---
title: MiniMax-H3_tests
canonical_url: "https://www.modelscope.cn/models/silveroxides/MiniMax-H3_tests"
md_url: "https://www.modelscope.cn/models/silveroxides/MiniMax-H3_tests.md"
repository: silveroxides/MiniMax-H3_tests
last_updated: 2026-11-01
base_model:
  - MiniMaxAI/MiniMax-H3
base_model_relation: adapter
parameters: 391.3B
tensor_type:
  - I8
  - U8
  - BF16
  - F32
  - F16
library_name:
  - lora
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 785
stars: 4
tags:
  - experiments
  - research
  - diffusion
  - lora
---

# MiniMax-H3_tests

> MiniMax-H3_tests - silveroxides 在 ModelScope 开源的模型。Repo for my experiments and tests with MiniMax H3

silveroxides/MiniMax-H3_tests 是 ModelScope 魔搭社区上的 391.3B 参数机器学习模型，基于 MiniMaxAI/MiniMax-H3 构建。

- **Repository**: silveroxides/MiniMax-H3_tests
- **Parameters**: 391.3B
- **Base model**: MiniMaxAI/MiniMax-H3
- **Tags**: experiments, research, diffusion, lora
- **Downloads**: 785
- **Stars**: 4
- **Last updated**: 2026-11-01

Source: https://www.modelscope.cn/models/silveroxides/MiniMax-H3_tests

---

# Repo for my experiments and tests with MiniMax H3

## Current full and pruned turbo LoRA is located at [silver_official_releases](./silver_official_releases)  
### The v2 ones have additional attention config layers optimized for inference.
### No need to use any attention backend patcher with those.