llama.rn / LlamaContext
- bench
- completion
- detokenize
- embedding
- getFormattedChat
- loadSession
- release
- saveSession
- stopCompletion
- tokenize
• new LlamaContext(«destructured»)
| Name | Type |
|---|---|
«destructured» |
NativeLlamaContext |
• gpu: boolean = false
• id: number
• model: Object = {}
| Name | Type |
|---|---|
isChatTemplateSupported? |
boolean |
• reasonNoGPU: string = ''
▸ bench(pp, tg, pl, nr): Promise<BenchResult>
| Name | Type |
|---|---|
pp |
number |
tg |
number |
pl |
number |
nr |
number |
Promise<BenchResult>
▸ completion(params, callback?): Promise<NativeCompletionResult>
| Name | Type |
|---|---|
params |
CompletionParams |
callback? |
(data: TokenData) => void |
Promise<NativeCompletionResult>
▸ detokenize(tokens): Promise<string>
| Name | Type |
|---|---|
tokens |
number[] |
Promise<string>
▸ embedding(text, params?): Promise<NativeEmbeddingResult>
| Name | Type |
|---|---|
text |
string |
params? |
NativeEmbeddingParams |
Promise<NativeEmbeddingResult>
▸ getFormattedChat(messages, template?): Promise<string>
| Name | Type |
|---|---|
messages |
RNLlamaOAICompatibleMessage[] |
template? |
string |
Promise<string>
▸ loadSession(filepath): Promise<NativeSessionLoadResult>
Load cached prompt & completion state from a file.
| Name | Type |
|---|---|
filepath |
string |
Promise<NativeSessionLoadResult>
▸ release(): Promise<void>
Promise<void>
▸ saveSession(filepath, options?): Promise<number>
Save current cached prompt & completion state to a file.
| Name | Type |
|---|---|
filepath |
string |
options? |
Object |
options.tokenSize |
number |
Promise<number>
▸ stopCompletion(): Promise<void>
Promise<void>
▸ tokenize(text): Promise<NativeTokenizeResult>
| Name | Type |
|---|---|
text |
string |
Promise<NativeTokenizeResult>