Used pytorch source compiled at commit c3466da to get Ampere support.
Software versions (see environment.yml for all software versions):
In [1]: import torch
...: print("Pytorch versionοΌ")
...: print(torch.__version__)
...: print("CUDA Version: ")
...: print(torch.version.cuda)
...: print("cuDNN version is :")
...: print(torch.backends.cudnn.version())
...: print("Arch version is :")
...: print(torch._C._cuda_getArchFlags())
Pytorch versionοΌ
1.8.0a0+c3466da
CUDA Version:
11.1
cuDNN version is :
8004
Arch version is :
sm_86 sm_75
Run download_glue_data.py to download data into ./glue_data
Platform details:
- AMD Ryzen Threadripper 1900X 8-Core Processor
- 128GB G.Skill 2400Mhz RAM (F4-2400C16-16GFT)
- Ubuntu 20.04.01
- NVME SSD Samsung 960 EVO 250GB
NB: GPU Fans were set to 99% using coolgpus
Led to GPU utilisation from 90-100%.
Run with fp16=True, per_gpu_train_batch_size=384.
Run with default power settings.
β― CUDA_VISIBLE_DEVICES=0 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [01:55<00:00, 2.27it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [01:53<00:00, 2.29it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [01:54<00:00, 2.29it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [05:43<00:00, 114.38s/it]
Run with default power settings.
β― CUDA_VISIBLE_DEVICES=1 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [02:01<00:00, 2.14it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [02:01<00:00, 2.15it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [02:01<00:00, 2.14it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [06:05<00:00, 121.68s/it]
Run with:
sudo nvidia-smi -i 1 --cuda-clocks=OVERRIDE
sudo nvidia-smi -i 1 -pl 320
β― CUDA_VISIBLE_DEVICES=1 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [01:55<00:00, 2.25it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [01:55<00:00, 2.25it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [01:56<00:00, 2.25it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [05:47<00:00, 115.96s/it]
Run with fp16=False, per_gpu_train_batch_size=384.
Run with default power settings.
Pytorch defaults to TF32 being enabled on Ampere so the number below is with TF32 enabled.
β― CUDA_VISIBLE_DEVICES=0 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [02:35<00:00, 1.68it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [02:33<00:00, 1.70it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [02:34<00:00, 1.69it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [07:43<00:00, 154.39s/it]
Run with fp16=False, per_gpu_train_batch_size=384.
Run with default power settings.
Run with TF32 disabled (ie. regular FP32):
torch.backends.cuda.matmul.allow_tf32 = False
torch.backends.cudnn.allow_tf32 = Falseβ― CUDA_VISIBLE_DEVICES=0 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [04:02<00:00, 1.08it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [04:03<00:00, 1.07it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [04:02<00:00, 1.07it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [12:08<00:00, 242.92s/it]
Run with default power settings.
CUDA_VISIBLE_DEVICES=1 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [05:05<00:00, 1.17s/it]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [05:06<00:00, 1.17s/it]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [05:06<00:00, 1.17s/it]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [15:17<00:00, 305.84s/it]
Run with:
sudo nvidia-smi -i 1 --cuda-clocks=OVERRIDE
sudo nvidia-smi -i 1 -pl 320
CUDA_VISIBLE_DEVICES=1 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [04:52<00:00, 1.12s/it]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [04:54<00:00, 1.13s/it]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 261/261 [04:55<00:00, 1.13s/it]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [14:42<00:00, 294.21s/it]
Led to GPU utilisation of around 85-90%.
Training code adapted from huggingface fine tuning examples
Run with fp16=True, per_gpu_train_batch_size=32. Logging and checkpointing was disabled.
Run with default power settings.
β― CUDA_VISIBLE_DEVICES=0 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:13<00:00, 16.14it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:13<00:00, 16.13it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:12<00:00, 16.21it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [09:40<00:00, 193.35s/it]
Run with default power settings.
β― CUDA_VISIBLE_DEVICES=1 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:25<00:00, 15.23it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:23<00:00, 15.35it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:23<00:00, 15.37it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [10:12<00:00, 204.01s/it]
Run with:
sudo nvidia-smi -i 1 --cuda-clocks=OVERRIDE
sudo nvidia-smi -i 1 -pl 320
β― CUDA_VISIBLE_DEVICES=1 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:17<00:00, 15.80it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:17<00:00, 15.85it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:18<00:00, 15.76it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [09:53<00:00, 197.78s/it]
Run with fp16=False, per_gpu_train_batch_size=32.
Run with default power settings.
Pytorch defaults to TF32 being enabled on Ampere so the number below is with TF32 enabled.
β― CUDA_VISIBLE_DEVICES=0 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:59<00:00, 13.06it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:56<00:00, 13.19it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [03:58<00:00, 13.10it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [11:54<00:00, 238.24s/it]
Run with fp16=False, per_gpu_train_batch_size=32. Logging and checkpointing was disabled.
Run with default power settings.
Run with TF32 disabled (ie. regular FP32):
torch.backends.cuda.matmul.allow_tf32 = False
torch.backends.cudnn.allow_tf32 = Falseβ― CUDA_VISIBLE_DEVICES=0 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [05:35<00:00, 9.33it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [05:35<00:00, 9.31it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [05:35<00:00, 9.32it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [16:45<00:00, 335.24s/it]
Run with default power settings.
