Class: LlamaModel

Properties

tokenizer

readonly tokenizer: Tokenizer;

Defined in

evaluator/LlamaModel/LlamaModel.ts:160

---

onDispose

readonly onDispose: EventRelay<void>;

Defined in

evaluator/LlamaModel/LlamaModel.ts:161

Accessors

disposed

get disposed(): boolean

Returns

boolean

Defined in

evaluator/LlamaModel/LlamaModel.ts:266

---

tokens

get tokens(): LlamaModelTokens

Returns

LlamaModelTokens

Defined in

evaluator/LlamaModel/LlamaModel.ts:270

---

filename

get filename(): undefined | string

Returns

undefined | string

Defined in

evaluator/LlamaModel/LlamaModel.ts:274

---

fileInfo

get fileInfo(): GgufFileInfo

Returns

GgufFileInfo

Defined in

evaluator/LlamaModel/LlamaModel.ts:278

---

fileInsights

get fileInsights(): GgufInsights

Returns

GgufInsights

Defined in

evaluator/LlamaModel/LlamaModel.ts:282

---

gpuLayers

get gpuLayers(): number

Number of layers offloaded to the GPU. If GPU support is disabled, this will always be 0.

Returns

number

Defined in

evaluator/LlamaModel/LlamaModel.ts:290

---

size

get size(): number

Total model size in memory in bytes

Returns

number

Defined in

evaluator/LlamaModel/LlamaModel.ts:297

---

flashAttentionSupported

get flashAttentionSupported(): boolean

Returns

boolean

Defined in

evaluator/LlamaModel/LlamaModel.ts:303

---

defaultContextFlashAttention

get defaultContextFlashAttention(): boolean

Returns

boolean

Defined in

evaluator/LlamaModel/LlamaModel.ts:307

---

trainContextSize

get trainContextSize(): number

The context size the model was trained on

Returns

number

Defined in

evaluator/LlamaModel/LlamaModel.ts:589

---

embeddingVectorSize

get embeddingVectorSize(): number

The size of an embedding vector the model can produce

Returns

number

Defined in

evaluator/LlamaModel/LlamaModel.ts:599

---

vocabularyType

get vocabularyType(): LlamaVocabularyType

Returns

LlamaVocabularyType

Defined in

evaluator/LlamaModel/LlamaModel.ts:608

Methods

dispose()

dispose(): Promise<void>

Returns

Promise<void>

Defined in

evaluator/LlamaModel/LlamaModel.ts:252

---

tokenize()

tokenize(text, specialTokens, options)

tokenize(
   text: string, 
   specialTokens?: boolean, 
   options?: "trimLeadingSpace"): Token[]

Transform text into tokens that can be fed to the model

Parameters

Parameter	Type	Description
text	string	the text to tokenize
specialTokens?	boolean	if set to true, text that correspond to special tokens will be tokenized to those tokens. For example, <s> will be tokenized to the BOS token if specialTokens is set to true, otherwise it will be tokenized to tokens that corresponds to the plaintext <s> string.
options?	"trimLeadingSpace"	additional options for tokenization. If set to "trimLeadingSpace", a leading space will be trimmed from the tokenized output if the output has an additional space at the beginning.

Returns

Token[]

Defined in

evaluator/LlamaModel/LlamaModel.ts:321

tokenize(text, specialTokens)

tokenize(text: BuiltinSpecialTokenValue, specialTokens: "builtin"): Token[]

Parameters

Parameter	Type
text	BuiltinSpecialTokenValue
specialTokens	"builtin"

Returns

Token[]

Defined in

evaluator/LlamaModel/LlamaModel.ts:322

---

detokenize()

detokenize(
   tokens: readonly Token[], 
   specialTokens?: boolean, 
   lastTokens?: readonly Token[]): string

Transform tokens into text

Parameters

Parameter	Type	Default value	Description
tokens	readonly Token[]	undefined	the tokens to detokenize.
specialTokens?	boolean	false	if set to true, special tokens will be detokenized to their corresponding token text representation. Recommended for debugging purposes only. > Note: there may be additional spaces around special tokens that were not present in the original text - this is not a bug, this is how the tokenizer is supposed to work. Defaults to false.
lastTokens?	readonly Token[]	undefined	the last few tokens that preceded the tokens to detokenize. If provided, the last few tokens will be used to determine whether a space has to be added before the current tokens or not, and apply other detokenizer-specific heuristics to provide the correct text continuation to the existing tokens. Using it may have no effect with some models, but it is still recommended.

Returns

string

Defined in

evaluator/LlamaModel/LlamaModel.ts:435

---

getTokenAttributes()

getTokenAttributes(token: Token): TokenAttributes

Parameters

Parameter	Type
token	Token

Returns

TokenAttributes

Defined in

evaluator/LlamaModel/LlamaModel.ts:456

---

isSpecialToken()

isSpecialToken(token: undefined | Token): boolean

Check whether the given token is a special token (a control-type token or a token with no normal text representation)

Parameters

Parameter	Type
token	undefined | Token

Returns

boolean

Defined in

evaluator/LlamaModel/LlamaModel.ts:467

---

iterateAllTokens()

iterateAllTokens(): Generator<Token, void, unknown>

Returns

Generator<Token, void, unknown>

Defined in

evaluator/LlamaModel/LlamaModel.ts:482

---

isEogToken()

isEogToken(token: undefined | Token): boolean

Check whether the given token is an EOG (End Of Generation) token, like EOS or EOT.

Parameters

Parameter	Type
token	undefined | Token

Returns

boolean

Defined in

evaluator/LlamaModel/LlamaModel.ts:495

---

createContext()

createContext(options: LlamaContextOptions): Promise<LlamaContext>

Parameters

Parameter	Type
options	LlamaContextOptions

Returns

Promise<LlamaContext>

Defined in

evaluator/LlamaModel/LlamaModel.ts:502

---

createEmbeddingContext()

createEmbeddingContext(options: LlamaEmbeddingContextOptions): Promise<LlamaEmbeddingContext>

Parameters

Parameter	Type
options	LlamaEmbeddingContextOptions

Returns

Promise<LlamaEmbeddingContext>

Defined in

evaluator/LlamaModel/LlamaModel.ts:516

---

getWarnings()

getWarnings(): string[]

Get warnings about the model file that would affect its usage.

These warnings include all the warnings generated by GgufInsights, but are more comprehensive.

Returns

string[]

Defined in

evaluator/LlamaModel/LlamaModel.ts:535