Downloads · 30 days
21
32% of all-time downloads
Phi-3.5-MoE-instruct-W4A16-AutoRound
Kris2017/Phi-3.5-MoE-instruct-W4A16-AutoRound
Phi-3.5-MoE-instruct-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 mit.
This is the verified W4A16 AutoRound checkpoint used in the DynaExQ paper experiments. It is derived from microsoft/Phi-3.5-MoE-instruct.
All-time downloads
66
Public
Parameters
5.8B
22.1 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors22.1 GB · 100%
Parameter types
How the weights are stored.
I325.2B · 90%
Try a prompt
Base models
Model card
From the Hugging Face model README
Phi-3.5-MoE-instruct W4A16 AutoRound
This is the verified W4A16 AutoRound checkpoint used in the DynaExQ paper experiments. It is derived from microsoft/Phi-3.5-MoE-instruct.
The checkpoint contains 11 safetensors shards, its weight index, tokenizer files, PhiMoE modeling code, and quantization provenance. Loading custom PhiMoE code may require trust_remote_code=True; review remote code before enabling it.
from transformers import AutoModelForCausalLM, AutoTokenizer
model_id = "Kris2017/Phi-3.5-MoE-instruct-W4A16-AutoRound"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
model_id,
device_map="auto",
torch_dtype="auto",
trust_remote_code=True,
)
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
- phimoe
- License
- mit
- Created
- Aug 3, 2026
- Updated
- Aug 3, 2026