Spaces:
Sleeping
Sleeping
Upload app.py with huggingface_hub
Browse files
app.py
CHANGED
@@ -1,14 +1,9 @@
|
|
1 |
-
#
|
2 |
-
|
3 |
## required lib, required "pip install"
|
4 |
-
# import transformers
|
5 |
-
# import accelerate
|
6 |
-
import openai
|
7 |
import torch
|
8 |
import cryptography
|
9 |
import cryptography.fernet
|
10 |
-
|
11 |
-
import gradio
|
12 |
import huggingface_hub
|
13 |
import huggingface_hub.hf_api
|
14 |
## standard libs, no need to install
|
@@ -22,14 +17,37 @@ import sys
|
|
22 |
import psutil
|
23 |
import threading
|
24 |
import socket
|
25 |
-
|
26 |
-
|
27 |
import matplotlib
|
28 |
-
|
29 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
30 |
# initialize the object
|
31 |
def __init__(self, name="Pluto",*args, **kwargs):
|
32 |
-
super(
|
33 |
self.author = "Duc Haba"
|
34 |
self.name = name
|
35 |
self._ph()
|
@@ -39,38 +57,38 @@ class HFace_Pluto(object):
|
|
39 |
self._ph()
|
40 |
#
|
41 |
# define class var for stable division
|
42 |
-
|
43 |
-
self._steps = [3,8,21,55,89,144]
|
44 |
-
self._guidances = [1.1,3.0,5.0,8.0,13.0,21.0]
|
45 |
-
self._xkeyfile = '.xoxo'
|
46 |
-
self._models = []
|
47 |
-
self._seed = 667 # sum of walnut in ascii (or Angle 667)
|
48 |
-
self._width = 512
|
49 |
-
self._height = 512
|
50 |
-
self._step = 50
|
51 |
-
self._guidances = 7.5
|
52 |
-
#self._generator = torch.Generator(device='cuda')
|
53 |
-
self.pipes = []
|
54 |
-
self.prompts = []
|
55 |
-
self.images = []
|
56 |
-
self.seeds = []
|
57 |
self.fname_id = 0
|
58 |
self.dname_img = "img_colab/"
|
59 |
-
self.
|
60 |
-
self.
|
61 |
-
self.
|
62 |
-
|
63 |
-
self.
|
64 |
-
self.
|
65 |
-
self.
|
66 |
-
self.
|
67 |
-
self.
|
68 |
-
self.
|
69 |
-
self.
|
|
|
70 |
return
|
71 |
#
|
72 |
# pretty print output name-value line
|
73 |
def _pp(self, a, b,is_print=True):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
74 |
# print("%34s : %s" % (str(a), str(b)))
|
75 |
x = f'{"%34s" % str(a)} : {str(b)}'
|
76 |
y = None
|
@@ -82,6 +100,16 @@ class HFace_Pluto(object):
|
|
82 |
#
|
83 |
# pretty print the header or footer lines
|
84 |
def _ph(self,is_print=True):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
85 |
x = f'{"-"*34} : {"-"*34}'
|
86 |
y = None
|
87 |
if (is_print):
|
@@ -95,23 +123,55 @@ class HFace_Pluto(object):
|
|
95 |
hf_names,
|
96 |
hf_space="duchaba/monty",
|
97 |
local_dir="/content/"):
|
98 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
99 |
try:
|
100 |
for f in hf_names:
|
101 |
lo = local_dir + f
|
102 |
-
huggingface_hub.hf_hub_download(repo_id=hf_space,
|
103 |
-
|
|
|
|
|
104 |
force_filename=lo)
|
105 |
except:
|
106 |
self._pp("*Error", f)
|
107 |
-
|
108 |
-
|
109 |
#
|
|
|
110 |
def push_hface_files(self,
|
111 |
hf_names,
|
112 |
hf_space="duchaba/skin_cancer_diagnose",
|
113 |
local_dir="/content/"):
|
114 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
115 |
try:
|
116 |
for f in hf_names:
|
117 |
lo = local_dir + f
|
@@ -122,10 +182,96 @@ class HFace_Pluto(object):
|
|
122 |
repo_type=huggingface_hub.REPO_TYPE_SPACE)
|
123 |
except Exception as e:
|
124 |
self._pp("*Error", e)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
125 |
return
|
126 |
#
|
127 |
# Define a function to display available CPU and RAM
|
128 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
129 |
s=''
|
130 |
# Get CPU usage as a percentage
|
131 |
cpu_usage = psutil.cpu_percent()
|
@@ -135,134 +281,515 @@ class HFace_Pluto(object):
|
|
135 |
mem_total_gb = mem.total / (1024 ** 3)
|
136 |
mem_available_gb = mem.available / (1024 ** 3)
|
137 |
mem_used_gb = mem.used / (1024 ** 3)
|
138 |
-
#
|
139 |
-
s += f"CPU usage: {cpu_usage}%\n"
|
140 |
s += f"Total memory: {mem_total_gb:.2f} GB\n"
|
141 |
s += f"Available memory: {mem_available_gb:.2f} GB\n"
|
142 |
# print(f"Used memory: {mem_used_gb:.2f} GB")
|
143 |
s += f"Memory usage: {mem_used_gb/mem_total_gb:.2f}%\n"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
144 |
return s
|
145 |
#
|
146 |
-
|
147 |
-
|
148 |
-
|
149 |
-
|
150 |
-
|
151 |
-
|
152 |
-
|
153 |
-
|
154 |
-
|
155 |
-
|
156 |
-
|
157 |
-
|
158 |
-
|
159 |
-
#
|
160 |
-
def fetch_gpu_info(self):
|
161 |
s=''
|
|
|
|
|
162 |
try:
|
163 |
-
|
164 |
-
|
165 |
-
|
166 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
167 |
except Exception as e:
|
168 |
s += f'**Warning, No GPU: {e}'
|
169 |
return s
|
170 |
#
|
171 |
-
|
172 |
-
|
173 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
174 |
s=open(self._xkeyfile, "rb").read()
|
|
|
175 |
return s
|
176 |
#
|
177 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
178 |
key = cryptography.fernet.Fernet.generate_key()
|
179 |
with open(self._xkeyfile, "wb") as key_file:
|
180 |
-
key_file.write(key)
|
181 |
-
return
|
182 |
#
|
183 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
184 |
y = self._fetch_crypt()
|
185 |
f = cryptography.fernet.Fernet(y)
|
186 |
m = f.decrypt(x)
|
187 |
return m.decode()
|
188 |
#
|
189 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
190 |
key = self._fetch_crypt()
|
191 |
p = x.encode()
|
192 |
f = cryptography.fernet.Fernet(key)
|
193 |
y = f.encrypt(p)
|
194 |
return y
|
195 |
#
|
196 |
-
|
197 |
-
|
198 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
199 |
self._ph()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
200 |
return
|
201 |
#
|
202 |
-
|
203 |
-
|
204 |
-
|
205 |
-
|
206 |
-
|
207 |
-
s += f"{'huggingface_hub: 0.14.1,':<28} Actual: {huggingface_hub.__version__}\n"
|
208 |
-
s += f"{'gradio: 3.32.0,':<28} Actual: {gradio.__version__}\n"
|
209 |
-
s += f"{'cryptography: 40.0.2,':<28} cryptography: {gradio.__version__}\n"
|
210 |
|
211 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
212 |
#
|
213 |
-
def _fetch_host_ip(self):
|
214 |
-
s=''
|
215 |
-
hostname = socket.gethostname()
|
216 |
-
ip_address = socket.gethostbyname(hostname)
|
217 |
-
s += f"Hostname: {hostname}\n"
|
218 |
-
s += f"IP Address: {ip_address}\n"
|
219 |
-
return s
|
220 |
#
|
221 |
-
def
|
222 |
-
write_to_file=True, fname_override=None):
|
223 |
"""
|
224 |
-
Reads a
|
225 |
-
|
|
|
226 |
|
227 |
Parameters:
|
228 |
-
|
229 |
-
|
230 |
-
|
231 |
-
|
232 |
-
|
233 |
-
- fname_override (str): If provided, overrides the output filename. The default is None.
