Skip to content

Latest commit

 

History

History

API

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
classes
classes
 
 
interfaces
interfaces
 
 
README.md
README.md
 
 

README.md

llama.rn

llama.rn

Table of contents

Classes

Interfaces

Type Aliases

Functions

Type Aliases

BenchResult

Ƭ BenchResult: Object

Type declaration

Name Type
modelDesc string
modelNParams number
modelSize number
ppAvg number
ppStd number
tgAvg number
tgStd number

Defined in

index.ts:138

CompletionBaseParams

Ƭ CompletionBaseParams: Object

Type declaration

Name Type
chatTemplate? string
chat_template? string
jinja? boolean
messages? RNLlamaOAICompatibleMessage[]
parallel_tool_calls? object
prompt? string
response_format? CompletionResponseFormat
tool_choice? string
tools? object

Defined in

index.ts:121

CompletionParams

Ƭ CompletionParams: Omit<NativeCompletionParams, "emit_partial_completion" | "prompt"> & CompletionBaseParams

Defined in

index.ts:132

CompletionResponseFormat

Ƭ CompletionResponseFormat: Object

Type declaration

Name Type
json_schema? { schema: object ; strict?: boolean }
json_schema.schema object
json_schema.strict? boolean
schema? object
type "text" | "json_object" | "json_schema"

Defined in

index.ts:112

ContextParams

Ƭ ContextParams: Omit<NativeContextParams, "cache_type_k" | "cache_type_v" | "pooling_type"> & { cache_type_k?: "f16" | "f32" | "q8_0" | "q4_0" | "q4_1" | "iq4_nl" | "q5_0" | "q5_1" ; cache_type_v?: "f16" | "f32" | "q8_0" | "q4_0" | "q4_1" | "iq4_nl" | "q5_0" | "q5_1" ; pooling_type?: "none" | "mean" | "cls" | "last" | "rank" }

Defined in

index.ts:85

EmbeddingParams

Ƭ EmbeddingParams: NativeEmbeddingParams

Defined in

index.ts:110

JinjaFormattedChatResult

Ƭ JinjaFormattedChatResult: Object

Type declaration

Name Type
additional_stops? string[]
chat_format? number
grammar? string
grammar_lazy? boolean
grammar_triggers? { at_start: boolean ; word: string }[]
preserved_tokens? string[]
prompt string

Defined in

NativeRNLlama.ts:309

NativeCompletionParams

Ƭ NativeCompletionParams: Object

Type declaration

Name Type Description
chat_format? number -
dry_allowed_length? number Tokens that extend repetition beyond this receive exponentially increasing penalty: multiplier * base ^ (length of repeating sequence before token - allowed length). Default: 2
dry_base? number Set the DRY repetition penalty base value. Default: 1.75
dry_multiplier? number Set the DRY (Don't Repeat Yourself) repetition penalty multiplier. Default: 0.0, which is disabled.
dry_penalty_last_n? number How many tokens to scan for repetitions. Default: -1, where 0 is disabled and -1 is context size.
dry_sequence_breakers? string[] Specify an array of sequence breakers for DRY sampling. Only a JSON array of strings is accepted. Default: ['\n', ':', '"', '*']
emit_partial_completion boolean -
grammar? string Set grammar for grammar-based sampling. Default: no grammar
grammar_lazy? boolean Lazy grammar sampling, trigger by grammar_triggers. Default: false
grammar_triggers? { at_start: boolean ; word: string }[] Lazy grammar triggers. Default: []
ignore_eos? boolean Ignore end of stream token and continue generating. Default: false
json_schema? string JSON schema for convert to grammar for structured JSON output. It will be override by grammar if both are set.
logit_bias? number[][] Modify the likelihood of a token appearing in the generated text completion. For example, use "logit_bias": [[15043,1.0]] to increase the likelihood of the token 'Hello', or "logit_bias": [[15043,-1.0]] to decrease its likelihood. Setting the value to false, "logit_bias": [[15043,false]] ensures that the token Hello is never produced. The tokens can also be represented as strings, e.g.[["Hello, World!",-0.5]] will reduce the likelihood of all the individual tokens that represent the string Hello, World!, just like the presence_penalty does. Default: []
min_p? number The minimum probability for a token to be considered, relative to the probability of the most likely token. Default: 0.05
mirostat? number Enable Mirostat sampling, controlling perplexity during text generation. Default: 0, where 0 is disabled, 1 is Mirostat, and 2 is Mirostat 2.0.
mirostat_eta? number Set the Mirostat learning rate, parameter eta. Default: 0.1
mirostat_tau? number Set the Mirostat target entropy, parameter tau. Default: 5.0
n_predict? number Set the maximum number of tokens to predict when generating text. Note: May exceed the set limit slightly if the last token is a partial multibyte character. When 0,no tokens will be generated but the prompt is evaluated into the cache. Default: -1, where -1 is infinity.
n_probs? number If greater than 0, the response also contains the probabilities of top N tokens for each generated token given the sampling settings. Note that for temperature < 0 the tokens are sampled greedily but token probabilities are still being calculated via a simple softmax of the logits without considering any other sampler settings. Default: 0
n_threads? number -
penalty_freq? number Repeat alpha frequency penalty. Default: 0.0, which is disabled.
penalty_last_n? number Last n tokens to consider for penalizing repetition. Default: 64, where 0 is disabled and -1 is ctx-size.
penalty_present? number Repeat alpha presence penalty. Default: 0.0, which is disabled.
penalty_repeat? number Control the repetition of token sequences in the generated text. Default: 1.0
preserved_tokens? string[] -
prompt string -
seed? number Set the random number generator (RNG) seed. Default: -1, which is a random seed.
stop? string[] Specify a JSON array of stopping strings. These words will not be included in the completion, so make sure to add them to the prompt for the next iteration. Default: []
temperature? number Adjust the randomness of the generated text. Default: 0.8
top_k? number Limit the next token selection to the K most probable tokens. Default: 40
top_p? number Limit the next token selection to a subset of tokens with a cumulative probability above a threshold P. Default: 0.95
typical_p? number Enable locally typical sampling with parameter p. Default: 1.0, which is disabled.
xtc_probability? number Set the chance for token removal via XTC sampler. Default: 0.0, which is disabled.
xtc_threshold? number Set a minimum probability threshold for tokens to be removed via XTC sampler. Default: 0.1 (> 0.5 disables XTC)

