Dev giga chkhikvadze

backend/fastapi_app.py

@@ -1,13 +1,9 @@
-import asyncio
-import json
-
-from fastapi import FastAPI, HTTPException, Request
+from fastapi import FastAPI
 from fastapi.middleware.cors import CORSMiddleware
-from fastapi.responses import StreamingResponse
-
-from neurons.api import initialize, query_synapse_image, query_synapse_text
+from routes.chat import router as chat_router
 
 app = FastAPI()
+
 app.add_middleware(
     CORSMiddleware,
     allow_origins=["*"],
@@ -17,32 +13,6 @@
     max_age=3600,
 )
 
-_, _, subtensor, dendrite, metagraph = initialize()
-
-
-@app.post("/generate-image")
-async def get_image(prompt_dict: dict):
-    prompt = prompt_dict.get('prompt')
-    if not prompt:
-        raise HTTPException(status_code=400, detail="Prompt is required")
-
-    generated_images = asyncio.run(query_synapse_image(dendrite, metagraph, subtensor, prompt))
-    try:
-        image_url = generated_images[0].deserialize().get("url")
-    except Exception:
-        raise HTTPException(status_code=500, detail="Can't generate image. Please try again")
-    return image_url
-
-
-@app.post("/generate-text")
-async def get_text(request: Request):
-    try:
-        body = await request.json()
-        prompt = body.get('prompt')
-    except json.JSONDecodeError:
-        raise HTTPException(status_code=500, detail="Cant decode JSON")
-    if not prompt:
-        raise HTTPException(status_code=400, detail="Prompt is required")
 
-    return StreamingResponse(query_synapse_text(dendrite, metagraph, subtensor, prompt), media_type='text/event-stream')
 
+app.include_router(chat_router)

backend/neurons/api.py → backend/neurons/chat.py

@@ -3,7 +3,10 @@
 import traceback
 import asyncio
 from template.protocol import StreamPrompting, ImageResponse
+from neurons.conversation_history import ConversationHistory
 
+conversation_history = ConversationHistory()
+MAX_TOKENS = 4096
 
 def initialize():
     parser = argparse.ArgumentParser()
@@ -18,8 +21,6 @@ def initialize():
     return config, wallet, subtensor, dendrite, metagraph
 
 
-# axons = [36, 80, 255]
-
 async def query_synapse_image(dendrite, metagraph, subtensor, prompt):
     try:
         axon = metagraph.axons[87]  # 87, 91, 98
@@ -44,17 +45,31 @@ async def main():
         bt.logging.error(f"General exception at step: {e}\n{traceback.format_exc()}")
 
 
-async def query_synapse_text(dendrite, metagraph, subtensor, prompt):
+
+async def query_synapse_text(dendrite, metagraph, subtensor, conversation_id, prompt):
     try:
         axon = metagraph.axons[87]
-        syn = StreamPrompting(messages=[{"role": "user", "content": prompt}], engine="gpt-4-1106-preview", seed=1234)
+        # Add user prompt to the conversation history
+        conversation_history.add_user_message(conversation_id, prompt)
+
+        # Get the conversation history within the token limit
+        history_within_limit = conversation_history.get_history_within_limit(conversation_id, MAX_TOKENS)  # Set MAX_TOKENS appropriately
+
+        syn = StreamPrompting(messages=history_within_limit, engine="gpt-4-1106-preview", seed=1234)
 
         responses = await dendrite([axon], syn, deserialize=False, streaming=True)
+        merged_chunks = ""  # Initialize merged_chunks
         for resp in responses:
             async for chunk in resp:
                 if isinstance(chunk, list):
+                    # Merge each streaming chunk together in the background
+                    merged_chunks += chunk[0]
+
+                    # Yield the current chunk as a response
                     yield f"data: {chunk[0]}\n\n"
 
+        # After all chunks have been processed, add the merged chunks as a single AI message in the conversation history
+        conversation_history.add_ai_message(conversation_id, merged_chunks)
     except Exception as e:
         bt.logging.error(f"General exception at step: {e}\n{traceback.format_exc()}")
 

backend/neurons/config.py

@@ -2,7 +2,6 @@
 import argparse
 import os
 
-
 def check_config(cls, config: "bt.Config"):
     bt.axon.check_config(config)
     bt.logging.check_config(config)
@@ -18,7 +17,6 @@ def check_config(cls, config: "bt.Config"):
     if not os.path.exists(config.miner.full_path):
         os.makedirs(config.miner.full_path)
 