|
234 |
|
235 |
Returns:
|
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 |
return
|
265 |
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
266 |
# add module/method
|
267 |
#
|
268 |
import functools
|
@@ -275,57 +802,48 @@ def add_method(cls):
|
|
275 |
return func # returning func means func can still be used normally
|
276 |
return decorator
|
277 |
#
|
278 |
-
|
279 |
-
|
280 |
-
|
281 |
-
|
282 |
-
|
283 |
-
|
284 |
-
|
285 |
-
monty
|
286 |
-
|
287 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
288 |
# %%write -a app.py
|
289 |
|
290 |
# client.moderations.create()
|
291 |
-
|
292 |
-
# %%
|
293 |
|
294 |
-
|
295 |
-
|
296 |
-
#
|
297 |
-
# #openai.Moderation.create()
|
298 |
-
# omod = openai.Moderation.create(p)
|
299 |
-
# r = omod.results[0].category_scores
|
300 |
-
# jmod = json.loads(str(r))
|
301 |
-
# #
|
302 |
-
# max_key = max(jmod, key=jmod.get)
|
303 |
-
# max_value = jmod[max_key]
|
304 |
-
# sum_value = sum(jmod.values())
|
305 |
-
# #
|
306 |
-
# jmod["is_safer_flagged"] = False
|
307 |
-
# if (max_value >= safer):
|
308 |
-
# jmod["is_safer_flagged"] = True
|
309 |
-
# jmod["is_flagged"] = omod.results[0].flagged
|
310 |
-
# jmod['max_key'] = max_key
|
311 |
-
# jmod['max_value'] = max_value
|
312 |
-
# jmod['sum_value'] = sum_value
|
313 |
-
# jmod['safer_value'] = safer
|
314 |
-
# jmod['message'] = p
|
315 |
-
# return jmod
|
316 |
-
#
|
317 |
-
# openai.api_key = monty._decrypt_it(monty._gpt_key)
|
318 |
#
|
319 |
# # for openai version 1.3.8
|
320 |
-
@add_method(
|
321 |
-
#
|
322 |
def _fetch_moderate_engine(self):
|
323 |
-
self.ai_client = openai.OpenAI(api_key=self.
|
324 |
self.text_model = "text-moderation-latest"
|
325 |
return
|
326 |
#
|
327 |
-
@add_method(
|
328 |
-
#
|
329 |
def _censor_me(self, p, safer=0.0005):
|
330 |
self._fetch_moderate_engine()
|
331 |
resp_orig = self.ai_client.moderations.create(input=p, model=self.text_model)
|
@@ -347,29 +865,29 @@ def _censor_me(self, p, safer=0.0005):
|
|
347 |
v1['message'] = p
|
348 |
return v1
|
349 |
#
|
350 |
-
@add_method(
|
351 |
def _draw_censor(self,data):
|
352 |
self._color_mid_gray = '#6c757d'
|
353 |
exp = (0.01, 0.01)
|
354 |
-
x = [data['max_value'], (
|
355 |
-
title=
|
356 |
-
lab = [data['max_key'], 'Other
|
357 |
if (data['is_flagged']):
|
358 |
-
col=[self.
|
359 |
elif (data['is_safer_flagged']):
|
360 |
-
col=[self.
|
361 |
-
lab = ['Relative Score:\n'+data['max_key'], 'Other
|
362 |
-
title=
|
363 |
else:
|
364 |
-
col=[self.
|
365 |
-
lab = ['False Negative:\n'+data['max_key'], 'Other
|
366 |
-
title='\
|
367 |
canvas = self._draw_donut(x, lab, col, exp,title)
|
368 |
return canvas
|
369 |
#
|
370 |
-
@add_method(
|
371 |
def _draw_donut(self,data,labels,col, exp,title):
|
372 |
-
# col = [self.
|
373 |
# exp = (0.01, 0.01)
|
374 |
# Create a pie chart
|
375 |
canvas, pic = matplotlib.pyplot.subplots()
|
@@ -392,41 +910,142 @@ def _draw_donut(self,data,labels,col, exp,title):
|
|
392 |
# canvas.show()
|
393 |
return canvas
|
394 |
#
|
395 |
-
@add_method(
|
396 |
-
def censor_me(self, msg, safer=0.
|
|
|
|
|
397 |
yjson = self._censor_me(msg,safer)
|
398 |
_canvas = self._draw_censor(yjson)
|
399 |
_yjson = json.dumps(yjson, indent=4)
|
400 |
-
return (_canvas, _yjson)
|
401 |
-
|
|
|
|
|
402 |
|
403 |
-
|
404 |
-
|
405 |
-
|
406 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
407 |
#
|
408 |
-
|
409 |
-
|
410 |
-
|
411 |
-
|
412 |
-
|
413 |
-
|
414 |
-
|
415 |
-
|
416 |
-
|
417 |
-
|
418 |
-
|
419 |
-
|
420 |
-
|
421 |
-
|
422 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
423 |
|
424 |
-
|
425 |
-
|
426 |
-
|
427 |
-
|
428 |
-
|
429 |
-
|
430 |
-
|
431 |
-
|
432 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# [BEGIN OF pluto_happy]
|
|
|
2 |
## required lib, required "pip install"
|
|
|
|
|
|
|
3 |
import torch
|
4 |
import cryptography
|
5 |
import cryptography.fernet
|
6 |
+
from flopth import flopth
|
|
|
7 |
import huggingface_hub
|
8 |
import huggingface_hub.hf_api
|
9 |
## standard libs, no need to install
|
|
|
17 |
import psutil
|
18 |
import threading
|
19 |
import socket
|
20 |
+
import PIL
|
21 |
+
import pandas
|
22 |
import matplotlib
|
23 |
+
import numpy
|
24 |
+
import importlib.metadata
|
25 |
+
import types
|
26 |
+
import cpuinfo
|
27 |
+
import pynvml
|
28 |
+
import pathlib
|
29 |
+
import re
|
30 |
+
import subprocess
|
31 |
+
# define class Pluto_Happy
|
32 |
+
class Pluto_Happy(object):
|
33 |
+
"""
|
34 |
+
The Pluto projects starts with fun AI hackings and become a part of my
|
35 |
+
first book "Data Augmentation with Python" with Packt Publishing.
|
36 |
+
|
37 |
+
In particular, Pluto_Happy is a clean and lite kernel of a simple class,
|
38 |
+
and using @add_module decoractor to add in specific methods to be a new class,
|
39 |
+
such as Pluto_HFace with a lot more function on HuggingFace, LLM and Transformers.
|
40 |
+
|
41 |
+
Args:
|
42 |
+
name (str): the display name, e.g. "Hanna the seeker"
|
43 |
+
|
44 |
+
Returns:
|
45 |
+
(object): the class instance.
