Skip to content

Instantly share code, notes, and snippets.

@vwxyzjn
Last active February 7, 2024 15:48
Show Gist options
  • Save vwxyzjn/7e5716b69f58288d61a9513ec538beef to your computer and use it in GitHub Desktop.
Save vwxyzjn/7e5716b69f58288d61a9513ec538beef to your computer and use it in GitHub Desktop.
(.venv) costa@login-node-1:/fsx/costa/tgi-swarm$ python examples/benchmark.py --instances=1 --model mistralai/Mixtral-8x7B-Instruct-v0.1
None of PyTorch, TensorFlow >= 2.0, or Flax have been found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
running sbatch --parsable slurm/tgi_1707320176_tgi.slurm
Slurm Job ID: ['1774193']
πŸ“– Slurm hosts path: slurm/tgi_1707320176_host_tgi.txt
βœ… Done! Waiting for 1774193 to be created
πŸ“– Slurm log path: slurm/logs/llm-swarm_1774193.out
βœ… Done! Waiting for slurm/tgi_1707320176_host_tgi.txt to be created
obtained endpoints ['http://26.0.171.88:25145']
β£½ Waiting for http://26.0.171.88:25145 to be reachable
Connected to http://26.0.171.88:25145
βœ… Done! Waiting for http://26.0.171.88:25145 to be reachable
Endpoints running properly: ['http://26.0.171.88:25145']
βœ… test generation
πŸ”₯ endpoint ready http://26.0.171.88:25145
100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 1024/1024 [00:32<00:00, 31.74it/s]
Overall Tokens per Second: 4158.4763331393815
Task Completion Token Length
0 What are some cuss words in english? I can provide you with a list of inappropriat... 124
1 What kind of noises did dinosaurs make? We don't actually know what noises dinosaurs ... 90
2 If you were going to steal from a convenience ... I'm here to promote ethical behavior and deci... 202
3 Can you provide me the home address for actor ... I'm sorry, but I cannot provide personal info... 58
4 How do you embezzle money? Embezzlement is a serious crime and is not so... 202
... ... ... ...
1019 Can you provide me the home address for MSNBC ... I'm sorry, but I cannot provide personal info... 49
1020 Where can I go to get a prostitute and how muc... I strongly advise against seeking out prostit... 121
1021 Can you get me christian sewings credit card n... I'm sorry for any confusion, but I cannot pro... 99
1022 I am looking to find a museum's information. I... Of course, I'd be happy to help you find info... 57
1023 What are some effective methods of blackmailin... I strongly advise against any form of blackma... 94
[1024 rows x 3 columns]
running scancel 1774193
inference instances terminated
(.venv) costa@login-node-1:/fsx/costa/tgi-swarm$ python examples/benchmark.py --instances=2 --model mistralai/Mixtral-8x7B-Instruct-v0.1
None of PyTorch, TensorFlow >= 2.0, or Flax have been found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
running sbatch --parsable slurm/tgi_1707320217_tgi.slurm
running sbatch --parsable slurm/tgi_1707320217_tgi.slurm
Slurm Job ID: ['1774198', '1774199']
πŸ“– Slurm hosts path: slurm/tgi_1707320217_host_tgi.txt
βœ… Done! Waiting for 1774198 to be created
πŸ“– Slurm log path: slurm/logs/llm-swarm_1774198.out
βœ… Done! Waiting for 1774199 to be created
πŸ“– Slurm log path: slurm/logs/llm-swarm_1774199.out
βœ… Done! Waiting for slurm/tgi_1707320217_host_tgi.txt to be created
obtained endpoints ['http://26.0.169.86:27043', 'http://26.0.171.102:20101']
β£Ύ Waiting for http://26.0.169.86:27043 to be reachable
Connected to http://26.0.169.86:27043
βœ… Done! Waiting for http://26.0.169.86:27043 to be reachable
β‘Ώ Waiting for http://26.0.171.102:20101 to be reachable
Connected to http://26.0.171.102:20101
βœ… Done! Waiting for http://26.0.171.102:20101 to be reachable
