - Details 📚
-- **WebText:** A dataset of web pages crawled from domains on the Alexa top 5,000 list. This dataset was used to pretrain GPT-2.
-  - [WebText: A Large-Scale Unsupervised Text Corpus by Radford et al.](https://paperswithcode.com/dataset/webtext)
-- **Common Crawl:** A dataset of web pages from a variety of domains, which is updated regularly. This dataset was used to pretrain GPT-3.
-  - [Language Models are Few-Shot Learners](https://paperswithcode.com/dataset/common-crawl) by Brown et al.
-- **BooksCorpus:** A dataset of over 11,000 books from a variety of genres.
-  - [Scalable Methods for 8 Billion Token Language Modeling](https://paperswithcode.com/dataset/bookcorpus) by Zhu et al.
-- **English Wikipedia:** A dump of the English-language Wikipedia as of 2018, with articles from 2001-2017.
-  - [Improving Language Understanding by Generative Pre-Training](https://huggingface.co/spaces/awacke1/WikipediaUltimateAISearch?logs=build) Space for Wikipedia Search
-- **Toronto Books Corpus:** A dataset of over 7,000 books from a variety of genres, collected by the University of Toronto.
-  - [Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond](https://paperswithcode.com/dataset/bookcorpus) by Schwenk and Douze.
-- **OpenWebText:** A dataset of web pages that were filtered to remove content that was likely to be low-quality or spammy. This dataset was used to pretrain GPT-3.
-  - [Language Models are Few-Shot Learners](https://paperswithcode.com/dataset/openwebtext) by Brown et al.
-"""
-
 
-#Chatbot2 Save Results
-def SaveResult(text, outputfileName):
-    basedir = os.path.dirname(__file__)
-    savePath = outputfileName
-    print("Saving: " + text + " to " + savePath)
-    from os.path import exists
-    file_exists = exists(savePath)
-    if file_exists:
-        with open(outputfileName, "a") as f: #append
-            f.write(str(text.replace("\n","  ")))
-            f.write('\n')
-    else:
-        with open(outputfileName, "w") as f: #write
-            f.write(str("time, message, text\n")) # one time only to get column headers for CSV file
-            f.write(str(text.replace("\n","  ")))
-            f.write('\n')
-    return
-
-#Chatbot2 Store Message
-def store_message(name: str, message: str, outputfileName: str):
-    basedir = os.path.dirname(__file__)
-    savePath = outputfileName
-    
-    # if file doesnt exist, create it with labels
-    from os.path import exists
-    file_exists = exists(savePath)
-    
-    if (file_exists==False):
-        with open(savePath, "w") as f: #write
-            f.write(str("time, message, text\n")) # one time only to get column headers for CSV file
-            if name and message:
-                writer = csv.DictWriter(f, fieldnames=["time", "message", "name"])
-                writer.writerow(
-                    {"time": str(datetime.now()), "message": message.strip(), "name": name.strip()  }
-                )
-        df = pd.read_csv(savePath)
-        df = df.sort_values(df.columns[0],ascending=False)
-    else:
-        if name and message:
-            with open(savePath, "a") as csvfile:
-                writer = csv.DictWriter(csvfile, fieldnames=[ "time", "message", "name", ])
-                writer.writerow(
-                    {"time": str(datetime.now()), "message": message.strip(), "name": name.strip()  }
-                )
-        df = pd.read_csv(savePath)
-        df = df.sort_values(df.columns[0],ascending=False)
-    return df
-
-#Chatbot2 get base directory of saves
-def get_base(filename): 
-        basedir = os.path.dirname(__file__)
-        print(basedir)
-        loadPath = basedir + filename 
-        print(loadPath)
-        return loadPath
-
-#Chatbot2 - History
-def chat(message, history):
-    history = history or []
-    if history: 
-        history_useful = ['</s> <s>'.join([str(a[0])+'</s> <s>'+str(a[1]) for a in history])]
-    else:
-        history_useful = []
-    history_useful = add_note_to_history(message, history_useful)
-    inputs = tokenizer(history_useful, return_tensors="pt")
-    inputs, history_useful, history = take_last_tokens(inputs, history_useful, history)
-    reply_ids = model.generate(**inputs)
-    response = tokenizer.batch_decode(reply_ids, skip_special_tokens=True)[0]
-    history_useful = add_note_to_history(response, history_useful)
-    list_history = history_useful[0].split('</s> <s>')
-    history.append((list_history[-2], list_history[-1]))  
-    df=pd.DataFrame()
-    if UseMemory: 
-        outputfileName = 'ChatbotMemory3.csv' # Test first time file create
-        df = store_message(message, response, outputfileName) # Save to dataset
-        basedir = get_base(outputfileName)
-    return history, df, basedir
-
-
-
-#ChatGPT predict
 def predict(inputs, top_p, temperature, chat_counter, chatbot=[], history=[]):  #repetition_penalty, top_k
 
