> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/nemo/automodel/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/nemo/automodel/_mcp/server.

# diffusiongemma-26B-A4B-it

> Use diffusiongemma-26B-A4B-it with NeMo AutoModel for diffusion language model fine-tuning, including checkpoints, runnable recipes, setup, and reference details.

[DiffusionGemma](https://huggingface.co/google) is a block-diffusion language model from Google. Instead of generating tokens left-to-right, it denoises a fixed-length canvas of tokens in parallel: a causal encoder reads the prompt and a bidirectional decoder iteratively refines the response canvas. The released checkpoint is a Mixture-of-Experts model with 26B total parameters and \~4B active per token.

Set up NeMo AutoModel with the [latest container](https://catalog.ngc.nvidia.com/orgs/nvidia/containers/nemo-automodel) or follow the [installation instructions](/get-started/installation).

## Fine-Tune diffusiongemma-26B-A4B-it

Install the repository's locked environment:

```bash
uv sync --locked --all-groups --extra all
```

Generate the GSM8K chat dataset required by both DiffusionGemma recipes. Run this from the repository root; it writes `gsm8k_chat_train.jsonl`, the path used by both YAML files:

```bash
uv run python examples/dllm_sft/prep_gsm8k.py
```

> **Note**
>
> This recipe was validated with **Expert Parallelism (EP=8)** on a single 8xH100 node. See the [Launcher Guide](/job-launchers/slurm-cluster) for multi-node setup.

Then run either recipe:

```bash
# Full SFT
uv run automodel examples/dllm_sft/diffusion_gemma_sft.yaml --nproc-per-node 8

# LoRA SFT
uv run automodel examples/dllm_sft/diffusion_gemma_lora.yaml --nproc-per-node 8
```

## Choose a Workflow

| Goal                                                                                    | Start Here                                                                                                                         |
| --------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------- |
| Full SFT - DiffusionGemma 26B-A4B with FSDP2 + Expert Parallelism                       | Use [diffusion\_gemma\_sft.yaml](https://github.com/NVIDIA-NeMo/Automodel/blob/main/examples/dllm_sft/diffusion_gemma_sft.yaml).   |
| Low-rank adaptation (LoRA) SFT - DiffusionGemma 26B-A4B with FSDP2 + Expert Parallelism | Use [diffusion\_gemma\_lora.yaml](https://github.com/NVIDIA-NeMo/Automodel/blob/main/examples/dllm_sft/diffusion_gemma_lora.yaml). |

## Model Reference

### Model Architecture

| Property                  | Value                                   |
| ------------------------- | --------------------------------------- |
| Task                      | Text Generation (Block Diffusion, MoE)  |
| Architecture              | `DiffusionGemmaForBlockDiffusion`       |
| Parameters                | 26B total / \~4B active                 |
| Hugging Face Organization | [google](https://huggingface.co/google) |

* `DiffusionGemmaForBlockDiffusion` - block-diffusion MoE (causal prompt encoder + bidirectional canvas decoder).

### Available Models

| Model                     | HF ID                                                                                         |
| ------------------------- | --------------------------------------------------------------------------------------------- |
| DiffusionGemma 26B-A4B-it | [`google/diffusiongemma-26B-A4B-it`](https://huggingface.co/google/diffusiongemma-26B-A4B-it) |

## Related Resources

* [DiffusionGemma Fine-Tuning Guide](/recipes-e2e-examples/diffusiongemma)