Model Intelligence Sheet
noctrex/Ling-3.0-flash-heretic-MXFP4_MOE-GGUF overview
This is a MXFP4 MOE quantization of the model trohrbaugh / Ling 3.0 flash heretic https://huggingface.co/trohrbaugh/Ling 3.0 flash heretic Quick Start 1. Downl…
Runs locally from ~13.64 GB disk (16 GB VRAM class GPUs with llama.cpp / guIDE).
Repository Files & Downloads
4 GGUF files detected
Direct downloads for local inference
| File | Type | Quantization | Size | Link |
|---|---|---|---|---|
| Ling-3.0-flash-heretic-MXFP4_MOE-00001-of-00004.gguf | GGUF | GGUF | 17.44 GB | Download |
| Ling-3.0-flash-heretic-MXFP4_MOE-00002-of-00004.gguf | GGUF | GGUF | 17.43 GB | Download |
| Ling-3.0-flash-heretic-MXFP4_MOE-00003-of-00004.gguf | GGUF | GGUF | 17.44 GB | Download |
| Ling-3.0-flash-heretic-MXFP4_MOE-00004-of-00004.gguf | GGUF | GGUF | 13.64 GB | Download |
Model Details
| Model ID | noctrex/Ling-3.0-flash-heretic-MXFP4_MOE-GGUF |
|---|---|
| Author | noctrex |
| Pipeline | text-generation |
| License | — |
| Base model | trohrbaugh/Ling-3.0-flash-heretic |
| Last modified | 2026-08-17T20:58:27.000Z |
Model README
---
pipeline_tag: text-generation
base_model:
- trohrbaugh/Ling-3.0-flash-heretic
---
This is a MXFP4_MOE quantization of the model trohrbaugh / Ling-3.0-flash-heretic
Quick Start
- Download the latest release of llama.cpp.
Recommended parameters from inclusionAI:
- temperature=0.6
- top_p=0.95
- top_k=20
Performance
| Metric | This model | Original model |
| :----- | :--------: | :---------------------------: |
| KL divergence | 0.0526 | 0 (by definition) |
| Refusals | 0/100 | 87/100 |
Run noctrex/Ling-3.0-flash-heretic-MXFP4_MOE-GGUF with guIDE
Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.
Source: Hugging Face · Compare models