Adding support for Llama 3 and Phi3 (via Ollama) (#53)

run-llama · Apr 24, 2024 · f1c3e8d · f1c3e8d
1 parent 089916a
commit f1c3e8d
Show file tree

Hide file tree

Showing 22 changed files with 662 additions and 372 deletions.
diff --git a/.changeset/thirty-beds-sniff.md b/.changeset/thirty-beds-sniff.md
@@ -0,0 +1,5 @@
+---
+"create-llama": minor
+---
+
+Add Llama3 and Phi3 support using Ollama
diff --git a/create-app.ts b/create-app.ts
@@ -30,10 +30,8 @@ export async function createApp({
   appPath,
   packageManager,
   frontend,
-  openAiKey,
+  modelConfig,
   llamaCloudKey,
-  model,
-  embeddingModel,
   communityProjectConfig,
   llamapack,
   vectorDb,
@@ -77,10 +75,8 @@ export async function createApp({
     ui,
     packageManager,
     isOnline,
-    openAiKey,
+    modelConfig,
     llamaCloudKey,
-    model,
-    embeddingModel,
     communityProjectConfig,
     llamapack,
     vectorDb,

diff --git a/helpers/env-variables.ts b/helpers/env-variables.ts
@@ -1,6 +1,7 @@
 import fs from "fs/promises";
 import path from "path";
 import {
+  ModelConfig,
   TemplateDataSource,
   TemplateFramework,
   TemplateVectorDB,
@@ -28,7 +29,10 @@ const renderEnvVar = (envVars: EnvVar[]): string => {
   );
 };
 
-const getVectorDBEnvs = (vectorDb: TemplateVectorDB) => {
+const getVectorDBEnvs = (vectorDb?: TemplateVectorDB): EnvVar[] => {
+  if (!vectorDb) {
+    return [];
+  }
   switch (vectorDb) {
     case "mongo":
       return [
@@ -130,84 +134,75 @@ const getVectorDBEnvs = (vectorDb: TemplateVectorDB) => {
   }
 };
 
