90e0f4c985
feat: migrate to pure Rust backend with WGPU/Vulkan support
...
- Remove libtorch dependency and related features
- Switch to ndarray backend as default for CPU execution
- Update README to reflect new WGPU/Vulkan backend usage
- Simplify device selection to use only CPU backend
- Enable WGPU backend via feature flag for GPU acceleration
2026-03-03 22:17:25 +01:00
rasmus
6a73e7b27e
updated to the newest version of burn
2026-03-03 22:17:25 +01:00
Hermes
75f0cedd9f
Update to burn v0.14.0 and switch to .mpk model file
2026-03-03 22:17:25 +01:00
Gadersd
ccbf062514
Add custom backend to enable flash attention
2026-03-03 22:17:25 +01:00
Gadersd
167e45fc30
Use torch by default
2026-03-03 22:17:25 +01:00
Gadersd
d4afd71fda
Use wgpu by default and ndarray for convert
2026-03-03 22:17:25 +01:00
Gadersd
effd3916b5
Update Cargo.toml
...
Remove unnecessary dependency
2026-03-03 22:17:25 +01:00
Gadersd
68d4310aa8
Switch to burn 0.9.0 to gain fast model io without the need for a custom recorder
2026-03-03 22:17:25 +01:00
Gadersd
c596eab0e3
Add wgpu option
2026-03-03 22:17:25 +01:00
Gadersd
ebf7c32e88
Fix allocation bug
2026-03-03 22:17:25 +01:00
Gadersd
a84825838d
Use buffer to increase model loading/saving speed tremendously
2026-03-03 22:17:25 +01:00
Gadersd
7a8d13d23c
Add files via upload
...
Add initial project files
2026-03-03 22:17:25 +01:00
