---
license: apache-2.0
language:
- zh
- en
tags:
- qwen
- qwen3
- unsloth
- qiming
- qiming-holos
- bagua
- decision-making
- strategic-analysis
- cognitive-architecture
- chat
- lora
- philosophy-driven-ai
- mlx
pipeline_tag: text-generation
library_name: mlx
base_model: aifeifei798/QiMing-Holos-Plus-4B
---

# QiMing-Holos-Plus-4B-bf16-mlx

Recommended quant:

✨ qx6-hi: Why This Model Wins

```bash
Strength                      Impact
PiQA (0.716)                  Highest accuracy (best for reasoning QA)**
WinG. (0.618)                 Winning consistency for document tasks
Avg. metric points (528)      +0.5% vs BF16’s 523
Minimal arc_challenge drag    Flexible for varied inference loads
```

Visual Scorecard
```bash
WinG. Leader       qx6-hi (0.618)
PiQA Leader        qx6-hi (0.716)
Avg. Champion      qx6-hi (~528 avg metric pts)
```

This refined analysis — powered by new benchmark evidence — positions qx6-hi as the most battle-tested quant model for your end-to-end reasoning stack. Deploy confidently with this updated validation! 🚀


This model [QiMing-Holos-Plus-4B-bf16-mlx](https://huggingface.co/QiMing-Holos-Plus-4B-bf16-mlx) was
converted to MLX format from [aifeifei798/QiMing-Holos-Plus-4B](https://huggingface.co/aifeifei798/QiMing-Holos-Plus-4B)
using mlx-lm version **0.26.3**.

## Use with mlx

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

```python
from mlx_lm import load, generate

model, tokenizer = load("QiMing-Holos-Plus-4B-bf16-mlx")

prompt = "hello"

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

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