Add LoRA adapter weights and config

LoRA adapter for meta-llama/Llama-3.1-8B-Instruct.

This model is fine-tuned for generating contrastive explanations for OWL reasoning tasks.

Method:
- LoRA fine-tuning (r=8, alpha=16)
- Trained using domain-specific OWL reasoning datasets

Usage:

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base_model = "meta-llama/Llama-3.1-8B-Instruct"

tokenizer = AutoTokenizer.from_pretrained(base_model)
base = AutoModelForCausalLM.from_pretrained(base_model)

model = PeftModel.from_pretrained(base, "your-username/your-repo-name")

Notes:
- This repository contains only LoRA adapter weights
- Base model must be loaded separately

adapter_config.json

@@ -0,0 +1 @@
+{"r": 8, "lora_alpha": 16, "target_modules": ["q_proj", "v_proj", "o_proj", "gate_proj", "down_proj", "up_proj"], "peft_type": "LORA", "base_model_name_or_path": "meta-llama/Meta-Llama-3.1-8B-Instruct"}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4db42dac6d70ee851de4ea8c11e99c328bc4dd801701e65e8024639d1e0106ba
+size 39372544