BOBsssss/gemma2-2b-customer-support-lora

Gemma-2-2B LoRA -- Customer Support

LoRA adapter fine-tuned from google/gemma-2-2b-it on the Bitext customer support dataset (26k+ single-turn support instructions/responses, 11 categories / 27 intents). Trained in fp16/bf16 (no quantization), loss masked to response tokens only.

Usage (Transformers + PEFT)

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base = AutoModelForCausalLM.from_pretrained("google/gemma-2-2b-it")
model = PeftModel.from_pretrained(base, "BOBsssss/gemma2-2b-customer-support-lora")
tokenizer = AutoTokenizer.from_pretrained("BOBsssss/gemma2-2b-customer-support-lora")

Usage (Ollama)

huggingface-cli download BOBsssss/gemma2-2b-customer-support-lora --local-dir ./adapter
ollama pull gemma2:2b
cat > Modelfile <<EOF
FROM gemma2:2b
ADAPTER ./adapter
EOF
ollama create support-bot -f Modelfile
ollama run support-bot "I need to cancel order 12345"