File size: 5,349 Bytes
aa9ad8b
e9ddca7
 
938737a
e9ddca7
 
 
938737a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
aa9ad8b
e9ddca7
 
 
 
 
 
 
 
 
 
 
 
 
 
7f0b9aa
 
e9ddca7
 
 
 
 
 
 
 
7f0b9aa
 
567194d
b0894a1
7f0b9aa
e9ddca7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
938737a
 
 
 
 
 
 
 
 
 
 
 
 
4b4521b
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
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
---
language:
- en
license: cc-by-nc-sa-4.0
datasets:
- kyujinpy/orca_math_dpo
pipeline_tag: text-generation
model-index:
- name: Sakura-SOLRCA-Math-Instruct-DPO-v2
  results:
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: AI2 Reasoning Challenge (25-Shot)
      type: ai2_arc
      config: ARC-Challenge
      split: test
      args:
        num_few_shot: 25
    metrics:
    - type: acc_norm
      value: 71.25
      name: normalized accuracy
    source:
      url: https://huggingface.co./spaces/HuggingFaceH4/open_llm_leaderboard?query=kyujinpy/Sakura-SOLRCA-Math-Instruct-DPO-v2
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: HellaSwag (10-Shot)
      type: hellaswag
      split: validation
      args:
        num_few_shot: 10
    metrics:
    - type: acc_norm
      value: 88.52
      name: normalized accuracy
    source:
      url: https://huggingface.co./spaces/HuggingFaceH4/open_llm_leaderboard?query=kyujinpy/Sakura-SOLRCA-Math-Instruct-DPO-v2
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: MMLU (5-Shot)
      type: cais/mmlu
      config: all
      split: test
      args:
        num_few_shot: 5
    metrics:
    - type: acc
      value: 66.13
      name: accuracy
    source:
      url: https://huggingface.co./spaces/HuggingFaceH4/open_llm_leaderboard?query=kyujinpy/Sakura-SOLRCA-Math-Instruct-DPO-v2
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: TruthfulQA (0-shot)
      type: truthful_qa
      config: multiple_choice
      split: validation
      args:
        num_few_shot: 0
    metrics:
    - type: mc2
      value: 72.16
    source:
      url: https://huggingface.co./spaces/HuggingFaceH4/open_llm_leaderboard?query=kyujinpy/Sakura-SOLRCA-Math-Instruct-DPO-v2
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: Winogrande (5-shot)
      type: winogrande
      config: winogrande_xl
      split: validation
      args:
        num_few_shot: 5
    metrics:
    - type: acc
      value: 83.03
      name: accuracy
    source:
      url: https://huggingface.co./spaces/HuggingFaceH4/open_llm_leaderboard?query=kyujinpy/Sakura-SOLRCA-Math-Instruct-DPO-v2
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: GSM8k (5-shot)
      type: gsm8k
      config: main
      split: test
      args:
        num_few_shot: 5
    metrics:
    - type: acc
      value: 63.91
      name: accuracy
    source:
      url: https://huggingface.co./spaces/HuggingFaceH4/open_llm_leaderboard?query=kyujinpy/Sakura-SOLRCA-Math-Instruct-DPO-v2
      name: Open LLM Leaderboard
---

# **Sakura-SOLRCA-Math-Instruct-DPO-v2**  
<img src='./sakura.png' width=512>

## Model Details

**Model Developers** Kyujin Han (kyujinpy)

**Method**  
Using DPO method.  
With [Intel/orca_dpo_pairs](https://huggingface.co./datasets/Intel/orca_dpo_pairs) and [argilla/distilabel-math-preference-dpo](https://huggingface.co./datasets/argilla/distilabel-math-preference-dpo).  

I shared the merge version [kyujinpy/orca_math_dpo](https://huggingface.co./datasets/kyujinpy/orca_math_dpo).  
     
I shared the information about my model. (training and code)  
Please see: ⭐[Sakura-SOLAR](https://github.com/KyujinHan/Sakura-SOLAR-DPO).  

# **Model Benchmark**  

## Open leaderboard
- Follow up as [link](https://huggingface.co./spaces/HuggingFaceH4/open_llm_leaderboard).  

| Model | Average | ARC | HellaSwag | MMLU | TruthfulQA | Winogrande | GSM8K |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Sakura-SOLRCA-Math-Instruct-DPO-v2 | 74.17 | 71.25 | 88.52 | 66.13 | 72.16 | 83.03 | 63.91 |
| Sakura-SOLRCA-Math-Instruct-DPO-v1 | 74.13 | 71.25 | 88.48 | 66.21 | 72.12 | 82.87 | 63.84 |
| Sakura-SOLRCA-Instruct-DPO | 74.05 | 71.16 | 88.49 | 66.17 | 72.10 | 82.95 | 63.46 |
| Sakura-SOLAR-Instruct-DPO-v2 | 74.14 | 70.90 | 88.41 | 66.48 | 71.86 | 83.43 | 63.76 |
| [kyujinpy/Sakura-SOLAR-Instruct](https://huggingface.co./kyujinpy/Sakura-SOLAR-Instruct) | 74.40 | 70.99 | 88.42 | 66.33 | 71.79 | 83.66 | 65.20 |

   
# Implementation Code
```python
### KO-Platypus
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

repo = "kyujinpy/Sakura-SOLRCA-Math-Instruct-DPO-v2"
OpenOrca = AutoModelForCausalLM.from_pretrained(
        repo,
        return_dict=True,
        torch_dtype=torch.float16,
        device_map='auto'
)
OpenOrca_tokenizer = AutoTokenizer.from_pretrained(repo)
```

---
# [Open LLM Leaderboard Evaluation Results](https://huggingface.co./spaces/HuggingFaceH4/open_llm_leaderboard)
Detailed results can be found [here](https://huggingface.co./datasets/open-llm-leaderboard/details_kyujinpy__Sakura-SOLRCA-Math-Instruct-DPO-v2)

|             Metric              |Value|
|---------------------------------|----:|
|Avg.                             |74.17|
|AI2 Reasoning Challenge (25-Shot)|71.25|
|HellaSwag (10-Shot)              |88.52|
|MMLU (5-Shot)                    |66.13|
|TruthfulQA (0-shot)              |72.16|
|Winogrande (5-shot)              |83.03|
|GSM8k (5-shot)                   |63.91|