nohurry/Opus-4.6-Reasoning-3000x-filtered
Viewer • Updated • 2.33k • 5.34k • 590
How to use avalon2244/Qwen3.5-4B-Claude-Opus-4.6-Distilled with Unsloth Studio:
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 avalon2244/Qwen3.5-4B-Claude-Opus-4.6-Distilled to start chatting
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 avalon2244/Qwen3.5-4B-Claude-Opus-4.6-Distilled to start chatting
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for avalon2244/Qwen3.5-4B-Claude-Opus-4.6-Distilled to start chatting
pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
model_name="avalon2244/Qwen3.5-4B-Claude-Opus-4.6-Distilled",
max_seq_length=2048,
)This terribly named model was a quick finetune of Qwen3.5-4B on the nohurry/Opus-4.6-Reasoning-3000x-filtered dataset. It tends to have cleaner reasoning traces than the original Qwen3.5-4B, and is around as accurate. I haven't tested it, though. This model was finetuned and converted to GGUF format using Unsloth.
It's a bit inconsistent with reasoning. It's far less likely to enter endless loops, and is uses far fewer tokens than the original model. But it's still a 4B model that's been finetuned on one dataset, so it's not fantastic.
You can find a few GGUF quants of this model here.
Qwen3.5-4B.Q5_K_M.ggufQwen3.5-4B.Q8_0.ggufQwen3.5-4B.Q4_K_M.ggufQwen3.5-4B.BF16-mmproj.gguf
This was trained 2x faster with Unsloth
