File size: 20,625 Bytes
16a3ff9 |
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 |
## Setup Notes
For this model, a VM with 2 T4 GPUs was used.
To get the training to work on the 2 GPUs (utilize both GPUS simultaneously), the following command was used to initiate training.
WORLD_SIZE=2 CUDA_VISIBLE_DEVICES=0,1 torchrun --nproc_per_node=2 --master_port=1234 finetune.py --base_model 'decapoda-research/llama-7b-hf' --data_path 'b-mc2/sql-create-context' --output_dir './lora-alpaca' --num_epochs 1 --micro_batch_size 16
Note 1. Micro batch size was increased from the default 4 to 16. Note that increasing it further is possible based on other training that has been performed. This was a first attempt.
Note 2. Output directory was initially lora-alpaca and then contents were moved to new folder when initializing git repository.
## Log
(sqltest) chrisdono@deep-learning-duo-t4-3:~/alpaca-lora$ WORLD_SIZE=2 CUDA_VISIBLE_DEVICES=0,1 torchrun --nproc_per_node=2 --master_port=1234 finetune.py --base_model 'decapoda-research/lla$
a-7b-hf' --data_path 'b-mc2/sql-create-context' --output_dir './lora-alpaca' --num_epochs 1 --micro_batch_size 16
WARNING:torch.distributed.run:
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your appli
cation as needed.
*****************************************
===================================BUG REPORT===================================
Welcome to bitsandbytes. For bug reports, please submit your error trace to: https://github.com/TimDettmers/bitsandbytes/issues
================================================================================
===================================BUG REPORT===================================
Welcome to bitsandbytes. For bug reports, please submit your error trace to: https://github.com/TimDettmers/bitsandbytes/issues
================================================================================
/opt/conda/envs/sqltest/lib/python3.10/site-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: /opt/conda/envs/sqltest did not contain libcudart.so as expected! Searching further path
s...
warn(msg)
CUDA SETUP: CUDA runtime path found: /usr/local/cuda/lib64/libcudart.so
CUDA SETUP: Highest compute capability among GPUs detected: 7.5
CUDA SETUP: Detected CUDA version 113
CUDA SETUP: Loading binary /opt/conda/envs/sqltest/lib/python3.10/site-packages/bitsandbytes/libbitsandbytes_cuda113.so...
/opt/conda/envs/sqltest/lib/python3.10/site-packages/bitsandbytes/cuda_setup/main.py:136: UserWarning: /opt/conda/envs/sqltest did not contain libcudart.so as expected! Searching further path
s...
warn(msg)
CUDA SETUP: CUDA runtime path found: /usr/local/cuda/lib64/libcudart.so
CUDA SETUP: Highest compute capability among GPUs detected: 7.5
CUDA SETUP: Detected CUDA version 113
CUDA SETUP: Loading binary /opt/conda/envs/sqltest/lib/python3.10/site-packages/bitsandbytes/libbitsandbytes_cuda113.so...
Training Alpaca-LoRA model with params:
base_model: decapoda-research/llama-7b-hf
data_path: b-mc2/sql-create-context
output_dir: ./lora-alpaca
batch_size: 128
micro_batch_size: 16
num_epochs: 1
learning_rate: 0.0003
cutoff_len: 256
val_set_size: 2000
lora_r: 8
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules: ['q_proj', 'v_proj']
train_on_inputs: True
add_eos_token: False
group_by_length: False
wandb_project:
wandb_run_name:
wandb_watch:
wandb_log_model:
resume_from_checkpoint: False
prompt template: alpaca
Loading checkpoint shards: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 33/33 [01:24<00:00, 2.57s/it]
Loading checkpoint shards: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 33/33 [01:24<00:00, 2.57s/it]
The tokenizer class you load from this checkpoint is not the same type as the class this function is called from. It may result in unexpected tokenization.
The tokenizer class you load from this checkpoint is 'LLaMATokenizer'.
The class this function is called from is 'LlamaTokenizer'.
The tokenizer class you load from this checkpoint is not the same type as the class this function is called from. It may result in unexpected tokenization.
The tokenizer class you load from this checkpoint is 'LLaMATokenizer'.
The class this function is called from is 'LlamaTokenizer'.
