<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
gpt-expt-sp-v3-K-600-MA-kmeans-v1
This model is a fine-tuned version of gpt2 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0165
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: 0.0005
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 512
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 1000
- num_epochs: 500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.1526 | 18.31 | 5000 | 0.0965 |
| 0.0728 | 36.63 | 10000 | 0.0381 |
| 0.0244 | 54.94 | 15000 | 0.0198 |
| 0.0204 | 73.26 | 20000 | 0.0183 |
| 0.023 | 91.57 | 25000 | 0.0173 |
| 0.0184 | 109.89 | 30000 | 0.0173 |
| 0.0182 | 128.2 | 35000 | 0.0172 |
| 0.0183 | 146.52 | 40000 | 0.0169 |
| 0.0175 | 164.83 | 45000 | 0.0170 |
| 0.0176 | 183.15 | 50000 | 0.0169 |
| 0.0174 | 201.46 | 55000 | 0.0170 |
| 0.0173 | 219.78 | 60000 | 0.0169 |
| 0.0172 | 238.1 | 65000 | 0.0168 |
| 0.0171 | 256.41 | 70000 | 0.0167 |
| 0.0171 | 274.72 | 75000 | 0.0167 |
| 0.017 | 293.04 | 80000 | 0.0167 |
| 0.0169 | 311.35 | 85000 | 0.0167 |
| 0.0169 | 329.67 | 90000 | 0.0166 |
| 0.0168 | 347.98 | 95000 | 0.0166 |
| 0.0168 | 366.3 | 100000 | 0.0166 |
| 0.0167 | 384.61 | 105000 | 0.0166 |
| 0.0167 | 402.93 | 110000 | 0.0166 |
| 0.0167 | 421.24 | 115000 | 0.0166 |
| 0.0166 | 439.56 | 120000 | 0.0165 |
| 0.0166 | 457.87 | 125000 | 0.0165 |
| 0.0166 | 476.19 | 130000 | 0.0165 |
| 0.0166 | 494.5 | 135000 | 0.0165 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1.post200
- Datasets 2.9.0
- Tokenizers 0.13.2
