-
Notifications
You must be signed in to change notification settings - Fork 0
/
launch2.py
241 lines (204 loc) · 12.7 KB
/
launch2.py
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
import sys
import time
from collections import Counter
import argparse
# import boto3
import ray
from ray.runtime_env import RuntimeEnv
# from img2dataset import download
# import img2dataset
import os
import pandas as pd
from ray.job_submission import JobSubmissionClient, JobStatus
import time
# from train import GetTrainConfigurations, GetTrainConfigurations3, GetInferenceConfiguration, GetControlNetConfiguration
# from inference import create_inference_endpoint
# import utils
import ray
# from logger import job_status_db
import datetime
import uuid
# import global_config
import os
import shutil
# from logger import Logger
# from config import settings
# from logger import eta_loss_db
class JobScheduler:
def __init__(self, address=f'ray://{os.environ.get("RAY_HOST", "http://127.0.0.1")}:{os.environ.get("RAY_PORT", "8265")}',
dashboard_address = f'http://{os.environ.get("RAY_HOST", "http://127.0.0.1")}:{os.environ.get("RAY_DASHBOARD_PORT", "8265")}'
) -> None:
# if not ray.is_initialized():
# print(f'initializing sing ray address {address} dashboard address {dashboard_address}')
# ray.init(address=f'ray://{address}')
print(f'Using ray address {address} dashboard address {dashboard_address}')
self.client = JobSubmissionClient(dashboard_address)
self.jobs = {}
def schedule_job(self, config, args):
import shutil
# config["job_id"] = job_id
job_name = config["job_name"]
scriptcmd = config['cmd']
scriptfile = scriptcmd.split(" ")[1]
log_dir = f'output/logs/{job_name}/'
if args.env=="pixart":
print("copying tree ")
shutil.copytree('PixArt-alpha/', log_dir+'/')
else:
try:
os.mkdir(log_dir, exists_ok=True)
os.mkdir(log_dir+'results/')
except:
print('dir already exists')
# log_dir = log_dir = f'/opt/logs/foundation/'
if "python -m" not in scriptcmd:
shutil.copy(scriptfile, log_dir)
# script_command = f"python raycomposer_at.py"
# script_command = f"python mosaic_cifar10.py --address 127.0.0.1:8265 -n 1 --use-gpu"
# runtime_env = "training_scripts/khoya_minimal/requirements_kohya.txt"
print(f'existing command {scriptcmd}')
if args.env !="pixart":
runtime_env= { "working_dir": log_dir,
# "pip": ["git+https://github.com/rakataprime/at1.git@main#egg=llava&subdirectory=LLaVA", "setuptools", "git+https://github.com/fsspec/s3fs"
# "numpy","pandas", "fastparquet", "b2", "bs4", "fastapi", "seaborn", "matplotlib", #"torch==2.0.1",
# 'git+https://github.com/fsspec/s3fs' ,
# "shortuuid"
# ]
"conda": {"dependencies": [ "pip", {"pip": [
"git+https://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@github.com/rakataprime/at1.git@main#egg=llava&subdirectory=LLaVA" , "setuptools", "git+https://github.com/fsspec/s3fs"
,"pandas", "fastparquet", "numpy" ,
'b2',
# 'git+https://github.com/fsspec/s3fs',
'fsspec', #'s3fs',
# "git+https://github.com/rom1504/img2dataset",
"s3transfer", #"fsspec[full]==2023.10.0",
# "fastapi",
"boto3", 'bs4',
# "bs4",
"shortuuid",
# "uvicorn",
# "img2dataset",
"more-itertools"
]}]}
}# "git+https://github.co
else:
runtime_env= { "config":{"setup_timeout_seconds":3600}, "working_dir": log_dir,