-
 def get_config() -> "bt.Config":
     parser = argparse.ArgumentParser()
     parser.add_argument(

backend/neurons/conversation_history.py

@@ -0,0 +1,78 @@
+# conversation_history.py
+import uuid
+from sumy.parsers.plaintext import PlaintextParser
+from sumy.nlp.tokenizers import Tokenizer
+from sumy.summarizers.lsa import LsaSummarizer
+
+
+temp_conversation_id= str(uuid.uuid4())
+class ConversationHistory:
+    def __init__(self):
+        self.history = {}
+
+    def get_history(self, conversation_id):
+        # Retrieve or initialize conversation history for the given conversation_id
+        return self.history.setdefault(conversation_id, [])
+
+    def get_history_within_limit(self, conversation_id, max_tokens, use_summaries=True, max_messages=None):
+        # Retrieve or initialize conversation history for the given conversation_id
+        history = self.history.setdefault(conversation_id, [])
+
+        if use_summaries:
+            # Calculate the total tokens in the conversation history
+            total_tokens = sum(len(message["content"].split()) for message in history)
+
+            # If it exceeds the token limit, calculate the number of messages needed
+            if total_tokens > max_tokens:
+                num_messages = 0
+                current_tokens = 0
+                for message in history:
+                    current_tokens += len(message["content"].split())
+                    num_messages += 1
+                    if current_tokens > max_tokens:
+                        break
+            else:
+                num_messages = len(history)
+
+            # Return the history within the calculated number of messages
+            return history[-num_messages:]
+        elif max_messages is not None:
+            # Get the last 'max_messages' messages from the history
+            return history[-max_messages:]
+        else:
+            # Return the entire history
+            return history
+
+    def get_summarized_history(self, conversation_id, max_tokens):
+        # Retrieve or initialize conversation history for the given conversation_id
+        history = self.history.setdefault(conversation_id, [])
+
+        # Combine all messages into a single document
+        document = " ".join(message["content"] for message in history)
+
+        # Summarize the document
+        summarizer = LsaSummarizer()
+        parser = PlaintextParser.from_string(document, Tokenizer("english"))
+        summary = summarizer(parser.document, 1)  # Get a single sentence summary
+
+        # Return the summarized history
+        return [{"role": "summarized", "content": str(sentence)} for sentence in summary]
+
+    def add_user_message(self, conversation_id, prompt):
+        # Add user prompt to the conversation history
+        history = self.get_history(conversation_id)
+        history.append({"role": "user", "content": prompt})
+
+    def add_ai_message(self, conversation_id, response):
+        # Add AI response to the conversation history
+        history = self.get_history(conversation_id)
+        history.append({"role": "assistant", "content": response})
+
+    def clear_history(self, conversation_id):
+        # Clear conversation history for the given conversation_id
+        if conversation_id in self.history:
+            del self.history[conversation_id]
+
+    def generate_conversation_id(self):
+        # Function to generate a unique conversation_id for each request
+        return temp_conversation_id

backend/neurons/miner.py

@@ -11,8 +11,7 @@
 from openai import OpenAI
 from openai import AsyncOpenAI
 import bittensor as bt
-from transformers import GPT2Tokenizer
-from typing import List, Dict, Tuple, Union, Callable, Awaitable
+from typing import Dict, Tuple
 from template.protocol import StreamPrompting, IsAlive, ImageResponse
 from config import get_config, check_config
 

backend/neurons/validator.py

@@ -1,24 +1,15 @@
 import os
-import re
-import json
-import math
 import time
-import torch
 import wandb
 import random
-import string
 import asyncio
 import template
-import requests
 import argparse
 import datetime
 import traceback
 import bittensor as bt
-import concurrent.futures
-from PIL import Image
 from io import BytesIO
 from openai import AsyncOpenAI
-from typing import List, Optional
 from template.protocol import StreamPrompting, IsAlive, ImageResponse
 
