Model Intelligence Sheet
magnitudedev/Qwen3.5-4B-DFlash-GGUF overview
Qwen3.5 4B DFlash GGUF Q8 0 GGUF conversion of z lab/Qwen3.5 4B DFlash https://huggingface.co/z lab/Qwen3.5 4B DFlash for use with Magnitude https://github.com…
Runs locally from ~653.4 MB disk (4 GB VRAM class GPUs with llama.cpp / guIDE).
Repository Files & Downloads
1 GGUF files detected
Direct downloads for local inference
| File | Type | Quantization | Size | Link |
|---|---|---|---|---|
| Qwen3.5-4B-DFlash-Q8_0.gguf | GGUF | Q8_0 | 653.4 MB | Download |
Model Details
| Model ID | magnitudedev/Qwen3.5-4B-DFlash-GGUF |
|---|---|
| Author | magnitudedev |
| Pipeline | — |
| License | apache-2.0 |
| Base model | z-lab/Qwen3.5-4B-DFlash |
| Last modified | 2026-10-07T12:35:30.000Z |
Model README
---
base_model:
- z-lab/Qwen3.5-4B-DFlash
license: apache-2.0
library_name: llama.cpp
tags:
- gguf
- dflash
- speculative-decoding
- magnitude
---
Qwen3.5-4B-DFlash GGUF
Q8_0 GGUF conversion of
for use with Magnitude.
This is a DFlash draft model for speculative decoding with
Qwen/Qwen3.5-4B; the tokenizer metadata comes from that target.
Converted with
at revision 43fe9c64281ef735046adc025e9e7559a1f659a5.
Artifact
- File:
Qwen3.5-4B-DFlash-Q8_0.gguf - Quantization:
Q8_0 - SHA-256:
8864ab4b1d147278c3cf0806ec9aa9f0db644987f084856317e3186bf433280a - Source revision:
9a1996ccf887b79ab3af4fcbf8c1d1f4b5658bcf
Draft metadata
dflash.attention.causal = [true, true, true, true, true, false]: the first five layers use causal attention; the last uses bidirectional attention.dflash.sample_from_anchor = false: proposals come from mask rows, starting after the anchor.- Derived from the checkpoint configuration and Z-Lab’s reference implementation; verified against reference outputs.
Run magnitudedev/Qwen3.5-4B-DFlash-GGUF with guIDE
Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.
Source: Hugging Face · Compare models