Found cached dataset json (/home/chrisdono/.cache/huggingface/datasets/b-mc2___json/b-mc2--sql-create-context-d62c31544f758e00/0.0.0/fe5dd6ea2639a6df622901539cb550cf8797e5a6b2dd7af1cf934bed8e
233e6e)
0%| | 0/1 [00:00<?, ?it/s]
Found cached dataset json (/home/chrisdono/.cache/huggingface/datasets/b-mc2___json/b-mc2--sql-create-context-d62c31544f758e00/0.0.0/fe5dd6ea2639a6df622901539cb550cf8797e5a6b2dd7af1cf934bed8e
233e6e)
100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 9.30it/s]
100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 7.83it/s]
trainable params: 4194304 || all params: 6742609920 || trainable%: 0.06220594176090199
trainable params: 4194304 || all params: 6742609920 || trainable%: 0.06220594176090199
Loading cached split indices for dataset at /home/chrisdono/.cache/huggingface/datasets/b-mc2___json/b-mc2--sql-create-context-d62c31544f758e00/0.0.0/fe5dd6ea2639a6df622901539cb550cf8797e5a6b
2dd7af1cf934bed8e233e6e/cache-5a5ac0bd39fc20e0.arrow and /home/chrisdono/.cache/huggingface/datasets/b-mc2___json/b-mc2--sql-create-context-d62c31544f758e00/0.0.0/fe5dd6ea2639a6df622901539cb5
50cf8797e5a6b2dd7af1cf934bed8e233e6e/cache-782fec259d4b8f6a.arrow
Loading cached split indices for dataset at /home/chrisdono/.cache/huggingface/datasets/b-mc2___json/b-mc2--sql-create-context-d62c31544f758e00/0.0.0/fe5dd6ea2639a6df622901539cb550cf8797e5a6b
2dd7af1cf934bed8e233e6e/cache-5a5ac0bd39fc20e0.arrow and /home/chrisdono/.cache/huggingface/datasets/b-mc2___json/b-mc2--sql-create-context-d62c31544f758e00/0.0.0/fe5dd6ea2639a6df622901539cb5
50cf8797e5a6b2dd7af1cf934bed8e233e6e/cache-782fec259d4b8f6a.arrow
{'loss': 2.7003, 'learning_rate': 2.9999999999999997e-05, 'epoch': 0.02}
{'loss': 2.566, 'learning_rate': 5.9999999999999995e-05, 'epoch': 0.03}
{'loss': 2.2648, 'learning_rate': 8.999999999999999e-05, 'epoch': 0.05}
{'loss': 1.657, 'learning_rate': 0.00011099999999999999, 'epoch': 0.07}
{'loss': 1.1599, 'learning_rate': 0.00014099999999999998, 'epoch': 0.08}
{'loss': 0.9037, 'learning_rate': 0.00017099999999999998, 'epoch': 0.1}
{'loss': 0.8137, 'learning_rate': 0.000201, 'epoch': 0.12}
{'loss': 0.7827, 'learning_rate': 0.00023099999999999998, 'epoch': 0.13}
{'loss': 0.7554, 'learning_rate': 0.000261, 'epoch': 0.15}
{'loss': 0.7357, 'learning_rate': 0.00029099999999999997, 'epoch': 0.17}
{'loss': 0.6893, 'learning_rate': 0.0002957831325301205, 'epoch': 0.18}
{'loss': 0.6606, 'learning_rate': 0.00028975903614457827, 'epoch': 0.2}
{'loss': 0.6506, 'learning_rate': 0.0002837349397590361, 'epoch': 0.22}
{'loss': 0.6462, 'learning_rate': 0.00027771084337349395, 'epoch': 0.23} [215/1857]
{'loss': 0.6315, 'learning_rate': 0.0002716867469879518, 'epoch': 0.25}
{'loss': 0.6337, 'learning_rate': 0.0002656626506024096, 'epoch': 0.27}
{'loss': 0.6223, 'learning_rate': 0.00025963855421686746, 'epoch': 0.28}
{'loss': 0.6136, 'learning_rate': 0.00025361445783132525, 'epoch': 0.3}
{'loss': 0.6198, 'learning_rate': 0.00024759036144578314, 'epoch': 0.32}
{'loss': 0.6084, 'learning_rate': 0.00024156626506024095, 'epoch': 0.33}
{'eval_loss': 0.608456552028656, 'eval_runtime': 123.856, 'eval_samples_per_second': 16.148, 'eval_steps_per_second': 1.009, 'epoch': 0.33}
{'loss': 0.6021, 'learning_rate': 0.00023554216867469876, 'epoch': 0.35}
