node-llama-cpp 2.8.0 → 3.0.0-beta.1

package/llama/binariesGithubRelease.json

@@ -1,3 +1,3 @@
 {
-    "release": "b1492"
+    "release": "b1567"
 }

package/llama/grammars/README.md

@@ -55,7 +55,7 @@ The order of symbols in a sequence matter. For example, in `"1. " move " " move
 
 Alternatives, denoted by `|`, give different sequences that are acceptable. For example, in `move ::= pawn | nonpawn | castle`, `move` can be a `pawn` move, a `nonpawn` move, or a `castle`.
 
-Parentheses `()` can be used to group sequences, which allows for embedding alternatives in a larger rule or applying repetition and optptional symbols (below) to a sequence.
+Parentheses `()` can be used to group sequences, which allows for embedding alternatives in a larger rule or applying repetition and optional symbols (below) to a sequence.
 
 ## Repetition and Optional Symbols
 
@@ -67,7 +67,7 @@ Parentheses `()` can be used to group sequences, which allows for embedding alte
 
 Comments can be specified with `#`:
 ```
-# defines optional whitspace
+# defines optional whitespace
 ws ::= [ \t\n]+
 ```
 

package/llamaBins/mac-arm64/ggml-metal.metal

@@ -792,7 +792,7 @@ kernel void kernel_mul_mv_f32_f32(
         constant   int64_t & ne0,
         constant   int64_t & ne1,
         uint3 tgpig[[threadgroup_position_in_grid]],
-        uint tiisg[[thread_index_in_simdgroup]]) {
+        uint  tiisg[[thread_index_in_simdgroup]]) {
 
     const int64_t r0 = tgpig.x;
     const int64_t rb = tgpig.y*N_F32_F32;
@@ -844,6 +844,79 @@ kernel void kernel_mul_mv_f32_f32(
     }
 }
 
+#define N_F16_F16 4
+
+kernel void kernel_mul_mv_f16_f16(
+        device const  char * src0,
+        device const  char * src1,
+        device       float * dst,
+        constant   int64_t & ne00,
+        constant   int64_t & ne01,
+        constant   int64_t & ne02,
+        constant  uint64_t & nb00,
+        constant  uint64_t & nb01,
+        constant  uint64_t & nb02,
+        constant   int64_t & ne10,
+        constant   int64_t & ne11,
+        constant   int64_t & ne12,
+        constant  uint64_t & nb10,
+        constant  uint64_t & nb11,
+        constant  uint64_t & nb12,
+        constant   int64_t & ne0,
+        constant   int64_t & ne1,
+        uint3 tgpig[[threadgroup_position_in_grid]],
+        uint  tiisg[[thread_index_in_simdgroup]]) {
+
+    const int64_t r0 = tgpig.x;
+    const int64_t rb = tgpig.y*N_F16_F16;
+    const int64_t im = tgpig.z;
+
+    device const half * x = (device const half *) (src0 + r0*nb01 + im/(ne12/ne02)*nb02);
+
+    if (ne00 < 128) {
+        for (int row = 0; row < N_F16_F16; ++row) {
+            int r1 = rb + row;
+            if (r1 >= ne11) {
+                break;
+            }
+
+            device const half * y = (device const half *) (src1 + r1*nb11 + im*nb12);
+
+            float sumf = 0;
+            for (int i = tiisg; i < ne00; i += 32) {
+                sumf += (half) x[i] * (half) y[i];
+            }
+
+            float all_sum = simd_sum(sumf);
+            if (tiisg == 0) {
+                dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
+            }
+        }
+    } else {
+        device const half4 * x4 = (device const half4 *)x;
+        for (int row = 0; row < N_F16_F16; ++row) {
+            int r1 = rb + row;
+            if (r1 >= ne11) {
+                break;
+            }
+
+            device const half  * y  = (device const half  *) (src1 + r1*nb11 + im*nb12);
+            device const half4 * y4 = (device const half4 *) y;
+
+            float sumf = 0;
+            for (int i = tiisg; i < ne00/4; i += 32) {
+                for (int k = 0; k < 4; ++k) sumf += (half) x4[i][k] * y4[i][k];
+            }
+
+            float all_sum = simd_sum(sumf);
+            if (tiisg == 0) {
+                for (int i = 4*(ne00/4); i < ne00; ++i) all_sum += (half) x[i] * y[i];
+                dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
+            }
+        }
+    }
+}
+
 kernel void kernel_mul_mv_f16_f32_1row(
         device const  char * src0,
         device const  char * src1,
@@ -1229,6 +1302,39 @@ kernel void kernel_rope(
 template [[host_name("kernel_rope_f32")]] kernel rope_t kernel_rope<float>;
 template [[host_name("kernel_rope_f16")]] kernel rope_t kernel_rope<half>;
 
+kernel void kernel_im2col_f16(
+        device const float * x,
+        device       half * dst,
+        constant   int32_t & ofs0,
+        constant   int32_t & ofs1,
+        constant   int32_t & IW,
+        constant   int32_t & IH,
+        constant   int32_t & CHW,
+        constant   int32_t & s0,
+        constant   int32_t & s1,
+        constant   int32_t & p0,
+        constant   int32_t & p1,
+        constant   int32_t & d0,
+        constant   int32_t & d1,
+        uint3 tgpig[[threadgroup_position_in_grid]],
+        uint3  tgpg[[threadgroups_per_grid]],
+        uint3 tpitg[[thread_position_in_threadgroup]],
+        uint3   ntg[[threads_per_threadgroup]]) {
+    const int32_t iiw = tgpig[2] * s0 + tpitg[2] * d0 - p0;
+    const int32_t iih = tgpig[1] * s1 + tpitg[1] * d1 - p1;
+
+    const int32_t offset_dst =
+        (tpitg[0] * tgpg[1] * tgpg[2] + tgpig[1] * tgpg[2] + tgpig[2]) * CHW +
+        (tgpig[0] * (ntg[1] * ntg[2]) + tpitg[1] * ntg[2] + tpitg[2]);
+
+    if (iih < 0 || iih >= IH || iiw < 0 || iiw >= IW) {
+        dst[offset_dst] = 0.0f;
+    } else {
+        const int32_t offset_src = tpitg[0] * ofs0 + tgpig[0] * ofs1;
+        dst[offset_dst] = x[offset_src + iih * IW + iiw];
+    }
+}
+
 kernel void kernel_cpy_f16_f16(
         device const half * src0,
         device       half * dst,

package/llamaBins/mac-x64/ggml-metal.metal

@@ -792,7 +792,7 @@ kernel void kernel_mul_mv_f32_f32(
         constant   int64_t & ne0,
         constant   int64_t & ne1,
         uint3 tgpig[[threadgroup_position_in_grid]],
-        uint tiisg[[thread_index_in_simdgroup]]) {
+        uint  tiisg[[thread_index_in_simdgroup]]) {
 
     const int64_t r0 = tgpig.x;
     const int64_t rb = tgpig.y*N_F32_F32;
@@ -844,6 +844,79 @@ kernel void kernel_mul_mv_f32_f32(
     }
 }
 
+#define N_F16_F16 4
+
+kernel void kernel_mul_mv_f16_f16(
+        device const  char * src0,
+        device const  char * src1,
+        device       float * dst,
+        constant   int64_t & ne00,
+        constant   int64_t & ne01,
+        constant   int64_t & ne02,
+        constant  uint64_t & nb00,
+        constant  uint64_t & nb01,
+        constant  uint64_t & nb02,
+        constant   int64_t & ne10,
+        constant   int64_t & ne11,
+        constant   int64_t & ne12,
+        constant  uint64_t & nb10,
+        constant  uint64_t & nb11,
+        constant  uint64_t & nb12,
+        constant   int64_t & ne0,
+        constant   int64_t & ne1,
+        uint3 tgpig[[threadgroup_position_in_grid]],
+        uint  tiisg[[thread_index_in_simdgroup]]) {
+
+    const int64_t r0 = tgpig.x;
+    const int64_t rb = tgpig.y*N_F16_F16;
+    const int64_t im = tgpig.z;
+
+    device const half * x = (device const half *) (src0 + r0*nb01 + im/(ne12/ne02)*nb02);
+
+    if (ne00 < 128) {
+        for (int row = 0; row < N_F16_F16; ++row) {
+            int r1 = rb + row;
+            if (r1 >= ne11) {
+                break;
+            }
+
+            device const half * y = (device const half *) (src1 + r1*nb11 + im*nb12);
+
+            float sumf = 0;
+            for (int i = tiisg; i < ne00; i += 32) {
+                sumf += (half) x[i] * (half) y[i];
+            }
+
+            float all_sum = simd_sum(sumf);
+            if (tiisg == 0) {
+                dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
+            }
+        }
+    } else {
+        device const half4 * x4 = (device const half4 *)x;
+        for (int row = 0; row < N_F16_F16; ++row) {
+            int r1 = rb + row;
+            if (r1 >= ne11) {
+                break;
+            }
+
+            device const half  * y  = (device const half  *) (src1 + r1*nb11 + im*nb12);
+            device const half4 * y4 = (device const half4 *) y;
+
+            float sumf = 0;
+            for (int i = tiisg; i < ne00/4; i += 32) {
+                for (int k = 0; k < 4; ++k) sumf += (half) x4[i][k] * y4[i][k];
+            }
+
+            float all_sum = simd_sum(sumf);
+            if (tiisg == 0) {
+                for (int i = 4*(ne00/4); i < ne00; ++i) all_sum += (half) x[i] * y[i];
+                dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
+            }
+        }
+    }
+}
+
 kernel void kernel_mul_mv_f16_f32_1row(
         device const  char * src0,
         device const  char * src1,
@@ -1229,6 +1302,39 @@ kernel void kernel_rope(
 template [[host_name("kernel_rope_f32")]] kernel rope_t kernel_rope<float>;
 template [[host_name("kernel_rope_f16")]] kernel rope_t kernel_rope<half>;
 
+kernel void kernel_im2col_f16(
+        device const float * x,
+        device       half * dst,
+        constant   int32_t & ofs0,
+        constant   int32_t & ofs1,
+        constant   int32_t & IW,
+        constant   int32_t & IH,
+        constant   int32_t & CHW,
+        constant   int32_t & s0,
+        constant   int32_t & s1,
+        constant   int32_t & p0,
+        constant   int32_t & p1,
+        constant   int32_t & d0,
+        constant   int32_t & d1,
+        uint3 tgpig[[threadgroup_position_in_grid]],
+        uint3  tgpg[[threadgroups_per_grid]],
+        uint3 tpitg[[thread_position_in_threadgroup]],
+        uint3   ntg[[threads_per_threadgroup]]) {
+    const int32_t iiw = tgpig[2] * s0 + tpitg[2] * d0 - p0;
+    const int32_t iih = tgpig[1] * s1 + tpitg[1] * d1 - p1;
+
+    const int32_t offset_dst =
+        (tpitg[0] * tgpg[1] * tgpg[2] + tgpig[1] * tgpg[2] + tgpig[2]) * CHW +
+        (tgpig[0] * (ntg[1] * ntg[2]) + tpitg[1] * ntg[2] + tpitg[2]);
+
+    if (iih < 0 || iih >= IH || iiw < 0 || iiw >= IW) {
+        dst[offset_dst] = 0.0f;
+    } else {
+        const int32_t offset_src = tpitg[0] * ofs0 + tgpig[0] * ofs1;
+        dst[offset_dst] = x[offset_src + iih * IW + iiw];
+    }
+}
+
 kernel void kernel_cpy_f16_f16(
         device const half * src0,
         device       half * dst,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "node-llama-cpp",
-  "version": "2.8.0",
+  "version": "3.0.0-beta.1",
   "description": "Run AI models locally on your machine with node.js bindings for llama.cpp. Force a JSON schema on the model output on the generation level",
   "main": "dist/index.js",
   "type": "module",
@@ -97,6 +97,10 @@
   "bugs": {
     "url": "https://github.com/withcatai/node-llama-cpp/issues"
   },
+  "funding": {
+    "type": "github",
+    "url": "https://github.com/sponsors/giladgd"
+  },
   "homepage": "https://withcatai.github.io/node-llama-cpp/",
   "devDependencies": {
     "@commitlint/cli": "^17.7.1",
@@ -114,17 +118,18 @@
     "@vitest/coverage-v8": "^0.34.6",
     "eslint": "^8.46.0",
     "eslint-plugin-import": "^2.28.0",
-    "eslint-plugin-node": "github:giladgd/eslint-plugin-node#dev/giladgd/fixImportExtentionFixingInTypeScript",
+    "eslint-plugin-jsdoc": "^46.9.0",
+    "eslint-plugin-n": "^16.3.1",
     "husky": "^8.0.3",
     "rimraf": "^5.0.1",
-    "semantic-release": "^21.0.7",
+    "semantic-release": "^22.0.8",
     "ts-node": "^10.9.1",
     "tslib": "^2.6.1",
-    "typedoc": "^0.25.1",
-    "typedoc-plugin-markdown": "^4.0.0-next.22",
-    "typedoc-plugin-mdn-links": "^3.1.0",
+    "typedoc": "^0.25.3",
+    "typedoc-plugin-markdown": "4.0.0-next.30",
+    "typedoc-plugin-mdn-links": "^3.1.5",
     "typedoc-vitepress-theme": "^1.0.0-next.3",
-    "typescript": "^5.1.6",
+    "typescript": "^5.2.2",
     "vitepress": "^1.0.0-rc.20",
     "vitest": "^0.34.6",
     "zx": "^7.2.3"
@@ -138,6 +143,7 @@
     "cross-spawn": "^7.0.3",
     "env-var": "^7.3.1",
     "fs-extra": "^11.1.1",
+    "lifecycle-utils": "^1.1.3",
     "log-symbols": "^5.1.0",
     "node-addon-api": "^7.0.0",
     "octokit": "^3.1.0",

package/dist/chatWrappers/createChatWrapperByBos.d.ts

@@ -1,2 +0,0 @@
-import { LlamaChatPromptWrapper } from "./LlamaChatPromptWrapper.js";
-export declare function getChatWrapperByBos(bos: string | undefined | null): typeof LlamaChatPromptWrapper | null;

package/dist/chatWrappers/createChatWrapperByBos.js

@@ -1,14 +0,0 @@
-import { LlamaChatPromptWrapper } from "./LlamaChatPromptWrapper.js";
-import { ChatMLChatPromptWrapper } from "./ChatMLChatPromptWrapper.js";
-export function getChatWrapperByBos(bos) {
-    if (bos === "" || bos == null)
-        return null;
-    if ("<s>[INST] <<SYS>>\n".startsWith(bos)) {
-        return LlamaChatPromptWrapper;
-    }
-    else if ("<|im_start|>system\n".startsWith(bos)) {
-        return ChatMLChatPromptWrapper;
-    }
-    return null;
-}
-//# sourceMappingURL=createChatWrapperByBos.js.map

package/dist/chatWrappers/createChatWrapperByBos.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"createChatWrapperByBos.js","sourceRoot":"","sources":["../../src/chatWrappers/createChatWrapperByBos.ts"],"names":[],"mappings":"AAAA,OAAO,EAAC,sBAAsB,EAAC,MAAM,6BAA6B,CAAC;AACnE,OAAO,EAAC,uBAAuB,EAAC,MAAM,8BAA8B,CAAC;AAErE,MAAM,UAAU,mBAAmB,CAAC,GAA8B;IAC9D,IAAI,GAAG,KAAK,EAAE,IAAI,GAAG,IAAI,IAAI;QACzB,OAAO,IAAI,CAAC;IAEhB,IAAI,qBAAqB,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE;QACvC,OAAO,sBAAsB,CAAC;KACjC;SAAM,IAAI,sBAAsB,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE;QAC/C,OAAO,uBAAuB,CAAC;KAClC;IAED,OAAO,IAAI,CAAC;AAChB,CAAC"}

package/dist/llamaEvaluator/LlamaContext.d.ts

@@ -1,100 +0,0 @@
-import { Token } from "../types.js";
-import { LlamaModel } from "./LlamaModel.js";
-import { LlamaGrammarEvaluationState } from "./LlamaGrammarEvaluationState.js";
-import { LlamaGrammar } from "./LlamaGrammar.js";
-export type LlamaContextOptions = {
-    model: LlamaModel;
-    prependBos?: boolean;
-    /**
-     * @deprecated use the `grammar` option on `LlamaChatSession`'s `prompt` function
-     * or the `grammarEvaluationState` option on `LlamaContext`'s `evaluate` function instead
-     * @hidden
-     */
-    grammar?: LlamaGrammar;
-    /** If null, a random seed will be used */
-    seed?: number | null;
-    /** text context size */
-    contextSize?: number;
-    /** prompt processing batch size */
-    batchSize?: number;
-    /** use fp16 for KV cache */
-    f16Kv?: boolean;
-    /** the llama_eval() call computes all logits, not just the last one */
-    logitsAll?: boolean;
-    /** embedding mode only */
-    embedding?: boolean;
-    /** number of threads to use to evaluate tokens */
-    threads?: number;
-};
-export type LlamaContextRepeatPenalty = {
-    /** Tokens to lower the predication probability of to be the next predicted token */
-    punishTokens: Uint32Array | (() => Uint32Array);
-    /**
-     * The relative amount to lower the probability of the tokens in `punishTokens` by
-     * Defaults to `1.1`.
-     * Set to `1` to disable.
-     */
-    penalty?: number;
-    /**
-     * For n time a token is in the `punishTokens` array, lower its probability by `n * frequencyPenalty`
-     * Disabled by default (`0`).
-     * Set to a value between `0` and `1` to enable.
-     */
-    frequencyPenalty?: number;
-    /**
-     * Lower the probability of all the tokens in the `punishTokens` array by `presencePenalty`
-     * Disabled by default (`0`).
-     * Set to a value between `0` and `1` to enable.
-     */
-    presencePenalty?: number;
-};
-export declare class LlamaContext {
-    private readonly _model;
-    private readonly _ctx;
-    private readonly _prependBos;
-    private _prependTokens;
-    /**
-     * @param {LlamaContextOptions} options
-     */
-    constructor({ model, prependBos, grammar, seed, contextSize, batchSize, f16Kv, logitsAll, embedding, threads }: LlamaContextOptions);
-    encode(text: string): Uint32Array;
-    decode(tokens: Uint32Array | Token[]): string;
-    get prependBos(): boolean;
-    /**
-     * @returns {Token | null} The BOS (Beginning Of Sequence) token.
-     */
-    getBosToken(): Token | null;
-    /**
-     * @returns {Token | null} The EOS (End Of Sequence) token.
-     */
-    getEosToken(): Token | null;
-    /**
-     * @returns {Token | null} The NL (New Line) token.
-     */
-    getNlToken(): Token | null;
-    /**
-     * @returns {string | null} The BOS (Beginning Of Sequence) token as a string.
-     */
-    getBosString(): string | null;
-    /**
-     * @returns {string | null} The EOS (End Of Sequence) token as a string.
-     */
-    getEosString(): string | null;
-    /**
-     * @returns {string | null} The NL (New Line) token as a string.
-     */
-    getNlString(): string | null;
-    getContextSize(): number;
-    /**
-     * @param {Uint32Array} tokens
-     * @param {object} options
-     * @returns {AsyncGenerator<Token, void>}
-     */
-    evaluate(tokens: Uint32Array, { temperature, topK, topP, grammarEvaluationState, repeatPenalty }?: {
-        temperature?: number;
-        topK?: number;
-        topP?: number;
-        grammarEvaluationState?: LlamaGrammarEvaluationState;
-        repeatPenalty?: LlamaContextRepeatPenalty;
-    }): AsyncGenerator<Token, void>;
-}

package/dist/llamaEvaluator/LlamaContext.js

@@ -1,141 +0,0 @@
-import { removeNullFields } from "../utils/removeNullFields.js";
-import { LLAMAContext } from "./LlamaBins.js";
-export class LlamaContext {
-    _model;
-    _ctx;
-    _prependBos;
-    _prependTokens;
-    /** @internal */
-    _chatGrammar;
-    /**
-     * @param {LlamaContextOptions} options
-     */
-    constructor({ model, prependBos = true, grammar, seed = model._contextOptions.seed, contextSize = model._contextOptions.contextSize, batchSize = model._contextOptions.batchSize, f16Kv = model._contextOptions.f16Kv, logitsAll = model._contextOptions.logitsAll, embedding = model._contextOptions.embedding, threads = model._contextOptions.threads }) {
-        this._model = model;
-        this._ctx = new LLAMAContext(model._model, removeNullFields({
-            seed: seed != null ? Math.max(-1, seed) : undefined,
-            contextSize,
-            batchSize,
-            f16Kv,
-            logitsAll,
-            embedding,
-            threads
-        }));
-        this._prependBos = prependBos;
-        this._prependTokens = [];
-        this._chatGrammar = grammar;
-        if (prependBos) {
-            this._prependTokens.unshift(this._ctx.tokenBos());
-        }
-    }
-    encode(text) {
-        if (text === "")
-            return new Uint32Array();
-        return this._ctx.encode(text);
-    }
-    decode(tokens) {
-        if (tokens.length === 0)
-            return "";
-        if (tokens instanceof Uint32Array)
-            return this._ctx.decode(tokens);
-        return this._ctx.decode(Uint32Array.from(tokens));
-    }
-    get prependBos() {
-        return this._prependBos;
-    }
-    /**
-     * @returns {Token | null} The BOS (Beginning Of Sequence) token.
-     */
-    getBosToken() {
-        const bosToken = this._ctx.tokenBos();
-        if (bosToken === -1)
-            return null;
-        return bosToken;
-    }
-    /**
-     * @returns {Token | null} The EOS (End Of Sequence) token.
-     */
-    getEosToken() {
-        const eosToken = this._ctx.tokenEos();
-        if (eosToken === -1)
-            return null;
-        return eosToken;
-    }
-    /**
-     * @returns {Token | null} The NL (New Line) token.
-     */
-    getNlToken() {
-        const nlToken = this._ctx.tokenNl();
-        if (nlToken === -1)
-            return null;
-        return nlToken;
-    }
-    /**
-     * @returns {string | null} The BOS (Beginning Of Sequence) token as a string.
-     */
-    getBosString() {
-        const bosToken = this.getBosToken();
-        if (bosToken == null)
-            return null;
-        return this._ctx.getTokenString(bosToken);
-    }
-    /**
-     * @returns {string | null} The EOS (End Of Sequence) token as a string.
-     */
-    getEosString() {
-        const eosToken = this.getEosToken();
-        if (eosToken == null)
-            return null;
-        return this._ctx.getTokenString(eosToken);
-    }
-    /**
-     * @returns {string | null} The NL (New Line) token as a string.
-     */
-    getNlString() {
-        const nlToken = this.getNlToken();
-        if (nlToken == null)
-            return null;
-        return this._ctx.getTokenString(nlToken);
-    }
-    getContextSize() {
-        return this._ctx.getContextSize();
-    }
-    /**
-     * @param {Uint32Array} tokens
-     * @param {object} options
-     * @returns {AsyncGenerator<Token, void>}
-     */
-    async *evaluate(tokens, { temperature = this._model._evaluationOptions.temperature, topK = this._model._evaluationOptions.topK, topP = this._model._evaluationOptions.topP, grammarEvaluationState, repeatPenalty } = {}) {
-        let evalTokens = tokens;
-        if (this._prependTokens.length > 0) {
-            const tokenArray = this._prependTokens.concat(Array.from(tokens));
-            evalTokens = Uint32Array.from(tokenArray);
-            this._prependTokens = [];
-        }
-        if (evalTokens.length === 0)
-            return;
-        // eslint-disable-next-line no-constant-condition
-        while (true) {
-            // Evaluate to get the next token.
-            const nextToken = await this._ctx.eval(evalTokens, removeNullFields({
-                temperature,
-                topK,
-                topP,
-                repeatPenalty: repeatPenalty?.penalty,
-                repeatPenaltyTokens: repeatPenalty?.punishTokens instanceof Function
-                    ? repeatPenalty.punishTokens()
-                    : repeatPenalty?.punishTokens,
-                repeatPenaltyPresencePenalty: repeatPenalty?.presencePenalty,
-                repeatPenaltyFrequencyPenalty: repeatPenalty?.frequencyPenalty,
-                grammarEvaluationState: grammarEvaluationState?._state
-            }));
-            // the assistant finished answering
-            if (nextToken === this._ctx.tokenEos())
-                break;
-            yield nextToken;
-            // Create tokens for the next eval.
-            evalTokens = Uint32Array.from([nextToken]);
-        }
-    }
-}
-//# sourceMappingURL=LlamaContext.js.map

package/dist/llamaEvaluator/LlamaContext.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"LlamaContext.js","sourceRoot":"","sources":["../../src/llamaEvaluator/LlamaContext.ts"],"names":[],"mappings":"AAAA,OAAO,EAAC,gBAAgB,EAAC,MAAM,8BAA8B,CAAC;AAE9D,OAAO,EAAC,YAAY,EAAC,MAAM,gBAAgB,CAAC;AAiE5C,MAAM,OAAO,YAAY;IACJ,MAAM,CAAa;IACnB,IAAI,CAAe;IACnB,WAAW,CAAU;IAC9B,cAAc,CAAU;IAEhC,gBAAgB;IACA,YAAY,CAAgB;IAG5C;;OAEG;IACH,YAAmB,EACf,KAAK,EACL,UAAU,GAAG,IAAI,EACjB,OAAO,EACP,IAAI,GAAG,KAAK,CAAC,eAAe,CAAC,IAAI,EACjC,WAAW,GAAG,KAAK,CAAC,eAAe,CAAC,WAAW,EAC/C,SAAS,GAAG,KAAK,CAAC,eAAe,CAAC,SAAS,EAC3C,KAAK,GAAG,KAAK,CAAC,eAAe,CAAC,KAAK,EACnC,SAAS,GAAG,KAAK,CAAC,eAAe,CAAC,SAAS,EAC3C,SAAS,GAAG,KAAK,CAAC,eAAe,CAAC,SAAS,EAC3C,OAAO,GAAG,KAAK,CAAC,eAAe,CAAC,OAAO,EACrB;QAClB,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC;QACpB,IAAI,CAAC,IAAI,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,MAAM,EAAE,gBAAgB,CAAC;YACxD,IAAI,EAAE,IAAI,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,SAAS;YACnD,WAAW;YACX,SAAS;YACT,KAAK;YACL,SAAS;YACT,SAAS;YACT,OAAO;SACV,CAAC,CAAC,CAAC;QACJ,IAAI,CAAC,WAAW,GAAG,UAAU,CAAC;QAC9B,IAAI,CAAC,cAAc,GAAG,EAAE,CAAC;QACzB,IAAI,CAAC,YAAY,GAAG,OAAO,CAAC;QAE5B,IAAI,UAAU,EAAE;YACZ,IAAI,CAAC,cAAc,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,CAAC;SACrD;IACL,CAAC;IAEM,MAAM,CAAC,IAAY;QACtB,IAAI,IAAI,KAAK,EAAE;YACX,OAAO,IAAI,WAAW,EAAE,CAAC;QAE7B,OAAO,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;IAClC,CAAC;IAEM,MAAM,CAAC,MAA6B;QACvC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;YACnB,OAAO,EAAE,CAAC;QAEd,IAAI,MAAM,YAAY,WAAW;YAC7B,OAAO,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;QAEpC,OAAO,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;IACtD,CAAC;IAED,IAAW,UAAU;QACjB,OAAO,IAAI,CAAC,WAAW,CAAC;IAC5B,CAAC;IAED;;OAEG;IACI,WAAW;QACd,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC;QAEtC,IAAI,QAAQ,KAAK,CAAC,CAAC;YACf,OAAO,IAAI,CAAC;QAEhB,OAAO,QAAQ,CAAC;IACpB,CAAC;IAED;;OAEG;IACI,WAAW;QACd,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC;QAEtC,IAAI,QAAQ,KAAK,CAAC,CAAC;YACf,OAAO,IAAI,CAAC;QAEhB,OAAO,QAAQ,CAAC;IACpB,CAAC;IAED;;OAEG;IACI,UAAU;QACb,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC;QAEpC,IAAI,OAAO,KAAK,CAAC,CAAC;YACd,OAAO,IAAI,CAAC;QAEhB,OAAO,OAAO,CAAC;IACnB,CAAC;IAED;;OAEG;IACI,YAAY;QACf,MAAM,QAAQ,GAAG,IAAI,CAAC,WAAW,EAAE,CAAC;QAEpC,IAAI,QAAQ,IAAI,IAAI;YAChB,OAAO,IAAI,CAAC;QAEhB,OAAO,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC;IAC9C,CAAC;IAED;;OAEG;IACI,YAAY;QACf,MAAM,QAAQ,GAAG,IAAI,CAAC,WAAW,EAAE,CAAC;QAEpC,IAAI,QAAQ,IAAI,IAAI;YAChB,OAAO,IAAI,CAAC;QAEhB,OAAO,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,QAAQ,CAAC,CAAC;IAC9C,CAAC;IAED;;OAEG;IACI,WAAW;QACd,MAAM,OAAO,GAAG,IAAI,CAAC,UAAU,EAAE,CAAC;QAElC,IAAI,OAAO,IAAI,IAAI;YACf,OAAO,IAAI,CAAC;QAEhB,OAAO,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,OAAO,CAAC,CAAC;IAC7C,CAAC;IAEM,cAAc;QACjB,OAAO,IAAI,CAAC,IAAI,CAAC,cAAc,EAAE,CAAC;IACtC,CAAC;IAED;;;;OAIG;IACI,KAAK,CAAC,CAAC,QAAQ,CAAC,MAAmB,EAAE,EACxC,WAAW,GAAG,IAAI,CAAC,MAAM,CAAC,kBAAkB,CAAC,WAAW,EACxD,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,kBAAkB,CAAC,IAAI,EAC1C,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,kBAAkB,CAAC,IAAI,EAC1C,sBAAsB,EACtB,aAAa,KAIb,EAAE;QACF,IAAI,UAAU,GAAG,MAAM,CAAC;QAExB,IAAI,IAAI,CAAC,cAAc,CAAC,MAAM,GAAG,CAAC,EAAE;YAChC,MAAM,UAAU,GAAY,IAAI,CAAC,cAAc,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;YAE3E,UAAU,GAAG,WAAW,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;YAC1C,IAAI,CAAC,cAAc,GAAG,EAAE,CAAC;SAC5B;QAED,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC;YACvB,OAAO;QAEX,iDAAiD;QACjD,OAAO,IAAI,EAAE;YACT,kCAAkC;YAClC,MAAM,SAAS,GAAU,MAAM,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,gBAAgB,CAAC;gBACvE,WAAW;gBACX,IAAI;gBACJ,IAAI;gBACJ,aAAa,EAAE,aAAa,EAAE,OAAO;gBACrC,mBAAmB,EAAE,aAAa,EAAE,YAAY,YAAY,QAAQ;oBAChE,CAAC,CAAC,aAAa,CAAC,YAAY,EAAE;oBAC9B,CAAC,CAAC,aAAa,EAAE,YAAY;gBACjC,4BAA4B,EAAE,aAAa,EAAE,eAAe;gBAC5D,6BAA6B,EAAE,aAAa,EAAE,gBAAgB;gBAC9D,sBAAsB,EAAE,sBAAsB,EAAE,MAAM;aACzD,CAAC,CAAC,CAAC;YAEJ,mCAAmC;YACnC,IAAI,SAAS,KAAK,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE;gBAClC,MAAM;YAEV,MAAM,SAAS,CAAC;YAEhB,mCAAmC;YACnC,UAAU,GAAG,WAAW,CAAC,IAAI,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC;SAC9C;IACL,CAAC;CACJ"}

