Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Update README_train.md #278

Closed
wants to merge 1 commit into from
Closed

Update README_train.md #278

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 7 additions & 7 deletions apps/protein_folding/helixfold/README_train.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,7 +47,7 @@ mkdir tools && mv lddt tm_score tools && chmod +x ./tools/lddt && chmod +x ./too
After installing all the above required dependencies and downloading the demo dataset, you can have a try by running `gpu_train.sh`. We provide multiple training modes in one script, which includes intial training and finetune modes on `single node, single GPU`, `single node, multiple GPUs` and `multiple nodes, multiple GPUs`. Note that you need to set `PADDLE_NNODES=number of devices` and `PADDLE_MASTER="xxx.xxx.xxx.xxx:port"` according to your network environment. The details of each parameter are included in the script `gpu_train.sh`.

```bash
sh gpu_train.sh [demo_initial_N1C1, demo_finetune_N1C1, demo_initial_N1C8, demo_finetune_N1C8, demo_initial_N8C64, demo_finetune_N8C64, demo_initial_N8C64_dp16_bp2_dap2, demo_initial_N8C64_dp32_bp1_dap2, demo_initial_N8C64_dp32_bp2_dap1]
bash gpu_train.sh [demo_initial_N1C1, demo_finetune_N1C1, demo_initial_N1C8, demo_finetune_N1C8, demo_initial_N8C64, demo_finetune_N8C64, demo_initial_N8C64_dp16_bp2_dap2, demo_initial_N8C64_dp32_bp1_dap2, demo_initial_N8C64_dp32_bp2_dap1]
```

Following are few examples:
Expand All @@ -56,34 +56,34 @@ Following are few examples:

* Train on a single node with 1 GPU in initial training mode:
```bash
sh gpu_train.sh demo_initial_N1C1
bash gpu_train.sh demo_initial_N1C1
```

* Train on a single node with 8 GPUs in finetune mode:
```bash
sh gpu_train.sh demo_finetune_N1C8
bash gpu_train.sh demo_finetune_N1C8
```

* Train on 8 nodes with 64 GPUs in initial training mode:
```bash
sh gpu_train.sh demo_initial_N8C64
bash gpu_train.sh demo_initial_N8C64
```

**DAP/BP/DP-DAP-BP mode**

* Train on 8 nodes with 64 GPUs, DP=16, BP=2, DAP=2 in initial training mode:
```bash
sh gpu_train.sh demo_initial_N8C64_dp16_bp2_dap2
bash gpu_train.sh demo_initial_N8C64_dp16_bp2_dap2
```

* Train on 8 nodes with 64 GPUs, DP=32, BP=1, DAP=2 in initial training mode:
```bash
sh gpu_train.sh demo_initial_N8C64_dp32_bp1_dap2
bash gpu_train.sh demo_initial_N8C64_dp32_bp1_dap2
```

* Train on 8 nodes with 64 GPUs, DP=32, BP=2, DAP=1 in initial training mode:
```bash
sh gpu_train.sh demo_initial_N8C64_dp32_bp2_dap1
bash gpu_train.sh demo_initial_N8C64_dp32_bp2_dap1
```

## Copyright
Expand Down