File size: 4,380 Bytes
cd11314
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7e65596
 
 
 
 
 
 
 
cd11314
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7e65596
cd11314
 
 
 
 
 
 
 
7e65596
cd11314
 
 
 
 
7e65596
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cd11314
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
---
base_model: roneneldan/TinyStories-33M
library_name: Distily
tags:
- generated_from_trainer
model-index:
- name: distily_bench_obj_cross_v2.10
  results: []
---

# distily_bench_obj_cross_v2.10

This student model is distilled from the teacher model [roneneldan/TinyStories-33M](https://huggingface.co/roneneldan/TinyStories-33M) 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: 12766.3359
- eval_frwikippl: 57742.3438
- eval_zhwikippl: 65334.25
- eval_tinystoriesppl: 4770.0942
- eval_loss: 5.2085
- eval_runtime: 13.0328
- eval_samples_per_second: 76.73
- eval_steps_per_second: 9.591

<!-- 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: 1e-06
- train_batch_size: 1
- eval_batch_size: 8
- 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: 6.6048 GB

### Eval-Phase Metrics
| step | epoch | enwikippl | frwikippl | loss | runtime | samples_per_second | steps_per_second | tinystoriesppl | zhwikippl |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| **teacher eval** |  | 169.9865 | 47377.9414 |  |  |  |  | 3.9789 | 4998.1294 |
| 0 | 0 | 61801.5039 | 81001.6719 | 6.4680 | 13.0128 | 76.847 | 9.606 | 44522.7852 | 75358.2109 |
| 5000 | 0.0505 | 12766.3359 | 57742.3438 | 5.2085 | 12.9999 | 76.923 | 9.615 | 4771.6733 | 65264.5664 |
| 10000 | 0.1010 | 12766.3359 | 57742.3438 | 5.2085 | 13.0144 | 76.838 | 9.605 | 4768.5161 | 65334.25 |
| 15000 | 0.1515 | 12766.3359 | 57742.3438 | 5.2085 | 13.0239 | 76.782 | 9.598 | 4770.0942 | 65334.25 |
| 20000 | 0.2020 | 12766.3359 | 57742.3438 | 5.2085 | 12.9909 | 76.977 | 9.622 | 4769.3076 | 65334.25 |
| 25000 | 0.2525 | 12766.3359 | 57709.8086 | 5.2083 | 13.1403 | 76.102 | 9.513 | 4768.5161 | 65334.25 |
| 30000 | 0.3030 | 12766.3359 | 57709.8086 | 5.2083 | 13.0382 | 76.698 | 9.587 | 4768.5161 | 65334.25 |
| 35000 | 0.3535 | 12766.3359 | 57742.3438 | 5.2083 | 13.0826 | 76.438 | 9.555 | 4770.0942 | 65334.25 |
| 40000 | 0.4040 | 12766.3359 | 57742.3438 | 5.2085 | 13.0472 | 76.645 | 9.581 | 4769.3076 | 65334.25 |
| 45000 | 0.4545 | 12766.3359 | 57742.3438 | 5.2085 | 13.1664 | 75.951 | 9.494 | 4770.0942 | 65334.25 |
| 50000 | 0.5051 | 12766.3359 | 57742.3438 | 5.2083 | 13.047 | 76.646 | 9.581 | 4768.5161 | 65334.25 |
| 55000 | 0.5556 | 12766.3359 | 57742.3438 | 5.2083 | 13.2134 | 75.681 | 9.46 | 4768.5161 | 65334.25 |
| 60000 | 0.6061 | 12766.3359 | 57742.3438 | 5.2087 | 13.0275 | 76.761 | 9.595 | 4769.3076 | 65334.25 |
| 65000 | 0.6566 | 12766.3359 | 57742.3438 | 5.2083 | 13.1101 | 76.277 | 9.535 | 4768.5161 | 65334.25 |
| 70000 | 0.7071 | 12766.3359 | 57742.3438 | 5.2085 | 13.0485 | 76.637 | 9.58 | 4771.6733 | 65334.25 |
| 75000 | 0.7576 | 12766.3359 | 57742.3438 | 5.2085 | 13.0209 | 76.8 | 9.6 | 4768.5161 | 65299.4297 |
| 80000 | 0.8081 | 12766.3359 | 57742.3438 | 5.2085 | 13.0587 | 76.577 | 9.572 | 4771.6733 | 65334.25 |
| 85000 | 0.8586 | 12766.3359 | 57742.3438 | 5.2085 | 13.0404 | 76.685 | 9.586 | 4770.0942 | 65299.4297 |
| 90000 | 0.9091 | 12766.3359 | 57742.3438 | 5.2087 | 13.0082 | 76.874 | 9.609 | 4770.0942 | 65334.25 |
| 95000 | 0.9596 | 12766.3359 | 57742.3438 | 5.2085 | 13.0077 | 76.878 | 9.61 | 4769.3076 | 65334.25 |
| 99000 | 1.0 | 12766.3359 | 57742.3438 | 5.2085 | 13.0328 | 76.73 | 9.591 | 4770.0942 | 65334.25 |

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