Downloads · 30 days
8
3% of all-time downloads
psudiis/payoon-gemma4-31b-cpt-fp8
payoon-gemma4-31b-cpt-fp8 is a machine learning model from psudiis. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
This variant uses LLM Compressor's FP8DYNAMIC scheme in the vLLM compressed-tensors format: linear weights are stored as FP8 with per-channel scales and input activations are quantized dynamically per token. The visio…
Downloads · 30 days
8
3% of all-time downloads
All-time downloads
290
Public
Parameters
32.7B
36.1 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors36.1 GB · 100%
How the weights are stored.
F8_E4M329.3B · 90%
From the Hugging Face model README
This variant uses LLM Compressor's FP8_DYNAMIC scheme in the vLLM compressed-tensors format: linear weights are stored as FP8 with per-channel scales and input activations are quantized dynamically per token. The vision tower, lm_head, and token embeddings remain in their original precision. Quantization was data-free and did not modify the merged source checkpoint.