# "conda": {"channels":["conda-forge"], "dependencies": [ "s5cmd", "s3fs", 'pandas', 'fastparquet', 'b2' # "pytorch" "torchvision" "torchaudio","pytorch-cuda=11.8",
# "pip" , {"pip": [
# # "git+https://github.com/rakataprime/at1.git@main#egg=pixart-alpha&subdirectory=PixArt-alpha" ,
# "setuptools", #"git+https://github.com/fsspec/s3fs",
# # "numpy" ,
# # 'fsspec', #'s3fs',
# # "git+https://github.com/rom1504/img2dataset",
# # "s3transfer", #"fsspec[full]==2023.10.0",
# # "fastapi",
# "boto3", 'bs4',
# # "bs4",
# "shortuuid",
# # "uvicorn",
# # "img2dataset",
# "more-itertools", "timm==0.6.12", "diffusers", "accelerate", "mmcv==1.7.0", "optimum",
# "tensorboard","transformers==4.26.1", "sentencepiece",
# "ftfy>=6.1.1", "beautifulsoup4", "opencv-python","einops", "cloudpathlib[s3]","xformers",
# "--index-url https://download.pytorch.org/whl/cu118", "torch", "--index-url https://download.pytorch.org/whl/cu118", "torchvision",
# ]}]}}
"pip": [ "pandas", "fastparquet", "b2", "bs4", "fastapi", "uvicorn", "torch==2.0.1", "boto3", "s3fs" , "shortuuid" , "tensorboard", "transformers==4.26.1", "sentencepiece", "ftfy>=6.1.1", "beautifulsoup4", "opencv-python","einops", "cloudpathlib[s3]","xformers",
"more-itertools", "timm==0.6.12", "diffusers", "accelerate", "mmcv==1.7.0", "optimum",]}
# "tensorboard","transformers==4.26.1", "sentencepiece",
# "ftfy>=6.1.1", "beautifulsoup4", "opencv-python","einops", "cloudpathlib[s3]","xformers", ]}
# "pip": ["git+https://github.com/rakataprime/at1.git@main#egg=llava&subdirectory=LLaVA", "pandas", "fastparquet", "b2", "bs4", "fastapi", "torch==2.0.1", "boto3", "s3fs" , "shortuuid" ]
if args.entrypoint_num_cpus >0 & args.entrypoint_memory >0 :
if args.entrypoint_num_gpus> 0:
ray_job_id = self.client.submit_job(
entrypoint=scriptcmd, # 'pip3 install -r requirements.txt;'+
entrypoint_num_cpus= max(2, args.entrypoint_num_cpus),
entrypoint_num_gpus= args.entrypoint_num_gpus,
# entrypoint_memory= args.entrypoint_memory,
runtime_env=runtime_env
)
else:
ray_job_id = self.client.submit_job(
entrypoint=scriptcmd, # 'pip3 install -r requirements.txt;'+
entrypoint_num_cpus= args.entrypoint_num_cpus,
entrypoint_memory= args.entrypoint_memory,
runtime_env=runtime_env
)
else:
ray_job_id = self.client.submit_job(
entrypoint=scriptcmd, # 'pip3 install -r requirements.txt;'+
runtime_env=runtime_env
)
# bitsandbytes 0.41.2
print(ray_job_id)
def TerminateJob(self, job_id):
# rid = self.job_tracker_db.get_rayjobid(job_id)
# if rid is None:
# return 'Invalid Job id'
# self.jobs.pop(job_id)
return self.client.stop_job(job_id)
def GetJobStatus(self, job_id):
rid = self.job_tracker_db.get_rayjobid(job_id)
if rid is None:
return 'Invalid Job id'
return self.client.get_job_status(job_id)
def MonitorJob(self, job_id):
rid = self.job_tracker_db.get_rayjobid(job_id)
if rid is None:
return 'Invalid Job id'
metrics = self.job_tracker_db.read_metrics(job_id)
return metrics
if __name__ == "__main__":
parser = argparse.ArgumentParser()
# parser.add_argument("--ray-address", default='https://provider.akash-ai.com:31506', type=str)
# parser.add_argument("--dash-address" , default='https://provider.akash-ai.com:31610', type=str)
# parser.add_argument("--ray-address", default='provider.akash-ai.com:30808', type=str)
# parser.add_argument("--dash-address" , default='http://provider.akash-ai.com:30311', type=str) #8265 default port
# parser.add_argument("--ray-address", default='provider.akash-ai.com:31856', type=str)
# parser.add_argument("--dash-address" , default='http://provider.akash-ai.com:30728', type=str) #8265 default port
# parser.add_argument("--ray-address", default='provider.akash-ai.com:30516', type=str)
# parser.add_argument("--dash-address" , default='http://provider.akash-ai.com:31816', type=str) #8265 default port
# parser.add_argument("--ray-address", default='provider.akash-ai.com:31930', type=str)
# parser.add_argument("--dash-address" , default='http://provider.akash-ai.com:32446', type=str) #8265 default port
# parser.add_argument("--ray-address", default='provider.akash-ai.com:32481', type=str)
# parser.add_argument("--dash-address" , default='http://provider.akash-ai.com:31917', type=str) #8265 