     # 1. Set up a payload
@@ -203,17 +71,18 @@ def predict(inputs, top_p, temperature, chat_counter, chatbot=[], history=[]):
     # TODO - make this parse out markdown so we can have similar interface
     counter=0
     for chunk in response.iter_lines():
-
+        #Skipping first chunk
         if counter == 0:
           counter+=1
           continue
-
+        #counter+=1
+        # check whether each line is non-empty
         if chunk.decode() :
-          
-            chunk = chunk.decode()
-
-            if len(chunk) > 12 and "content" in json.loads(chunk[6:])['choices'][0]['delta']:
-
+          chunk = chunk.decode()
+          # decode each line as response data is in bytes
+          if len(chunk) > 12 and "content" in json.loads(chunk[6:])['choices'][0]['delta']:
+              #if len(json.loads(chunk.decode()[6:])['choices'][0]["delta"]) == 0:
+              #  break
               partial_words = partial_words + json.loads(chunk[6:])['choices'][0]["delta"]["content"]
               if token_counter == 0:
                 history.append(" " + partial_words)
@@ -221,61 +90,50 @@ def predict(inputs, top_p, temperature, chat_counter, chatbot=[], history=[]):
                 history[-1] = partial_words
               chat = [(history[i], history[i + 1]) for i in range(0, len(history) - 1, 2) ]  # convert to tuples of list
               token_counter+=1
-
-              
-              
-              df=pd.DataFrame()
-              if UseMemory: 
-                 outputfileName = 'ChatGPT-RLHF-Memory.csv' # Test first time file create
-                 df = store_message(chat, history, outputfileName) # Save to dataset
-                 basedir = get_base(outputfileName)
-              #return history, df, basedir
-
-              
-              
               yield chat, history, chat_counter  # resembles {chatbot: chat, state: history}  
+                   
 
-def take_last_tokens(inputs, note_history, history):
-    if inputs['input_ids'].shape[1] > 128:
-        inputs['input_ids'] = torch.tensor([inputs['input_ids'][0][-128:].tolist()])
-        inputs['attention_mask'] = torch.tensor([inputs['attention_mask'][0][-128:].tolist()])
-        note_history = ['</s> <s>'.join(note_history[0].split('</s> <s>')[2:])]
-        history = history[1:]
-    return inputs, note_history, history
-    
-def add_note_to_history(note, note_history):# good example of non async since we wait around til we know it went okay.
-    note_history.append(note)
-    note_history = '</s> <s>'.join(note_history)
-    return [note_history]  
-        
-# ChatGPT clear
 def reset_textbox():
     return gr.update(value='')
 
