Spaces:
Running
Running
Daniel Fried
commited on
Commit
•
d3c5d0d
1
Parent(s):
2e513e8
remove logging and simplify structure
Browse files- modules/app.py → app.py +0 -18
- modules/cloud_logging.py +0 -21
- start.py +1 -1
modules/app.py → app.py
RENAMED
@@ -9,10 +9,6 @@ import pprint
|
|
9 |
from huggingface_hub import Repository
|
10 |
from text_generation import Client
|
11 |
|
12 |
-
import logging
|
13 |
-
logging.basicConfig(level=logging.INFO)
|
14 |
-
import modules.cloud_logging
|
15 |
-
|
16 |
# from flask import Flask, request, render_template
|
17 |
# from flask_cors import CORS
|
18 |
# app = Flask(__name__, static_folder='static')
|
@@ -80,8 +76,6 @@ def generate(prefix, suffix=None, temperature=0.9, max_new_tokens=256, top_p=0.9
|
|
80 |
|
81 |
if suffix is not None:
|
82 |
prompt = f"{FIM_PREFIX}{prefix}{FIM_SUFFIX}{suffix}{FIM_MIDDLE}"
|
83 |
-
print("----prompt----")
|
84 |
-
print(prompt)
|
85 |
else:
|
86 |
prompt = prefix
|
87 |
output = client.generate(prompt, **generate_kwargs)
|
@@ -117,11 +111,6 @@ async def generate_maybe(info: str):
|
|
117 |
prompt = form['prompt']
|
118 |
length_limit = int(form['length'])
|
119 |
temperature = float(form['temperature'])
|
120 |
-
logging.info(json.dumps({
|
121 |
-
'length': length_limit,
|
122 |
-
'temperature': temperature,
|
123 |
-
'prompt': prompt,
|
124 |
-
}))
|
125 |
try:
|
126 |
generation = generate(prompt, temperature=temperature, max_new_tokens=length_limit, top_p=0.95, repetition_penalty=1.0)
|
127 |
if generation['truncated']:
|
@@ -131,7 +120,6 @@ async def generate_maybe(info: str):
|
|
131 |
return {'result': 'success', 'type': 'generate', 'prompt': prompt, 'text': generation['text'], 'message': message}
|
132 |
except Exception as e:
|
133 |
traceback.print_exception(*sys.exc_info())
|
134 |
-
logging.error(e)
|
135 |
return {'result': 'error', 'type': 'generate', 'prompt': prompt, 'message': f'Error: {e}.'}
|
136 |
|
137 |
@app.get('/infill')
|
@@ -147,11 +135,6 @@ async def infill_maybe(info: str):
|
|
147 |
temperature = float(form['temperature'])
|
148 |
max_retries = 1
|
149 |
extra_sentinel = True
|
150 |
-
logging.info(json.dumps({
|
151 |
-
'length': length_limit,
|
152 |
-
'temperature': temperature,
|
153 |
-
'parts_joined': '<infill>'.join(form['parts']),
|
154 |
-
}))
|
155 |
try:
|
156 |
if len(form['parts']) > 2:
|
157 |
return {'result': 'error', 'text': ''.join(form['parts']), 'type': 'infill', 'message': f"error: Only a single infill is supported!"}
|
@@ -166,7 +149,6 @@ async def infill_maybe(info: str):
|
|
166 |
# return {'result': 'success', 'prefix': prefix, 'suffix': suffix, 'text': generation['text']}
|
167 |
except Exception as e:
|
168 |
traceback.print_exception(*sys.exc_info())
|
169 |
-
logging.error(e)
|
170 |
return {'result': 'error', 'type': 'infill', 'message': f'Error: {e}.'}
|
171 |
|
172 |
|
|
|
9 |
from huggingface_hub import Repository
|
10 |
from text_generation import Client
|
11 |
|
|
|
|
|
|
|
|
|
12 |
# from flask import Flask, request, render_template
|
13 |
# from flask_cors import CORS
|
14 |
# app = Flask(__name__, static_folder='static')
|
|
|
76 |
|
77 |
if suffix is not None:
|
78 |
prompt = f"{FIM_PREFIX}{prefix}{FIM_SUFFIX}{suffix}{FIM_MIDDLE}"
|
|
|
|
|
79 |
else:
|
80 |
prompt = prefix
|
81 |
output = client.generate(prompt, **generate_kwargs)
|
|
|
111 |
prompt = form['prompt']
|
112 |
length_limit = int(form['length'])
|
113 |
temperature = float(form['temperature'])
|
|
|
|
|
|
|
|
|
|
|
114 |
try:
|
115 |
generation = generate(prompt, temperature=temperature, max_new_tokens=length_limit, top_p=0.95, repetition_penalty=1.0)
|
116 |
if generation['truncated']:
|
|
|
120 |
return {'result': 'success', 'type': 'generate', 'prompt': prompt, 'text': generation['text'], 'message': message}
|
121 |
except Exception as e:
|
122 |
traceback.print_exception(*sys.exc_info())
|
|
|
123 |
return {'result': 'error', 'type': 'generate', 'prompt': prompt, 'message': f'Error: {e}.'}
|
124 |
|
125 |
@app.get('/infill')
|
|
|
135 |
temperature = float(form['temperature'])
|
136 |
max_retries = 1
|
137 |
extra_sentinel = True
|
|
|
|
|
|
|
|
|
|
|
138 |
try:
|
139 |
if len(form['parts']) > 2:
|
140 |
return {'result': 'error', 'text': ''.join(form['parts']), 'type': 'infill', 'message': f"error: Only a single infill is supported!"}
|
|
|
149 |
# return {'result': 'success', 'prefix': prefix, 'suffix': suffix, 'text': generation['text']}
|
150 |
except Exception as e:
|
151 |
traceback.print_exception(*sys.exc_info())
|
|
|
152 |
return {'result': 'error', 'type': 'infill', 'message': f'Error: {e}.'}
|
153 |
|
154 |
|
modules/cloud_logging.py
DELETED
@@ -1,21 +0,0 @@
|
|
1 |
-
import os
|
2 |
-
def make_logging_client():
|
3 |
-
cred_filename = os.environ.get('GOOGLE_APPLICATION_CREDENTIALS')
|
4 |
-
if not cred_filename:
|
5 |
-
return None
|
6 |
-
print("cred filename:", cred_filename)
|
7 |
-
cred_string = os.environ.get('GOOGLE_APPLICATION_CREDENTIALS_STRING')
|
8 |
-
print("cred string:", bool(cred_string))
|
9 |
-
if not os.path.exists(cred_filename):
|
10 |
-
if cred_string:
|
11 |
-
print(f"writing cred string to {cred_filename}")
|
12 |
-
with open(cred_filename, 'w') as f:
|
13 |
-
f.write(cred_string)
|
14 |
-
else:
|
15 |
-
return None
|
16 |
-
from google.cloud import logging
|
17 |
-
logging_client = logging.Client()
|
18 |
-
logging_client.setup_logging()
|
19 |
-
return logging_client
|
20 |
-
|
21 |
-
logging_client = make_logging_client()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
start.py
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
import subprocess
|
2 |
|
3 |
-
subprocess.run("uvicorn
|
|
|
1 |
import subprocess
|
2 |
|
3 |
+
subprocess.run("uvicorn app:app --timeout-keep-alive 300 --host 0.0.0.0 --port 7860", shell=True)
|