---
title: gemma-4-e4b-it-4bit-MAD
canonical_url: "https://www.modelscope.cn/models/mlx-community/gemma-4-e4b-it-4bit-MAD"
md_url: "https://www.modelscope.cn/models/mlx-community/gemma-4-e4b-it-4bit-MAD.md"
repository: mlx-community/gemma-4-e4b-it-4bit-MAD
last_updated: 2026-05-21
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - gemma4
architectures:
  - Gemma4ForConditionalGeneration
parameters: 1.1B
tensor_type:
  - BF16
  - U32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
downloads: 18
stars: 0
tags:
  - mlx
  - gemma
  - quantized
  - 4bit-MAD
---

# gemma-4-e4b-it-4bit-MAD

> gemma-4-e4b-it-4bit-MAD - mlx-community 在 ModelScope 开源的模型。Mixed-precision Alternating-least-squares DWQ export for the local Gemma 4 E4B instruction model.

- **Repository**: mlx-community/gemma-4-e4b-it-4bit-MAD
- **Tasks**: text-generation
- **Parameters**: 1.1B
- **Tags**: mlx, gemma, quantized, 4bit-MAD
- **Downloads**: 18
- **Stars**: 0
- **Last updated**: 2026-05-21

Source: https://www.modelscope.cn/models/mlx-community/gemma-4-e4b-it-4bit-MAD

---

# 4bit-MAD

Mixed-precision Alternating-least-squares DWQ export for the local Gemma 4 E4B
instruction model.

This is a full standalone MLX model export, not a scale/bias-only checkpoint.
It was produced from the dynamic `4b128/6b64` mixed quant at approximately
`4.5 bpw`, initialized with alternating round + least-squares for 30 iterations,
then trained with the n1024 cosine DWQ recipe.

Canonical old-heldout KL against `/Users/natebreslow/Documents/gemmaOpt/gemma-4-E4B-it`:

```text
avg_kl=0.095879
total_kl=514.967500
total_tokens=5371
```

The eval log is included as `oldheldout_kl.log`.