Instructions to use gittyeric/MiniMax-M2.7-oQ8e with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use gittyeric/MiniMax-M2.7-oQ8e 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("gittyeric/MiniMax-M2.7-oQ8e")

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 gittyeric/MiniMax-M2.7-oQ8e with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "gittyeric/MiniMax-M2.7-oQ8e"

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": "gittyeric/MiniMax-M2.7-oQ8e"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

OpenClaw new

How to use gittyeric/MiniMax-M2.7-oQ8e with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "gittyeric/MiniMax-M2.7-oQ8e"

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 "gittyeric/MiniMax-M2.7-oQ8e" \
  --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"

MLX LM

How to use gittyeric/MiniMax-M2.7-oQ8e with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "gittyeric/MiniMax-M2.7-oQ8e"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "gittyeric/MiniMax-M2.7-oQ8e"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "gittyeric/MiniMax-M2.7-oQ8e",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use gittyeric/MiniMax-M2.7-oQ8e 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 "gittyeric/MiniMax-M2.7-oQ8e"

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 gittyeric/MiniMax-M2.7-oQ8e

Run Hermes

hermes

Request for oQ6e and oQ4e variants

by TheOlRazzleDazzle - opened 17 days ago

Discussion

TheOlRazzleDazzle

17 days ago

Hi @gittyeric ,

Thanks for your contribution to the oQe family of models!

I share a similar experience in that Minimax-M2.7 is still exceptional in this class.

I am running a smaller EXO cluster so a bit tighter on room for KV cache with this current precision - any chance you would be willing to create the oQ6e and/or oQ4e variants?

Much appreciated and happy generating🫶

gittyeric

Owner 17 days ago

I suppose I can get around to this once some batch stuff I'm running is down, but I mostly added the oQ8 quant because there were no oQ's at all, there were however oQ4's and oQ6's already if you search "https://huggingface.co/models?sort=trending&search=minimax+oq". To be honest, I'm not really noticing any benefit from the -e imatrix bit but just feelsies. In any case I can generate the biggest one you can fit and you can use those to get a check on what fits for you.

gittyeric

Owner 16 days ago

•

edited 16 days ago

Here they are! oQ4e and oQ6e are now under my account. Pretty soon I'll run terminal and SWE benchmarks against all 3 quants since I'm a bit curious how they all hold up and append to the READMEs.

Shoot me a reply on how the different quants work for ya since the benchmarks are sorta made up anyway, curious how the oQ bit + the e bit works for you since on paper it should top any other quants on HF.

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