Spaces:
Build error
Build error
File size: 11,800 Bytes
8dbf509 7b924b8 8dbf509 e042085 66854bf af4feb8 2ca300b eb88ab8 7b924b8 8dbf509 af4feb8 a1c598c eb88ab8 7b924b8 eb88ab8 af4feb8 e042085 eb88ab8 a12fc17 eb88ab8 0308f3c 70c5bc9 0308f3c 70c5bc9 0308f3c 70c5bc9 0308f3c 70c5bc9 0308f3c 70c5bc9 0308f3c 70c5bc9 0308f3c 70c5bc9 0308f3c 70c5bc9 0308f3c 70c5bc9 66854bf 70c5bc9 0308f3c 66854bf 70c5bc9 66854bf 70c5bc9 66854bf 70c5bc9 66854bf |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 |
# main.py
import spaces
import torch
import torch.nn.functional as F
from torch.nn import DataParallel
from torch import Tensor
from transformers import AutoTokenizer, AutoModel
import threading
import queue
import os
import json
import numpy as np
import gradio as gr
from huggingface_hub import InferenceClient
import openai
from openai import OpenAI
from globalvars import API_BASE, intention_prompt, tasks , system_message, model_name
from dotenv import load_dotenv
import re
from utils import load_env_variables
import chromadb
from chromadb import Documents, EmbeddingFunction, Embeddings
from chromadb.config import Settings
from chromadb import HttpClient
from langchain_community.document_loaders import UnstructuredFileLoader
from utils import load_env_variables , parse_and_route
os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb:30'
os.environ['CUDA_LAUNCH_BLOCKING'] = '1'
os.environ['CUDA_CACHE_DISABLE'] = '1'
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
hf_token, yi_token = load_env_variables()
def clear_cuda_cache():
torch.cuda.empty_cache()
client = OpenAI(
api_key=yi_token,
base_url=API_BASE
)
class EmbeddingGenerator:
def __init__(self, model_name: str, token: str, intention_client):
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
self.tokenizer = AutoTokenizer.from_pretrained(model_name, token=token, trust_remote_code=True)
self.model = AutoModel.from_pretrained(model_name, token=token, trust_remote_code=True).to(self.device)
self.intention_client = intention_client
def clear_cuda_cache(self):
torch.cuda.empty_cache()
@spaces.GPU
def compute_embeddings(self, input_text: str):
# Get the intention
intention_completion = self.intention_client.chat.completions.create(
model="yi-large",
messages=[
{"role": "system", "content": intention_prompt},
{"role": "user", "content": input_text}
]
)
intention_output = intention_completion.choices[0].message['content']
# Parse and route the intention
parsed_task = parse_and_route(intention_output)
selected_task = list(parsed_task.keys())[0]
# Construct the prompt
try:
task_description = tasks[selected_task]
except KeyError:
print(f"Selected task not found: {selected_task}")
return f"Error: Task '{selected_task}' not found. Please select a valid task."
query_prefix = f"Instruct: {task_description}\nQuery: "
queries = [input_text]
# Get the embeddings
with torch.no_grad():
inputs = self.tokenizer(queries, return_tensors='pt', padding=True, truncation=True, max_length=4096).to(self.device)
outputs = self.model(**inputs)
query_embeddings = outputs.last_hidden_state.mean(dim=1)
# Normalize embeddings
query_embeddings = F.normalize(query_embeddings, p=2, dim=1)
embeddings_list = query_embeddings.detach().cpu().numpy().tolist()
self.clear_cuda_cache()
return embeddings_list
class MyEmbeddingFunction(EmbeddingFunction):
def __init__(self, embedding_generator: EmbeddingGenerator):
self.embedding_generator = embedding_generator
def __call__(self, input: Documents) -> Embeddings:
embeddings = [self.embedding_generator.compute_embeddings(doc) for doc in input]
embeddings = [item for sublist in embeddings for item in sublist]
return embeddings
# main.py
import os
import uuid
import gradio as gr
import torch
import torch.nn.functional as F
from torch.nn import DataParallel
from torch import Tensor
from transformers import AutoTokenizer, AutoModel
from huggingface_hub import InferenceClient
from openai import OpenAI
from langchain_community.document_loaders import UnstructuredFileLoader
from chromadb import Documents, EmbeddingFunction, Embeddings
from chromadb.config import Settings
from chromadb import HttpClient
from langchain_chroma import Chroma
from utils import load_env_variables, parse_and_route
from globalvars import API_BASE, intention_prompt, tasks, system_message, model_name
from langchain_core.embeddings import Embeddings
from chromadb.api.types import EmbeddingFunction, Documents
os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb:30'
os.environ['CUDA_LAUNCH_BLOCKING'] = '1'
os.environ['CUDA_CACHE_DISABLE'] = '1'
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
### Utils
hf_token, yi_token = load_env_variables()
def clear_cuda_cache():
torch.cuda.empty_cache()
client = OpenAI(api_key=yi_token, base_url=API_BASE)
class EmbeddingGenerator:
def __init__(self, model_name: str, token: str, intention_client):
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
self.tokenizer = AutoTokenizer.from_pretrained(model_name, token=token, trust_remote_code=True)
self.model = AutoModel.from_pretrained(model_name, token=token, trust_remote_code=True).to(self.device)
self.intention_client = intention_client
def clear_cuda_cache(self):
torch.cuda.empty_cache()
@spaces.GPU
def compute_embeddings(self, input_text: str):
# Get the intention
intention_completion = self.intention_client.chat.completions.create(
model="yi-large",
messages=[
{"role": "system", "content": intention_prompt},
{"role": "user", "content": input_text}
]
)
intention_output = intention_completion.choices[0].message['content']
# Parse and route the intention
parsed_task = parse_and_route(intention_output)
selected_task = list(parsed_task.keys())[0]
# Construct the prompt
try:
task_description = tasks[selected_task]
except KeyError:
print(f"Selected task not found: {selected_task}")
return f"Error: Task '{selected_task}' not found. Please select a valid task."
