Bemærk
Adgang til denne side kræver godkendelse. Du kan prøve at logge på eller ændre mapper.
Adgang til denne side kræver godkendelse. Du kan prøve at ændre mapper.
These examples fine-tune and post-train large language models (LLMs) on AI Runtime. They cover parameter-efficient methods like Low-Rank Adaptation (LoRA) and full supervised fine-tuning across libraries including TRL, Unsloth, Axolotl, and LLM Foundry. For batch inference examples, see Batch inference.
| Example | Interface | Description |
|---|---|---|
| Reinforcement learning of Gemma4-2B | Notebook | Post-train Gemma4-2B with Group Relative Policy Optimization (GRPO) and Low-Rank Adaptation (LoRA) on a single H100 GPU using Unsloth. |
| Supervised Fine-tuning (Full) of Qwen3-4B | Notebook | Full-weight fine-tune Qwen3-4B on a single H100 GPU with TRL, using BF16 mixed precision and gradient checkpointing. |
| Supervised Fine-tuning (LoRA) of Llama-3.2-3B | Notebook | LoRA fine-tune Llama-3.2-3B on a single GPU with the Unsloth library. |
| Supervised Fine-tuning (LoRA) of GPT-OSS 20B | Notebook | LoRA fine-tune GPT-OSS 20B on 8 H100 GPUs with TRL and distributed data parallelism. |
| Supervised Fine-tuning (Full) of Llama-3.2-1B | Notebook | Full-weight fine-tune Llama-3.2-1B on 8 H100 GPUs with TRL and DeepSpeed ZeRO Stage 3. |
| Supervised Fine-tuning (LoRA) of Olmo3 7B | Notebook | LoRA fine-tune Olmo3 7B across multiple GPUs with the Axolotl library. |
| Supervised fine-tuning (Full) and serving of Qwen3.5-0.8B | Notebook | Fine-tune Qwen3.5-0.8B on a single H100 GPU with TRL, then serve it behind a vLLM Model Serving endpoint. |
| Supervised Fine-tuning of Llama-3.2-3B on multi-GPUs | Notebook | LoRA fine-tune Llama-3.2-3B across multiple GPUs with the Unsloth library. |
| Supervised Fine-tuning (LoRA) of GPT-OSS 120B | Notebook | LoRA fine-tune GPT-OSS 120B on 8 H100 GPUs with TRL, using DDP and FSDP. |
| Training transformers on multi-GPUs | Notebook | Train Transformer models using PyTorch Fully Sharded Data Parallel (FSDP) to shard model parameters across multiple GPUs. |
| Full-weight fine-tuning of Qwen3-8B for invoice extraction | CLI | Fine-tune Qwen3-8B for invoice and receipt entity extraction with TRL and DeepSpeed ZeRO-3 on one 8x H100 node. |
| Supervised fine-tuning of Qwen2.5-3B with Ray Train | CLI | Use Ray Train to fine-tune Qwen2.5-3B with one distributed data parallel worker per GPU on an 8x H100 node. |
| Longformer fine-tuning for long-document classification | CLI | Fine-tune Longformer with LoRA for long-document sentiment classification on one GPU and save the adapter to a Unity Catalog volume. |
| ModernBERT fine-tuning on one GPU | CLI | Fine-tune ModernBERT for topic classification on one A10 GPU and register the model in Unity Catalog. |
| ModernBERT fine-tuning with DDP | CLI | Fine-tune ModernBERT for topic classification with PyTorch DDP on an 8x H100 node and register the model in Unity Catalog. |
| LoRA fine-tuning of Nemotron 3.5 Lightning | CLI | Fine-tune Nemotron 3.5 Lightning with NeMo AutoModel, FSDP2, and expert parallelism on an 8x H100 node. |
| LoRA fine-tuning of Nemotron 3.5 Super VL | CLI | Fine-tune the Nemotron 3.5 Super VL vision-language model with NeMo AutoModel and LoRA on an 8x H100 node. |
| Fine-tune and serve Qwen2.5-0.5B | CLI | Fine-tune Qwen2.5-0.5B with LoRA, register the merged model in Unity Catalog, and deploy it to Custom Model Serving. |
Video demo
This video walks through the Supervised Fine-tuning (LoRA) of Llama-3.2-3B example notebook in detail (12 minutes).