Downloads · 30 days
9
9% of all-time downloads
hassanshka/Qwen3-32B-FP8-to-BF16
Qwen3-32B-FP8-to-BF16 is a text generation model from hassanshka. 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 a dequantized BF16 version of Qwen3-32B-FP8. It has been converted from FP8 → BF16 format.
Downloads · 30 days
9
9% of all-time downloads
All-time downloads
105
Public
Parameters
32.8B
65.6 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors65.5 GB · 100%
From the Hugging Face model README
This is a dequantized BF16 version of Qwen3-32B-FP8. It has been converted from FP8 → BF16 format.
This "Bridge Model" serves several purposes:
| Parameter | Value |
|---|---|
| Source | Qwen3-32B-FP8 |
| Target Dtype | BFloat16 |
| Method | FP8 → BF16 dequantization |
⚠️ This model is NOT identical to the original BF16 model.
The dequantization process recovers an approximation:
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
model = AutoModelForCausalLM.from_pretrained(
"hassanshka/Qwen3-32B-FP8-to-BF16",
device_map="auto",
torch_dtype=torch.bfloat16,
trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("hassanshka/Qwen3-32B-FP8-to-BF16")
# Use for inference or fine-tuning
Apache 2.0
If you use this model, please cite the original Qwen3 model.