from huggingface_hub import notebook_login
import torch
import transformers
from transformers import AutoModelForMaskedLM, AutoTokenizer
from transformers import pipeline
from transformers import LlamaTokenizer, LlamaForCausalLM
import time
import csv
import locale
locale.getpreferredencoding = lambda: "UTF-8"


notebook_login()
checkpoint = "ziqingyang/chinese-alpaca-2-7b"
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
model = AutoModelForMaskedLM.from_pretrained(checkpoint)

!CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
    --stage sft \
    --model_name_or_path model \
    --do_train \
    --dataset mydata \
    --template chatglm2 \
    --finetuning_type lora \
    --lora_target q_proj,v_proj \
    --output_dir ckpoint \
    --overwrite_cache \
    --auto_find_batch_size=False \
    --per_device_train_batch_size 4 \
    --gradient_accumulation_steps 4 \
    --lr_scheduler_type cosine \
    --logging_steps 10 \
    --save_steps 10 \
    --learning_rate 5e-5 \
    --num_train_epochs 3.0 \
    --plot_loss \
    --fp16


model.push_to_hub("gutalk_classification") 
tokenizer.push_to_hub("gutalk_classification")