# Guide de permutation à chaud LoRA

### :shaved\_ice: vLLM LoRA Hot Swapping / LoRA dynamiques

Pour activer le service LoRA pour au plus 4 LoRA à la fois (qui sont échangés à chaud / modifiés), définissez d'abord le drapeau d'environnement pour autoriser l'échange à chaud :

```bash
export VLLM_ALLOW_RUNTIME_LORA_UPDATING=True
```

Ensuite, servez-le avec la prise en charge de LoRA :

```bash
export VLLM_ALLOW_RUNTIME_LORA_UPDATING=True
vllm serve unsloth/Llama-3.1-8B-Instruct \
    --quantization fp8 \
    --kv-cache-dtype fp8
    --gpu-memory-utilization 0.8 \
    --max-model-len 65536 \
    --enable-lora \
    --max-loras 4 \
    --max-lora-rank 64
```

Pour charger un LoRA dynamiquement (définissez également le nom du lora), faites :

```bash
curl -X POST http://localhost:8000/v1/load_lora_adapter \
    -H "Content-Type: application/json" \
    -d '{
        "lora_name": "LORA_NAME",
        "lora_path": "/path/to/LORA"
    }'
```

Pour le retirer du pool :

```bash
curl -X POST http://localhost:8000/v1/unload_lora_adapter \
    -H "Content-Type: application/json" \
    -d '{
        "lora_name": "LORA_NAME"
    }'
```

Par exemple lors du fine-tuning avec Unsloth :

{% code overflow="wrap" %}

```python
from unsloth import FastLanguageModel
import torch
model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = "unsloth/Llama-3.1-8B-Instruct",
    max_seq_length = 2048,
    load_in_4bit = True,
)
model = FastLanguageModel.get_peft_model(model)
```

{% endcode %}

Puis après l'entraînement, nous sauvegardons les LoRA :

```python
model.save_pretrained("finetuned_lora")
tokenizer.save_pretrained("finetuned_lora")
```

Nous pouvons ensuite charger le LoRA :

{% code overflow="wrap" %}

```bash
curl -X POST http://localhost:8000/v1/load_lora_adapter \
    -H "Content-Type: application/json" \
    -d '{
        "lora_name": "LORA_NAME_finetuned_lora",
        "lora_path": "finetuned_lora"
    }'
```

{% endcode %}


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://unsloth.ai/docs/fr/bases/inference-and-deployment/vllm-guide/lora-hot-swapping-guide.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.