|
46 |
+
"""
|
47 |
+
|
48 |
# initialize the object
|
49 |
def __init__(self, name="Pluto",*args, **kwargs):
|
50 |
+
super(Pluto_Happy, self).__init__(*args, **kwargs)
|
51 |
self.author = "Duc Haba"
|
52 |
self.name = name
|
53 |
self._ph()
|
|
|
57 |
self._ph()
|
58 |
#
|
59 |
# define class var for stable division
|
60 |
+
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
61 |
self.fname_id = 0
|
62 |
self.dname_img = "img_colab/"
|
63 |
+
self.flops_per_sec_gcolab_cpu = 4887694725 # 925,554,209 | 9,276,182,810 | 1,722,089,747 | 5,287,694,725
|
64 |
+
self.flops_per_sec_gcolab_gpu = 6365360673 # 1,021,721,764 | 9,748,048,188 | 2,245,406,502 | 6,965,360,673
|
65 |
+
self.fname_requirements = './pluto_happy/requirements.txt'
|
66 |
+
#
|
67 |
+
self.color_primary = '#2780e3' #blue
|
68 |
+
self.color_secondary = '#373a3c' #dark gray
|
69 |
+
self.color_success = '#3fb618' #green
|
70 |
+
self.color_info = '#9954bb' #purple
|
71 |
+
self.color_warning = '#ff7518' #orange
|
72 |
+
self.color_danger = '#ff0039' #red
|
73 |
+
self.color_mid_gray = '#495057'
|
74 |
+
self._xkeyfile = '.xoxo'
|
75 |
return
|
76 |
#
|
77 |
# pretty print output name-value line
|
78 |
def _pp(self, a, b,is_print=True):
|
79 |
+
|
80 |
+
"""
|
81 |
+
Pretty print output name-value line
|
82 |
+
|
83 |
+
Args:
|
84 |
+
a (str) :
|
85 |
+
b (str) :
|
86 |
+
is_print (bool): whether to print the header or footer lines to console or return a str.
|
87 |
+
|
88 |
+
Returns:
|
89 |
+
y : None or output as (str)
|
90 |
+
|
91 |
+
"""
|
92 |
# print("%34s : %s" % (str(a), str(b)))
|
93 |
x = f'{"%34s" % str(a)} : {str(b)}'
|
94 |
y = None
|
|
|
100 |
#
|
101 |
# pretty print the header or footer lines
|
102 |
def _ph(self,is_print=True):
|
103 |
+
"""
|
104 |
+
Pretty prints the header or footer lines.
|
105 |
+
|
106 |
+
Args:
|
107 |
+
is_print (bool): whether to print the header or footer lines to console or return a str.
|
108 |
+
|
109 |
+
Return:
|
110 |
+
y : None or output as (str)
|
111 |
+
|
112 |
+
"""
|
113 |
x = f'{"-"*34} : {"-"*34}'
|
114 |
y = None
|
115 |
if (is_print):
|
|
|
123 |
hf_names,
|
124 |
hf_space="duchaba/monty",
|
125 |
local_dir="/content/"):
|
126 |
+
"""
|
127 |
+
Given a list of huggingface file names, download them from the provided huggingface space.
|
128 |
+
|
129 |
+
Args:
|
130 |
+
hf_names: (list) list of huggingface file names to download
|
131 |
+
hf_space: (str) huggingface space to download from.
|
132 |
+
local_dir: (str) local directory to store the files.
|
133 |
+
|
134 |
+
Returns:
|
135 |
+
status: (bool) True if download was successful, False otherwise.
|
136 |
+
"""
|
137 |
+
status = True
|
138 |
+
# f = str(hf_names) + " is not iteratable, type: " + str(type(hf_names))
|
139 |
try:
|
140 |
for f in hf_names:
|
141 |
lo = local_dir + f
|
142 |
+
huggingface_hub.hf_hub_download(repo_id=hf_space,
|
143 |
+
filename=f,
|
144 |
+
use_auth_token=True,
|
145 |
+
repo_type=huggingface_hub.REPO_TYPE_SPACE,
|
146 |
force_filename=lo)
|
147 |
except:
|
148 |
self._pp("*Error", f)
|
149 |
+
status = False
|
150 |
+
return status
|
151 |
#
|
152 |
+
# push files to huggingface
|
153 |
def push_hface_files(self,
|
154 |
hf_names,
|
155 |
hf_space="duchaba/skin_cancer_diagnose",
|
156 |
local_dir="/content/"):
|
157 |
+
# push files to huggingface space
|
158 |
+
|
159 |
+
"""
|
160 |
+
Pushes files to huggingface space.
|
161 |
+
|
162 |
+
The function takes a list of file names as a
|
163 |
+
paramater and pushes to the provided huggingface space.
|
164 |
+
|
165 |
+
Args:
|
166 |
+
hf_names: list(of strings), list of file names to be pushed.
|
167 |
+
hf_space: (str), the huggingface space to push to.
|
168 |
+
local_dir: (str), the local directory where the files
|
169 |
+
are stored.
|
170 |
+
|
171 |
+
Returns:
|
172 |
+
status: (bool) True if successfully pushed else False.
|
173 |
+
"""
|
174 |
+
status = True
|
175 |
try:
|
176 |
for f in hf_names:
|
177 |
lo = local_dir + f
|
|
|
182 |
repo_type=huggingface_hub.REPO_TYPE_SPACE)
|
183 |
except Exception as e:
|
184 |
self._pp("*Error", e)
|
185 |
+
status = False
|
186 |
+
return status
|
187 |
+
#
|
188 |
+
# push the folder to huggingface space
|
189 |
+
def push_hface_folder(self,
|
190 |
+
hf_folder,
|
191 |
+
hf_space_id,
|
192 |
+
hf_dest_folder=None):
|
193 |
+
|
194 |
+
"""
|
195 |
+
|
196 |
+
This function pushes the folder to huggingface space.
|
197 |
+
|
198 |
+
Args:
|
199 |
+
hf_folder: (str). The path to the folder to push.
|
200 |
+
hf_space_id: (str). The space id to push the folder to.
|
201 |
+
hf_dest_folder: (str). The destination folder in the space. If not specified,
|
202 |
+
the folder name will be used as the destination folder.
|
203 |
+
|
204 |
+
Returns:
|
205 |
+
status: (bool) True if the folder is pushed successfully, otherwise False.
|
206 |
+
"""
|
207 |
+
|
208 |
+
status = True
|
209 |
+
try:
|
210 |
+
api = huggingface_hub.HfApi()
|
211 |
+
api.upload_folder(folder_path=hf_folder,
|
212 |
+
repo_id=hf_space_id,
|
213 |
+
path_in_repo=hf_dest_folder,
|
214 |
+
repo_type="space")
|
215 |
+
except Exception as e:
|
216 |
+
self._pp("*Error: ",e)
|
217 |
+
status = False
|
218 |
+
return status
|
219 |
+
#
|
220 |
+
# automatically restart huggingface space
|
221 |
+
def restart_hface_periodically(self):
|
222 |
+
|
223 |
+
"""
|
224 |
+
This function restarts the huggingface space automatically in random
|
225 |
+
periodically.
|
226 |
+
|
227 |
+
Args:
|
228 |
+
None
|
229 |
+
|
230 |
+
Returns:
|
231 |
+
None
|
232 |
+
"""
|
233 |
+
|
234 |
+
while True:
|
235 |
+
random_time = random.randint(15800, 21600)
|
236 |
+
time.sleep(random_time)
|
237 |
+
os.execl(sys.executable, sys.executable, *sys.argv)
|
238 |
+
return
|
239 |
+
#
|
240 |
+
# log into huggingface
|
241 |
+
def login_hface(self, key=None):
|
242 |
+
|
243 |
+
"""
|
244 |
+
Log into HuggingFace.
|
245 |
+
|
246 |
+
Args:
|
247 |
+
key: (str, optional) If key is set, this key will be used to log in,
|
248 |
+
otherwise the key will be decrypted from the key file.