β― CUDA_VISIBLE_DEVICES=1 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [06:28<00:00, 8.04it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [06:32<00:00, 7.97it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [06:31<00:00, 7.98it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [19:32<00:00, 390.78s/it]
Run with:
sudo nvidia-smi -i 1 --cuda-clocks=OVERRIDE
sudo nvidia-smi -i 1 -pl 320
β― CUDA_VISIBLE_DEVICES=1 python 01_text_classification.py
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [06:18<00:00, 8.25it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [06:18<00:00, 8.26it/s]
Iteration: 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββ| 3125/3125 [06:20<00:00, 8.21it/s]
Epoch: 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 3/3 [18:57<00:00, 379.21s/it]
DistilBertForSequenceClassification(
(distilbert): DistilBertModel(
(embeddings): Embeddings(
(word_embeddings): Embedding(28996, 768, padding_idx=0)
(position_embeddings): Embedding(512, 768)
(LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
(dropout): Dropout(p=0.1, inplace=False)
)
(transformer): Transformer(
(layer): ModuleList(
(0): TransformerBlock(
(attention): MultiHeadSelfAttention(
(dropout): Dropout(p=0.1, inplace=False)
(q_lin): Linear(in_features=768, out_features=768, bias=True)
(k_lin): Linear(in_features=768, out_features=768, bias=True)
(v_lin): Linear(in_features=768, out_features=768, bias=True)
(out_lin): Linear(in_features=768, out_features=768, bias=True)
)
(sa_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
(ffn): FFN(
(dropout): Dropout(p=0.1, inplace=False)
(lin1): Linear(in_features=768, out_features=3072, bias=True)
(lin2): Linear(in_features=3072, out_features=768, bias=True)
)
(output_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
)
(1): TransformerBlock(
(attention): MultiHeadSelfAttention(
(dropout): Dropout(p=0.1, inplace=False)
(q_lin): Linear(in_features=768, out_features=768, bias=True)
(k_lin): Linear(in_features=768, out_features=768, bias=True)
(v_lin): Linear(in_features=768, out_features=768, bias=True)
(out_lin): Linear(in_features=768, out_features=768, bias=True)
)
(sa_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
(ffn): FFN(
(dropout): Dropout(p=0.1, inplace=False)
(lin1): Linear(in_features=768, out_features=3072, bias=True)
(lin2): Linear(in_features=3072, out_features=768, bias=True)
)
(output_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
)
(2): TransformerBlock(
(attention): MultiHeadSelfAttention(
(dropout): Dropout(p=0.1, inplace=False)
(q_lin): Linear(in_features=768, out_features=768, bias=True)
(k_lin): Linear(in_features=768, out_features=768, bias=True)
(v_lin): Linear(in_features=768, out_features=768, bias=True)
(out_lin): Linear(in_features=768, out_features=768, bias=True)
)
(sa_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
(ffn): FFN(
(dropout): Dropout(p=0.1, inplace=False)
(lin1): Linear(in_features=768, out_features=3072, bias=True)
(lin2): Linear(in_features=3072, out_features=768, bias=True)
)
(output_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
)
(3): TransformerBlock(
(attention): MultiHeadSelfAttention(
(dropout): Dropout(p=0.1, inplace=False)
(q_lin): Linear(in_features=768, out_features=768, bias=True)
(k_lin): Linear(in_features=768, out_features=768, bias=True)
(v_lin): Linear(in_features=768, out_features=768, bias=True)
(out_lin): Linear(in_features=768, out_features=768, bias=True)
)
(sa_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
(ffn): FFN(
(dropout): Dropout(p=0.1, inplace=False)
(lin1): Linear(in_features=768, out_features=3072, bias=True)
(lin2): Linear(in_features=3072, out_features=768, bias=True)
)
(output_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
)
(4): TransformerBlock(
(attention): MultiHeadSelfAttention(
(dropout): Dropout(p=0.1, inplace=False)
(q_lin): Linear(in_features=768, out_features=768, bias=True)
(k_lin): Linear(in_features=768, out_features=768, bias=True)
(v_lin): Linear(in_features=768, out_features=768, bias=True)
(out_lin): Linear(in_features=768, out_features=768, bias=True)
)
(sa_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
(ffn): FFN(
(dropout): Dropout(p=0.1, inplace=False)
(lin1): Linear(in_features=768, out_features=3072, bias=True)
(lin2): Linear(in_features=3072, out_features=768, bias=True)
)
(output_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
)
(5): TransformerBlock(
(attention): MultiHeadSelfAttention(
(dropout): Dropout(p=0.1, inplace=False)
(q_lin): Linear(in_features=768, out_features=768, bias=True)
(k_lin): Linear(in_features=768, out_features=768, bias=True)
(v_lin): Linear(in_features=768, out_features=768, bias=True)
(out_lin): Linear(in_features=768, out_features=768, bias=True)
)
(sa_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
(ffn): FFN(
(dropout): Dropout(p=0.1, inplace=False)
(lin1): Linear(in_features=768, out_features=3072, bias=True)
(lin2): Linear(in_features=3072, out_features=768, bias=True)
)
(output_layer_norm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
)
)
)
)
(pre_classifier): Linear(in_features=768, out_features=768, bias=True)
(classifier): Linear(in_features=768, out_features=3, bias=True)
(dropout): Dropout(p=0.2, inplace=False)
)