Downloads · 30 days
10
19% of all-time downloads
Qwen3-30B-A3B-Instruct-2507-W4A16-AutoRound
Kris2017/Qwen3-30B-A3B-Instruct-2507-W4A16-AutoRound
Qwen3-30B-A3B-Instruct-2507-W4A16-AutoRound is a text generation model from Kris2017. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as apache-2.0.
This is the mixed-precision INT4/W4A16 AutoRound checkpoint used in the DynaExQ paper experiments. It is derived from Qwen/Qwen3-30B-A3B-Instruct-2507.
All-time downloads
54
Public
Parameters
635M
17.8 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors17.8 GB · 100%
Parameter types
How the weights are stored.
I323.9B · 78%
Try a prompt
Base models
Model card
From the Hugging Face model README
Qwen3-30B-A3B-Instruct-2507 W4A16 AutoRound
This is the mixed-precision INT4/W4A16 AutoRound checkpoint used in the DynaExQ paper experiments. It is derived from Qwen/Qwen3-30B-A3B-Instruct-2507.
The checkpoint uses symmetric expert-weight quantization with mixed higher-precision fallbacks recorded in quantization_config.json.
from transformers import AutoModelForCausalLM, AutoTokenizer
model_id = "Kris2017/Qwen3-30B-A3B-Instruct-2507-W4A16-AutoRound"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
model_id,
device_map="auto",
torch_dtype="auto",
)
Code, experiment scripts, manifests, and paper sources are available in DynaQuant. The original model license and usage restrictions continue to apply.
- Task
- Text Generation
- Library
- transformers
- Type
- qwen3_moe
- License
- apache-2.0
- Created
- Aug 3, 2026
- Updated
- Aug 3, 2026