gemma2_on_korean_conv-ltm
This model is a fine-tuned version of beomi/gemma-ko-2b on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.1680
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 10
- total_train_batch_size: 20
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- training_steps: 2000
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
1.2653 | 0.2563 | 100 | 1.2545 |
1.1568 | 0.5126 | 200 | 1.1327 |
1.0709 | 0.7688 | 300 | 1.0871 |
1.0143 | 1.0251 | 400 | 1.0468 |
0.898 | 1.2814 | 500 | 1.0379 |
0.8809 | 1.5377 | 600 | 1.0323 |
0.8741 | 1.7940 | 700 | 1.0204 |
0.7439 | 2.0502 | 800 | 1.0383 |
0.7114 | 2.3065 | 900 | 1.0412 |
0.747 | 2.5628 | 1000 | 1.0355 |
0.7272 | 2.8191 | 1100 | 1.0286 |
0.5777 | 3.0753 | 1200 | 1.0777 |
0.5715 | 3.3316 | 1300 | 1.0852 |
0.6144 | 3.5879 | 1400 | 1.0826 |
0.5737 | 3.8442 | 1500 | 1.0921 |
0.463 | 4.1005 | 1600 | 1.1487 |
0.4711 | 4.3567 | 1700 | 1.1612 |
0.4734 | 4.6130 | 1800 | 1.1612 |
0.4605 | 4.8693 | 1900 | 1.1645 |
0.3862 | 5.1256 | 2000 | 1.1680 |
Framework versions
- PEFT 0.11.1
- Transformers 4.41.1
- Pytorch 2.3.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1
- Downloads last month
- 0
Model tree for ghost613/gemma2_on_korean_conv-base-ltm
Base model
beomi/gemma-ko-2b