File size: 2,172 Bytes
f870355
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
library_name: transformers
base_model: sam-paech/gutenberg3-orpo-exp02
tags:
- alignment-handbook
- generated_from_trainer
datasets:
- sam-paech/gutenberg3-generalfiction-scifi-fantasy-romance-adventure-dpo
model-index:
- name: gemma-2-9b-it-gutenberg3
  results: []
---

<!-- 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. -->

# gemma-2-9b-it-gutenberg3

This model is a fine-tuned version of [sam-paech/gutenberg3-orpo-exp02](https://huggingface.co/sam-paech/gutenberg3-orpo-exp02) on the sam-paech/gutenberg3-generalfiction-scifi-fantasy-romance-adventure-dpo dataset.
It achieves the following results on the evaluation set:
- Loss: 20.3722
- Rewards/chosen: -27.5820
- Rewards/rejected: -11.3463
- Rewards/accuracies: 0.0
- Rewards/margins: -16.2357
- Logps/rejected: -1.1346
- Logps/chosen: -2.7582

## 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:
- learning_rate: 5.8e-08
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- num_devices: 5
- gradient_accumulation_steps: 2
- total_train_batch_size: 10
- total_eval_batch_size: 5
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1

### Training results

| Training Loss | Epoch  | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen |
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|
| 20.2621       | 0.7092 | 400  | 20.3722         | -27.5820       | -11.3463         | 0.0                | -16.2357        | -1.1346        | -2.7582      |


### Framework versions

- Transformers 4.45.2
- Pytorch 2.2.0+cu121
- Datasets 3.0.1
- Tokenizers 0.20.1