Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,20 +1,15 @@
|
|
1 |
import os
|
2 |
-
import json
|
3 |
-
import subprocess
|
4 |
-
from threading import Thread
|
5 |
import logging
|
6 |
from logging.handlers import RotatingFileHandler
|
7 |
|
8 |
import torch
|
9 |
-
import spaces
|
10 |
import gradio as gr
|
11 |
-
from transformers import
|
12 |
from langchain_huggingface import ChatHuggingFace
|
13 |
from langchain.prompts import PromptTemplate
|
14 |
from langchain.chains import LLMChain
|
15 |
|
16 |
-
|
17 |
-
|
18 |
log_file = '/tmp/app_debug.log'
|
19 |
logger = logging.getLogger(__name__)
|
20 |
logger.setLevel(logging.DEBUG)
|
@@ -47,14 +42,12 @@ template = """<|im_start|>system
|
|
47 |
"""
|
48 |
prompt = PromptTemplate(template=template, input_variables=["system_prompt", "history", "human_input"])
|
49 |
|
50 |
-
# Format the conversation history
|
51 |
def format_history(history):
|
52 |
formatted = ""
|
53 |
for human, ai in history:
|
54 |
formatted += f"<|im_start|>user\n{human}\n<|im_end|>\n<|im_start|>assistant\n{ai}\n<|im_end|>\n"
|
55 |
return formatted
|
56 |
|
57 |
-
@spaces.GPU()
|
58 |
def predict(message, history, system_prompt, temperature, max_new_tokens, top_k, repetition_penalty, top_p):
|
59 |
logger.debug(f"Received prediction request: message='{message}', system_prompt='{system_prompt}'")
|
60 |
|
@@ -88,7 +81,6 @@ chat_model = ChatHuggingFace(
|
|
88 |
model_kwargs={
|
89 |
"device_map": "auto",
|
90 |
"quantization_config": quantization_config,
|
91 |
-
"attn_implementation": "flash_attention_2",
|
92 |
},
|
93 |
tokenizer=tokenizer
|
94 |
)
|
@@ -105,11 +97,6 @@ gr.ChatInterface(
|
|
105 |
["How can I reverse a string in JavaScript?"],
|
106 |
["Create a C++ function to find the factorial of a number."],
|
107 |
["Write a Python list comprehension to generate a list of squares of numbers from 1 to 10."],
|
108 |
-
["How do I implement a binary search algorithm in C?"],
|
109 |
-
["Write a Ruby script to read a file and count the number of lines in it."],
|
110 |
-
["Create a Swift class to represent a bank account with deposit and withdrawal methods."],
|
111 |
-
["How do I find the maximum element in an array using Kotlin?"],
|
112 |
-
["Write a Rust program to generate the Fibonacci sequence up to the 10th number."]
|
113 |
],
|
114 |
additional_inputs=[
|
115 |
gr.Textbox("You are a code assistant.", label="System prompt"),
|
|
|
1 |
import os
|
|
|
|
|
|
|
2 |
import logging
|
3 |
from logging.handlers import RotatingFileHandler
|
4 |
|
5 |
import torch
|
|
|
6 |
import gradio as gr
|
7 |
+
from transformers import AutoTokenizer, BitsAndBytesConfig
|
8 |
from langchain_huggingface import ChatHuggingFace
|
9 |
from langchain.prompts import PromptTemplate
|
10 |
from langchain.chains import LLMChain
|
11 |
|
12 |
+
# Настройка логирования
|
|
|
13 |
log_file = '/tmp/app_debug.log'
|
14 |
logger = logging.getLogger(__name__)
|
15 |
logger.setLevel(logging.DEBUG)
|
|
|
42 |
"""
|
43 |
prompt = PromptTemplate(template=template, input_variables=["system_prompt", "history", "human_input"])
|
44 |
|
|
|
45 |
def format_history(history):
|
46 |
formatted = ""
|
47 |
for human, ai in history:
|
48 |
formatted += f"<|im_start|>user\n{human}\n<|im_end|>\n<|im_start|>assistant\n{ai}\n<|im_end|>\n"
|
49 |
return formatted
|
50 |
|
|
|
51 |
def predict(message, history, system_prompt, temperature, max_new_tokens, top_k, repetition_penalty, top_p):
|
52 |
logger.debug(f"Received prediction request: message='{message}', system_prompt='{system_prompt}'")
|
53 |
|
|
|
81 |
model_kwargs={
|
82 |
"device_map": "auto",
|
83 |
"quantization_config": quantization_config,
|
|
|
84 |
},
|
85 |
tokenizer=tokenizer
|
86 |
)
|
|
|
97 |
["How can I reverse a string in JavaScript?"],
|
98 |
["Create a C++ function to find the factorial of a number."],
|
99 |
["Write a Python list comprehension to generate a list of squares of numbers from 1 to 10."],
|
|
|
|
|
|
|
|
|
|
|
100 |
],
|
101 |
additional_inputs=[
|
102 |
gr.Textbox("You are a code assistant.", label="System prompt"),
|