metadata
license: apache-2.0
library_name: peft
tags:
- trl
- sft
- generated_from_trainer
base_model: HuggingFaceTB/SmolLM-1.7B-Instruct
datasets:
- generator
model-index:
- name: SmolLM_1_7B_Instruct_qlora_nf4_merged
results: []
SmolLM_1_7B_Instruct_qlora_nf4_merged
This model is a fine-tuned version of HuggingFaceTB/SmolLM-1.7B-Instruct on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 1.6129
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: 0.001
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 16
- gradient_accumulation_steps: 4
- total_train_batch_size: 256
- total_eval_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.03
- num_epochs: 20
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.088 | 0.9524 | 10 | 1.9222 |
| 1.8671 | 2.0 | 21 | 1.7931 |
| 1.7735 | 2.9524 | 31 | 1.7340 |
| 1.7236 | 4.0 | 42 | 1.6932 |
| 1.6739 | 4.9524 | 52 | 1.6680 |
| 1.652 | 6.0 | 63 | 1.6494 |
| 1.6354 | 6.9524 | 73 | 1.6379 |
| 1.6139 | 8.0 | 84 | 1.6288 |
| 1.5938 | 8.9524 | 94 | 1.6233 |
| 1.5828 | 10.0 | 105 | 1.6189 |
| 1.5722 | 10.9524 | 115 | 1.6164 |
| 1.5588 | 12.0 | 126 | 1.6149 |
| 1.5539 | 12.9524 | 136 | 1.6141 |
| 1.5506 | 14.0 | 147 | 1.6134 |
| 1.5437 | 14.9524 | 157 | 1.6132 |
| 1.5427 | 16.0 | 168 | 1.6130 |
| 1.5407 | 16.9524 | 178 | 1.6130 |
| 1.5386 | 18.0 | 189 | 1.6130 |
| 1.5373 | 18.9524 | 199 | 1.6130 |
| 1.5397 | 19.0476 | 200 | 1.6129 |
Framework versions
- PEFT 0.10.0
- Transformers 4.40.0
- Pytorch 2.1.0
- Datasets 2.18.0
- Tokenizers 0.19.1