+title = """<h1 align="center">Memory Chat Story Generator ChatGPT</h1>"""
+description = """
+## ChatGPT Datasets 📚
+- WebText
+- Common Crawl
+- BooksCorpus
+- English Wikipedia
+- Toronto Books Corpus
+- OpenWebText
+## ChatGPT Datasets - Details 📚
+- **WebText:** A dataset of web pages crawled from domains on the Alexa top 5,000 list. This dataset was used to pretrain GPT-2.
+  - [WebText: A Large-Scale Unsupervised Text Corpus by Radford et al.](https://paperswithcode.com/dataset/webtext)
+- **Common Crawl:** A dataset of web pages from a variety of domains, which is updated regularly. This dataset was used to pretrain GPT-3.
+  - [Language Models are Few-Shot Learners](https://paperswithcode.com/dataset/common-crawl) by Brown et al.
+- **BooksCorpus:** A dataset of over 11,000 books from a variety of genres.
+  - [Scalable Methods for 8 Billion Token Language Modeling](https://paperswithcode.com/dataset/bookcorpus) by Zhu et al.
+- **English Wikipedia:** A dump of the English-language Wikipedia as of 2018, with articles from 2001-2017.
+  - [Improving Language Understanding by Generative Pre-Training](https://huggingface.co/spaces/awacke1/WikipediaUltimateAISearch?logs=build) Space for Wikipedia Search
+- **Toronto Books Corpus:** A dataset of over 7,000 books from a variety of genres, collected by the University of Toronto.
+  - [Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond](https://paperswithcode.com/dataset/bookcorpus) by Schwenk and Douze.
+- **OpenWebText:** A dataset of web pages that were filtered to remove content that was likely to be low-quality or spammy. This dataset was used to pretrain GPT-3.
+  - [Language Models are Few-Shot Learners](https://paperswithcode.com/dataset/openwebtext) by Brown et al.
+  
+  """
+
 # 6. Use Gradio to pull it all together
-with gr.Blocks(css = """#col_container {width: 1000px; margin-left: auto; margin-right: auto;}   #chatbot {height: 520px; overflow: auto;}""") as demo:
-                   
-    gr.HTML(title)
+with gr.Blocks(css = """#col_container {width: 1000px; margin-left: auto; margin-right: auto;}
+                #chatbot {height: 520px; overflow: auto;}""") as demo:
 
-    # Chat bot memory - dataframe
-    gr.Markdown("<h1><center>🍰Gradio chatbot backed by dataframe CSV memory🎨</center></h1>")
-    with gr.Row():
-       t1 = gr.Textbox(lines=1, default="", label="Chat Text:")
-       b1 = gr.Button("🍰 Respond and Retrieve Messages")
-    with gr.Row(): # inputs and buttons
-       s1 = gr.State([])
-       df1 = gr.Dataframe(wrap=True, max_rows=1000, overflow_row_behaviour= "paginate")
-    with gr.Row(): # inputs and buttons
-       file = gr.File(label="File")
-       s2 = gr.Markdown()
-    #b1.click(fn=chat, inputs=[t1, s1], outputs=[s1, df1, file]) 
+                    
+    gr.HTML(title)
 
                     
     with gr.Column(elem_id = "col_container"):
-        chatbot = gr.Chatbot(elem_id='chatbot')
-        inputs = gr.Textbox(placeholder= "There is only one real true reward in life and this is existence versus nonexistence.  Everything else is a corollary.", label= "Type an input and press Enter") #t
-        state = gr.State([]) 
-        gpt = gr.Button()
+        chatbot = gr.Chatbot(elem_id='chatbot') #c
+        inputs = gr.Textbox(placeholder= "Hi there!", label= "Type an input and press Enter") #t
+        state = gr.State([]) #s
+        b1 = gr.Button()
     
         with gr.Accordion("Parameters", open=False):
             top_p = gr.Slider( minimum=-0, maximum=1.0, value=1.0, step=0.05, interactive=True, label="Top-p (nucleus sampling)",)
@@ -283,13 +141,9 @@ with gr.Blocks(css = """#col_container {width: 1000px; margin-left: auto; margin
             chat_counter = gr.Number(value=0, visible=False, precision=0)
 
     inputs.submit( predict, [inputs, top_p, temperature,chat_counter, chatbot, state], [chatbot, state, chat_counter],)
-                    
-    gpt.click(predict, [inputs, top_p, temperature, chat_counter, chatbot, state], [chatbot, state, chat_counter],)
-    gpt.click(reset_textbox, [], [inputs])                  
+    b1.click( predict, [inputs, top_p, temperature, chat_counter, chatbot, state], [chatbot, state, chat_counter],)
+    b1.click(reset_textbox, [], [inputs])
     inputs.submit(reset_textbox, [], [inputs])
-
-    # Show ChatGPT Datasets information
+                    
     gr.Markdown(description)
-
-    # Kickoff
     demo.queue().launch(debug=True)