Skip to content

Add documentation for bfloat16 (git commit 648f7bfa5009484b822064d0c2… #1516

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

Merged
jeffra merged 3 commits into from
Nov 13, 2021
Merged

Add documentation for bfloat16 (git commit 648f7bfa5009484b822064d0c2… #1516

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
9fa7139
Add documentation for bfloat16 (git commit 648f7bfa5009484b822064d0c2…
manuelciosici Nov 3, 2021
8397f29
Merge branch 'master' into bfloat_docs
jeffra Nov 5, 2021
feecbe1
Merge branch 'master' into bfloat_docs
jeffra Nov 12, 2021
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
30 changes: 30 additions & 0 deletions docs/_pages/config-json.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -233,6 +233,36 @@ Example of <i>**scheduler**</i>
| ----------------------------------------------------------------------------------------------------- | ------- |
| <i>**min_loss_scale**</i> is a **fp16** parameter representing the minimum dynamic loss scale value. | `1000` |

### BFLOAT16 options

**Note:** this mode cannot be combined with the `amp` mode described below.
{: .notice--warning}

**Note:** this mode cannot be combined with the `fp16` mode described above.
{: .notice--warning}

**Note:** this mode is only compatible with ZeRO stage 2.
{: .notice--warning}

<i>**bfloat16**</i>: [dictionary]

| Description | Default |
| ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------- |
| Configuration for using [bfloat16](https://en.wikipedia.org/wiki/Bfloat16_floating-point_format) floating-point format as an alternative to FP16. BFLOAT16 requires hardware support (e.g., NVIDIA A100). An example, including the available dictionary keys is illustrated below. Training with bfloat16 does not require loss scaling. | None |

```json
"bfloat16": {
"enabled": true
}
```

<i>**bfloat16:enabled**</i>: [boolean]

| Description | Default |
| ------------------------------------------------------------------------------------------- | ------- |
| <i>**enabled**</i> is a **bfloat16** parameter indicating whether or not BFLOAT16 training enabled. | `false` |


### Automatic mixed precision (AMP) training options

**Note:** this mode cannot be combined with the `fp16` mode described above. In addition this mode is not currently compatible with ZeRO.
Expand Down