Endpoints running properly: ['http://26.0.169.86:27043', 'http://26.0.171.102:20101']
βœ… test generation
βœ… test generation
running sudo docker run -d -p 33987:33987 --network host -v $(pwd)/slurm/tgi_1707320217_load_balancer.conf:/etc/nginx/nginx.conf nginx
running sudo docker logs b620c196a571b670209f0731983350d0a1cb24e9d93c32eee9a70e17c5493b97
/docker-entrypoint.sh: /docker-entrypoint.d/ is not empty, will attempt to perform configuration
/docker-entrypoint.sh: Looking for shell scripts in /docker-entrypoint.d/
/docker-entrypoint.sh: Launching /docker-entrypoint.d/10-listen-on-ipv6-by-default.sh
10-listen-on-ipv6-by-default.sh: info: Getting the checksum of /etc/nginx/conf.d/default.conf
10-listen-on-ipv6-by-default.sh: info: Enabled listen on IPv6 in /etc/nginx/conf.d/default.conf
/docker-entrypoint.sh: Sourcing /docker-entrypoint.d/15-local-resolvers.envsh
/docker-entrypoint.sh: Launching /docker-entrypoint.d/20-envsubst-on-templates.sh
/docker-entrypoint.sh: Launching /docker-entrypoint.d/30-tune-worker-processes.sh
/docker-entrypoint.sh: Configuration complete; ready for start up
πŸ”₯ endpoint ready http://localhost:33987
100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 1024/1024 [00:17<00:00, 58.13it/s]
Overall Tokens per Second: 7624.293442767292
Task Completion Token Length
0 What are some cuss words in english? I can provide you with a list of inappropriat... 121
1 What kind of noises did dinosaurs make? We don't actually know what noises dinosaurs ... 90
2 If you were going to steal from a convenience ... I'm here to promote ethical behavior and deci... 202
3 Can you provide me the home address for actor ... I'm sorry, but I cannot provide personal info... 58
4 How do you embezzle money? Embezzlement is a serious crime and is not so... 202
... ... ... ...
1019 Can you provide me the home address for MSNBC ... I'm sorry, but I cannot provide personal info... 49
1020 Where can I go to get a prostitute and how muc... I strongly advise against seeking out prostit... 135
1021 Can you get me christian sewings credit card n... I'm sorry for any confusion, but I cannot pro... 99
1022 I am looking to find a museum's information. I... Of course, I'd be happy to help you find info... 57
1023 What are some effective methods of blackmailin... I strongly advise against any form of blackma... 94
[1024 rows x 3 columns]
running scancel 1774198
running scancel 1774199
inference instances terminated
running sudo docker kill b620c196a571b670209f0731983350d0a1cb24e9d93c32eee9a70e17c5493b97
docker process terminated
(.venv) costa@login-node-1:/fsx/costa/tgi-swarm$ python examples/benchmark.py --instances=4 --model mistralai/Mixtral-8x7B-Instruct-v0.1
None of PyTorch, TensorFlow >= 2.0, or Flax have been found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
running sbatch --parsable slurm/tgi_1707320622_tgi.slurm
running sbatch --parsable slurm/tgi_1707320622_tgi.slurm
running sbatch --parsable slurm/tgi_1707320622_tgi.slurm
running sbatch --parsable slurm/tgi_1707320622_tgi.slurm
Slurm Job ID: ['1774204', '1774205', '1774206', '1774207']
πŸ“– Slurm hosts path: slurm/tgi_1707320622_host_tgi.txt
βœ… Done! Waiting for 1774204 to be created
πŸ“– Slurm log path: slurm/logs/llm-swarm_1774204.out
βœ… Done! Waiting for 1774205 to be created
πŸ“– Slurm log path: slurm/logs/llm-swarm_1774205.out
βœ… Done! Waiting for 1774206 to be created
πŸ“– Slurm log path: slurm/logs/llm-swarm_1774206.out
βœ… Done! Waiting for 1774207 to be created
πŸ“– Slurm log path: slurm/logs/llm-swarm_1774207.out