 AsyncOpenAI.api_key = os.environ.get('OPENAI_API_KEY')

backend/requirements.txt

@@ -1,128 +1,14 @@
 aiohttp==3.8.5
-aiosignal==1.3.1
-ansible==6.7.0
-ansible-core==2.13.13
-ansible-vault==2.1.0
-anyio==3.7.1
-appdirs==1.4.4
-async-timeout==4.0.3
-attrs==23.1.0
-backoff==2.2.1
-base58==2.1.1
 bittensor==6.3.0
-black==23.7.0
-certifi==2023.11.17
-cffi==1.16.0
-charset-normalizer==3.3.2
-click==8.1.7
-cryptography==41.0.3
-cytoolz==0.12.2
-ddt==1.6.0
-decorator==5.1.1
-distro==1.8.0
-docker-pycreds==0.4.0
-ecdsa==0.18.0
-eth-hash==0.5.2
-eth-keys==0.4.0
-eth-typing==3.5.2
-eth-utils==2.3.1
-exceptiongroup==1.2.0
 fastapi==0.99.1
-filelock==3.13.1
-frozenlist==1.4.0
-fsspec==2023.10.0
-fuzzywuzzy==0.18.0
-gitdb==4.0.11
-GitPython==3.1.40
-h11==0.14.0
-httpcore==1.0.2
-httpx==0.25.1
-huggingface-hub==0.19.4
-idna==3.4
-iniconfig==2.0.0
-Jinja2==3.1.2
-joblib==1.3.2
-Levenshtein==0.23.0
-loguru==0.7.0
-markdown-it-py==3.0.0
-MarkupSafe==2.1.3
-mdurl==0.1.2
-more-itertools==10.1.0
-mpmath==1.3.0
-msgpack==1.0.7
-msgpack-numpy-opentensor==0.5.0
-multidict==6.0.4
-munch==2.5.0
-mypy-extensions==1.0.0
-nest-asyncio==1.5.8
-netaddr==0.9.0
-networkx==3.2.1
-numpy==1.26.2
-nvidia-cublas-cu12==12.1.3.1
-nvidia-cuda-cupti-cu12==12.1.105
-nvidia-cuda-nvrtc-cu12==12.1.105
-nvidia-cuda-runtime-cu12==12.1.105
-nvidia-cudnn-cu12==8.9.2.26
-nvidia-cufft-cu12==11.0.2.54
-nvidia-curand-cu12==10.3.2.106
-nvidia-cusolver-cu12==11.4.5.107
-nvidia-cusparse-cu12==12.1.0.106
-nvidia-nccl-cu12==2.18.1
-nvidia-nvjitlink-cu12==12.3.101
-nvidia-nvtx-cu12==12.1.105
 openai>=1.3.2
-packaging==23.2
-password-strength==0.0.3.post2
-pathspec==0.11.2
-Pillow==10.1.0
-platformdirs==4.0.0
-pluggy==1.3.0
-protobuf==4.25.1
-psutil==5.9.6
 py==1.11.0
-py-bip39-bindings==0.1.11
-py-ed25519-zebra-bindings==1.0.1
-py-sr25519-bindings==0.2.0
-pycparser==2.21
-pycryptodome==3.19.0
 pydantic==1.10.13
-Pygments==2.17.2
-PyNaCl==1.5.0
-pytest==7.4.3
-pytest-asyncio==0.21.1
-python-Levenshtein==0.23.0
-PyYAML==6.0.1
-rapidfuzz==3.5.2
-regex==2023.10.3
 requests==2.31.0
-resolvelib==0.8.1
-retry==0.9.2
-rich==13.7.0
-safetensors==0.4.0
-scalecodec==1.2.0
-scikit-learn==1.3.2
-scipy==1.11.4
-sentry-sdk==1.36.0
-setproctitle==1.3.3
-six==1.16.0
-smmap==5.0.1
-sniffio==1.3.0
 starlette==0.27.0
-substrate-interface==1.5.0
-sympy==1.12
-termcolor==2.3.0
-threadpoolctl==3.2.0
-tokenizers==0.15.0
-tomli==2.0.1
-toolz==0.12.0
 torch==2.1.1
-tqdm==4.66.1
 transformers==4.35.2
-triton==2.1.0
-typing_extensions==4.8.0
-urllib3==2.1.0
 uvicorn==0.22.0
-wandb==0.16.0
-websocket-client==1.6.4
-xxhash==3.4.1
-yarl==1.9.3
+scikit-learn==1.3.2
+Pillow==10.1.0
+sumy==0.11.0