README.md

---
base_model: AlignmentResearch/Llama-3.3-Tiny-Instruct
---
# Random LoRA Adapter for tiny-random-Llama-3

This is a randomly initialized LoRA adapter for the `AlignmentResearch/Llama-3.3-Tiny-Instruct` model.

## Details
- **Base model**: AlignmentResearch/Llama-3.3-Tiny-Instruct
- **Seed**: 1
- **LoRA rank**: 16
- **LoRA alpha**: 32
- **Target modules**: q_proj, v_proj, k_proj, o_proj

## Usage

```python
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

# Load base model
base_model = AutoModelForCausalLM.from_pretrained("AlignmentResearch/Llama-3.3-Tiny-Instruct")
tokenizer = AutoTokenizer.from_pretrained("AlignmentResearch/Llama-3.3-Tiny-Instruct")

# Load LoRA adapter
model = PeftModel.from_pretrained(base_model, "AlignmentResearch/Llama-3.3-Tiny-Instruct-lora-1")
```

This adapter was created for testing purposes and contains random weights.