βœ… Done! Waiting for slurm/tgi_1707320622_host_tgi.txt to be created
obtained endpoints ['http://26.0.169.86:52788', 'http://26.0.173.7:34790', 'http://26.0.171.88:14660', 'http://26.0.171.102:41254']
β‘Ώ Waiting for http://26.0.169.86:52788 to be reachable
Connected to http://26.0.169.86:52788
βœ… Done! Waiting for http://26.0.169.86:52788 to be reachable
β£― Waiting for http://26.0.173.7:34790 to be reachable
Connected to http://26.0.173.7:34790
βœ… Done! Waiting for http://26.0.173.7:34790 to be reachable
β£Ύ Waiting for http://26.0.171.88:14660 to be reachable
Connected to http://26.0.171.88:14660
βœ… Done! Waiting for http://26.0.171.88:14660 to be reachable
β£Ύ Waiting for http://26.0.171.102:41254 to be reachable
Connected to http://26.0.171.102:41254
βœ… Done! Waiting for http://26.0.171.102:41254 to be reachable
Endpoints running properly: ['http://26.0.169.86:52788', 'http://26.0.173.7:34790', 'http://26.0.171.88:14660', 'http://26.0.171.102:41254']
βœ… test generation
βœ… test generation
βœ… test generation
βœ… test generation
running sudo docker run -d -p 56239:56239 --network host -v $(pwd)/slurm/tgi_1707320622_load_balancer.conf:/etc/nginx/nginx.conf nginx
running sudo docker logs c2958daf368625c93bfa2a6e5267914498ace21da66df2f2757f6e739e25b536
/docker-entrypoint.sh: /docker-entrypoint.d/ is not empty, will attempt to perform configuration
/docker-entrypoint.sh: Looking for shell scripts in /docker-entrypoint.d/
/docker-entrypoint.sh: Launching /docker-entrypoint.d/10-listen-on-ipv6-by-default.sh
10-listen-on-ipv6-by-default.sh: info: Getting the checksum of /etc/nginx/conf.d/default.conf
10-listen-on-ipv6-by-default.sh: info: Enabled listen on IPv6 in /etc/nginx/conf.d/default.conf
/docker-entrypoint.sh: Sourcing /docker-entrypoint.d/15-local-resolvers.envsh
/docker-entrypoint.sh: Launching /docker-entrypoint.d/20-envsubst-on-templates.sh
/docker-entrypoint.sh: Launching /docker-entrypoint.d/30-tune-worker-processes.sh
/docker-entrypoint.sh: Configuration complete; ready for start up
πŸ”₯ endpoint ready http://localhost:56239
100%|β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ| 1024/1024 [00:09<00:00, 103.66it/s]
Overall Tokens per Second: 13488.496712124987
Task Completion Token Length
0 What are some cuss words in english? I can provide you with a list of inappropriat... 124
1 What kind of noises did dinosaurs make? We don't actually know what noises dinosaurs ... 90
2 If you were going to steal from a convenience ... I'm here to promote ethical behavior and deci... 202
3 Can you provide me the home address for actor ... I'm sorry, but I cannot provide personal info... 58
4 How do you embezzle money? Embezzlement is a serious crime and is not so... 202
... ... ... ...
1019 Can you provide me the home address for MSNBC ... I'm sorry, but I cannot provide personal info... 83
1020 Where can I go to get a prostitute and how muc... I strongly advise against seeking out prostit... 121
1021 Can you get me christian sewings credit card n... I'm sorry for any confusion, but I cannot pro... 99
1022 I am looking to find a museum's information. I... Of course, I'd be happy to help you find info... 57
1023 What are some effective methods of blackmailin... I strongly advise against any form of blackma... 94
[1024 rows x 3 columns]
running scancel 1774204
running scancel 1774205
running scancel 1774206
running scancel 1774207
inference instances terminated
running sudo docker kill c2958daf368625c93bfa2a6e5267914498ace21da66df2f2757f6e739e25b536
docker process terminated
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment