llama2-2t-asstop1-lr2-e5-cos-ep3-instruct-v4
This model is a fine-tuned version of TinyLlama/TinyLlama-1.1B-intermediate-step-955k-token-2T on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.4769
Too many warmup steps as a result loss did not go down much. This model is suboptimal.
Training Script
https://github.com/habanoz/llm-sft/
Training Setup
Trained on Sagemaker ml.g5.xlarge for 1 hour(s), 43 minute(s), and 30 second(s).
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: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_steps: 100
- training_steps: 0
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.6795 | 1.0 | 202 | 1.5372 |
| 1.5387 | 2.0 | 404 | 1.4769 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.0.0
- Datasets 2.12.0
- Tokenizers 0.15.0