|
249 |
+
|
250 |
+
Returns:
|
251 |
+
None
|
252 |
+
"""
|
253 |
+
|
254 |
+
if (key is None):
|
255 |
+
x = self._decrypt_it(self._huggingface_crkey)
|
256 |
+
else:
|
257 |
+
x = key
|
258 |
+
huggingface_hub.login(x, add_to_git_credential=True) # non-blocking login
|
259 |
+
self._ph()
|
260 |
return
|
261 |
#
|
262 |
# Define a function to display available CPU and RAM
|
263 |
+
def fetch_info_system(self):
|
264 |
+
|
265 |
+
"""
|
266 |
+
Fetches system information, such as CPU usage and memory usage.
|
267 |
+
|
268 |
+
Args:
|
269 |
+
None.
|
270 |
+
|
271 |
+
Returns:
|
272 |
+
s: (str) A string containing the system information.
|
273 |
+
"""
|
274 |
+
|
275 |
s=''
|
276 |
# Get CPU usage as a percentage
|
277 |
cpu_usage = psutil.cpu_percent()
|
|
|
281 |
mem_total_gb = mem.total / (1024 ** 3)
|
282 |
mem_available_gb = mem.available / (1024 ** 3)
|
283 |
mem_used_gb = mem.used / (1024 ** 3)
|
284 |
+
# save the results
|
|
|
285 |
s += f"Total memory: {mem_total_gb:.2f} GB\n"
|
286 |
s += f"Available memory: {mem_available_gb:.2f} GB\n"
|
287 |
# print(f"Used memory: {mem_used_gb:.2f} GB")
|
288 |
s += f"Memory usage: {mem_used_gb/mem_total_gb:.2f}%\n"
|
289 |
+
try:
|
290 |
+
cpu_info = cpuinfo.get_cpu_info()
|
291 |
+
s += f'CPU type: {cpu_info["brand_raw"]}, arch: {cpu_info["arch"]}\n'
|
292 |
+
s += f'Number of CPU cores: {cpu_info["count"]}\n'
|
293 |
+
s += f"CPU usage: {cpu_usage}%\n"
|
294 |
+
s += f'Python version: {cpu_info["python_version"]}'
|
295 |
+
except Exception as e:
|
296 |
+
s += f'CPU type: Not accessible, Error: {e}'
|
297 |
return s
|
298 |
#
|
299 |
+
# fetch GPU RAM info
|
300 |
+
def fetch_info_gpu(self):
|
301 |
+
|
302 |
+
"""
|
303 |
+
Function to fetch GPU RAM info
|
304 |
+
|
305 |
+
Args:
|
306 |
+
None.
|
307 |
+
|
308 |
+
Returns:
|
309 |
+
s: (str) GPU RAM info in human readable format.
|
310 |
+
"""
|
311 |
+
|
|
|
|
|
312 |
s=''
|
313 |
+
mtotal = 0
|
314 |
+
mfree = 0
|
315 |
try:
|
316 |
+
nvml_handle = pynvml.nvmlInit()
|
317 |
+
devices = pynvml.nvmlDeviceGetCount()
|
318 |
+
for i in range(devices):
|
319 |
+
device = pynvml.nvmlDeviceGetHandleByIndex(i)
|
320 |
+
memory_info = pynvml.nvmlDeviceGetMemoryInfo(device)
|
321 |
+
mtotal += memory_info.total
|
322 |
+
mfree += memory_info.free
|
323 |
+
mtotal = mtotal / 1024**3
|
324 |
+
mfree = mfree / 1024**3
|
325 |
+
# print(f"GPU {i}: Total Memory: {memory_info.total/1024**3} GB, Free Memory: {memory_info.free/1024**3} GB")
|
326 |
+
s += f'GPU type: {torch.cuda.get_device_name(0)}\n'
|
327 |
+
s += f'GPU ready staus: {torch.cuda.is_available()}\n'
|
328 |
+
s += f'Number of GPUs: {devices}\n'
|
329 |
+
s += f'Total Memory: {mtotal:.2f} GB\n'
|
330 |
+
s += f'Free Memory: {mfree:.2f} GB\n'
|
331 |
+
s += f'GPU allocated RAM: {round(torch.cuda.memory_allocated(0)/1024**3,2)} GB\n'
|
332 |
+
s += f'GPU reserved RAM {round(torch.cuda.memory_reserved(0)/1024**3,2)} GB\n'
|
333 |
except Exception as e:
|
334 |
s += f'**Warning, No GPU: {e}'
|
335 |
return s
|
336 |
#
|
337 |
+
# fetch info about host ip
|
338 |
+
def fetch_info_host_ip(self):
|
339 |
+
"""
|
340 |
+
Function to fetch current host name and ip address
|
341 |
+
|
342 |
+
Args:
|
343 |
+
None.
|
344 |
+
|
345 |
+
Returns:
|
346 |
+
s: (str) host name and ip info in human readable format.
|
347 |
+
"""
|
348 |
+
s=''
|
349 |
+
try:
|
350 |
+
hostname = socket.gethostname()
|
351 |
+
ip_address = socket.gethostbyname(hostname)
|
352 |
+
s += f"Hostname: {hostname}\n"
|
353 |
+
s += f"IP Address: {ip_address}\n"
|
354 |
+
except Exception as e:
|
355 |
+
s += f"**Warning, No hostname: {e}"
|
356 |
+
return s
|
357 |
+
#
|
358 |
+
# fetch files name
|
359 |
+
def fetch_file_names(self,directory, file_extension=None):
|
360 |
+
"""
|
361 |
+
This function gets all the filenames with a given extension.
|
362 |
+
Args:
|
363 |
+
directory (str):
|
364 |
+
directory path to scan for files in.
|
365 |
+
file_extension (list):
|
366 |
+
file extension to look for or "None" (default) to get all files.
|
367 |
+
Returns:
|
368 |
+
filenames (list):
|
369 |
+
list of strings containing the filenames with the given extension.
|
370 |
+
"""
|
371 |
+
filenames = []
|
372 |
+
for (root, subFolders, files) in os.walk(directory):
|
373 |
+
for fname in files:
|
374 |
+
if (file_extension is None):
|
375 |
+
filenames.append(os.path.join(root, fname))
|
376 |
+
else:
|
377 |
+
for ext in file_extension:
|
378 |
+
if fname.endswith(ext):
|
379 |
+
filenames.append(os.path.join(root, fname))
|
380 |
+
return filenames
|
381 |
+
#
|
382 |
+
# fetch the crypto key
|
383 |
+
def _fetch_crypt(self,has_new_key=False):
|
384 |
+
|
385 |
+
"""
|
386 |
+
This function fetches the crypto key from the file or from the
|
387 |
+
variable created previously in the class.
|
388 |
+
Args:
|
389 |
+
has_new_key (bool):
|
390 |
+
is_generate flag to indicate whether the key should be
|
391 |
+
use as-is or fetch from the file.
|
392 |
+
Returns:
|
393 |
+
s (str):
|
394 |
+
string value containing the crypto key.
|
395 |
+
"""
|
396 |
+
if self._fkey == 'your_key_goes_here':
|
397 |
+
raise Exception('Cryto Key is not correct!')
|
398 |
+
#
|
399 |
+
s=self._fkey[::-1]
|
400 |
+
if (has_new_key):
|
401 |
s=open(self._xkeyfile, "rb").read()
|
402 |
+
self._fkey = s[::-1]
|
403 |
return s
|
404 |
#
|
405 |
+
# generate new cryto key
|
406 |
+
def gen_key(self):
|
407 |
+
"""
|
408 |
+
This function generates a new cryto key and saves it to a file
|
409 |
+
|
410 |
+
Args:
|
411 |
+
None
|
412 |
+
|
413 |
+
Returns:
|
414 |
+
(str) crypto key
|
415 |
+
"""
|
416 |
+
|
417 |
key = cryptography.fernet.Fernet.generate_key()
|
418 |
with open(self._xkeyfile, "wb") as key_file:
|
419 |
+
key_file.write(key[::-1]) # write in reversed
|
420 |
+
return key
|
421 |
#
|
422 |
+
# decrypt message
|
423 |
+
def decrypt_it(self, x):
|
424 |
+
"""
|
425 |
+
Decrypts the encrypted string using the stored crypto key.
