Instructions to use ruv/ruvltra-medium with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ruv/ruvltra-medium with MambaSSM:

from mamba_ssm import MambaLMHeadModel

model = MambaLMHeadModel.from_pretrained("ruv/ruvltra-medium")

llama-cpp-python

How to use ruv/ruvltra-medium with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="ruv/ruvltra-medium",
	filename="ruvltra-1.1b-q4_k_m.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use ruv/ruvltra-medium with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf ruv/ruvltra-medium:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf ruv/ruvltra-medium:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf ruv/ruvltra-medium:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf ruv/ruvltra-medium: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 ruv/ruvltra-medium:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf ruv/ruvltra-medium: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 ruv/ruvltra-medium:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf ruv/ruvltra-medium:Q4_K_M

Use Docker

docker model run hf.co/ruv/ruvltra-medium:Q4_K_M

LM Studio
Jan

vLLM

How to use ruv/ruvltra-medium with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ruv/ruvltra-medium"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ruv/ruvltra-medium",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ruv/ruvltra-medium:Q4_K_M

Ollama
How to use ruv/ruvltra-medium with Ollama:
```
ollama run hf.co/ruv/ruvltra-medium:Q4_K_M
```

Unsloth Studio new

How to use ruv/ruvltra-medium 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 ruv/ruvltra-medium 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 ruv/ruvltra-medium to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for ruv/ruvltra-medium to start chatting

Docker Model Runner
How to use ruv/ruvltra-medium with Docker Model Runner:
```
docker model run hf.co/ruv/ruvltra-medium:Q4_K_M
```

Lemonade

How to use ruv/ruvltra-medium with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull ruv/ruvltra-medium:Q4_K_M

Run and chat with the model

lemonade run user.ruvltra-medium-Q4_K_M

List all available models

lemonade list

ruv commited on Mar 28

Commit

a353bb7

verified ·

1 Parent(s): c861fb5

Add default TurboQuant KV-cache profile (ADR-129)

Browse files

Files changed (1) hide show

default.turboquant.json +57 -0

default.turboquant.json ADDED Viewed

	@@ -0,0 +1,57 @@

+{
+  "version": 1,
+  "model": "ruv/ruvltra-medium",
+  "default_bits": "3.5",
+  "default_eviction": "h2o",
+  "use_qjl": true,
+  "per_layer_config": {
+    "layer_0": {
+      "bits": "4.0",
+      "reason": "boundary layer \u2014 higher precision for input/output"
+    },
+    "layer_1": {
+      "bits": "4.0",
+      "reason": "boundary layer \u2014 higher precision for input/output"
+    },
+    "layer_2": {
+      "bits": "3.5",
+      "reason": "early layer \u2014 moderate compression"
+    },
+    "layer_3": {
+      "bits": "3.5",
+      "reason": "early layer \u2014 moderate compression"
+    },
+    "layer_4": {
+      "bits": "3.5",
+      "reason": "early layer \u2014 moderate compression"
+    },
+    "layer_5": {
+      "bits": "3.5",
+      "reason": "early layer \u2014 moderate compression"
+    },
+    "layer_6": {
+      "bits": "3.5",
+      "reason": "early layer \u2014 moderate compression"
+    },
+    "layer_7": {
+      "bits": "3.5",
+      "reason": "early layer \u2014 moderate compression"
+    },
+    "layer_8": {
+      "bits": "3.5",
+      "reason": "early layer \u2014 moderate compression"
+    },
+    "layer_9": {
+      "bits": "3.5",
+      "reason": "early layer \u2014 moderate compression"
+    },
+    "layer_40": {
+      "bits": "4.0",
+      "reason": "boundary layer \u2014 higher precision for input/output"
+    },
+    "layer_41": {
+      "bits": "4.0",
+      "reason": "boundary layer \u2014 higher precision for input/output"
+    }
+  }
+}