---
base_model: gpt2
library_name: Distily
license: mit
tags:
- generated_from_trainer
model-index:
- name: distily_bench_obj_cross_v2.10_gpt2
  results: []
---

# distily_bench_obj_cross_v2.10_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: 401.6902
- eval_frwikippl: 385.9396
- eval_zhwikippl: 137.9653
- eval_tinystoriesppl: 881.4292
- eval_loss: 0.7112
- eval_runtime: 21.2483
- eval_samples_per_second: 47.063
- eval_steps_per_second: 11.766

<!-- 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: 4e-05
- train_batch_size: 1
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 1.0

### Resource Usage
Peak GPU Memory: 3.9285 GB

### Eval-Phase Metrics
| step | epoch | enwikippl | frwikippl | loss | runtime | samples_per_second | steps_per_second | tinystoriesppl | zhwikippl |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| **teacher eval** |  | 270.2348 | 76.8142 |  |  |  |  | 671.1238 | 22.8030 |
| 0 | 0 | 120078.375 | 1867851235328.0 | 18.7920 | 21.2125 | 47.142 | 11.786 | 72.8770 | 4013754155008.0 |
| 5000 | 0.0505 | 621.5149 | 991.7020 | 1.3528 | 21.2177 | 47.13 | 11.783 | 980.0922 | 399.9691 |
| 10000 | 0.1010 | 574.4407 | 664.8521 | 1.1590 | 21.2225 | 47.12 | 11.78 | 1036.6780 | 493.8460 |
| 15000 | 0.1515 | 543.0890 | 635.0353 | 1.0360 | 21.2351 | 47.092 | 11.773 | 1033.2988 | 145.9157 |
| 20000 | 0.2020 | 509.8121 | 599.6746 | 0.9759 | 21.2099 | 47.148 | 11.787 | 985.1690 | 251.1274 |
| 25000 | 0.2525 | 448.2854 | 486.9003 | 0.8334 | 21.2284 | 47.107 | 11.777 | 923.3450 | 171.9567 |
| 30000 | 0.3030 | 420.2149 | 441.8981 | 0.7741 | 21.2742 | 47.005 | 11.751 | 893.9037 | 129.4944 |
| 35000 | 0.3535 | 417.6187 | 442.7548 | 0.7695 | 21.5924 | 46.313 | 11.578 | 884.2755 | 140.6411 |
| 40000 | 0.4040 | 419.8570 | 418.2776 | 0.7678 | 21.23 | 47.103 | 11.776 | 893.9774 | 162.6632 |
| 45000 | 0.4545 | 420.1905 | 413.8966 | 0.7576 | 21.2355 | 47.091 | 11.773 | 905.9177 | 154.8089 |
| 50000 | 0.5051 | 420.9561 | 426.7430 | 0.7544 | 21.2196 | 47.126 | 11.782 | 906.1800 | 147.5501 |
| 55000 | 0.5556 | 417.3034 | 409.1867 | 0.7509 | 21.2021 | 47.165 | 11.791 | 902.3304 | 143.7327 |
| 60000 | 0.6061 | 418.3230 | 413.0230 | 0.7525 | 21.2367 | 47.088 | 11.772 | 894.0145 | 156.6996 |
| 65000 | 0.6566 | 404.0308 | 404.5305 | 0.7221 | 21.2003 | 47.169 | 11.792 | 878.4468 | 136.2006 |
| 70000 | 0.7071 | 406.0154 | 392.1317 | 0.7194 | 21.2119 | 47.143 | 11.786 | 891.9106 | 137.0481 |
| 75000 | 0.7576 | 400.8665 | 383.9604 | 0.7188 | 21.2118 | 47.144 | 11.786 | 871.7914 | 140.4630 |
| 80000 | 0.8081 | 402.5625 | 387.4647 | 0.7168 | 21.2234 | 47.118 | 11.779 | 882.3771 | 141.0827 |
| 85000 | 0.8586 | 399.3479 | 385.9124 | 0.7123 | 21.2047 | 47.159 | 11.79 | 875.1130 | 140.0700 |
| 90000 | 0.9091 | 401.2549 | 386.7830 | 0.7117 | 21.2316 | 47.1 | 11.775 | 881.0649 | 138.5555 |
| 95000 | 0.9596 | 401.4725 | 386.1842 | 0.7112 | 21.2217 | 47.122 | 11.78 | 880.2640 | 138.0389 |
| 99000 | 1.0 | 401.6902 | 385.9396 | 0.7112 | 21.2483 | 47.063 | 11.766 | 881.4292 | 137.9653 |

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