---
base_model: gpt2
library_name: Distily
license: mit
tags:
- generated_from_trainer
model-index:
- name: distily_bench_obj_cross_v2.13_gpt2
  results: []
---

# distily_bench_obj_cross_v2.13_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: 1248.0
- eval_frwikippl: 6560.0
- eval_zhwikippl: 24448.0
- eval_tinystoriesppl: 968.0
- eval_loss: 2.4413
- eval_runtime: 12.6655
- eval_samples_per_second: 47.373
- eval_steps_per_second: 11.843

<!-- 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: 0.0001
- train_batch_size: 8
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.5
- num_epochs: 1.0

### Resource Usage
Peak GPU Memory: 7.9388 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 | 1065151889408.0 | 117097988358144.0 | 20.4033 | 12.6433 | 47.456 | 11.864 | 4362076160.0 | 27762668601344.0 |
| 750 | 0.1010 | 1248.0 | 6560.0 | 2.4413 | 12.6655 | 47.373 | 11.843 | 968.0 | 24448.0 |
| 1500 | 0.2020 | 500.0 | 3712.0 | 1.8019 | 12.6495 | 47.433 | 11.858 | 352.0 | 820.0 |
| 2250 | 0.3030 | 342.0 | 1832.0 | 1.5659 | 12.681 | 47.315 | 11.829 | 276.0 | 294.0 |
| 3000 | 0.4040 | 245.0 | 932.0 | 1.3644 | 12.6925 | 47.272 | 11.818 | 198.0 | 228.0 |
| 3750 | 0.5051 | 191.0 | 680.0 | 1.2038 | 12.6837 | 47.305 | 11.826 | 159.0 | 219.0 |
| 4500 | 0.6061 | 148.0 | 596.0 | 1.0541 | 12.6874 | 47.291 | 11.823 | 127.5 | 188.0 |
| 5250 | 0.7071 | 129.0 | 442.0 | 0.9331 | 12.6692 | 47.359 | 11.84 | 102.5 | 126.5 |
| 6000 | 0.8081 | 117.0 | 412.0 | 0.8757 | 12.6912 | 47.277 | 11.819 | 94.5 | 124.5 |
| 6750 | 0.9091 | 112.0 | 398.0 | 0.8483 | 12.8574 | 46.666 | 11.666 | 90.0 | 119.5 |
| 7425 | 1.0 | 110.5 | 392.0 | 0.8433 | 12.717 | 47.181 | 11.795 | 89.0 | 119.0 |

### Framework versions
- Distily 0.2.0
- Transformers 4.44.0
- Pytorch 2.3.0
- Datasets 2.21.0