Files changed (1) hide show
  1. README.md +118 -2
README.md CHANGED
@@ -1,7 +1,110 @@
1
  ---
2
- license: apache-2.0
3
  language:
4
  - en
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5
  ---
6
 
7
  dpo-phi2 is an instruction-tuned model from microsoft/phi-2. Direct preference optimization (DPO) is used for fine-tuning on argilla/distilabel-intel-orca-dpo-pairs dataset.
@@ -20,4 +123,17 @@ dpo-phi2 is an instruction-tuned model from microsoft/phi-2. Direct preference o
20
 
21
  * Toxicity: Despite being trained with carefully selected data, the model can still produce harmful content if explicitly prompted or instructed to do so. We chose to release the model for research purposes only -- We hope to help the open-source community develop the most effective ways to reduce the toxicity of a model directly after pretraining.
22
 
23
- * Verbosity: Phi-2 being a base model often produces irrelevant or extra text and responses following its first answer to user prompts within a single turn. This is due to its training dataset being primarily textbooks, which results in textbook-like responses.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
 
2
  language:
3
  - en
4
+ license: apache-2.0
5
+ model-index:
6
+ - name: dpo-phi2
7
+ results:
8
+ - task:
9
+ type: text-generation
10
+ name: Text Generation
11
+ dataset:
12
+ name: AI2 Reasoning Challenge (25-Shot)
13
+ type: ai2_arc
14
+ config: ARC-Challenge
15
+ split: test
16
+ args:
17
+ num_few_shot: 25
18
+ metrics:
19
+ - type: acc_norm
20
+ value: 61.69
21
+ name: normalized accuracy
22
+ source:
23
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=amu/dpo-phi2
24
+ name: Open LLM Leaderboard
25
+ - task:
26
+ type: text-generation
27
+ name: Text Generation
28
+ dataset:
29
+ name: HellaSwag (10-Shot)
30
+ type: hellaswag
31
+ split: validation
32
+ args:
33
+ num_few_shot: 10
34
+ metrics:
35
+ - type: acc_norm
36
+ value: 75.13
37
+ name: normalized accuracy
38
+ source:
39
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=amu/dpo-phi2
40
+ name: Open LLM Leaderboard
41
+ - task:
42
+ type: text-generation
43
+ name: Text Generation
44
+ dataset:
45
+ name: MMLU (5-Shot)
46
+ type: cais/mmlu
47
+ config: all
48
+ split: test
49
+ args:
50
+ num_few_shot: 5
51
+ metrics:
52
+ - type: acc
53
+ value: 58.1
54
+ name: accuracy
55
+ source:
56
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=amu/dpo-phi2
57
+ name: Open LLM Leaderboard
58
+ - task:
59
+ type: text-generation
60
+ name: Text Generation
61
+ dataset:
62
+ name: TruthfulQA (0-shot)
63
+ type: truthful_qa
64
+ config: multiple_choice
65
+ split: validation
66
+ args:
67
+ num_few_shot: 0
68
+ metrics:
69
+ - type: mc2
70
+ value: 43.99
71
+ source:
72
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=amu/dpo-phi2
73
+ name: Open LLM Leaderboard
74
+ - task:
75
+ type: text-generation
76
+ name: Text Generation
77
+ dataset:
78
+ name: Winogrande (5-shot)
79
+ type: winogrande
80
+ config: winogrande_xl
81
+ split: validation
82
+ args:
83
+ num_few_shot: 5
84
+ metrics:
85
+ - type: acc
86
+ value: 74.19
87
+ name: accuracy
88
+ source:
89
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=amu/dpo-phi2
90
+ name: Open LLM Leaderboard
91
+ - task:
92
+ type: text-generation
93
+ name: Text Generation
94
+ dataset:
95
+ name: GSM8k (5-shot)
96
+ type: gsm8k
97
+ config: main
98
+ split: test
99
+ args:
100
+ num_few_shot: 5
101
+ metrics:
102
+ - type: acc
103
+ value: 54.44
104
+ name: accuracy
105
+ source:
106
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=amu/dpo-phi2
107
+ name: Open LLM Leaderboard
108
  ---
109
 
110
  dpo-phi2 is an instruction-tuned model from microsoft/phi-2. Direct preference optimization (DPO) is used for fine-tuning on argilla/distilabel-intel-orca-dpo-pairs dataset.
 
123
 
124
  * Toxicity: Despite being trained with carefully selected data, the model can still produce harmful content if explicitly prompted or instructed to do so. We chose to release the model for research purposes only -- We hope to help the open-source community develop the most effective ways to reduce the toxicity of a model directly after pretraining.
125
 
126
+ * Verbosity: Phi-2 being a base model often produces irrelevant or extra text and responses following its first answer to user prompts within a single turn. This is due to its training dataset being primarily textbooks, which results in textbook-like responses.
127
+ # [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
128
+ Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_amu__dpo-phi2)
129
+
130
+ | Metric |Value|
131
+ |---------------------------------|----:|
132
+ |Avg. |61.26|
133
+ |AI2 Reasoning Challenge (25-Shot)|61.69|
134
+ |HellaSwag (10-Shot) |75.13|
135
+ |MMLU (5-Shot) |58.10|
136
+ |TruthfulQA (0-shot) |43.99|
137
+ |Winogrande (5-shot) |74.19|
138
+ |GSM8k (5-shot) |54.44|
139
+