Getting TS2304: Cannot find name 'PretrainedProcessorOptions'. error

[ShibleNK]

System Info

transformers.js - 3.7.2
github codespaces
node - 22.17.0
Next.js 15 for frontend
Node.js for backend
typescript - 5.8.3

Environment/Platform

• Website/web-app
• Browser extension
• Server-side (e.g., Node.js, Deno, Bun)
• Desktop app (e.g., Electron)
• Other (e.g., VSCode extension)

Description

I was trying to use the gemma-3-1b-it-ONNX-GQA for giving prompt to generate text.

Reproduction

package.json

{
  "devDependencies": {
    "@huggingface/transformers": "^3.7.2"
  }
}

tsconfig.json

{
  "compilerOptions": {
    "target": "ES6",
    "baseUrl": "./src",
    "rootDir": "./src",
    "module": "commonjs",
    "paths": {
      "@/*": ["*"]
    },
    "typeRoots": ["./src/types", "./node_modules/@types"],
    "strict": true,
    "outDir": "./dist",
    "esModuleInterop": true
  },
  "include": ["./src/**/*"]
}

main.ts

import { pipeline } from "@huggingface/transformers";

executing tsc returns -

32     static from_pretrained(pretrained_model_name_or_path: string, options?: PretrainedProcessorOptions): Promise<Processor>;
                                                                               ~~~~~~~~~~~~~~~~~~~~~~~~~~

../node_modules/.pnpm/@huggingface+transformers@3.7.2/node_modules/@huggingface/transformers/types/models/mgp_str/processing_mgp_str.d.ts:52:5 - error TS2416: Property 'batch_decode' in type 'MgpstrProcessor' is not assignable to the same property in base type 'Processor'.
  Type '([char_logits, bpe_logits, wp_logits]: Tensor[]) => { generated_text: string[]; scores: number[]; char_preds: string[]; bpe_preds: string[]; wp_preds: string[]; }' is not assignable to type '(batch: number[][] | Tensor, decode_args?: any) => string[]'.
    Types of parameters '__0' and 'batch' are incompatible.
      Type 'number[][] | Tensor' is not assignable to type 'Tensor[]'.
        Type 'number[][]' is not assignable to type 'Tensor[]'.
          Type 'number[]' is missing the following properties from type 'Tensor': dims, type, data, size, and 47 more.

52     batch_decode([char_logits, bpe_logits, wp_logits]: import("../../utils/tensor.js").Tensor[]): {
       ~~~~~~~~~~~~

../node_modules/.pnpm/@huggingface+transformers@3.7.2/node_modules/@huggingface/transformers/types/utils/maths.d.ts:116:125 - error TS2304: Cannot find name 'Float16Array'.

116 export type TypedArray = Int8Array | Uint8Array | Uint8ClampedArray | Int16Array | Uint16Array | Int32Array | Uint32Array | Float16Array | Float32Array | Float64Array;
                                                                                                                                ~~~~~~~~~~~~

../node_modules/.pnpm/@huggingface+transformers@3.7.2/node_modules/@huggingface/transformers/types/utils/tensor.d.ts:160:39 - error TS2552: Cannot find name 'Float16ArrayConstructor'. Did you mean 'Float64ArrayConstructor'?

160     float16: Uint16ArrayConstructor | Float16ArrayConstructor;
                                          ~~~~~~~~~~~~~~~~~~~~~~~


Found 4 errors in 4 files.

Errors  Files
     1  ../node_modules/.pnpm/@huggingface+transformers@3.7.2/node_modules/@huggingface/transformers/types/models/auto/processing_auto.d.ts:32
     1  ../node_modules/.pnpm/@huggingface+transformers@3.7.2/node_modules/@huggingface/transformers/types/models/mgp_str/processing_mgp_str.d.ts:52
     1  ../node_modules/.pnpm/@huggingface+transformers@3.7.2/node_modules/@huggingface/transformers/types/utils/maths.d.ts:116
     1  ../node_modules/.pnpm/@huggingface+transformers@3.7.2/node_modules/@huggingface/transformers/types/utils/tensor.d.ts:160
 ELIFECYCLE  Command failed with exit code 2.```

[vishnusureshperumbavoor]

Have you tried out adding custom typescript definition??

