Sign up

AtomicChat/Qwen3-4B-DFlash-GGUF

🤗 On Hugging Facetext-generationmit1.7 GBGGUF✓ HF checksums availableupdated today
Magnet

Base model: z-lab/Qwen3-4B-DFlash-b16

Qwen3 4B Dflash b16, self-quantized to GGUF by Atomic Chat. Built straight from Z Lab's original weights with a per-tensor importance matrix, so this is not a repack of somebody else's files. Runs fully offline.

Highlights

  • 0.5B parameters: the weights this repo quantizes.
  • Context length: 40,960 tokens (40K), as published by Z Lab.
  • 5 layers: Dense decoder.
  • Full imatrix ladder: every quant is calibrated with an importance matrix.
[!NOTE]
These GGUFs are self-quantized from the original weights, not a repack. The importance matrix keeps low-bit quants closer to the full-precision model.
[!IMPORTANT]
Always pass --jinja so the Qwen3 4B Dflash b16 chat template is applied. Without it the model can emit malformed turns.

Model Overview

| Property | Value |

|---|---|

| Base model | z-lab/Qwen3-4B-DFlash-b16 |

| Parameters | 0.5B |

| Layers | 5 |

| Context length | 40,960 tokens (40K) |

| Vocabulary | 151,936 |

| Modalities | Text |

| Architecture | Dense decoder, 32 attention heads over 8 KV heads, DFlashDraftModel |

| This repo | GGUF quants (imatrix). Quants: Q8_0 |

Choosing a quant

| Quant | Size | Notes |

|---|---|---|

| Q8_0 | 0.6 GB | Effectively lossless, reference quality. |

[!TIP]
Pick the largest file that fits your (V)RAM with room for context. Q8_0 is the sweet spot for most setups; Q6_K or Q8_0 for maximum fidelity.

Get started

Run Qwen3 4B Dflash b16 locally with:

  • Atomic Chat: the easiest path. Open the app, search AtomicChat/Qwen3-4B-DFlash-GGUF, pick a quant, hit Use this model.
  • llama.cpp: llama-server -hf AtomicChat/Qwen3-4B-DFlash-GGUF:Q8_0 --jinja -c 8192
  • Ollama: ollama run hf.co/AtomicChat/Qwen3-4B-DFlash-GGUF:Q8_0
  • LM Studio / Jan: search the repo id, download any quant.

Best practices

| Parameter | Value |

|---|---|

| temperature | 0.0 |

Z Lab's recommended sampling configuration for z-lab/Qwen3-4B-DFlash-b16.

Run in llama.cpp

git clone https://github.com/ggml-org/llama.cpp
cmake llama.cpp -B llama.cpp/build -DBUILD_SHARED_LIBS=OFF -DGGML_CUDA=ON
cmake --build llama.cpp/build --config Release -j --target llama-cli llama-server
./llama.cpp/build/bin/llama-server \
    -hf AtomicChat/Qwen3-4B-DFlash-GGUF:Q8_0 \
    --jinja -ngl 99 -c 8192 -fa on

How these were made

1. Download z-lab/Qwen3-4B-DFlash-b16 (original weights).

2. Convert to f16 GGUF with llama.cpp.

3. Build an importance matrix over our calibration corpus.

4. Quantize the ladder with --imatrix.

License

Original model by Z Lab, released under the MIT license. Quantized by Atomic Chat.