Instructions to use programmer-666/Ornith-1.0-35B-MLX-oQ7 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use programmer-666/Ornith-1.0-35B-MLX-oQ7 with MLX:

# Make sure mlx-lm is installed
# pip install --upgrade mlx-lm

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("programmer-666/Ornith-1.0-35B-MLX-oQ7")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

text = generate(model, tokenizer, prompt=prompt, verbose=True)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use programmer-666/Ornith-1.0-35B-MLX-oQ7 with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "programmer-666/Ornith-1.0-35B-MLX-oQ7"

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "programmer-666/Ornith-1.0-35B-MLX-oQ7"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use programmer-666/Ornith-1.0-35B-MLX-oQ7 with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "programmer-666/Ornith-1.0-35B-MLX-oQ7"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "programmer-666/Ornith-1.0-35B-MLX-oQ7"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "programmer-666/Ornith-1.0-35B-MLX-oQ7",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use programmer-666/Ornith-1.0-35B-MLX-oQ7 with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "programmer-666/Ornith-1.0-35B-MLX-oQ7"

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default programmer-666/Ornith-1.0-35B-MLX-oQ7

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use programmer-666/Ornith-1.0-35B-MLX-oQ7 with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "programmer-666/Ornith-1.0-35B-MLX-oQ7"

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "programmer-666/Ornith-1.0-35B-MLX-oQ7" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

MTP in oMLX does not work

by bambusek - opened Jul 3

Discussion

bambusek

Jul 3

Hello, thanks for this quant.

However when I try to enable MTP in oMLX it gives this warning:

Maybe you know why is that a issue?
Thanks

programmer-666

Owner Jul 3

•

edited Jul 3

Hi, thanks for your feedback. I've identified an inconsistency in the safetensor group size values, and I've begun working on requantizing the model. I will be providing information on the progress.

blackfan23

Jul 12

I had problems quantizing using oQ method and MTP heads. Only succeeded once with oQ4. Interested in hearing how this is going for you.

programmer-666

Owner Jul 12

I had problems quantizing using oQ method and MTP heads. Only succeeded once with oQ4. Interested in hearing how this is going for you.

Yeah, I hit the exact same wall. I had to drop MTP for this release because the auxiliary head just doesn't pair correctly with the oQ7 mixed-precision weights. I removed the -mtp tag from the repo name to avoid confusion, but left the mtp.safetensors file in for future tinkering.

programmer-666 changed discussion status to closed Jul 12

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment