Qwen3-8B-Engineer7

Qwen3-8B-Engineer7

This is a task-arithmetic merge of the models:

• qingy2024/Qwen3-VLTO-8B-Thinking
• TeichAI/Qwen3-8B-Gemini-3-Pro-Preview-Distill
• TeichAI/Qwen3-8B-Gemini-2.5-Flash-Distill

Brainwave

          arc   arc/e boolq hswag obkqa piqa  wino
qx86-hi   0.526,0.710,0.820,0.700,0.426,0.784,0.682

Qwen3-VLTO-8B-Thinking
qx86x-hi  0.475,0.599,0.706,0.638,0.402,0.765,0.684

Qwen3-8B-Gemini-2.5-Flash-Distill
qx86-hi   0.398,0.430,0.420,0.608,0.364,0.753,0.615

Qwen3-8B-Gemini-3-Pro-Preview-Distill
qx86-hi   0.379,0.440,0.602,0.592,0.344,0.717,0.595

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("Qwen3-8B-Engineer7")

prompt = "hello"

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

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