package/dist/utils/withLock.d.ts

@@ -1 +0,0 @@
-export declare function withLock<ReturnType>(scope: any, key: string, callback: () => Promise<ReturnType>): Promise<ReturnType>;

package/dist/utils/withLock.js

@@ -1,19 +0,0 @@
-const locks = new Map();
-export async function withLock(scope, key, callback) {
-    while (locks.get(scope)?.has(key)) {
-        await locks.get(scope)?.get(key);
-    }
-    const promise = callback();
-    if (!locks.has(scope))
-        locks.set(scope, new Map());
-    locks.get(scope).set(key, promise);
-    try {
-        return await promise;
-    }
-    finally {
-        locks.get(scope)?.delete(key);
-        if (locks.get(scope)?.size === 0)
-            locks.delete(scope);
-    }
-}
-//# sourceMappingURL=withLock.js.map

package/dist/utils/withLock.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"withLock.js","sourceRoot":"","sources":["../../src/utils/withLock.ts"],"names":[],"mappings":"AAAA,MAAM,KAAK,GAAG,IAAI,GAAG,EAAkC,CAAC;AAExD,MAAM,CAAC,KAAK,UAAU,QAAQ,CAAa,KAAU,EAAE,GAAW,EAAE,QAAmC;IACnG,OAAO,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,GAAG,CAAC,EAAE;QAC/B,MAAM,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,GAAG,CAAC,CAAC;KACpC;IAED,MAAM,OAAO,GAAG,QAAQ,EAAE,CAAC;IAE3B,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC;QACjB,KAAK,CAAC,GAAG,CAAC,KAAK,EAAE,IAAI,GAAG,EAAE,CAAC,CAAC;IAEhC,KAAK,CAAC,GAAG,CAAC,KAAK,CAAE,CAAC,GAAG,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;IAEpC,IAAI;QACA,OAAO,MAAM,OAAO,CAAC;KACxB;YAAS;QACN,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,MAAM,CAAC,GAAG,CAAC,CAAC;QAE9B,IAAI,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,IAAI,KAAK,CAAC;YAC5B,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;KAC3B;AACL,CAAC"}