10Eros-Max H3 — INT8 ConvRot

INT8 ConvRot conversion of the 10Eros-Max H3 diffusion model.

Source

Original model: TenStrip/10Eros-Max
Source file: 10Eros_Max_h3_fl2va_beta2_pruned.safetensors
Source dtype: BF16

Conversion

Converted with Comfy-Org/comfy-model-tools using Comfy Kitchen’s native INT8 tensorwise ConvRot format.

Quantized layers: 208
ConvRot groupsize: 256
Quantization: per-channel absmax
Native format marker: int8_tensorwise with ConvRot enabled
Output size: approximately 20.2 GB

Place the safetensors file in ComfyUI’s models/diffusion_models directory and load it with a ComfyUI version that supports native INT8 ConvRot models.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for cicalooo/10Eros-Max-h3-int8-convrot

Base model

MiniMaxAI/MiniMax-H3

Finetuned

TenStrip/10Eros-Max

Finetuned

(2)

this model