References:

  • Data Persistence
  • Create and manage a repository
  • Managing local and online repositories
  • git-lfs tutorial
In [2]:
from huggingface_hub import Repository
In [10]:
%cd /workspace/llm-playground
/workspace/llm-playground
In [4]:
import os
HF_TOKEN = os.environ.get("HUGGINGFACE_TOKEN")
In [5]:
# https://huggingface.co/datasets/utensil/storage
repo = Repository(local_dir="storage", clone_from="utensil/storage", repo_type='dataset', skip_lfs_files=True, use_auth_token=HF_TOKEN)
/workspace/llm-playground/storage is already a clone of https://huggingface.co/datasets/utensil/storage. Make sure you pull the latest changes with `repo.git_pull()`.
In [6]:
repo.git_pull()
In [7]:
!date --rfc-3339=seconds > storage/time.txt
In [8]:
!cat storage/time.txt
2023-04-17 17:09:58+00:00
In [19]:
!pwd
/workspaces/llm-playground
In [9]:
!python helper/upload.py
Working directory changed to: /workspace/llm-playground/helper/..
/workspace/llm-playground/storage is already a clone of https://huggingface.co/datasets/utensil/storage. Make sure you pull the latest changes with `repo.git_pull()`.
To https://huggingface.co/datasets/utensil/storage
   febf6ec..74662c3  main -> main

Upload succeeded.
In [11]:
!ls /workspace/code/saved_models/llama-7b-hf_alpaca/
checkpoint-117	checkpoint-195	checkpoint-273	checkpoint-351	checkpoint-78
checkpoint-156	checkpoint-234	checkpoint-312	checkpoint-39
In [12]:
!mkdir -p storage/saved_models/llama-7b-hf_alpaca/
In [13]:
!cp -r /workspace/code/saved_models/llama-7b-hf_alpaca/checkpoint-351 storage/saved_models/llama-7b-hf_alpaca/
In [14]:
!ls storage/saved_models/llama-7b-hf_alpaca/
checkpoint-351
In [16]:
!ls -lhta storage/saved_models/llama-7b-hf_alpaca/checkpoint-351
total 49M
drwxr-xr-x 2 root root  160 Apr 17 17:16 .
-rw-r--r-- 1 root root  33M Apr 17 17:16 optimizer.pt
-rw-r--r-- 1 root root  15K Apr 17 17:16 rng_state.pth
-rw-r--r-- 1 root root  557 Apr 17 17:16 scaler.pt
-rw-r--r-- 1 root root  627 Apr 17 17:16 scheduler.pt
-rw-r--r-- 1 root root 4.2K Apr 17 17:16 trainer_state.json
-rw-r--r-- 1 root root  17M Apr 17 17:16 pytorch_model.bin
-rw-r--r-- 1 root root 3.5K Apr 17 17:16 training_args.bin
drwxr-xr-x 3 root root   28 Apr 17 17:16 ..
In [17]:
!python helper/upload.py
Working directory changed to: /workspace/llm-playground/helper/..
/workspace/llm-playground/storage is already a clone of https://huggingface.co/datasets/utensil/storage. Make sure you pull the latest changes with `repo.git_pull()`.
Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/optimizer.pt:   0%| |
Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/training_args.bin:   

Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/pytorch_model.bin:   


Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/scheduler.pt:   0%| |



Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/scaler.pt:   0%| | 1.




Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/optimizer.pt:  48%|▍|

Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/optimizer.pt:  95%|▉|

Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/optimizer.pt: 45.8MB [00:04, 12.1MB/s]MB/s]To https://huggingface.co/datasets/utensil/storage
   74662c3..68e82e4  main -> main

Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/optimizer.pt: 100%|█|                 

Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/training_args.bin: 10
Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/training_args.bin: 10


Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/pytorch_model.bin: 10                      



Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/scheduler.pt: 100%|█|


Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/scheduler.pt: 100%|█|




Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/scaler.pt: 100%|█| 55



Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/scaler.pt: 100%|█| 55





Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/rng_state.pth: 100%|█




Upload file saved_models/llama-7b-hf_alpaca/checkpoint-351/rng_state.pth: 100%|█
Upload succeeded.
In [18]:
!cp -r /workspace/code/saved_models/llama-7b-hf_alpaca/checkpoint-390 storage/saved_models/llama-7b-hf_alpaca/
In [19]:
!ls -lhta storage/saved_models/llama-7b-hf_alpaca/checkpoint-390
total 49M
drwxr-xr-x 2 root root  160 Apr 17 17:34 .
-rw-r--r-- 1 root root  33M Apr 17 17:34 optimizer.pt
-rw-r--r-- 1 root root  15K Apr 17 17:34 rng_state.pth
-rw-r--r-- 1 root root  557 Apr 17 17:34 scaler.pt
-rw-r--r-- 1 root root  627 Apr 17 17:34 scheduler.pt
-rw-r--r-- 1 root root 4.7K Apr 17 17:34 trainer_state.json
-rw-r--r-- 1 root root  17M Apr 17 17:34 pytorch_model.bin
-rw-r--r-- 1 root root 3.5K Apr 17 17:34 training_args.bin
drwxr-xr-x 4 root root   50 Apr 17 17:34 ..
In [20]:
!cp /workspace/code/saved_models/llama-7b-hf_alpaca/adapter* storage/saved_models/llama-7b-hf_alpaca/
In [21]:
!ls -lhta storage/saved_models/llama-7b-hf_alpaca/
total 8.0K
drwxr-xr-x 4 root root 102 Apr 17 17:35 .
-rw-r--r-- 1 root root 350 Apr 17 17:35 adapter_config.json
-rw-r--r-- 1 root root 443 Apr 17 17:35 adapter_model.bin
drwxr-xr-x 2 root root 160 Apr 17 17:34 checkpoint-390
drwxr-xr-x 2 root root 160 Apr 17 17:16 checkpoint-351
drwxr-xr-x 3 root root  32 Apr 17 17:15 ..
In [22]:
!ls -lhta /workspace/code/saved_models/llama-7b-hf_alpaca/
total 12K
drwxr-xr-x 12 root root 4.0K Apr 17 17:33 .
-rw-r--r--  1 root root  350 Apr 17 17:33 adapter_config.json
-rw-r--r--  1 root root  443 Apr 17 17:33 adapter_model.bin
drwxr-xr-x  2 root root  160 Apr 17 17:33 checkpoint-390
drwxr-xr-x  2 root root  160 Apr 17 17:12 checkpoint-351
drwxr-xr-x  2 root root  160 Apr 17 16:51 checkpoint-312
drwxr-xr-x  2 root root  160 Apr 17 16:30 checkpoint-273
drwxr-xr-x  2 root root  160 Apr 17 16:10 checkpoint-234
drwxr-xr-x  2 root root  160 Apr 17 15:49 checkpoint-195
drwxr-xr-x  2 root root  160 Apr 17 15:28 checkpoint-156
drwxr-xr-x  2 root root  160 Apr 17 15:07 checkpoint-117
drwxr-xr-x  2 root root  160 Apr 17 14:46 checkpoint-78
drwxr-xr-x  2 root root  160 Apr 17 14:25 checkpoint-39
drwxr-xr-x  3 root root   32 Apr 17 14:05 ..
In [23]:
!python helper/upload.py
Working directory changed to: /workspace/llm-playground/helper/..
/workspace/llm-playground/storage is already a clone of https://huggingface.co/datasets/utensil/storage. Make sure you pull the latest changes with `repo.git_pull()`.
Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/optimizer.pt:   0%| |
Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/scheduler.pt:   0%| |

Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/scaler.pt:   0%| | 1.


Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/pytorch_model.bin:   



Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/rng_state.pth:   0%| 

Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/scaler.pt: 15.3MB [00:01, 16.0MB/s]
Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/scheduler.pt: 15.3MB [00:01, 16.0MB/s]



Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/rng_state.pth: 15.3MB [00:01, 16.0MB/s]


Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/optimizer.pt:  48%|▍|


Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/optimizer.pt: 45.8MB [00:05, 10.3MB/s]MB/s]To https://huggingface.co/datasets/utensil/storage
   68e82e4..3b73be1  main -> main

Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/optimizer.pt: 100%|█|                 

Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/scheduler.pt: 100%|█|                 


Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/scaler.pt: 100%|█| 55              



Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/pytorch_model.bin: 10                      




Upload file saved_models/llama-7b-hf_alpaca/checkpoint-390/rng_state.pth: 100%|█                  
Upload succeeded.
In [ ]: