@empiricalrun/llm - npm Package Compare versions

Comparing version 0.9.6 to 0.9.9

CHANGELOG.md

 # @empiricalrun/llm
 
+## 0.9.9
+
+### Patch Changes
+
+- b5379b8: feat: added capability to select model in vision api's
+
+## 0.9.8
+
+### Patch Changes
+
+- e33656e: test: added structural output tests in vision
+
+## 0.9.7
+
+### Patch Changes
+
+- cf1740a: fix: move bbox to separate export path
+
 ## 0.9.6
@@ -4,0 +22,0 @@

dist/vision/index.d.ts

@@ -1,2 +0,1 @@
-export { getBoundingBox } from "./bbox";
 export { solveRecaptchaGrid } from "./captcha";
@@ -3,0 +2,0 @@ export { query } from "./query";

dist/vision/index.js

 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.imageFormatForProvider = exports.query = exports.solveRecaptchaGrid = exports.getBoundingBox = void 0;
-var bbox_1 = require("./bbox");
-Object.defineProperty(exports, "getBoundingBox", { enumerable: true, get: function () { return bbox_1.getBoundingBox; } });
+exports.imageFormatForProvider = exports.query = exports.solveRecaptchaGrid = void 0;
 var captcha_1 = require("./captcha");
@@ -7,0 +5,0 @@ Object.defineProperty(exports, "solveRecaptchaGrid", { enumerable: true, get: function () { return captcha_1.solveRecaptchaGrid; } });

dist/vision/query/index.d.ts

 import { z, ZodType } from "zod";
+import { LLMModel } from "../..";
 type ExtractType<T> = T extends ZodType ? z.infer<T> : never;
-export declare function query<T extends z.ZodType>(base64Image: string, instruction: string, responseFormat?: T): Promise<ExtractType<T>>;
+export declare function query<T extends z.ZodType>(base64Image: string, instruction: string, options?: {
+    responseFormat?: T;
+    model?: LLMModel;
+}): Promise<ExtractType<T>>;
 export {};
 //# sourceMappingURL=index.d.ts.map

dist/vision/query/index.js

@@ -8,3 +8,7 @@ "use strict";
 const utils_1 = require("../utils");
-async function query(base64Image, instruction, responseFormat = zod_2.z.string()) {
+async function query(base64Image, instruction, options = {
+    responseFormat: zod_2.z.string(),
+    model: "gpt-4o-mini",
+}) {
+    const { responseFormat = zod_2.z.string(), model = "gpt-4o-mini" } = options;
     const llm = new __1.LLM({
@@ -52,3 +56,3 @@ provider: "openai",
         ],
-        model: "gpt-4o-mini",
+        model,
         modelParameters: {
@@ -55,0 +59,0 @@ temperature: 0.5,

package.json

 {
   "name": "@empiricalrun/llm",
-  "version": "0.9.6",
+  "version": "0.9.9",
   "main": "dist/index.js",
@@ -13,2 +13,6 @@ "exports": {
       "default": "./dist/vision/index.js"
+    },
+    "./vision/bbox": {
+      "types": "./dist/vision/bbox/index.d.ts",
+      "default": "./dist/vision/bbox/index.js"
     }
@@ -15,0 +19,0 @@ },

README.md

@@ -29,6 +29,7 @@ # llm
 // With Appium
-const data = await driver.saveScreenshot('dummy.png');
-const instruction = "Extract number of ATOM tokens from the image. Return only the number.";
+const data = await driver.saveScreenshot("dummy.png");
+const instruction =
+  "Extract number of ATOM tokens from the image. Return only the number.";
 
-const text = await query(data.toString('base64'), instruction);
+const text = await query(data.toString("base64"), instruction);
 // Example response: "0.01"
@@ -40,10 +41,11 @@ ```
 ```ts
-import { getBoundingBox } from "@empiricalrun/llm/vision";
+import { getBoundingBox } from "@empiricalrun/llm/vision/bbox";
 
 // With Appium
-const data = await driver.saveScreenshot('dummy.png');
+const data = await driver.saveScreenshot("dummy.png");
 // Give a line describing the screen and then the element that you want to find
-const instruction = "This screenshot shows a screen to send crypto tokens. What is the bounding box for the dropdown to select the token?";
+const instruction =
+  "This screenshot shows a screen to send crypto tokens. What is the bounding box for the dropdown to select the token?";
 
-const bbox = await getBoundingBox(data.toString('base64'), instruction);
+const bbox = await getBoundingBox(data.toString("base64"), instruction);
 const centerToTap = bbox.center; // { x: 342, y: 450 }
@@ -59,3 +61,5 @@
 ```ts
-const bbox = await getBoundingBox(data.toString('base64'), instruction, { debug: true });
+const bbox = await getBoundingBox(data.toString("base64"), instruction, {
+  debug: true,
+});
 console.log(bbox.annotatedImage);
@@ -62,0 +66,0 @@ ```

No alert changes

Improved metrics

Total package byte prevSize

1037329

increased by584.68%

Number of package files

34

increased by6.25%

Lines of code

544

increased by0.93%

Number of lines in readme file

70

increased by6.06%

No dependency changes