|
--- |
|
base_model: gpt2 |
|
library_name: Distily |
|
license: mit |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: distily_bench_obj_cross_v2.15_gpt2 |
|
results: [] |
|
--- |
|
|
|
# distily_bench_obj_cross_v2.15_gpt2 |
|
|
|
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: 5056.0 |
|
- eval_frwikippl: 3696.0 |
|
- eval_zhwikippl: 29312.0 |
|
- eval_tinystoriesppl: 4672.0 |
|
- eval_loss: 1.3042 |
|
- eval_runtime: 16.773 |
|
- eval_samples_per_second: 59.62 |
|
- eval_steps_per_second: 7.452 |
|
|
|
<!-- 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=0, loss_fn=None, 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: 0.0004 |
|
- 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 |
|
- lr_scheduler_warmup_ratio: 0.2 |
|
- num_epochs: 1.0 |
|
|
|
### Resource Usage |
|
Peak GPU Memory: 7.9368 GB |
|
|
|
### Eval-Phase Metrics |
|
| step | epoch | enwikippl | frwikippl | loss | runtime | samples_per_second | steps_per_second | tinystoriesppl | zhwikippl | |
|
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | |
|
| **teacher eval** | | 43.75 | 61.75 | | | | | 11.8125 | 19.125 | |
|
| 0 | 0 | 2336462209024.0 | 122045790683136.0 | 24.1200 | 16.7622 | 59.658 | 7.457 | 4429185024.0 | 25975962206208.0 | |
|
| 1000 | 0.0808 | 3312.0 | 3184.0 | 1.1229 | 16.7657 | 59.646 | 7.456 | 2720.0 | 2992.0 | |
|
| 2000 | 0.1616 | 5248.0 | 4048.0 | 1.2528 | 16.7825 | 59.586 | 7.448 | 5408.0 | 9984.0 | |
|
| 3000 | 0.2424 | 5600.0 | 3744.0 | 1.2812 | 16.7695 | 59.632 | 7.454 | 5312.0 | 23680.0 | |
|
| 4000 | 0.3232 | 5408.0 | 3920.0 | 1.2832 | 16.8694 | 59.279 | 7.41 | 5440.0 | 33280.0 | |
|
| 5000 | 0.4040 | 5376.0 | 3952.0 | 1.2841 | 16.7361 | 59.751 | 7.469 | 5408.0 | 27008.0 | |
|
| 6000 | 0.4848 | 5344.0 | 3680.0 | 1.2770 | 16.7635 | 59.653 | 7.457 | 5440.0 | 29312.0 | |
|
| 7000 | 0.5657 | 5024.0 | 3760.0 | 1.2800 | 16.7492 | 59.704 | 7.463 | 5184.0 | 39936.0 | |
|
| 8000 | 0.6465 | 4992.0 | 3712.0 | 1.2922 | 16.7445 | 59.721 | 7.465 | 5088.0 | 26752.0 | |
|
| 9000 | 0.7273 | 5056.0 | 3696.0 | 1.3042 | 16.773 | 59.62 | 7.452 | 4672.0 | 29312.0 | |
|
| 10000 | 0.8081 | 5824.0 | 3648.0 | 1.3192 | 16.7669 | 59.641 | 7.455 | 5312.0 | 24448.0 | |
|
| 11000 | 0.8889 | 5568.0 | 3872.0 | 1.3215 | 16.8215 | 59.448 | 7.431 | 5504.0 | 40704.0 | |
|
| 12000 | 0.9697 | 5440.0 | 3792.0 | 1.3263 | 16.7825 | 59.586 | 7.448 | 5120.0 | 72704.0 | |
|
| 12375 | 1.0 | 5696.0 | 3936.0 | 1.3389 | 16.7852 | 59.576 | 7.447 | 5696.0 | 40192.0 | |
|
|
|
### Framework versions |
|
- Distily 0.2.0 |
|
- Transformers 4.44.0 |
|
- Pytorch 2.3.0 |
|
- Datasets 2.21.0 |
|
|
