> 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.

# Devstral-2-123B-Instruct-2512

> Fine-tune Devstral-2-123B-Instruct-2512 for text generation with NeMo AutoModel using the checked-in single-node tensor-parallel LoRA recipe.

[Devstral-2-123B-Instruct-2512](https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512) is a text-only, code-focused checkpoint from Mistral AI. NeMo AutoModel includes a LoRA fine-tuning recipe for the checkpoint.

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 Devstral-2-123B-Instruct-2512

Run the checked-in recipe on one node with eight GPUs:

```bash
uv run automodel examples/llm_finetune/devstral/devstral2_123b_2512_squad_peft.yaml \
  --nproc-per-node 8
```

## Choose a Workflow

| Goal                      | Start Here                                                                                                                                                            |
| ------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| LoRA fine-tuning on SQuAD | Use [devstral2\_123b\_2512\_squad\_peft.yaml](https://github.com/NVIDIA-NeMo/Automodel/blob/main/examples/llm_finetune/devstral/devstral2_123b_2512_squad_peft.yaml). |

## Model Reference

### Model Architecture

| Property                  | Value                                         |
| ------------------------- | --------------------------------------------- |
| Task                      | Text Generation                               |
| Architecture              | `Ministral3ForCausalLM`                       |
| Strategy                  | FSDP2 with tensor parallelism 8 and LoRA      |
| Hugging Face Organization | [mistralai](https://huggingface.co/mistralai) |

### Available Models

* [`mistralai/Devstral-2-123B-Instruct-2512`](https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512)

## Related Resources

* [LLM Fine-Tuning Guide](/recipes-e2e-examples/sft-peft)