Instructions to use gghfez/MiMo-V2.5-Pro-unfused-test with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use gghfez/MiMo-V2.5-Pro-unfused-test with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="gghfez/MiMo-V2.5-Pro-unfused-test",
	filename="MiMo-V2.5-Pro-IQ2_S-unfused.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use gghfez/MiMo-V2.5-Pro-unfused-test with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S
# Run inference directly in the terminal:
llama-cli -hf gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S
# Run inference directly in the terminal:
llama-cli -hf gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S

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 gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S
# Run inference directly in the terminal:
./llama-cli -hf gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S

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 gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S
# Run inference directly in the terminal:
./build/bin/llama-cli -hf gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S

Use Docker

docker model run hf.co/gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S

LM Studio
Jan
Ollama
How to use gghfez/MiMo-V2.5-Pro-unfused-test with Ollama:
```
ollama run hf.co/gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S
```

Unsloth Studio new

How to use gghfez/MiMo-V2.5-Pro-unfused-test 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 gghfez/MiMo-V2.5-Pro-unfused-test 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 gghfez/MiMo-V2.5-Pro-unfused-test to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for gghfez/MiMo-V2.5-Pro-unfused-test to start chatting

Pi new

How to use gghfez/MiMo-V2.5-Pro-unfused-test with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use gghfez/MiMo-V2.5-Pro-unfused-test with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S

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 gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S

Run Hermes

hermes

Docker Model Runner
How to use gghfez/MiMo-V2.5-Pro-unfused-test with Docker Model Runner:
```
docker model run hf.co/gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S
```

Lemonade

How to use gghfez/MiMo-V2.5-Pro-unfused-test with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull gghfez/MiMo-V2.5-Pro-unfused-test:IQ2_S

Run and chat with the model

lemonade run user.MiMo-V2.5-Pro-unfused-test-IQ2_S

List all available models

lemonade list

ik_llama.cpp compatible quants of MiMo-V2.5-Pro

ik_llama.cpp doesn't support fused attn_qkv These are converted from mainline quants.

Tested with the main branch of ik_llama.cpp

Un-fused attn_qkv.weight -> attn_q.weight, attn_k.weight, attn_v.weight
Dropped MTP tensors

Tested and working with the latest ik_llama.cpp

MiMo-V2.5-Pro-IQ2_XXS-unfused.gguf - converted from bartowski/MiMo-V2.5-Pro-GGUF

MiMo-V2.5-Pro-IQ2_S-unfused.gguf - converted from AesSedai/MiMo-V2.5-Pro-GGUF

MiMo-V2.5-Pro-IQ3_S-unfused.gguf - converted from AesSedai/MiMo-V2.5-Pro-GGUF

architecture: mimo2
q_size=24576  head_dim=192  v_head_dim=128  layers=73  kv_heads=array  mtp=3  effective_layers=70
split 70 fused qkv tensors, dropped 36 mtp tensors

Perplexity Tests

iq2_s

Final estimate: PPL over 584 chunks for n_ctx=512 = 4.0549 +/- 0.02288

llama_print_timings:        load time =  191864.40 ms
llama_print_timings:      sample time =       0.00 ms /     1 runs   (    0.00 ms per token,      inf tokens per second)
llama_print_timings: prompt eval time = 4322704.22 ms / 299008 tokens (   14.46 ms per token,    69.17 tokens per second)
llama_print_timings:        eval time =       0.00 ms /     1 runs   (    0.00 ms per token,      inf tokens per second)
llama_print_timings:       total time = 4326930.78 ms / 299009 tokens

iq2_xxs

Final estimate: PPL over 584 chunks for n_ctx=512 = 4.5954 +/- 0.02684

llama_print_timings:        load time =  149084.44 ms
llama_print_timings:      sample time =       0.00 ms /     1 runs   (    0.00 ms per token,      inf tokens per second)
llama_print_timings: prompt eval time = 4160557.52 ms / 299008 tokens (   13.91 ms per token,    71.87 tokens per second)
llama_print_timings:        eval time =       0.00 ms /     1 runs   (    0.00 ms per token,      inf tokens per second)
llama_print_timings:       total time = 4164868.84 ms / 299009 tokens

Downloads last month: 348

GGUF

Model size

1T params

Architecture

mimo2

Hardware compatibility

2-bit

3-bit

Inference Providers NEW

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

Model tree for gghfez/MiMo-V2.5-Pro-unfused-test

Base model

XiaomiMiMo/MiMo-V2.5-Pro

Quantized

(7)

this model