Model Card for Model ID

  • mdim λͺ¨λΈμ„ ν•œκ΅­μ–΄ μ˜ν™” 리뷰 데이터셋(NSMC)을 ν•΄κ²°ν•˜λŠ” λͺ¨λΈμ΄ λ˜λ„λ‘ λ―Έμ„ΈνŠœλ‹
  • ν•œκ΅­ μ˜ν™” 리뷰의 긍정 λ˜λŠ” 뢀정을 νŒλ‹¨ν•˜λŠ” λͺ¨λΈμ„ ν•™μŠ΅

Model Details

λͺ¨λΈ: KT-AI/midm-bitext-S-7B-inst-v1 ν•™μŠ΅ 데이터: NSMC (넀이버 μ˜ν™” 리뷰 데이터셋) batch 크기: 1 μ‹œν€€μŠ€ 길이: 384 ν•™μŠ΅λ₯ : 1e-4 epoch: 1

정확도 ν–₯상 μΆ”κ°€ λ…Έλ ₯

  • epoch 300λΆ€ν„° μ‹œμž‘ν•˜μ—¬ 1000κΉŒμ§€ ν•™μŠ΅
  • 데이터 개수 2000λΆ€ν„° μ‹œμž‘ν•˜μ—¬ 3000κ°œκΉŒμ§€ ν•™μŠ΅

평가

  • (ν•™μŠ΅ 데이터) nsmc μƒμœ„ 3000개
  • (검증 데이터) nsmc μƒμœ„ 1000개
  • ν•™μŠ΅ κ²°κ³Ό: TrainOutput(global_step=1000, training_loss=0.9650133666992188, metrics={'train_runtime': 2982.9519, 'train_samples_per_second': 0.67, 'train_steps_per_second': 0.335, 'total_flos': 3.1051694997504e+16, 'train_loss': 0.9650133666992188, 'epoch': 0.67})
  • 정확도 ν…ŒμŠ€νŠΈ:

TP TN PP 477 79 PN 31 413

정확도: 0.89

Training procedure

The following bitsandbytes quantization config was used during training:

  • quant_method: bitsandbytes
  • load_in_8bit: False
  • load_in_4bit: True
  • llm_int8_threshold: 6.0
  • llm_int8_skip_modules: None
  • llm_int8_enable_fp32_cpu_offload: False
  • llm_int8_has_fp16_weight: False
  • bnb_4bit_quant_type: nf4
  • bnb_4bit_use_double_quant: False
  • bnb_4bit_compute_dtype: bfloat16

Framework versions

  • PEFT 0.7.0
Downloads last month
6
Inference API
Unable to determine this model’s pipeline type. Check the docs .

Model tree for haeun161/midm-7B-nsmc

Adapter
(5)
this model