---
title: "ReliquaryForge's qwen3-4b-base-dapo-v4 Brings 32B Reasoning Power to a Single GPU"
subtitle: "A community-trained Qwen3-4B base tuned with ByteDance's DAPO reinforcement learning recipe hits nearly 900K downloads, showing appetite for small open reasoning models."
url: "https://alphasignal.ai/news/reliquaryforge-s-qwen3-4b-base-dapo-v4-brings-32b-reasoning-power-to-a-single"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "AlphaSignal"
company_url: "https://alphasignal.ai/company/alphasignal"
published: "2026-08-18T17:50:59.000Z"
updated: "2026-09-10T04:04:03.914Z"
type: "Model"
topics:
  - "Llms"
  - "Training Infra"
subtopics:
  - "Mixture Of Experts"
  - "Pretraining"
  - "Scaling Laws"
upvotes: 3296
source_url: "https://huggingface.co/ReliquaryForge/qwen3-4b-base-dapo-v4"
comments: 0
truncated: true
---

# ReliquaryForge's qwen3-4b-base-dapo-v4 Brings 32B Reasoning Power to a Single GPU

> A community-trained Qwen3-4B base tuned with ByteDance's DAPO reinforcement learning recipe hits nearly 900K downloads, showing appetite for small open reasoning models.

A community fine-tuned reasoning model built on Qwen3-4B-Base has quietly become one of the most downloaded small models on Hugging Face. [qwen3-4b-base-dapo-v4](https://huggingface.co/ReliquaryForge/qwen3-4b-base-dapo-v4), published by ReliquaryForge, is the fourth iteration of a Qwen3-4B checkpoint further trained with DAPO, the open reinforcement learning recipe ByteDance Seed released earlier this year. It sits at 884k downloads and 3.1k likes, unusual traction for a hobbyist repo with no benchmark table of its own.

The model card itself mostly inherits the upstream Qwen3 documentation, so the interesting story is the recipe and the size class. A 4B dense model pushed through a proper large-scale RL loop is exactly the kind of artifact many teams want for on-device or single-GPU reasoning work, and until recently that combination was mostly locked inside frontier labs.

## What's actually in the box

The starting point is Qwen Team's Qwen3-4B-Base, a dense causal LM with [4.0B total parameters (3.6B non-embedding), 36 layers, grouped-query attention with 32 query heads and 8 KV heads, and a 32,768 token context window](https://qwenlm.github.io/blog/qwen3/). Qwen3 as a family was [pre-trained on 36 trillion tokens across 119 languages, roughly triple the language coverage of Qwen2.5, with a heavier mix of coding, STEM, reasoning, book, multilingual, and synthetic data](https://arxiv.org/abs/2505.09388).

On top of that base, the v4 checkpoint applies DAPO, short for

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [ReliquaryForge/qwen3-4b-base-dapo-v4](https://huggingface.co/ReliquaryForge/qwen3-4b-base-dapo-v4)
- [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://arxiv.org/abs/2503.14476)
- [Qwen3 Technical Report](https://arxiv.org/abs/2505.09388)
- [Qwen3 Blog](https://qwenlm.github.io/blog/qwen3/)