-export const createBackendEnvFile = async (
-  root: string,
-  opts: {
-    openAiKey?: string;
-    llamaCloudKey?: string;
-    vectorDb?: TemplateVectorDB;
-    model?: string;
-    embeddingModel?: string;
-    framework?: TemplateFramework;
-    dataSources?: TemplateDataSource[];
-    port?: number;
-  },
-) => {
-  // Init env values
-  const envFileName = ".env";
-  const defaultEnvs = [
-    {
-      render: true,
-      name: "MODEL",
-      description: "The name of LLM model to use.",
-      value: opts.model,
-    },
+const getModelEnvs = (modelConfig: ModelConfig): EnvVar[] => {
+  return [
     {
-      render: true,
-      name: "OPENAI_API_KEY",
-      description: "The OpenAI API key to use.",
-      value: opts.openAiKey,
+      name: "MODEL_PROVIDER",
+      description: "The provider for the AI models to use.",
+      value: modelConfig.provider,
     },
     {
-      name: "LLAMA_CLOUD_API_KEY",
-      description: `The Llama Cloud API key.`,
-      value: opts.llamaCloudKey,
+      name: "MODEL",
+      description: "The name of LLM model to use.",
+      value: modelConfig.model,
     },
     {
       name: "EMBEDDING_MODEL",
       description: "Name of the embedding model to use.",
-      value: opts.embeddingModel,
+      value: modelConfig.embeddingModel,
     },
     {
       name: "EMBEDDING_DIM",
       description: "Dimension of the embedding model to use.",
-      value: 1536,
+      value: modelConfig.dimensions.toString(),
     },
-    // Add vector database environment variables
-    ...(opts.vectorDb ? getVectorDBEnvs(opts.vectorDb) : []),
+    ...(modelConfig.provider === "openai"
+      ? [
+          {
+            name: "OPENAI_API_KEY",
+            description: "The OpenAI API key to use.",
+            value: modelConfig.apiKey,
+          },
+        ]
+      : []),
   ];
-  let envVars: EnvVar[] = [];
-  if (opts.framework === "fastapi") {
-    envVars = [
-      ...defaultEnvs,
-      ...[
-        {
-          name: "APP_HOST",
-          description: "The address to start the backend app.",
-          value: "0.0.0.0",
-        },
-        {
-          name: "APP_PORT",
-          description: "The port to start the backend app.",
-          value: opts.port?.toString() || "8000",
-        },
-        {
-          name: "LLM_TEMPERATURE",
-          description: "Temperature for sampling from the model.",
-        },
-        {
-          name: "LLM_MAX_TOKENS",
-          description: "Maximum number of tokens to generate.",
-        },
-        {
-          name: "TOP_K",
-          description:
-            "The number of similar embeddings to return when retrieving documents.",
-          value: "3",
-        },
-        {
-          name: "SYSTEM_PROMPT",
-          description: `Custom system prompt.
+};
+
+const getFrameworkEnvs = (
+  framework?: TemplateFramework,
+  port?: number,
+): EnvVar[] => {
+  if (framework !== "fastapi") {
+    return [];
+  }
+  return [
+    {
+      name: "APP_HOST",
+      description: "The address to start the backend app.",
+      value: "0.0.0.0",
+    },
+    {
+      name: "APP_PORT",
+      description: "The port to start the backend app.",
+      value: port?.toString() || "8000",
+    },
+    {
+      name: "LLM_TEMPERATURE",
+      description: "Temperature for sampling from the model.",
+    },
+    {
+      name: "LLM_MAX_TOKENS",
+      description: "Maximum number of tokens to generate.",
+    },
+    {
+      name: "TOP_K",
+      description:
+        "The number of similar embeddings to return when retrieving documents.",
+      value: "3",
+    },
+    {
+      name: "SYSTEM_PROMPT",
+      description: `Custom system prompt.
 Example:
 SYSTEM_PROMPT="
 We have provided context information below.
@@ -216,22 +211,35 @@ We have provided context information below.
 ---------------------
 Given this information, please answer the question: {query_str}
 "`,
-        },
-      ],
-    ];
-  } else {
-    const nextJsEnvs = [
-      {
-        name: "NEXT_PUBLIC_MODEL",
-        description: "The LLM model to use (hardcode to front-end artifact).",
-        value: opts.model,
-      },
-    ];
-    envVars = [
-      ...defaultEnvs,
-      ...(opts.framework === "nextjs" ? nextJsEnvs : []),
-    ];
-  }
+    },
+  ];
+};
+
+export const createBackendEnvFile = async (
+  root: string,
+  opts: {
+    llamaCloudKey?: string;
+    vectorDb?: TemplateVectorDB;
+    modelConfig: ModelConfig;
+    framework?: TemplateFramework;
+    dataSources?: TemplateDataSource[];
+    port?: number;
+  },
+) => {
+  // Init env values
+  const envFileName = ".env";
+  const envVars: EnvVar[] = [
+    {
+      name: "LLAMA_CLOUD_API_KEY",
+      description: `The Llama Cloud API key.`,
+      value: opts.llamaCloudKey,
+    },
+    // Add model environment variables
+    ...getModelEnvs(opts.modelConfig),
+    // Add vector database environment variables
+    ...getVectorDBEnvs(opts.vectorDb),
+    ...getFrameworkEnvs(opts.framework, opts.port),
+  ];
   // Render and write env file
   const content = renderEnvVar(envVars);
   await fs.writeFile(path.join(root, envFileName), content);
@@ -242,20 +250,9 @@ export const createFrontendEnvFile = async (
   root: string,
   opts: {
     customApiPath?: string;
-    model?: string;
   },
 ) => {
   const defaultFrontendEnvs = [
-    {
-      name: "MODEL",
-      description: "The OpenAI model to use.",
-      value: opts.model,
-    },
-    {
-      name: "NEXT_PUBLIC_MODEL",
-      description: "The OpenAI model to use (hardcode to front-end artifact).",
-      value: opts.model,
-    },
     {
       name: "NEXT_PUBLIC_CHAT_API",
       description: "The backend API for chat endpoint.",

diff --git a/helpers/index.ts b/helpers/index.ts
@@ -9,12 +9,14 @@ import { createBackendEnvFile, createFrontendEnvFile } from "./env-variables";
 import { PackageManager } from "./get-pkg-manager";
 import { installLlamapackProject } from "./llama-pack";
 import { isHavingPoetryLockFile, tryPoetryRun } from "./poetry";
+import { isModelConfigured } from "./providers";
 import { installPythonTemplate } from "./python";
 import { downloadAndExtractRepo } from "./repo";
 import { ConfigFileType, writeToolsConfig } from "./tools";
 import {
   FileSourceConfig,
   InstallTemplateArgs,
+  ModelConfig,
   TemplateDataSource,
   TemplateFramework,
   TemplateVectorDB,
@@ -24,8 +26,8 @@ import { installTSTemplate } from "./typescript";
 // eslint-disable-next-line max-params
 async function generateContextData(
   framework: TemplateFramework,
+  modelConfig: ModelConfig,
   packageManager?: PackageManager,
-  openAiKey?: string,
   vectorDb?: TemplateVectorDB,
   llamaCloudKey?: string,
   useLlamaParse?: boolean,
@@ -36,12 +38,12 @@ async function generateContextData(
         ? "poetry run generate"
         : `${packageManager} run generate`,
     )}`;
-    const openAiKeyConfigured = openAiKey || process.env["OPENAI_API_KEY"];
+    const modelConfigured = isModelConfigured(modelConfig);
     const llamaCloudKeyConfigured = useLlamaParse
       ? llamaCloudKey || process.env["LLAMA_CLOUD_API_KEY"]
       : true;
     const hasVectorDb = vectorDb && vectorDb !== "none";
-    if (openAiKeyConfigured && llamaCloudKeyConfigured && !hasVectorDb) {
+    if (modelConfigured && llamaCloudKeyConfigured && !hasVectorDb) {
       // If all the required environment variables are set, run the generate script
       if (framework === "fastapi") {
         if (isHavingPoetryLockFile()) {
@@ -63,7 +65,7 @@ async function generateContextData(
 
     // generate the message of what to do to run the generate script manually
     const settings = [];
-    if (!openAiKeyConfigured) settings.push("your OpenAI key");
+    if (!modelConfigured) settings.push("your model provider API key");
     if (!llamaCloudKeyConfigured) settings.push("your Llama Cloud key");
     if (hasVectorDb) settings.push("your Vector DB environment variables");
     const settingsMessage =
@@ -141,11 +143,9 @@ export const installTemplate = async (
 
     // Copy the environment file to the target directory.
     await createBackendEnvFile(props.root, {
-      openAiKey: props.openAiKey,
+      modelConfig: props.modelConfig,
       llamaCloudKey: props.llamaCloudKey,
       vectorDb: props.vectorDb,
-      model: props.model,
-      embeddingModel: props.embeddingModel,
       framework: props.framework,
       dataSources: props.dataSources,
       port: props.externalPort,
@@ -163,8 +163,8 @@ export const installTemplate = async (
       ) {
         await generateContextData(
           props.framework,
+          props.modelConfig,
           props.packageManager,
-          props.openAiKey,
           props.vectorDb,
           props.llamaCloudKey,
           props.useLlamaParse,
@@ -174,7 +174,6 @@ export const installTemplate = async (
   } else {
     // this is a frontend for a full-stack app, create .env file with model information
     await createFrontendEnvFile(props.root, {
-      model: props.model,
       customApiPath: props.customApiPath,
     });
   }