File size: 3,837 Bytes
b6cf0f6
 
 
 
5d13082
 
b6cf0f6
 
 
5d13082
 
b6cf0f6
 
 
 
 
 
 
 
 
 
 
5d13082
b6cf0f6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: mit
library_name: peft
tags:
- alignment-handbook
- generated_from_trainer
- trl
- dpo
- generated_from_trainer
datasets:
- HuggingFaceH4/ultrafeedback_binarized
base_model: microsoft/phi-2
model-index:
- name: phi-2-gpo-ultrafeedback-lora
  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. -->

# phi-2-gpo-ultrafeedback-lora

This model is a fine-tuned version of [lole25/phi-2-sft-ultrachat-lora](https://huggingface.co/lole25/phi-2-sft-ultrachat-lora) on the HuggingFaceH4/ultrafeedback_binarized dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0021
- Rewards/chosen: -0.0083
- Rewards/rejected: -0.0184
- Rewards/accuracies: 0.6920
- Rewards/margins: 0.0101
- Logps/rejected: -233.2711
- Logps/chosen: -261.0694
- Logits/rejected: 0.8833
- Logits/chosen: 0.7809

## 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: 5e-06
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 2

### Training results

| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
| 0.0026        | 0.21  | 100  | 0.0025          | 0.0001         | -0.0005          | 0.5080             | 0.0006          | -231.4896      | -260.2373    | 0.9175          | 0.8151        |
| 0.0023        | 0.42  | 200  | 0.0023          | -0.0015        | -0.0068          | 0.6560             | 0.0053          | -232.1152      | -260.3932    | 0.9120          | 0.8092        |
| 0.0022        | 0.63  | 300  | 0.0022          | -0.0067        | -0.0141          | 0.6700             | 0.0073          | -232.8447      | -260.9179    | 0.9022          | 0.7992        |
| 0.0021        | 0.84  | 400  | 0.0022          | -0.0092        | -0.0178          | 0.6640             | 0.0086          | -233.2157      | -261.1620    | 0.8914          | 0.7884        |
| 0.0022        | 1.05  | 500  | 0.0021          | -0.0094        | -0.0193          | 0.7100             | 0.0098          | -233.3614      | -261.1852    | 0.8853          | 0.7821        |
| 0.002         | 1.26  | 600  | 0.0021          | -0.0088        | -0.0185          | 0.6940             | 0.0097          | -233.2843      | -261.1207    | 0.8840          | 0.7815        |
| 0.0021        | 1.47  | 700  | 0.0021          | -0.0083        | -0.0182          | 0.7000             | 0.0099          | -233.2560      | -261.0788    | 0.8816          | 0.7790        |
| 0.0021        | 1.67  | 800  | 0.0021          | -0.0082        | -0.0184          | 0.6940             | 0.0102          | -233.2740      | -261.0643    | 0.8811          | 0.7781        |
| 0.0021        | 1.88  | 900  | 0.0021          | -0.0085        | -0.0178          | 0.6900             | 0.0093          | -233.2118      | -261.0922    | 0.8833          | 0.7806        |


### Framework versions

- PEFT 0.7.1
- Transformers 4.36.2
- Pytorch 2.1.2+cu118
- Datasets 2.14.6
- Tokenizers 0.15.2