Instructions to use Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF",
	filename="Slopus-4.65-XHigh-IQ3_M.gguf",
)

output = llm(
	"Once upon a time,",
	max_tokens=512,
	echo=True
)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M

Use Docker

docker model run hf.co/Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M

LM Studio
Jan
Ollama
How to use Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF with Ollama:
```
ollama run hf.co/Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M
```

Unsloth Studio

How to use Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF to start chatting

Atomic Chat new
Docker Model Runner
How to use Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF with Docker Model Runner:
```
docker model run hf.co/Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M
```

Lemonade

How to use Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Slopus-4.65-XHigh-Schmidhuber-GGUF-Q4_K_M

List all available models

lemonade list

Slopus 4.65 XHigh — Schmidhuber Edition — GGUF

GGUF quantizations del LoRA Slopus mergeado sobre Darwin-36B-Opus (Qwen3.6-35B-A3B + Claude-4.6-Opus distill).

Adapter source: Kukedlc/slopus-4.65-xhigh-schmidhuber-edition Base model: FINAL-Bench/Darwin-36B-Opus

Quantizations disponibles

Todos quantizados con importance matrix (imatrix) calibrada con bartowski calibration_datav3 (2481 lineas code + razonamiento + multilingual). Pipeline: llama.cpp release b8919.

Quant	Size	Uso recomendado
Q2_K	12.9 GB	recursos minimos, calidad notable degradada
IQ3_M	15.4 GB	low VRAM, mejor calidad que Q3_K_S
Q3_K_M	16.8 GB	low VRAM compromise standard
IQ4_XS	18.7 GB	mejor calidad/tamaño que Q4_K_M para algunos casos
Q4_K_M	21.2 GB	LA MAS USADA, default recomendado
Q5_K_M	24.7 GB	near-lossless, recomendado si VRAM permite
Q8_0	36.9 GB	reference quality, casi sin perdida vs F16

imatrix.dat incluido en el repo para re-quantizar si querés generar otros formatos.

Sampling oficial Qwen3.6

temperature       = 1.0
top_p             = 0.95
top_k             = 20
min_p             = 0.0
presence_penalty  = 1.5
repeat_penalty    = 1.0

llama-server

llama-server   --model Slopus-4.65-XHigh-Q4_K_M.gguf   --gpu-layers 999   --ctx-size 16384   --host 0.0.0.0 --port 8000

Speculative decoding

NO usar --model-draft sobre Slopus. El base Darwin NO tiene MTP heads (es merge sin esa optimización). Benchmarks comunitarios (RTX 3090, A6000) muestran que speculative tradicional con A3B MoE es net negativo. Correr solo.

Hecho por

Kukito (data scientist, Cordoba, Argentina) + zoo de Claude persona-typed instances. Pipeline en RunPod RTX A6000 48GB Secure. Mas detalles en el adapter README.

Downloads last month: 1,083

GGUF

Model size

35B params

Architecture

qwen35moe

Hardware compatibility

2-bit

3-bit

4-bit

5-bit

8-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Kukedlc/Slopus-4.65-XHigh-Schmidhuber-GGUF

Base model

FINAL-Bench/Darwin-36B-Opus

Quantized

(8)

this model