Defined in

NativeRNLlama.ts:74

NativeCompletionResult

Ƭ NativeCompletionResult: Object

Type declaration

Name Type
completion_probabilities? NativeCompletionTokenProb[]
stopped_eos boolean
stopped_limit number
stopped_word string
stopping_word string
text string
timings NativeCompletionResultTimings
tokens_cached number
tokens_evaluated number
tokens_predicted number
truncated boolean

Defined in

NativeRNLlama.ts:233

NativeCompletionResultTimings

Ƭ NativeCompletionResultTimings: Object

Type declaration

Name Type
predicted_ms number
predicted_n number
predicted_per_second number
predicted_per_token_ms number
prompt_ms number
prompt_n number
prompt_per_second number
prompt_per_token_ms number

Defined in

NativeRNLlama.ts:222

NativeCompletionTokenProb

Ƭ NativeCompletionTokenProb: Object

Type declaration

Name Type
content string
probs NativeCompletionTokenProbItem[]

Defined in

NativeRNLlama.ts:217

NativeCompletionTokenProbItem

Ƭ NativeCompletionTokenProbItem: Object

Type declaration

Name Type
prob number
tok_str string

Defined in

NativeRNLlama.ts:212

NativeContextParams

Ƭ NativeContextParams: Object

Type declaration

Name Type Description
cache_type_k? string KV cache data type for the K (Experimental in llama.cpp)
cache_type_v? string KV cache data type for the V (Experimental in llama.cpp)
chat_template? string Chat template to override the default one from the model.
embd_normalize? number -
embedding? boolean -
flash_attn? boolean Enable flash attention, only recommended in GPU device (Experimental in llama.cpp)
is_model_asset? boolean -
lora? string Single LoRA adapter path
lora_list? { path: string ; scaled?: number }[] LoRA adapter list
lora_scaled? number Single LoRA adapter scale
model string -
n_batch? number -
n_ctx? number -
n_gpu_layers? number Number of layers to store in VRAM (Currently only for iOS)
n_threads? number -
n_ubatch? number -
no_gpu_devices? boolean Skip GPU devices (iOS only)
pooling_type? number -
rope_freq_base? number -
rope_freq_scale? number -
use_mlock? boolean -
use_mmap? boolean -
use_progress_callback? boolean -
vocab_only? boolean -

Defined in

NativeRNLlama.ts:8

NativeEmbeddingParams

Ƭ NativeEmbeddingParams: Object

Type declaration

Name Type
embd_normalize? number

Defined in

NativeRNLlama.ts:4

NativeEmbeddingResult

Ƭ NativeEmbeddingResult: Object

Type declaration

Name Type
embedding number[]