// src/types/transformers.d.ts
export type PretrainedProcessorOptions = Record<string, any>;

[ShibleNK]

Have you tried out adding custom typescript definition??

// src/types/transformers.d.ts
export type PretrainedProcessorOptions = Record<string, any>;

Tried it, still no luck.

[xenova]

Thanks for the report 👍 This may just be a simple missing import somewhere. If you are able to, would you (or someone else) like to open a PR for it? :)

[ShibleNK]

So, basically it’s a language issue. As this model only works in English but i was trying to give prompt in Bengali.

[julijan]

Getting the same error:
"@huggingface/transformers": "^3.7.5"
Node.js v22.14.0
TypeScript 5.9.3
Type: module

tsconfig.json:

{
    "compilerOptions": {
        "alwaysStrict": true,
        "noImplicitAny": true,
        "strictNullChecks": true,
        "strictPropertyInitialization": true,
        "strictBindCallApply": true,
        "noUnusedLocals": true,
        "noImplicitReturns": true,
        "moduleResolution" : "node16",
        "module": "Node16",
        "outDir": "./build",
        "target": "ES2021",
        "allowSyntheticDefaultImports": true,
		"esModuleInterop": true,
        "removeComments": true,
        "baseUrl": ".",
        "rootDir": "."
    },
    "include": ["./app.ts"]
}

To reproduce:
npm install @huggingface/transformers@3.7.5

Code:

import { pipeline, ProgressInfo } from "@huggingface/transformers";

pipeline("automatic-speech-recognition", "openai/whisper-large-v3", {
	progress_callback: function(progress: ProgressInfo) {
		console.log(progress.status);
	},
	dtype: "fp16",
});

[julijan]

Adding:
type PretrainedProcessorOptions = import("../../base/processing_utils.js").PretrainedProcessorOptions;
to:
/node_modules/@huggingface/transformers/types/models/auto/processing_auto.d.ts
Solves that issue, of course, this is a dirty workaround, just putting it out there in case it helps with a proper fix.

[julijan]

I've been getting another error in addition to the one above which is for now solved with the dirty fix, second error is:
node_modules/@huggingface/transformers/types/models/mgp_str/processing_mgp_str.d.ts:52:5 - error TS2416: Property 'batch_decode' in type 'MgpstrProcessor' is not assignable to the same property in base type 'Processor'.
Type '([char_logits, bpe_logits, wp_logits]: Tensor[]) => { generated_text: string[]; scores: number[]; char_preds: string[]; bpe_preds: string[]; wp_preds: string[]; }' is not assignable to type '(batch: number[][] | Tensor, decode_args?: any) => string[]'.
Types of parameters '__0' and 'batch' are incompatible.
Type 'number[][] | Tensor' is not assignable to type 'Tensor[]'.
Type 'number[][]' is not assignable to type 'Tensor[]'.
Type 'number[]' is missing the following properties from type 'Tensor': dims, type, data, size, and 47 more.

52 batch_decode([char_logits, bpe_logits, wp_logits]: import("../../utils/tensor.js").Tensor[]): {

[julijan]

A dirty fix for this one, change batch_decode declaration (line 52) in:
node_modules/@huggingface/transformers/types/models/mgp_str/processing_mgp_str.d.ts
To:
batch_decode(...args: Parameters<PreTrainedTokenizer['batch_decode']>): ReturnType<PreTrainedTokenizer['batch_decode']>;

Wish I could be more useful by submitting actual PR's, but I'm unfamiliar with the project structure, hopefully these point someone who is familiar with it in the right direction.