default port
# parser.add_argument("--ray-address", default='provider.akash-ai.com:30140', type=str)
# parser.add_argument("--dash-address" , default='http://provider.akash-ai.com:32754', type=str) #8265 default port
# parser.add_argument("--ray-address", default='provider.akash-ai.com:31970', type=str)
# parser.add_argument("--dash-address" , default='http://provider.akash-ai.com:31031', type=str) #8265 default port
parser.add_argument("--ray-address", default='provider.akash-ai.com:30749', type=str)
parser.add_argument("--dash-address" , default='http://provider.akash-ai.com:32576', type=str) #8265 default port
parser.add_argument("--job", default='submit', type=str)
parser.add_argument("--jobname", default='test', type=str)
parser.add_argument("--cmd", default='python i2dataset.py --local --smoketest', type=str)
parser.add_argument("--env", default='llava', type=str)
parser.add_argument("--entrypoint_num_cpus", default=-1, type=int)
parser.add_argument("--entrypoint_num_gpus", default=-1, type=int)
parser.add_argument("--entrypoint_memory", default=-1, type=int)
args = parser.parse_args()
print(args)
# ray.init(address=f"ray://{args.ray_address}")
scheduler = JobScheduler(address=args.ray_address, dashboard_address=args.dash_address)
scheduler.schedule_job(config={
"job_name": args.jobname, #"test",
"cmd" : args.cmd,#"python3 i2dataset.py --smoketest "
"env": args.env,
},args= args)
# python3 launch.py --jobname create_dataset --cmd 'python i2dataset.py'
# python3 launch.py --jobname createdataset --cmd 'python i2dataset.py'
# python3 launch.py --jobname createdataset --cmd 'python i2dataset.py'
# python3 launch.py --jobname createdataset --entrypoint_num_cpus 40 --entrypoint_memory 40000 --cmd 'python i2dataset.py --local --n_processes 4'
# python -m llava.serve.cli_batch --model-path '/mnt/sabrent/llava-v1.5-7b' \
# --load-4bit \
# --max-new-tokens 128 \
# --user-prompt 'describe this image in a highly detailed manner' \
# --image-folder '/home/ray/train_data' \
# --output-csv '/home/ray/train_data'
# python3 launch.py --jobname caption --cmd 'python -m llava.serve.cli_batch --model-path "/home/ray/train_data/llava-v1.5-7b/" --load-4bit --max-new-tokens 128 --user-prompt "describe this image in a highly detailed manner" --image-folder "/home/ray/train_data" # --output-csv "/home/ray/train_data"'
# python3 launch.py --jobname caption --cmd 'python -m llava.serve.cli_batch --model-path "/home/ray/models/llava-v1.5-7b/" --max-new-tokens 128 --user-prompt "describe this image in a highly detailed manner" --image-folder "/home/ray/train_data" --output-csv "/home/ray/train_data/"'
# python3 launch.py --jobname download --cmd 'python download_models.py'
# "s3://akash-thumper-v1-training-images.s3.us-east-005.backblazeb2.com"
# python3 launch.py --jobname createdataset --entrypoint_num_cpus 40 --entrypoint_memory 40000 --cmd 'python i2dataset.py --local --n_processes 4 --out_folder s3://akash-thumper-v1-training-images.s3.us-east-005.backblazeb2.com'
# python3 launch.py --jobname createdataset --cmd 'pyt2dataset.py -n_processes 4 --out_folder s3://akash-thumper-v1-training-images.s3.us-east-005.backblazeb2.cohon im'
# python3 launch.py --jobname createdataset --entrypoint_num_cpus 140 --jobname createdataset --cmd 'b2 sync b2://akash-thumper-v1-training-images /home/ray/train_data/ --skipNewer --threads 148'
# python3 launch.py --jobname createdataset --entrypoint_num_cpus 140 --entrypoint_memory 40000 --cmd 'python i2dataset.py --local --n_processes 10 --out_folder /home/ray/s3fs-bucket1'
# python i2dataset.py --local --n_processes 120 --out_folder /home/ray/training_data