|
426 |
+
|
427 |
+
Args:
|
428 |
+
x: (str) to be decrypted.
|
429 |
+
|
430 |
+
Returns:
|
431 |
+
x: (str) decrypted version of x.
|
432 |
+
"""
|
433 |
y = self._fetch_crypt()
|
434 |
f = cryptography.fernet.Fernet(y)
|
435 |
m = f.decrypt(x)
|
436 |
return m.decode()
|
437 |
#
|
438 |
+
# encrypt message
|
439 |
+
def encrypt_it(self, x):
|
440 |
+
"""
|
441 |
+
encrypt message
|
442 |
+
|
443 |
+
Args:
|
444 |
+
x (str): message to encrypt
|
445 |
+
|
446 |
+
Returns:
|
447 |
+
str: encrypted message
|
448 |
+
"""
|
449 |
+
|
450 |
key = self._fetch_crypt()
|
451 |
p = x.encode()
|
452 |
f = cryptography.fernet.Fernet(key)
|
453 |
y = f.encrypt(p)
|
454 |
return y
|
455 |
#
|
456 |
+
# fetch import libraries
|
457 |
+
def _fetch_lib_import(self):
|
458 |
+
|
459 |
+
"""
|
460 |
+
This function fetches all the imported libraries that are installed.
|
461 |
+
|
462 |
+
Args:
|
463 |
+
None
|
464 |
+
|
465 |
+
Returns:
|
466 |
+
x (list):
|
467 |
+
list of strings containing the name of the imported libraries.
|
468 |
+
"""
|
469 |
+
|
470 |
+
x = []
|
471 |
+
for name, val in globals().items():
|
472 |
+
if isinstance(val, types.ModuleType):
|
473 |
+
x.append(val.__name__)
|
474 |
+
x.sort()
|
475 |
+
return x
|
476 |
+
#
|
477 |
+
# fetch lib version
|
478 |
+
def _fetch_lib_version(self,lib_name):
|
479 |
+
|
480 |
+
"""
|
481 |
+
This function fetches the version of the imported libraries.
|
482 |
+
|
483 |
+
Args:
|
484 |
+
lib_name (list):
|
485 |
+
list of strings containing the name of the imported libraries.
|
486 |
+
|
487 |
+
Returns:
|
488 |
+
val (list):
|
489 |
+
list of strings containing the version of the imported libraries.
|
490 |
+
"""
|
491 |
+
|
492 |
+
val = []
|
493 |
+
for x in lib_name:
|
494 |
+
try:
|
495 |
+
y = importlib.metadata.version(x)
|
496 |
+
val.append(f'{x}=={y}')
|
497 |
+
except Exception as e:
|
498 |
+
val.append(f'|{x}==unknown_*or_system')
|
499 |
+
val.sort()
|
500 |
+
return val
|
501 |
+
#
|
502 |
+
# fetch the lib name and version
|
503 |
+
def fetch_info_lib_import(self):
|
504 |
+
"""
|
505 |
+
This function fetches all the imported libraries name and version that are installed.
|
506 |
+
|
507 |
+
Args:
|
508 |
+
None
|
509 |
+
|
510 |
+
Returns:
|
511 |
+
x (list):
|
512 |
+
list of strings containing the name and version of the imported libraries.
|
513 |
+
"""
|
514 |
+
x = self._fetch_lib_version(self._fetch_lib_import())
|
515 |
+
return x
|
516 |
+
#
|
517 |
+
# write a file to local or cloud diskspace
|
518 |
+
def write_file(self,fname, in_data):
|
519 |
+
|
520 |
+
"""
|
521 |
+
Write a file to local or cloud diskspace or append to it if it already exists.
|
522 |
+
|
523 |
+
Args:
|
524 |
+
fname (str): The name of the file to write.
|
525 |
+
in_data (list): The
|
526 |
+
|
527 |
+
This is a utility function that writes a file to disk.
|
528 |
+
The file name and text to write are passed in as arguments.
|
529 |
+
The file is created, the text is written to it, and then the file is closed.
|
530 |
+
|
531 |
+
Args:
|
532 |
+
fname (str): The name of the file to write.
|
533 |
+
in_data (list): The text to write to the file.
|
534 |
+
|
535 |
+
Returns:
|
536 |
+
None
|
537 |
+
"""
|
538 |
+
|
539 |
+
if os.path.isfile(fname):
|
540 |
+
f = open(fname, "a")
|
541 |
+
else:
|
542 |
+
f = open(fname, "w")
|
543 |
+
f.writelines("\n".join(in_data))
|
544 |
+
f.close()
|
545 |
+
return
|
546 |
+
#
|
547 |
+
# fetch flops info
|
548 |
+
def fetch_info_flops(self,model, input_shape=(1, 3, 224, 224), device="cpu", max_epoch=1):
|
549 |
+
|
550 |
+
"""
|
551 |
+
Calculates the number of floating point operations (FLOPs).
|
552 |
+
|
553 |
+
Args:
|
554 |
+
model (torch.nn.Module): neural network model.
|
555 |
+
input_shape (tuple): input tensor size.
|
556 |
+
device (str): device to perform computation on.
|
557 |
+
max_epoch (int): number of times
|
558 |
+
|
559 |
+
Returns:
|
560 |
+
(float): number of FLOPs, average from epoch, default is 1 epoch.
|
561 |
+
(float): elapsed seconds
|
562 |
+
(list): of string for a friendly human readable output
|
563 |
+
"""
|
564 |
+
|
565 |
+
ttm_input = torch.rand(input_shape, dtype=torch.float32, device=device)
|
566 |
+
# ttm_input = torch.rand((1, 3, 224, 224), dtype=torch.float32, device=device)
|
567 |
+
tstart = time.time()
|
568 |
+
for i in range(max_epoch):
|
569 |
+
flops, params = flopth(model, inputs=(ttm_input,), bare_number=True)
|
570 |
+
tend = time.time()
|
571 |
+
etime = (tend - tstart)/max_epoch
|
572 |
+
|
573 |
+
# kilo = 10^3, maga = 10^6, giga = 10^9, tera=10^12, peta=10^15, exa=10^18, zetta=10^21
|
574 |
+
valstr = []
|
575 |
+
valstr.append(f'Tensors device: {device}')
|
576 |
+
valstr.append(f'flops: {flops:,}')
|
577 |
+
valstr.append(f'params: {params:,}')
|
578 |
+
valstr.append(f'epoch: {max_epoch}')
|
579 |
+
valstr.append(f'sec: {etime}')
|
580 |
+
# valstr += f'Tensors device: {device}, flops: {flops}, params: {params}, epoch: {max_epoch}, sec: {etime}\n'
|
581 |
+
x = flops/etime
|
582 |
+
y = (x/10**15)*86400
|
583 |
+
valstr.append(f'Flops/s: {x:,}')
|
584 |
+
valstr.append(f'PetaFlops/s: {x/10**15}')
|
585 |
+
valstr.append(f'PetaFlops/day: {y}')
|
586 |
+
valstr.append(f'1 PetaFlopsDay (on this system will take): {round(1/y, 2):,.2f} days')
|
587 |
+
return flops, etime, valstr
|
588 |
+
#
|
589 |
+
def print_petaflops(self):
|
590 |
+
|
591 |
+
"""
|
592 |
+
Prints the flops and peta-flops-day calculation.
|
593 |
+
**WARING**: This method will break/interfer with Stable Diffusion use of LoRA.
|
594 |
+
I can't debug why yet.