query_prefix = f"Instruct: {task_description}\nQuery: "
queries = [input_text]
# Get the embeddings
with torch.no_grad():
inputs = self.tokenizer(queries, return_tensors='pt', padding=True, truncation=True, max_length=4096).to(self.device)
outputs = self.model(**inputs)
query_embeddings = outputs.last_hidden_state.mean(dim=1)
# Normalize embeddings
query_embeddings = F.normalize(query_embeddings, p=2, dim=1)
embeddings_list = query_embeddings.detach().cpu().numpy().tolist()
self.clear_cuda_cache()
return embeddings_list
class ChromaEmbeddingsAdapter(Embeddings):
def __init__(self, ef: EmbeddingFunction):
self.ef = ef
def embed_documents(self, texts):
return self.ef(texts)
def embed_query(self, query):
return self.ef([query])[0]
class LangChainEmbeddingAdapter(EmbeddingFunction[Documents]):
def __init__(self, ef: Embeddings):
self.ef = ef
def __call__(self, input: Documents) -> Embeddings:
return self.ef.embed_documents(input)
class MyEmbeddingFunction(EmbeddingFunction):
def __init__(self, embedding_generator: EmbeddingGenerator):
self.embedding_generator = embedding_generator
def __call__(self, input: Documents) -> Embeddings:
embeddings = [self.embedding_generator.compute_embeddings(doc) for doc in input]
embeddings = [item for sublist in embeddings for item in sublist]
return embeddings
def load_documents(file_path: str, mode: str = "elements"):
loader = UnstructuredFileLoader(file_path, mode=mode)
docs = loader.load()
return [doc.page_content for doc in docs]
def initialize_chroma(collection_name: str, embedding_function: MyEmbeddingFunction):
client = Chroma.from_documents([], ChromaEmbeddingsAdapter(embedding_function)) # Initialize with no documents
return client
def add_documents_to_chroma(client, documents: list, embedding_function: MyEmbeddingFunction):
for doc in documents:
client.add(ids=[str(uuid.uuid1())], documents=[doc], embeddings=embedding_function([doc]))
def query_chroma(client, query_text: str):
result_docs = client.similarity_search(query_text)
return result_docs
# Initialize clients
intention_client = OpenAI(api_key=yi_token, base_url=API_BASE)
embedding_generator = EmbeddingGenerator(model_name=model_name, token=hf_token, intention_client=intention_client)
embedding_function = MyEmbeddingFunction(embedding_generator=embedding_generator)
chroma_client = initialize_chroma(collection_name="Tonic-instruct", embedding_function=embedding_function)
def respond(
message,
history: list[tuple[str, str]],
system_message,
max_tokens,
temperature,
top_p,
):
retrieved_text = query_documents(message)
messages = [{"role": "system", "content": system_message}]
for val in history:
if val[0]:
messages.append({"role": "user", "content": val[0]})
if val[1]:
messages.append({"role": "assistant", "content": val[1]})
messages.append({"role": "user", "content": f"{retrieved_text}\n\n{message}"})
response = ""
for message in intention_client.chat_completion(
messages,
max_tokens=max_tokens,
stream=True,
temperature=temperature,
top_p=top_p,
):
token = message.choices[0].delta.content
response += token
yield response
def upload_documents(files):
for file in files:
loader = UnstructuredFileLoader(file.name)
documents = loader.load()
add_documents_to_chroma(chroma_client, documents, embedding_function)
return "Documents uploaded and processed successfully!"
def query_documents(query):
results = query_chroma(chroma_client, query)
return "\n\n".join([result.page_content for result in results])
with gr.Blocks() as demo:
with gr.Tab("Upload Documents"):
with gr.Row():
document_upload = gr.File(file_count="multiple", file_types=["document"])
upload_button = gr.Button("Upload and Process")
upload_button.click(upload_documents, inputs=document_upload, outputs=gr.Text())
with gr.Tab("Ask Questions"):
with gr.Row():
chat_interface = gr.ChatInterface(
respond,
additional_inputs=[
gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
],
)
query_input = gr.Textbox(label="Query")
query_button = gr.Button("Query")
query_output = gr.Textbox()
query_button.click(query_documents, inputs=query_input, outputs=query_output)
if __name__ == "__main__":
demo.launch() |