kleverboots/nope-edge-llama-lora

NOPE-Edge LLaMA 3.2 3B LoRA Adapters

LoRA adapters for mental health crisis detection, trained on LLaMA 3.2 3B.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

# Load base model
base_model = AutoModelForCausalLM.from_pretrained("unsloth/Llama-3.2-3B-Instruct")
tokenizer = AutoTokenizer.from_pretrained("unsloth/Llama-3.2-3B-Instruct")

# Load LoRA adapters
model = PeftModel.from_pretrained(base_model, "kleverboots/nope-edge-llama-lora")
model = model.merge_and_unload()

Performance

Metric	Score
F1 Score	93.1%
Recall	92.4%
Precision	93.8%

Training Details

• LoRA Config: r=16, alpha=16, dropout=0.05
• Target Modules: q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
• Trainable Params: 24.3M (0.75% of 3.2B)