GraySoft
Projects Models Compare Cloud benchmarks FAQ Download guIDE →
Model Intelligence Sheet

Koshkasa/TheDrummer_Big-Alice-28B-v1-IQ4_KT-GGUF overview

What's that? The goal : Make a competitive 4 bit quality quant of TheDrummer/Big Alice 28B v1 using SOTA quant types from ik llama.cpp, allowing the resulting …

ik_llama.cppggufquantizediq4_kt4 bittrellisroleplaymixed precisiontext-generationbase_model:TheDrummer/Big-Alice-28B-v1base_model:quantized:TheDrummer/Big-Alice-28B-v1license:mitendpoints_compatibleregion:usimatrixconversational

Runs locally from ~14.50 GB disk (16 GB VRAM class GPUs with llama.cpp / guIDE).

Downloads
0
Likes
0
Pipeline
text-generation
Author

Repository Files & Downloads

1 GGUF files detected
Direct downloads for local inference
FileTypeQuantizationSizeLink
TheDrummer_Big-Alice-28B-v1-IQ4_KT.ggufGGUFIQ4_KT14.50 GBDownload

Model Details

Model IDKoshkasa/TheDrummer_Big-Alice-28B-v1-IQ4_KT-GGUF
AuthorKoshkasa
Pipelinetext-generation
Licensemit
Base modelTheDrummer/Big-Alice-28B-v1
Last modified2026-07-28T10:05:57.000Z

Model README

---

license: mit

base_model:

  • TheDrummer/Big-Alice-28B-v1

library_name: ik_llama.cpp

pipeline_tag: text-generation

tags:

  • gguf
  • quantized
  • ik_llama.cpp
  • iq4_kt
  • 4 bit
  • trellis
  • roleplay
  • mixed precision

quantized_by: Koshkasa

base_model_relation: quantized

---

What's that?

The goal: Make a competitive 4-bit quality quant of TheDrummer/Big-Alice-28B-v1 using SOTA quant types from ik_llama.cpp, allowing the resulting gguf to fit into 16gb VRAM without KVO, accounting for system overhead.

The result: Mixed precision quantization of TheDrummer/Big-Alice-28B-v1 utilizing 4-bit trellis quants for FFN tensors that is theoretically smarter and faster than IQ4_XS.

quantized with ik_llama.cpp build: 9d07d868

incompatible with mainline llama.cpp

Layout

| Layer | Dims | Dims | Quant |

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

| token\_embd | 5120 | 131072.0 | iq4\_k |

| | | | |

| | blk| 40| |

| attn\_k | 5120 | 1024 | iq5\_ks |

| attn\_norm | 5120 | 1 | f32 |

| attn\_q | 5120 | 4096 | iq5\_ks |

| attn\_v | 5120 | 1024 | iq6\_k |

| attn\_output | 4096 | 5120 | iq5\_k |

| ffn\_down | 14336 | 5120 | iq4\_kt |

| ffn\_gate | 5120 | 14336 | iq4\_kt |

| ffn\_norm | 5120 | 1 | f32 |

| ffn\_up | 5120 | 14336 | iq4\_kt |

| | | | |

| output | 5120 | 131072 | iq6\_k |

| output\_norm | 5120 | 1 | f32 |

Conclusion

In retrospect, it probably wasn't the best idea for 16GB VRAM - the PCI-E travel overhead from KV cache in RAM is MASSIVE for upscaled Apriel (especially for my 8-lane 5060 Ti.). This one remains an ok option at 20gb+ VRAM, or on 16gb GPUs with a less impotent PCI-E config. A smaller 3.44 bpw ffn + trellis 4.0 bpw attention quant is in the overn for strictly 16gb inference.

Regardless, this one is provided as is. WYSIWYG.

Cheers

SillyTilly - for the base model.

TheDrummer - for the finetuned model.

mradermacher - for the imatrix + the myriad of quants we all benefit from.

ikawrakow and contributors of ik_llama.cpp - I probably misused your ingenious creation.

Run Koshkasa/TheDrummer_Big-Alice-28B-v1-IQ4_KT-GGUF with guIDE

Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.

Download guIDE → · Browse 524k+ models · Compare models

Source: Hugging Face · Compare models