Downloads · 30 days
0
propagation/kogpt2-chatbot-lora
kogpt2-chatbot-lora is a machine learning model from propagation. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as cc-by-nc-nd-4.0.
Downloads · 30 days
0
Access
Public
Updated Nov 5, 2025
Repo size
324 MB
Likes
0
Public
Click a slice to open those files.
.safetensors324 MB · 99%
From the Hugging Face model README
r=16,
lora_alpha=32,
target_modules=["c_attn", "c_proj", "c_fc"],
lora_dropout=0.05,
bias="none",
task_type=TaskType.CAUSAL_LM
num_train_epochs=10,
per_device_train_batch_size=4
per_device_eval_batch_size=8,
gradient_accumulation_steps=4,
learning_rate=0.0002,
warmup_steps=100,
logging_steps=50,
eval_strategy= "epoch",
eval_steps=100,
save_strategy= "epoch",
save_steps=100,
load_best_model_at_end=True,
fp16=True,
report_to="none",
weight_decay=0.01,
from peft import PeftModel
# 베이스 모델 로드 (분류용)
print("베이스 모델 로딩")
base_model_reload = AutoModelForSequenceClassification.from_pretrained(
"klue/bert-base",
num_labels=2
)
# 업로드한 LoRA 어댑터 로드
print(f"LoRA 어댑터 로딩: propagation/kogpt2-chatbot-lora")
model_reload = PeftModel.from_pretrained(base_model_reload, model_name_upload)
tokenizer_reload = AutoTokenizer.from_pretrained(model_name_upload)
# GPU로 이동
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model_reload = model_reload.to(device)
model_reload.eval()
print("모델 로드 완료!")