|
595 |
+
|
596 |
+
Args:
|
597 |
+
None
|
598 |
+
|
599 |
+
Returns:
|
600 |
+
None
|
601 |
+
"""
|
602 |
+
self._pp('Model', 'TTM, Tiny Torch Model on: CPU')
|
603 |
+
mtoy = TTM()
|
604 |
+
# my_model = MyModel()
|
605 |
+
dev = torch.device("cuda:0")
|
606 |
+
a,b,c = self.fetch_info_flops(mtoy)
|
607 |
+
y = round((a/b)/self.flops_per_sec_gcolab_cpu * 100, 2)
|
608 |
+
self._pp('Flops', f'{a:,} flops')
|
609 |
+
self._pp('Total elapse time', f'{b:,} seconds')
|
610 |
+
self._pp('Flops compared', f'{y:,}% of Google Colab Pro')
|
611 |
+
for i, val in enumerate(c):
|
612 |
+
self._pp(f'Info {i}', val)
|
613 |
self._ph()
|
614 |
+
|
615 |
+
try:
|
616 |
+
self._pp('Model', 'TTM, Tiny Torch Model on: GPU')
|
617 |
+
dev = torch.device("cuda:0")
|
618 |
+
a2,b2,c2 = self.fetch_info_flops(mtoy, device=dev)
|
619 |
+
y2 = round((a2/b2)/self.flops_per_sec_gcolab_gpu * 100, 2)
|
620 |
+
self._pp('Flops', f'{a2:,} flops')
|
621 |
+
self._pp('Total elapse time', f'{b2:,} seconds')
|
622 |
+
self._pp('Flops compared', f'{y2:,}% of Google Colab Pro')
|
623 |
+
d2 = round(((a2/b2)/(a/b))*100, 2)
|
624 |
+
self._pp('Flops GPU compared', f'{d2:,}% of CPU (or {round(d2-100,2):,}% faster)')
|
625 |
+
for i, val in enumerate(c2):
|
626 |
+
self._pp(f'Info {i}', val)
|
627 |
+
except Exception as e:
|
628 |
+
self._pp('Error', e)
|
629 |
+
self._ph()
|
630 |
return
|
631 |
#
|
632 |
+
#
|
633 |
+
def fetch_installed_libraries(self):
|
634 |
+
"""
|
635 |
+
Retrieves and prints the names and versions of Python libraries installed by the user,
|
636 |
+
excluding the standard libraries.
|
|
|
|
|
|
|
637 |
|
638 |
+
Args:
|
639 |
+
-----
|
640 |
+
None
|
641 |
+
|
642 |
+
Returns:
|
643 |
+
--------
|
644 |
+
dictionary: (dict)
|
645 |
+
A dictionary where keys are the names of the libraries and values are their respective versions.
|
646 |
+
|
647 |
+
Examples:
|
648 |
+
---------
|
649 |
+
libraries = get_installed_libraries()
|
650 |
+
for name, version in libraries.items():
|
651 |
+
print(f"{name}: {version}")
|
652 |
+
"""
|
653 |
+
# List of standard libraries (this may not be exhaustive and might need updates based on the Python version)
|
654 |
+
# Run pip freeze command to get list of installed packages with their versions
|
655 |
+
result = subprocess.run(['pip', 'freeze'], stdout=subprocess.PIPE)
|
656 |
+
|
657 |
+
# Decode result and split by lines
|
658 |
+
packages = result.stdout.decode('utf-8').splitlines()
|
659 |
+
|
660 |
+
# Split each line by '==' to separate package names and versions
|
661 |
+
installed_libraries = {}
|
662 |
+
for package in packages:
|
663 |
+
try:
|
664 |
+
name, version = package.split('==')
|
665 |
+
installed_libraries[name] = version
|
666 |
+
except Exception as e:
|
667 |
+
#print(f'{package}: Error: {e}')
|
668 |
+
pass
|
669 |
+
return installed_libraries
|
670 |
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
671 |
#
|
672 |
+
def fetch_match_file_dict(self, file_path, reference_dict):
|
|
|
673 |
"""
|
674 |
+
Reads a file from the disk, creates an array with each line as an item,
|
675 |
+
and checks if each line exists as a key in the provided dictionary. If it exists,
|
676 |
+
the associated value from the dictionary is also returned.
|
677 |
|
678 |
Parameters:
|
679 |
+
-----------
|
680 |
+
file_path: str
|
681 |
+
Path to the file to be read.
|
682 |
+
reference_dict: dict
|
683 |
+
Dictionary against which the file content (each line) will be checked.
|
|
|
684 |
|
685 |
Returns:
|
686 |
+
--------
|
687 |
+
dict:
|
688 |
+
A dictionary where keys are the lines from the file and values are either
|
689 |
+
the associated values from the reference dictionary or None if the key
|
690 |
+
doesn't exist in the dictionary.
|
691 |
+
|
692 |
+
Raises:
|
693 |
+
-------
|
694 |
+
FileNotFoundError:
|
695 |
+
If the provided file path does not exist.
|
696 |
+
"""
|
697 |
+
|
698 |
+
if not os.path.exists(file_path):
|
699 |
+
raise FileNotFoundError(f"The file at {file_path} does not exist.")
|
700 |
+
|
701 |
+
with open(file_path, 'r') as file:
|
702 |
+
lines = file.readlines()
|
703 |
+
|
704 |
+
# Check if each line (stripped of whitespace and newline characters) exists in the reference dictionary.
|
705 |
+
# If it exists, fetch its value. Otherwise, set the value to None.
|
706 |
+
results = {line.strip(): reference_dict.get(line.strip().replace('_', '-'), None) for line in lines}
|
707 |
+
|
708 |
+
return results
|
709 |
+
# print fech_info about myself
|
710 |
+
def print_info_self(self):
|
711 |
+
|
712 |
+
"""
|
713 |
+
Prints information about the model/myself.
|
714 |
+
|
715 |
+
Args:
|
716 |
+
None
|
717 |
+
|
718 |
+
Returns:
|
719 |
+
None
|
720 |
+
"""
|
721 |
+
|
722 |
+
self._ph()
|
723 |
+
self._pp("Hello, I am", self.name)
|
724 |
+
self._pp("I will display", "Python, Jupyter, and system info.")
|
725 |
+
self._pp("For complete doc type", "help(pluto) ...or help(your_object_name)")
|
726 |
+
self._pp('.','.')
|
727 |
+
self._pp("...", "Β―\_(γ)_/Β―")
|
728 |
+
self._ph()
|
729 |
+
# system
|
730 |
+
self._pp('System', 'Info')
|
731 |
+
x = self.fetch_info_system()
|
732 |
+
print(x)
|
733 |
+
self._ph()
|
734 |
+
# gpu
|
735 |
+
self._pp('GPU', 'Info')
|
736 |
+
x = self.fetch_info_gpu()
|
737 |
+
print(x)
|
738 |
+
self._ph()
|
739 |
+
# lib used
|
740 |
+
self._pp('Installed lib from', self.fname_requirements)
|
741 |
+
self._ph()
|
742 |
+
x = self.fetch_match_file_dict(self.fname_requirements, self.fetch_installed_libraries())
|
743 |
+
for item, value in x.items():
|
744 |
+
self._pp(f'{item} version', value)
|
745 |
+
self._ph()
|
746 |
+
self._pp('Standard lib from', 'System')
|
747 |
+
self._ph()
|
748 |
+
self._pp('matplotlib version', matplotlib.__version__)
|
749 |
+
self._pp('numpy version', numpy.__version__)
|
750 |
+
self._pp('pandas version',pandas.__version__)
|
751 |
+
self._pp('PIL version', PIL.__version__)
|
752 |
+
self._pp('torch version', torch.__version__)
|
753 |
+
self._ph()
|
754 |
+
# host ip
|
755 |
+
self._pp('Host', 'Info')
|
756 |
+
x = self.fetch_info_host_ip()
|
757 |
+
print(x)
|
758 |
+
self._ph()
|
759 |
+
#
|
760 |
return
|
761 |
#
|
762 |
+
#
|
763 |
+
# define TTM for use in calculating flops
|
764 |
+
class TTM(torch.nn.Module):
|
765 |
+
|
766 |
+
"""
|
767 |
+
Tiny Torch Model (TTM)
|
768 |
+
|
769 |
+
This is a toy model consisting of four convolutional layers.
