llama3_darulm_20_05_24_part1-2_128000_bpe_part1_lr2e4_bs256
This model is a fine-tuned version of RefalMachine/llama3_darulm_20_05_24_part1-2_128000_bpe_mean_init_03_07_24 on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 2.3612
- Accuracy: 0.5178
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: 0.0002
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 32
- total_train_batch_size: 128
- total_eval_batch_size: 128
- optimizer: Adam with betas=(0.9,0.95) and epsilon=1e-05
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- num_epochs: 1.0
Training results
| Training Loss |
Epoch |
Step |
Validation Loss |
Accuracy |
| 2.7657 |
0.1 |
2000 |
2.5269 |
0.4967 |
| 2.7065 |
0.2 |
4000 |
2.4675 |
0.5033 |
| 2.6458 |
0.3 |
6000 |
2.4329 |
0.5078 |
| 2.6112 |
0.4 |
8000 |
2.4078 |
0.5109 |
| 2.5934 |
0.51 |
10000 |
2.3867 |
0.5140 |
| 2.5881 |
0.61 |
12000 |
2.3724 |
0.5161 |
| 2.5566 |
0.71 |
14000 |
2.3647 |
0.5174 |
| 2.5829 |
0.81 |
16000 |
2.3618 |
0.5178 |
| 2.5821 |
0.91 |
18000 |
2.3612 |
0.5179 |
Framework versions
- Transformers 4.37.2
- Pytorch 2.3.0a0+6ddf5cf85e.nv24.04
- Datasets 2.18.0
- Tokenizers 0.15.2
