add macOS MPS configs

This commit is contained in:
josc146
2023-06-06 22:42:38 +08:00
parent 1ecb0b444b
commit 88a5d11e15
6 changed files with 739 additions and 567 deletions

View File

@@ -30,7 +30,7 @@ export type ApiParameters = {
frequencyPenalty: number;
}
export type Device = 'CPU' | 'CUDA' | 'Custom';
export type Device = 'CPU' | 'CUDA' | 'MPS' | 'Custom';
export type Precision = 'fp16' | 'int8' | 'fp32';
export type ModelParameters = {
@@ -52,564 +52,6 @@ export type ModelConfig = {
modelParameters: ModelParameters
}
export const defaultModelConfigs: ModelConfig[] = [
{
name: 'GPU-2G-1B5-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-1B5-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 4,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-4G-1B5-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-1B5-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-4G-3B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 24,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-4G-3B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230527-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 24,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-4G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 8,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-4G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 8,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-6G-1B5-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-1B5-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-6G-3B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-6G-3B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230527-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-6G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 18,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-6G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 18,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-8G-3B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-8G-3B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230527-ctx4096.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-8G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 27,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-8G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 27,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-10G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-10G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-12G-14B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-14B-v12-Eng98%-Other2%-20230523-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 24,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-16G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-16G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-16G-14B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-14B-v12-Eng98%-Other2%-20230523-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 37,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-18G-14B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-14B-v12-Eng98%-Other2%-20230523-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-32G-14B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-14B-v12-Eng98%-Other2%-20230523-ctx8192.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'CPU-6G-1B5-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-1B5-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-12G-3B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-12G-3B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230527-ctx4096.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-28G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-28G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
}
];
export const Configs: FC = observer(() => {
const { t } = useTranslation();
const [selectedIndex, setSelectedIndex] = React.useState(commonStore.currentModelConfigIndex);
@@ -836,6 +278,7 @@ export const Configs: FC = observer(() => {
}
}}>
<Option value="CPU">CPU</Option>
{commonStore.platform === 'darwin' && <Option value="MPS">MPS</Option>}
<Option value="CUDA">CUDA</Option>
<Option value="Custom">{t('Custom')!}</Option>
</Dropdown>
@@ -903,7 +346,8 @@ export const Configs: FC = observer(() => {
onMouseEnter={() => setDisplayStrategyImg(true)}
onMouseLeave={() => setDisplayStrategyImg(false)}
content={
<Input className="grow" placeholder="cuda:0 fp16 *20 -> cuda:1 fp16"
<Input className="grow"
placeholder={commonStore.platform != 'darwin' ? 'cuda:0 fp16 *20 -> cuda:1 fp16' : 'mps fp32'}
value={selectedConfig.modelParameters.customStrategy}
onChange={(e, data) => {
setSelectedConfigModelParams({
@@ -914,7 +358,7 @@ export const Configs: FC = observer(() => {
}
{selectedConfig.modelParameters.device == 'Custom' && <div />}
{
selectedConfig.modelParameters.device != 'CPU' &&
selectedConfig.modelParameters.device != 'CPU' && selectedConfig.modelParameters.device != 'MPS' &&
<Labeled label={t('Use Custom CUDA kernel to Accelerate')}
desc={t('Enabling this option can greatly improve inference speed, but there may be compatibility issues. If it fails to start, please turn off this option.')}
content={

View File

@@ -0,0 +1,720 @@
import { ModelConfig } from './Configs';
export const defaultModelConfigsMac: ModelConfig[] = [
{
name: 'MAC-1B5-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-1B5-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'MPS',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: false,
customStrategy: 'mps fp32'
}
},
{
name: 'MAC-3B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'MPS',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: false,
customStrategy: 'mps fp32'
}
},
{
name: 'MAC-3B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230527-ctx4096.pth',
device: 'MPS',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: false,
customStrategy: 'mps fp32'
}
},
{
name: 'CPU-6G-1B5-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-1B5-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-12G-3B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-12G-3B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230527-ctx4096.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-28G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-28G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
}
];
export const defaultModelConfigs: ModelConfig[] = [
{
name: 'GPU-2G-1B5-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-1B5-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 4,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-4G-1B5-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-1B5-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-4G-3B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 24,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-4G-3B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230527-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 24,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-4G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 8,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-4G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 8,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-6G-1B5-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-1B5-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-6G-3B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-6G-3B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230527-ctx4096.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-6G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 18,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-6G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 18,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: true,
useCustomCuda: true
}
},
{
name: 'GPU-8G-3B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-8G-3B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230527-ctx4096.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-8G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 27,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-8G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 27,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-10G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-10G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-12G-14B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-14B-v12-Eng98%-Other2%-20230523-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 24,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-16G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-16G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-16G-14B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-14B-v12-Eng98%-Other2%-20230523-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 37,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-18G-14B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-14B-v12-Eng98%-Other2%-20230523-ctx8192.pth',
device: 'CUDA',
precision: 'int8',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'GPU-32G-14B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-14B-v12-Eng98%-Other2%-20230523-ctx8192.pth',
device: 'CUDA',
precision: 'fp16',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false,
useCustomCuda: true
}
},
{
name: 'CPU-6G-1B5-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-1B5-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-12G-3B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng98%-Other2%-20230520-ctx4096.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-12G-3B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-3B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230527-ctx4096.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-28G-7B-EN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
},
{
name: 'CPU-28G-7B-CN',
apiParameters: {
apiPort: 8000,
maxResponseToken: 4100,
temperature: 1.2,
topP: 0.5,
presencePenalty: 0.4,
frequencyPenalty: 0.4
},
modelParameters: {
modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
device: 'CPU',
precision: 'fp32',
storedLayers: 41,
maxStoredLayers: 41,
enableHighPrecisionForLastLayer: false
}
}
];