qwen3-4b-journalist-ONNX

An investigative journalism model fine-tuned on Qwen/Qwen3-4B, converted to ONNX for in-browser inference via Transformers.js and WebGPU.

Built by Buried Signals for local/browser OSINT assistant workflows.

Usage

import { AutoTokenizer, AutoModelForCausalLM } from "@huggingface/transformers";

const tokenizer = await AutoTokenizer.from_pretrained("tomvaillant/qwen3-4b-journalist-ONNX");
const model = await AutoModelForCausalLM.from_pretrained("tomvaillant/qwen3-4b-journalist-ONNX", {
  dtype: "q4f16",
  device: "webgpu",
});

Training

• Method: QLoRA with Unsloth / TRL SFT
• Data: 687 instruction/response pairs across investigative journalism and OSINT categories
• Task: browser-sized assistant for OSINT tool selection, investigation planning, verification, and source handling

ONNX Conversion

• Quantization: q4f16
• Runtime: Transformers.js via WebGPU
• Files: onnx/model_q4f16.onnx, external data files, tokenizer, and chat template

Sources And Attribution

Training data: tomvaillant/investigative-journalism-training — 687 instruction/response pairs synthesized by Claude Opus 4.6 (Anthropic) from the Buried Signals OSINT and investigative-journalism corpus: OSINT Navigator tool data, Indicator Media briefings, Buried Signals investigative skills, GIJN, Bellingcat, Verification Handbook 3, SPJ Code of Ethics, RCFP, and public manuals from UNESCO, Al Jazeera Media Institute, CiFAR, CIPE, and EJF/TEMPO Institute.

See the dataset card for the full source list, licenses, and per-partner attribution.

Intended Use

Use for local/browser investigative assistance. Generated findings, URLs, and legal conclusions require independent verification before reporting or operational use.