Defined in

NativeRNLlama.ts:253

NativeLlamaContext

Ƭ NativeLlamaContext: Object

Type declaration

Name Type Description
androidLib? string Loaded library name for Android
contextId number -
gpu boolean -
model { chatTemplates: { llamaChat: boolean ; minja: { default: boolean ; defaultCaps: { parallelToolCalls: boolean ; systemRole: boolean ; toolCallId: boolean ; toolCalls: boolean ; toolResponses: boolean ; tools: boolean } ; toolUse: boolean ; toolUseCaps: { parallelToolCalls: boolean ; systemRole: boolean ; toolCallId: boolean ; toolCalls: boolean ; toolResponses: boolean ; tools: boolean } } } ; desc: string ; isChatTemplateSupported: boolean ; metadata: Object ; nEmbd: number ; nParams: number ; size: number } -
model.chatTemplates { llamaChat: boolean ; minja: { default: boolean ; defaultCaps: { parallelToolCalls: boolean ; systemRole: boolean ; toolCallId: boolean ; toolCalls: boolean ; toolResponses: boolean ; tools: boolean } ; toolUse: boolean ; toolUseCaps: { parallelToolCalls: boolean ; systemRole: boolean ; toolCallId: boolean ; toolCalls: boolean ; toolResponses: boolean ; tools: boolean } } } -
model.chatTemplates.llamaChat boolean -
model.chatTemplates.minja { default: boolean ; defaultCaps: { parallelToolCalls: boolean ; systemRole: boolean ; toolCallId: boolean ; toolCalls: boolean ; toolResponses: boolean ; tools: boolean } ; toolUse: boolean ; toolUseCaps: { parallelToolCalls: boolean ; systemRole: boolean ; toolCallId: boolean ; toolCalls: boolean ; toolResponses: boolean ; tools: boolean } } -
model.chatTemplates.minja.default boolean -
model.chatTemplates.minja.defaultCaps { parallelToolCalls: boolean ; systemRole: boolean ; toolCallId: boolean ; toolCalls: boolean ; toolResponses: boolean ; tools: boolean } -
model.chatTemplates.minja.defaultCaps.parallelToolCalls boolean -
model.chatTemplates.minja.defaultCaps.systemRole boolean -
model.chatTemplates.minja.defaultCaps.toolCallId boolean -
model.chatTemplates.minja.defaultCaps.toolCalls boolean -
model.chatTemplates.minja.defaultCaps.toolResponses boolean -
model.chatTemplates.minja.defaultCaps.tools boolean -
model.chatTemplates.minja.toolUse boolean -
model.chatTemplates.minja.toolUseCaps { parallelToolCalls: boolean ; systemRole: boolean ; toolCallId: boolean ; toolCalls: boolean ; toolResponses: boolean ; tools: boolean } -
model.chatTemplates.minja.toolUseCaps.parallelToolCalls boolean -
model.chatTemplates.minja.toolUseCaps.systemRole boolean -
model.chatTemplates.minja.toolUseCaps.toolCallId boolean -
model.chatTemplates.minja.toolUseCaps.toolCalls boolean -
model.chatTemplates.minja.toolUseCaps.toolResponses boolean -
model.chatTemplates.minja.toolUseCaps.tools boolean -
model.desc string -
model.isChatTemplateSupported boolean -
model.metadata Object -
model.nEmbd number -
model.nParams number -
model.size number -
reasonNoGPU string -

Defined in

NativeRNLlama.ts:257

NativeSessionLoadResult

Ƭ NativeSessionLoadResult: Object

Type declaration

Name Type
prompt string
tokens_loaded number

Defined in

NativeRNLlama.ts:299

NativeTokenizeResult

Ƭ NativeTokenizeResult: Object

Type declaration

Name Type
tokens number[]

Defined in

NativeRNLlama.ts:249

RNLlamaMessagePart

Ƭ RNLlamaMessagePart: Object

Type declaration

Name Type
text? string

Defined in

chat.ts:3

RNLlamaOAICompatibleMessage

Ƭ RNLlamaOAICompatibleMessage: Object

Type declaration

Name Type
content? string | RNLlamaMessagePart[] | any
role string

Defined in

chat.ts:7

TokenData

Ƭ TokenData: Object

Type declaration

Name Type
completion_probabilities? NativeCompletionTokenProb[]
token string

Defined in

index.ts:75

Functions

addNativeLogListener

▸ addNativeLogListener(listener): Object

Parameters

Name Type
listener (level: string, text: string) => void

Returns

Object

Name Type
remove () => void

Defined in

index.ts:372

convertJsonSchemaToGrammar

▸ convertJsonSchemaToGrammar(«destructured»): string | Promise<string>

Parameters

Name Type
«destructured» Object
› allowFetch? boolean
› dotall? boolean
› propOrder? SchemaGrammarConverterPropOrder
› schema any

Returns

string | Promise<string>

Defined in

grammar.ts:829

initLlama

▸ initLlama(«destructured», onProgress?): Promise<LlamaContext>

Parameters

Name Type
«destructured» ContextParams
onProgress? (progress: number) => void

Returns

Promise<LlamaContext>

Defined in

index.ts:412

loadLlamaModelInfo

▸ loadLlamaModelInfo(model): Promise<Object>

Parameters

Name Type
model string

Returns

Promise<Object>

Defined in

index.ts:397

releaseAllLlama

▸ releaseAllLlama(): Promise<void>

Returns

Promise<void>

Defined in

index.ts:478

setContextLimit

▸ setContextLimit(limit): Promise<void>

Parameters

Name Type
limit number

Returns

Promise<void>

Defined in

index.ts:383

toggleNativeLog

▸ toggleNativeLog(enabled): Promise<void>

Parameters

Name Type
enabled boolean

Returns

Promise<void>

Defined in

index.ts:368