Model Card for Model ID

  • KT-AI/midm-bitext-S-7B-inst-v1

Model Details

Model Description

  • NSMC ์˜ํ™” ๋ฆฌ๋ทฐ ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•˜์—ฌ KT-AI/midm-bitext-S-7B-inst-v1 ๋ฏธ์„ธํŠœ๋‹.
  • ์ž…๋ ฅ ํ”„๋กฌํ”„ํŠธ๋ฅผ ์ด์šฉํ•˜์—ฌ ๋ฐ์ดํ„ฐ์…‹์˜ document(๋ฆฌ๋ทฐ)๊ฐ€ ๊ธ์ •์ ์ธ ๋‚ด์šฉ์ด๋ฉด '1'์„ ๋ถ€์ •์ ์ธ ๋‚ด์šฉ์ด๋ฉด '0'์„ ์˜ˆ์ธกํ•˜๋„๋ก ํ•จ.
  • train data: nsmc train ์ƒ์œ„ 2000๊ฐœ ์ƒ˜ํ”Œ ์ด์šฉ
  • test data: nsmc test ์ƒ์œ„ 2000๊ฐœ ์ƒ˜ํ”Œ ์ด์šฉ

Training Data

'nsmc'

  • ์ƒ์œ„ 2000๊ฐœ ๋ฐ์ดํ„ฐ ์ด์šฉ

Training Procedure

  • prepare_sample_text์— ๋ฆฌ๋ทฐ๋ฅผ ๊ธ์ •/๋ถ€์ •์œผ๋กœ ํŒ๋‹จํ•˜๋„๋ก ์ž…๋ ฅ ํ”„๋กฌํ”„ํŠธ ์ˆ˜์ •ํ•˜์˜€์Œ.

Training Hyperparameters

  • per_device_train_batch_size: 1
  • per_device_eval_batch_size: 1
  • learning_rate: 1e-4
  • gradient_accumulation_steps: 2
  • optimizer: paged_adamw_32bit
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_ratio: 0.03
  • training_args.logging_steps: 50
  • training_args.max_steps : 1000
  • trainable params: trainable params: 16,744,448 || all params: 7,034,347,520 || trainable%: 0.23803839591934178

Results

TrainOutput(global_step=1000, training_loss=1.0208648338317872, metrics={'train_runtime': 1128.0266, 'train_samples_per_second': 1.773, 'train_steps_per_second': 0.887, 'total_flos': 3.1051694997504e+16, 'train_loss': 1.0208648338317872, 'epoch': 1.0})

Accruacy

๋ฏธ์„ธํŠœ๋‹ ํ›„ ๋ชจ๋ธ์˜ ์ •ํ™•๋„:0.61

Downloads last month
3
Inference API
Unable to determine this modelโ€™s pipeline type. Check the docs .

Model tree for Rfy23/hw-llama-2-7B-nsmc

Adapter
(16)
this model