{'loss': 0.5949, 'learning_rate': 0.0002295180722891566, 'epoch': 0.37}
{'loss': 0.5972, 'learning_rate': 0.00022349397590361444, 'epoch': 0.38}
{'loss': 0.5922, 'learning_rate': 0.00021746987951807228, 'epoch': 0.4}
{'loss': 0.5876, 'learning_rate': 0.0002114457831325301, 'epoch': 0.42}
{'loss': 0.5788, 'learning_rate': 0.00020542168674698793, 'epoch': 0.43}
{'loss': 0.5894, 'learning_rate': 0.0001993975903614458, 'epoch': 0.45}
{'loss': 0.5877, 'learning_rate': 0.0001933734939759036, 'epoch': 0.47}
{'loss': 0.5835, 'learning_rate': 0.00018734939759036142, 'epoch': 0.48}
{'loss': 0.5791, 'learning_rate': 0.00018132530120481925, 'epoch': 0.5}
{'loss': 0.5841, 'learning_rate': 0.00017530120481927712, 'epoch': 0.52}
{'loss': 0.5728, 'learning_rate': 0.00016927710843373493, 'epoch': 0.53}
{'loss': 0.569, 'learning_rate': 0.00016325301204819274, 'epoch': 0.55}
{'loss': 0.5709, 'learning_rate': 0.00015722891566265058, 'epoch': 0.57}
{'loss': 0.5762, 'learning_rate': 0.00015120481927710845, 'epoch': 0.58}
{'loss': 0.5704, 'learning_rate': 0.00014518072289156626, 'epoch': 0.6}
{'loss': 0.5661, 'learning_rate': 0.0001391566265060241, 'epoch': 0.62}
{'loss': 0.5662, 'learning_rate': 0.00013313253012048193, 'epoch': 0.63}
{'loss': 0.5674, 'learning_rate': 0.00012710843373493975, 'epoch': 0.65}
{'loss': 0.5635, 'learning_rate': 0.00012108433734939758, 'epoch': 0.67}
{'eval_loss': 0.568750262260437, 'eval_runtime': 122.9061, 'eval_samples_per_second': 16.273, 'eval_steps_per_second': 1.017, 'epoch': 0.67}
{'loss': 0.5609, 'learning_rate': 0.00011506024096385541, 'epoch': 0.69}
{'loss': 0.5724, 'learning_rate': 0.00010903614457831325, 'epoch': 0.7}
{'loss': 0.5603, 'learning_rate': 0.00010301204819277107, 'epoch': 0.72}
{'loss': 0.5599, 'learning_rate': 9.698795180722891e-05, 'epoch': 0.74}
{'loss': 0.5655, 'learning_rate': 9.096385542168674e-05, 'epoch': 0.75}
{'loss': 0.5578, 'learning_rate': 8.493975903614457e-05, 'epoch': 0.77}
{'loss': 0.5577, 'learning_rate': 7.89156626506024e-05, 'epoch': 0.79}
{'loss': 0.5606, 'learning_rate': 7.289156626506024e-05, 'epoch': 0.8}
{'loss': 0.5496, 'learning_rate': 6.686746987951806e-05, 'epoch': 0.82}
{'loss': 0.5635, 'learning_rate': 6.08433734939759e-05, 'epoch': 0.84}
{'loss': 0.5522, 'learning_rate': 5.481927710843373e-05, 'epoch': 0.85}
{'loss': 0.5572, 'learning_rate': 4.879518072289156e-05, 'epoch': 0.87}
{'loss': 0.5454, 'learning_rate': 4.2771084337349395e-05, 'epoch': 0.89}
{'loss': 0.5485, 'learning_rate': 3.6746987951807227e-05, 'epoch': 0.9}
{'loss': 0.5592, 'learning_rate': 3.072289156626506e-05, 'epoch': 0.92}
{'loss': 0.5499, 'learning_rate': 2.469879518072289e-05, 'epoch': 0.94}
{'loss': 0.55, 'learning_rate': 1.867469879518072e-05, 'epoch': 0.95}
{'loss': 0.5511, 'learning_rate': 1.2650602409638553e-05, 'epoch': 0.97}
{'loss': 0.5531, 'learning_rate': 6.626506024096385e-06, 'epoch': 0.99}
100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 598/598 [4:45:30<00:00, 27.59s/it]
{'train_runtime': 17131.1027, 'train_samples_per_second': 4.47, 'train_steps_per_second': 0.035, 'train_loss': 0.7246327424129116, 'epoch': 1.0}
100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 598/598 [4:45:30<00:00, 28.65s/it]
|