---
title: quanttesting
canonical_url: "https://www.modelscope.cn/models/ApacheOne/quanttesting"
md_url: "https://www.modelscope.cn/models/ApacheOne/quanttesting.md"
repository: ApacheOne/quanttesting
last_updated: 2026-03-24
license: apache-2.0
model_type:
  - custom
base_model:
  - black-forest-labs/FLUX.2-klein-9B
base_model_relation: quantized
parameters: 10.8B
tensor_type:
  - F32
  - BF16
  - U8
  - F8_E4M3
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
downloads: 31
stars: 0
tags:
  - art
---

# quanttesting

> quanttesting - ApacheOne 在 ModelScope 开源的模型。Pin down what is needed between 50s,40s,30s cards

ApacheOne/quanttesting 是 ModelScope 魔搭社区上的 10.8B 参数机器学习模型，采用 apache-2.0 许可，基于 black-forest-labs/FLUX.2-klein-9B 构建。

- **Repository**: ApacheOne/quanttesting
- **License**: apache-2.0
- **Parameters**: 10.8B
- **Base model**: black-forest-labs/FLUX.2-klein-9B
- **Tags**: art
- **Downloads**: 31
- **Stars**: 0
- **Last updated**: 2026-03-24

Source: https://www.modelscope.cn/models/ApacheOne/quanttesting

---

# Just some testing

Pin down what is needed between 50s,40s,30s cards

Translate info for use in TensorRT fp4 converting / onnx 

Anyone that has made fp4 onnx before with NVIDIA’s Model Optimizer PTQ help in the Community tab is welcomed.

Currently working on correct Quantize/Dequantize nodes for onnx export to build fp4 RT engine for 50s cards. 

## BL Klein 9B NVFP4 Variants — `txt_attn` dtype notes

### bl_Klein9b-nvfp4.safetensors
- `txt_attn` weights: **NVFP4**

### bl_Klein9b-nvfp4_BF16.safetensors
- `txt_attn` weights: **BF16**