dnrops
/
burn
mirror of https://github.com/tracel-ai/burn.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Book: Fix typos in the name of MessagePack format (#1868) 

In the Record page, there are 4 places where "Pack" is written as
"Park." Also, the fix makes them the same as the official name, which
has no space and is in CamelCase.

Co-authored-by: towerpark <t56ouhw1d@mozmail.com>
This commit is contained in:
towerpark 2024-06-11 02:30:53 +09:00 committed by GitHub
parent de5b681b18
commit 9a32e53e65
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 12 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
burn-book/src/building-blocks/record.md
View File

@ -32,16 +32,16 @@ with Gzip compression).
Recorders are independent of the backend and serialize records with precision and a format. Note Recorders are independent of the backend and serialize records with precision and a format. Note
that the format can also be in-memory, allowing you to save the records directly into bytes. that the format can also be in-memory, allowing you to save the records directly into bytes.
| Recorder | Format | Compression | | Recorder | Format | Compression |
| ---------------------- | ------------------------- | ----------- | | ---------------------- | ------------------------ | ----------- |
| DefaultFileRecorder | File - Named Message Park | None | | DefaultFileRecorder | File - Named MessagePack | None |
| NamedMpkFileRecorder | File - Named Message Park | None | | NamedMpkFileRecorder | File - Named MessagePack | None |
| NamedMpkGzFileRecorder | File - Named Message Park | Gzip | | NamedMpkGzFileRecorder | File - Named MessagePack | Gzip |
| BinFileRecorder | File - Binary | None | | BinFileRecorder | File - Binary | None |
| BinGzFileRecorder | File - Binary | Gzip | | BinGzFileRecorder | File - Binary | Gzip |
| JsonGzFileRecorder | File - Json | Gzip | | JsonGzFileRecorder | File - Json | Gzip |
| PrettyJsonFileRecorder | File - Pretty Json | Gzip | | PrettyJsonFileRecorder | File - Pretty Json | Gzip |
| BinBytesRecorder | In Memory - Binary | None | | BinBytesRecorder | In Memory - Binary | None |
Each recorder supports precision settings decoupled from the precision used for training or Each recorder supports precision settings decoupled from the precision used for training or
inference. These settings allow you to define the floating-point and integer types that will be used inference. These settings allow you to define the floating-point and integer types that will be used
@ -60,8 +60,8 @@ and deserialization; otherwise, you will encounter loading errors.
**Which recorder should you use?** **Which recorder should you use?**
- If you want fast serialization and deserialization, choose a recorder without compression. The one - If you want fast serialization and deserialization, choose a recorder without compression. The one
with the lowest file size without compression is the binary format; otherwise, the named message with the lowest file size without compression is the binary format; otherwise, the named
park could be used. MessagePack could be used.
- If you want to save models for storage, you can use compression, but avoid using the binary - If you want to save models for storage, you can use compression, but avoid using the binary
format, as it may not be backward compatible. format, as it may not be backward compatible.
- If you want to debug your model's weights, you can use the pretty JSON format. - If you want to debug your model's weights, you can use the pretty JSON format.