Instructions to use Abiray/BitCPM4-CANN-8B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Abiray/BitCPM4-CANN-8B-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="Abiray/BitCPM4-CANN-8B-GGUF",
	filename="BitCPM4-CANN-8B-Q3_K_M.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 Abiray/BitCPM4-CANN-8B-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Abiray/BitCPM4-CANN-8B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Abiray/BitCPM4-CANN-8B-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 Abiray/BitCPM4-CANN-8B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Abiray/BitCPM4-CANN-8B-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 Abiray/BitCPM4-CANN-8B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf Abiray/BitCPM4-CANN-8B-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 Abiray/BitCPM4-CANN-8B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Abiray/BitCPM4-CANN-8B-GGUF:Q4_K_M

Use Docker

docker model run hf.co/Abiray/BitCPM4-CANN-8B-GGUF:Q4_K_M

LM Studio
Jan
Ollama
How to use Abiray/BitCPM4-CANN-8B-GGUF with Ollama:
```
ollama run hf.co/Abiray/BitCPM4-CANN-8B-GGUF:Q4_K_M
```

Unsloth Studio new

How to use Abiray/BitCPM4-CANN-8B-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 Abiray/BitCPM4-CANN-8B-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 Abiray/BitCPM4-CANN-8B-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Abiray/BitCPM4-CANN-8B-GGUF to start chatting

Docker Model Runner
How to use Abiray/BitCPM4-CANN-8B-GGUF with Docker Model Runner:
```
docker model run hf.co/Abiray/BitCPM4-CANN-8B-GGUF:Q4_K_M
```

Lemonade

How to use Abiray/BitCPM4-CANN-8B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Abiray/BitCPM4-CANN-8B-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.BitCPM4-CANN-8B-GGUF-Q4_K_M

List all available models

lemonade list

Abiray/BitCPM4-CANN-8B-GGUF

This repository contains quantized GGUF formats of the openbmb/BitCPM4-CANN-8B model, heavily optimized for local inference using llama.cpp, text-generation-webui, LM Studio, Ollama, and other compatible backend frameworks.

Model Information

Original Model: openbmb/BitCPM4-CANN-8B
Architecture: MiniCPM / BitCPM (8 Billion Parameters)

Available Files & Hardware Compatibility

The following quantization formats are available. As an 8-Billion parameter model, these variants offer excellent reasoning, coding, and comprehension capabilities while remaining small enough to run entirely on consumer GPUs (like the RTX 3060/4060 series) or modern system RAM.

Filename	Quant Type	File Size	Description
BitCPM4-CANN-8B-Q8_0.gguf	8-bit	8.70 GB	Extremely high fidelity. Practically identical to the unquantized base model. Recommended if you have 12GB+ of VRAM/RAM.
BitCPM4-CANN-8B-Q6_K.gguf	6-bit	6.72 GB	Exceptional performance with near-zero degradation. Highly stable for complex instructions.
BitCPM4-CANN-8B-Q5_K_M.gguf	5-bit	5.81 GB	Highly recommended balance of file size, text generation speed, and response accuracy.
BitCPM4-CANN-8B-Q5_K_S.gguf	5-bit	5.67 GB	A slightly lighter version of the 5-bit intermediate format, maximizing speed over minor edge-case logic.
BitCPM4-CANN-8B-Q4_K_M.gguf	4-bit	4.97 GB	Recommended. The absolute sweet spot for 8B models. Keeps the model under 5GB while preserving most of its native intelligence.
BitCPM4-CANN-8B-Q4_K_S.gguf	4-bit	4.72 GB	Optimized heavily for speed and low memory impact, perfect for constrained environments.
BitCPM4-CANN-8B-Q3_K_M.gguf	3-bit	4.02 GB	Maximum compression. Fits easily into lower-tier hardware, though some fallback in complex logic may occur.

How to Run

Using `llama.cpp` (Command Line)

If you have compiled llama.cpp, you can run the model directly from your terminal. Replace the filename with the specific version you downloaded:

./llama-cli \
  -m BitCPM4-CANN-8B-Q4_K_M.gguf \
  -p "Explain the concept of artificial intelligence to a five-year-old." \
  -n 256 \
  -c 2048 \
  --temp 0.7

Downloads last month: -

GGUF

Model size

8B params

Architecture

minicpm

Hardware compatibility

3-bit

4-bit

5-bit

6-bit

8-bit

Inference Providers NEW

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

Model tree for Abiray/BitCPM4-CANN-8B-GGUF

Base model

openbmb/BitCPM4-CANN-8B

Quantized

(3)

this model

Collection including Abiray/BitCPM4-CANN-8B-GGUF

BitCPM4-CANN

Collection

4 items • Updated about 17 hours ago