---
title: KernelLLM-6bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/KernelLLM-6bit"
md_url: "https://www.modelscope.cn/models/mlx-community/KernelLLM-6bit.md"
repository: mlx-community/KernelLLM-6bit
last_updated: 2025-05-23
license: other
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - llama
architectures:
  - LlamaForCausalLM
base_model:
  - facebook/KernelLLM
base_model_relation: quantized
parameters: 1.8B
tensor_type:
  - F16
  - U32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
inference_backends:
  - "deploy_task text/emb"
  - "lmdeploy 0.9.1"
  - "lmdeploy_turbomind 0.9.1"
  - "sglang 0.5.2"
  - "vllm 0.9.2"
downloads: 314
stars: 0
tags:
  - mlx
---

# KernelLLM-6bit

> KernelLLM-6bit - mlx-community 在 ModelScope 开源的模型。mlx-community/KernelLLM-6bit

mlx-community/KernelLLM-6bit 是 ModelScope 魔搭社区上的 1.8B 参数text-generation模型，采用 other 许可，基于 facebook/KernelLLM 构建，可用 deploy_task text/emb、lmdeploy 0.9.1、lmdeploy_turbomind 0.9.1 部署。

- **Repository**: mlx-community/KernelLLM-6bit
- **License**: other
- **Tasks**: text-generation
- **Parameters**: 1.8B
- **Base model**: facebook/KernelLLM
- **Inference backends**: deploy_task text/emb, lmdeploy 0.9.1, lmdeploy_turbomind 0.9.1, sglang 0.5.2, vllm 0.9.2
- **Tags**: mlx
- **Downloads**: 314
- **Stars**: 0
- **Last updated**: 2025-05-23

Source: https://www.modelscope.cn/models/mlx-community/KernelLLM-6bit

---

# mlx-community/KernelLLM-6bit

The Model [mlx-community/KernelLLM-6bit](https://huggingface.co/mlx-community/KernelLLM-6bit) was converted to MLX format from [facebook/KernelLLM](https://huggingface.co/facebook/KernelLLM) using mlx-lm version **0.22.3**.

## Use with mlx

```bash
pip install mlx-lm
```

```python
from mlx_lm import load, generate

model, tokenizer = load("mlx-community/KernelLLM-6bit")

prompt="hello"

if hasattr(tokenizer, "apply_chat_template") and tokenizer.chat_template is not None:
    messages = [{"role": "user", "content": prompt}]
    prompt = tokenizer.apply_chat_template(
        messages, tokenize=False, add_generation_prompt=True
    )

response = generate(model, tokenizer, prompt=prompt, verbose=True)
```