|
770 |
+
|
771 |
+
Args:
|
772 |
+
input_shape (tuple): input tensor size.
|
773 |
+
|
774 |
+
Returns:
|
775 |
+
(tensor): output of the model.
|
776 |
+
"""
|
777 |
+
|
778 |
+
def __init__(self, input_shape=(1, 3, 224, 224)):
|
779 |
+
super(TTM, self).__init__()
|
780 |
+
self.conv1 = torch.nn.Conv2d(3, 3, kernel_size=3, padding=1)
|
781 |
+
self.conv2 = torch.nn.Conv2d(3, 3, kernel_size=3, padding=1)
|
782 |
+
self.conv3 = torch.nn.Conv2d(3, 3, kernel_size=3, padding=1)
|
783 |
+
self.conv4 = torch.nn.Conv2d(3, 3, kernel_size=3, padding=1)
|
784 |
+
|
785 |
+
def forward(self, x1):
|
786 |
+
x1 = self.conv1(x1)
|
787 |
+
x1 = self.conv2(x1)
|
788 |
+
x1 = self.conv3(x1)
|
789 |
+
x1 = self.conv4(x1)
|
790 |
+
return x1
|
791 |
+
#
|
792 |
+
# (end of class TTM)
|
793 |
# add module/method
|
794 |
#
|
795 |
import functools
|
|
|
802 |
return func # returning func means func can still be used normally
|
803 |
return decorator
|
804 |
#
|
805 |
+
# [END OF pluto_happy]
|
806 |
+
## %%write app.py
|
807 |
+
import openai
|
808 |
+
import gradio
|
809 |
+
# %%write -a app.py
|
810 |
+
|
811 |
+
# wake up monty
|
812 |
+
monty = Pluto_Happy('Monty, shares or steal')
|
813 |
+
# %%write -a app.py
|
814 |
+
|
815 |
+
# check out my environments
|
816 |
+
|
817 |
+
# monty.fname_requirements = 'pluto_happy/requirements.txt'
|
818 |
+
# monty.print_info_self()
|
819 |
+
# %%write -a app.py
|
820 |
+
|
821 |
+
monty._huggingface_key=b'gAAAAABld_3fKLl7aPBJzfAq-th37t95pMu2bVbH9QccOSecaUnm33XrpKpCXP4GL6Wr23g3vtrKWli5JK1ZPh18ilnDb_Su6GoVvU92Vzba64k3gBQwKF_g5DoH2vWq2XM8vx_5mKJh'
|
822 |
+
monty._kaggle_key=b'gAAAAABld_4_B6rrRhFYyfl77dacu1RhR4ktaLU6heYhQBSIj4ELBm7y4DzU1R8-H4yPKd0w08s11wkFJ9AR7XyESxM1SsrMBzqQEeW9JKNbl6jAaonFGmqbhFblkQqH4XjsapZru0qX'
|
823 |
+
monty._fkey="fes_f8Im569hYnI1Tn6FqP-6hS4rdmNOJ6DWcRPOsvc="
|
824 |
+
monty._fkey=monty._fkey[::-1]
|
825 |
+
monty._ok=b'gAAAAABld_-y70otUll4Jwq3jEBXiw1tooSFo_gStRbkCyuu9_Dmdehc4M8lI_hFbum9CwyZuj9ZnXgxFIROebcPSF5qoA197VRvzUDQOMxY5zmHnImVROrsXVdZqXyIeYH_Q6cvXvFTX3rLBIKKWgvJmnpYGRaV6Q=='
|
826 |
+
|
827 |
# %%write -a app.py
|
828 |
|
829 |
# client.moderations.create()
|
830 |
+
ai_client = openai.OpenAI(api_key=monty.decrypt_it(monty._ok))
|
831 |
+
# %%write -a app.py
|
832 |
|
833 |
+
fname = 'toxic_data.csv'
|
834 |
+
monty.df_toxic_data = pandas.read_csv(fname)
|
835 |
+
# %%writefile -a app.py
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
836 |
#
|
837 |
# # for openai version 1.3.8
|
838 |
+
@add_method(Pluto_Happy)
|
839 |
+
#
|
840 |
def _fetch_moderate_engine(self):
|
841 |
+
self.ai_client = openai.OpenAI(api_key=self.decrypt_it(self._ok))
|
842 |
self.text_model = "text-moderation-latest"
|
843 |
return
|
844 |
#
|
845 |
+
@add_method(Pluto_Happy)
|
846 |
+
#
|
847 |
def _censor_me(self, p, safer=0.0005):
|
848 |
self._fetch_moderate_engine()
|
849 |
resp_orig = self.ai_client.moderations.create(input=p, model=self.text_model)
|
|
|
865 |
v1['message'] = p
|
866 |
return v1
|
867 |
#
|
868 |
+
@add_method(Pluto_Happy)
|
869 |
def _draw_censor(self,data):
|
870 |
self._color_mid_gray = '#6c757d'
|
871 |
exp = (0.01, 0.01)
|
872 |
+
x = [data['max_value'], (1-data['max_value'])]
|
873 |
+
title=f"\nUnsafe: {data['max_key']}: {(data['max_value']*100):.2f}% Confidence\n"
|
874 |
+
lab = [data['max_key'], 'Other 13 categories']
|
875 |
if (data['is_flagged']):
|
876 |
+
col=[self.color_danger, self.color_mid_gray]
|
877 |
elif (data['is_safer_flagged']):
|
878 |
+
col=[self.color_warning, self.color_mid_gray]
|
879 |
+
lab = ['Relative Score:\n'+data['max_key'], 'Other 13 categories']
|
880 |
+
title=f"\nPersonal Unsafe: {data['max_key']}: {(data['max_value']*100):.2f}% Confidence\n"
|
881 |
else:
|
882 |
+
col=[self.color_mid_gray, self.color_success]
|
883 |
+
lab = ['False Negative:\n'+data['max_key'], 'Other 13 categories']
|
884 |
+
title='\nSafe Message\n'
|
885 |
canvas = self._draw_donut(x, lab, col, exp,title)
|
886 |
return canvas
|
887 |
#
|
888 |
+
@add_method(Pluto_Happy)
|
889 |
def _draw_donut(self,data,labels,col, exp,title):
|
890 |
+
# col = [self.color_danger, self._color_secondary]
|
891 |
# exp = (0.01, 0.01)
|
892 |
# Create a pie chart
|
893 |
canvas, pic = matplotlib.pyplot.subplots()
|
|
|
910 |
# canvas.show()
|
911 |
return canvas
|
912 |
#
|
913 |
+
@add_method(Pluto_Happy)
|
914 |
+
# def censor_me(self, msg, safer=0.02, ibutton_1=0):
|
915 |
+
def censor_me(self, msg, safer):
|
916 |
+
# safer=0.2
|
917 |
yjson = self._censor_me(msg,safer)
|
918 |
_canvas = self._draw_censor(yjson)
|
919 |
_yjson = json.dumps(yjson, indent=4)
|
920 |
+
# return (_canvas, _yjson)
|
921 |
+
return(_canvas)
|
922 |
+
# %%write -a app.py
|
923 |
+
# result from a lot of prompt AI and old fashion try and error
|
924 |
|
925 |
+
# print(gradio.__version__)
|
926 |
+
import random
|
927 |
+
|
928 |
+
def say_hello(val):
|
929 |
+
return f"Hello: {val}"
|
930 |
+
def say_toxic():
|
931 |
+
return f"I am toxic"
|
932 |
+
def fetch_toxic_tweets(maxi=2):
|
933 |
+
sample_df = monty.df_toxic_data.sample(maxi)
|
934 |
+
is_true = random.choice([True, False])
|
935 |
+
c1 = "more_toxic"
|
936 |
+
if is_true:
|
937 |
+
c1 = "less_toxic"
|
938 |
+
toxic1 = sample_df[c1].iloc[0]
|
939 |
+
# toxic1 = "cat eats my homework."
|
940 |
+
return sample_df.to_html(index=False), toxic1
|
941 |
#
|
942 |
+
# define all gradio widget/components outside the block for easy to visualize the blocks structure
|
943 |
+
#
|
944 |
+
in1 = gradio.Textbox(lines=3, label="Enter Text:")
|
945 |
+
in2 = gradio.Slider(0.005, .1, value=0.02, step=.005,label="Personalize Safer Value: (larger value is less safe)")
|
946 |
+
out1 = gradio.Plot(label="Output:")
|
947 |
+
out2 = gradio.HTML(label="Real-world Toxic Posts/Tweets: *WARNING")
|
948 |
+
out3 = gradio.Textbox(lines=5, label="Output JSON:")
|
949 |
+
but1 = gradio.Button("Measure 14 Toxicity", variant="primary",size="sm")
|
950 |
+
but2 = gradio.Button("Fetch Toxic Text", variant="stop", size="sm")
|
951 |
+
#
|
952 |
+
txt1 = """
|
953 |
+
# π Welcome Friendly Text Moderation
|
954 |
+
|
955 |
+
### Identify 14 categories of text toxicity.
|
956 |
+
|
957 |
+
>The purpose of this NLP (Natural Language Processing) AI demonstration is to prevent profanity, vulgarity, hate speech, violence, sexism, and any other offensive language.
|
958 |
+
>It is **not an act of censorship**, as the final UI (User Interface) will give the reader, but not a young reader, the option to click on a label to read the toxic message.
|
959 |
+
>The goal is to create a safer and more respectful environment for you, your colleages, and your family.
|
960 |
+
---
|
961 |
+
### π΄ Helpful Instruction:
|
962 |
+
|
963 |
+
1. Enter your [harmful] message in the input box.
|
964 |
+
|
965 |
+
2. Click the "Measure 14 Toxicity" button.
|
966 |
+
3. View the result on the Donut plot.
|
967 |
+
4. (**Optional**) Click on the "Fetch Real World Toxic Dataset" below.
|
968 |
+
5. Please find below the explanation of additional options available.
|
969 |
+
"""
|
970 |
+
txt2 = """
|
971 |
+
## π» Author and Developer Notes:
|
972 |
+
---
|
973 |
+
- The demo uses the cutting-edge (2024) AI Natural Language Processing (NLP) model from OpenAI.
|
974 |
+
- It is not a Generative (GenAI) model, such as Google Gemini or GPT-4.
|
975 |
+
- The NLP understands the message context, nuance, innuendo, and not just swear words.
|
976 |
+
- We **challenge you** to trick it, i.e., write a toxic tweet or post, but our AI thinks it is safe. If you win, please send us your message.
|
977 |
+
- The 14 toxicity categories are as follows:
|
978 |
+
|
979 |
+
1. harassment
|
980 |
+
2. harassment threatening
|
981 |
+
3. harassment instructions
|
982 |
+
4. hate
|
983 |
+
5. hate threatening
|
984 |
+
6. hate instructions
|
985 |
+
7. self harm
|
986 |
+
8. self harm instructions
|
987 |
+
9. self harm intent
|
988 |
+
10. self harm minor
|
989 |
+
11. sexual
|
990 |
+
12. sexual minors
|
991 |
+
13. violence
|
992 |
+
14. violence graphic
|
993 |
+
|
994 |
+
- If the NLP model classifies the message as "safe," you can still limit the level of toxicity by using the "Personal Safe" slider.
|
995 |
+
- The smaller the personal-safe value, the stricter the limitation. It means that if you're a young or sensitive adult, you should choose a lower personal-safe value, less than 0.02, to ensure you're not exposed to harmful content.
|
996 |
+
- The color of the donut plot is as follows:
|
997 |
+
- Red is an "unsafe" message by the NLP model
|
998 |
+
- Green is a "safe" message
|
999 |
+
- Yellow is an "unsafe" message by your toxicity level
|
1000 |
+
|
1001 |
+
- The real-world dataset is from the Jigsaw Rate Severity of Toxic Comments on Kaggle. It has 30,108 records.
|
1002 |
+
- The intent is to share with Duc's friends and colleagues, but for those with nefarious intent, this Text Moderation model is governed by the GNU 3.0 License: https://www.gnu.org/licenses/gpl-3.0.en.html
|
1003 |
+
- Author: **Duc Haba, 2024**
|
1004 |
+
"""
|
1005 |
+
txt3 = """
|
1006 |
+
## π₯ WARNING: WARNING:
|
1007 |
+
---
|
1008 |
|
1009 |
+
- The following button will retrieve **real-world** offensive posts from Twitter and customer reviews from consumer companies.
|
1010 |
+
- The button will display four toxic messages at a time. **Click again** for four more random messages.
|
1011 |
+
- They contain **profanity, vulgarity, hate, violence, sexism, and other offensive language.**
|
1012 |
+
- After you fetch the toxic messages, Click on the **"Measure 14 Toxicity" button**.
|
1013 |
+
"""
|
1014 |
+
#reverse_button.click(process_text, inputs=text_input, outputs=reversed_text)
|
1015 |
+
#
|
1016 |
+
|
1017 |
+
with gradio.Blocks() as gradio_app:
|
1018 |
+
# title
|
1019 |
+
gradio.Markdown(txt1) # any html or simple mark up
|
1020 |
+
#
|
1021 |
+
# first row, has two columns 1/3 size and 2/3 size
|
1022 |
+
with gradio.Row(): # items inside rows are columns
|
1023 |
+
# left column
|
1024 |
+
with gradio.Column(scale=1): # items under columns are row, scale is 1/3 size
|
1025 |
+
# left column has two rows, text entry, and buttons
|
1026 |
+
in1.render()
|
1027 |
+
in2.render()
|
1028 |
+
but1.render()
|
1029 |
+
but1.click(monty.censor_me, inputs=[in1, in2], outputs=out1)
|
1030 |
+
|
1031 |
+
with gradio.Column(scale=2):
|
1032 |
+
out1.render()
|
1033 |
+
#
|
1034 |
+
# second row is warning text
|
1035 |
+
with gradio.Row():
|
1036 |
+
gradio.Markdown(txt3)
|
1037 |
+
|
1038 |
+
# third row is fetching toxic data
|
1039 |
+
with gradio.Row():
|
1040 |
+
with gradio.Column(scale=1):
|
1041 |
+
but2.render()
|
1042 |
+
but2.click(fetch_toxic_tweets, inputs=None, outputs=[out2, in1])
|
1043 |
+
with gradio.Column(scale=2):
|
1044 |
+
out2.render()
|
1045 |
+
|
1046 |
+
# fourth row is note text
|
1047 |
+
with gradio.Row():
|
1048 |
+
gradio.Markdown(txt2)
|
1049 |
+
# %%write -a app.py
|
1050 |
+
# open/launch it
|
1051 |
+
gradio_app.launch()
|