|
--- |
|
base_model: gpt2 |
|
library_name: Distily |
|
license: mit |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: distily_bench_gpt2_activation_loss_b |
|
results: [] |
|
--- |
|
|
|
# distily_bench_gpt2_activation_loss_b |
|
|
|
This student model is distilled from the teacher model [gpt2](https://huggingface.co/gpt2) using the dataset (unspecified). |
|
|
|
The [Distily](https://github.com/lapp0/distily) library was used for this distillation. |
|
|
|
It achieves the following results on the evaluation set: |
|
- eval_enwikippl: 210.2820 |
|
- eval_frwikippl: 1274.1346 |
|
- eval_zhwikippl: 583.2827 |
|
- eval_loss: 1.2965 |
|
- eval_runtime: 17.2526 |
|
- eval_samples_per_second: 57.962 |
|
- eval_steps_per_second: 7.245 |
|
|
|
<!-- 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. |
|
|
|
## 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: |
|
- distillation_objective: DistillationObjective(logits_loss_component=LossComponent(label=logits, weight=1, loss_fn=kl, layer_mapper=None, projector=None), hs_loss_component=LossComponent(label=hs, weight=2.0, loss_fn=mse, layer_mapper=None, projector=None), attn_loss_component=LossComponent(label=attn, weight=0, loss_fn=None, layer_mapper=None, projector=None)) |
|
- train_embeddings: True |
|
- learning_rate: 4e-05 |
|
- train_batch_size: 8 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: constant |
|
- num_epochs: 1.0 |
|
|
|
### Resource Usage |
|
Peak GPU Memory: 8.0904 GB |
|
|
|
### Eval-Phase Metrics |
|
| step | epoch | enwikippl | frwikippl | loss | runtime | samples_per_second | steps_per_second | zhwikippl | |
|
| --- | --- | --- | --- | --- | --- | --- | --- | --- | |
|
| **teacher eval** | | 30.2086 | 57.2728 | | | | | 18.1784 | |
|
| 0 | 0 | 58037.3203 | 58017.0117 | 6.0237 | 17.2607 | 57.935 | 7.242 | 56038.0625 | |
|
| 1000 | 0.0808 | 715.0994 | 4658.6846 | 2.0131 | 17.1734 | 58.23 | 7.279 | 16350.8623 | |
|
| 2000 | 0.1616 | 508.9246 | 3343.2109 | 1.8201 | 17.2004 | 58.138 | 7.267 | 3102.6990 | |
|
| 3000 | 0.2424 | 419.7101 | 2552.4004 | 1.7020 | 17.1441 | 58.329 | 7.291 | 1042.4126 | |
|
| 4000 | 0.3232 | 361.0421 | 2336.7490 | 1.6177 | 17.0616 | 58.611 | 7.326 | 911.8621 | |
|
| 5000 | 0.4040 | 313.2633 | 1815.2219 | 1.5316 | 17.1786 | 58.212 | 7.276 | 863.9713 | |
|
| 6000 | 0.4848 | 281.3860 | 1725.1301 | 1.4597 | 17.3168 | 57.747 | 7.218 | 705.6341 | |
|
| 7000 | 0.5657 | 253.9131 | 1485.6165 | 1.3999 | 17.1434 | 58.332 | 7.291 | 605.2624 | |
|
| 8000 | 0.6465 | 229.4073 | 1427.2965 | 1.3455 | 17.134 | 58.363 | 7.295 | 629.6656 | |
|
| 9000 | 0.7273 | 210.2820 | 1274.1346 | 1.2965 | 17.2526 | 57.962 | 7.245 | 583.2827 | |
|
| 10000 | 0.8081 | 194.6313 | 1199.3423 | 1.2490 | 17.1679 | 58.248 | 7.281 | 677.5621 | |
|
| 11000 | 0.8889 | 180.3274 | 1160.25 | 1.1980 | 17.1591 | 58.278 | 7.285 | 758.1945 | |
|
| 12000 | 0.9697 | 164.7045 | 1005.8066 | 1.1583 | 17.1824 | 58.199 | 7.275 | 600.1918 | |
|
| 12375 | 1.0 | 161.0243 | 969.7354 | 1.1403 | 17.1939 | 58.16 | 7.27 | 632.9536 | |
|
|
|
### Framework versions |
|
- Distily 0.2.0 |
|
- Transformers 4.44.0 |
|
- Pytorch 2.3.0 |
|
- Datasets 2.21.0 |
|
|