Model Card for Model ID
- mdim λͺ¨λΈμ νκ΅μ΄ μν 리뷰 λ°μ΄ν°μ (NSMC)μ ν΄κ²°νλ λͺ¨λΈμ΄ λλλ‘ λ―ΈμΈνλ
- νκ΅ μν 리뷰μ κΈμ λλ λΆμ μ νλ¨νλ λͺ¨λΈμ νμ΅
Model Details
λͺ¨λΈ: KT-AI/midm-bitext-S-7B-inst-v1 νμ΅ λ°μ΄ν°: NSMC (λ€μ΄λ² μν 리뷰 λ°μ΄ν°μ ) batch ν¬κΈ°: 1 μνμ€ κΈΈμ΄: 384 νμ΅λ₯ : 1e-4 epoch: 1
μ νλ ν₯μ μΆκ° λ Έλ ₯
- epoch 300λΆν° μμνμ¬ 1000κΉμ§ νμ΅
- λ°μ΄ν° κ°μ 2000λΆν° μμνμ¬ 3000κ°κΉμ§ νμ΅
νκ°
- (νμ΅ λ°μ΄ν°) nsmc μμ 3000κ°
- (κ²μ¦ λ°μ΄ν°) nsmc μμ 1000κ°
- νμ΅ 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})
- μ νλ ν μ€νΈ:
TP TN PP 477 79 PN 31 413
μ νλ: 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
Model tree for haeun161/midm-7B-nsmc
Base model
jangmin/midm-7b-safetensors-only