theluyuan/RWKV-Runner
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

allow setting quantizedLayers of WebGPU mode; chore

Browse Source
This commit is contained in:
josc146
2024-03-01 14:23:05 +08:00
parent c9513822c9
commit 887ba06bd6
8 changed files with 46 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
frontend/src/pages/Configs.tsx
View File

@@ -331,6 +331,26 @@ const Configs: FC = observer(() => {
}} />
} />
}
{selectedConfig.modelParameters.device.startsWith('WebGPU') && <div />}
{
selectedConfig.modelParameters.device.startsWith('WebGPU') &&
<Labeled label={t('Quantized Layers')}
desc={t('Number of the neural network layers quantized with current precision, the more you quantize, the lower the VRAM usage, but the quality correspondingly decreases.')}
content={
<ValuedSlider
disabled={selectedConfig.modelParameters.precision !== 'int8' && selectedConfig.modelParameters.precision !== 'nf4'}
value={selectedConfig.modelParameters.precision === 'int8' ? (selectedConfig.modelParameters.quantizedLayers || 31) :
selectedConfig.modelParameters.precision === 'nf4' ? (selectedConfig.modelParameters.quantizedLayers || 26) :
selectedConfig.modelParameters.maxStoredLayers
} min={0}
max={selectedConfig.modelParameters.maxStoredLayers} step={1} input
onChange={(e, data) => {
setSelectedConfigModelParams({
quantizedLayers: data.value
});
}} />
} />
}
{selectedConfig.modelParameters.device.startsWith('CUDA') && <div />}
{
displayStrategyImg &&