RDson commited on
Commit
c52ca11
·
verified ·
1 Parent(s): 18bf4a8

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +61 -0
README.md CHANGED
@@ -52,4 +52,65 @@ args = dict(
52
  preprocessing_num_workers=16,
53
  cutoff_len=max_seq_length,
54
  )
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
55
  ```
 
52
  preprocessing_num_workers=16,
53
  cutoff_len=max_seq_length,
54
  )
55
+ ```
56
+
57
+ System used:
58
+ ```
59
+ 'You are a helpful assistant. Please reason step by step inside the tags <think> and </think>. Conclude with **Answer** and put your final answer within \\boxed{}.'
60
+ ```
61
+
62
+ Custom template used in training:
63
+ ```
64
+ register_template(
65
+ name="custom_template",
66
+ format_user=StringFormatter(
67
+ slots=["<|User|>{{content}}"]
68
+ ),
69
+ format_assistant=StringFormatter(
70
+ slots=["<|Assistant|>{{content}}<|end▁of▁sentence|>"]
71
+ ),
72
+ format_system=StringFormatter(
73
+ slots=["{{content}}"]
74
+ ),
75
+ format_function=FunctionFormatter(
76
+ slots=[
77
+ "<|Assistant|><|tool▁calls▁begin|><|tool▁call▁begin|>{{type}}<|tool▁sep|>{{name}}\n```json\n{{arguments}}\n```<|tool▁call▁end|><|tool▁calls▁end|><|end▁of▁sentence|>"
78
+ ],
79
+ tool_format="qwen"
80
+ ),
81
+ format_observation=StringFormatter(
82
+ slots=[
83
+ "<|tool▁outputs▁begin|><|tool▁output_begin|>{{content}}<|tool▁output▁end|><|tool▁outputs▁end|>"
84
+ ]
85
+ ),
86
+ format_tools=ToolFormatter(tool_format="qwen"),
87
+ default_system="",
88
+ stop_words=["<|end▁of▁sentence|>"]
89
+ )
90
+ ```
91
+
92
+ In the dataset for variation, I randomly replaced the start of the string "Okay," with one of the following:
93
+ ```
94
+ starts = [
95
+ "Alright,",
96
+ "Well,",
97
+ "So,",
98
+ "Hmm,",
99
+ "Okay then,",
100
+ "Right,",
101
+ "Let's see,",
102
+ "Now,",
103
+ "Alrighty,",
104
+ "Thinking about it,",
105
+ "You know,",
106
+ "Well then,",
107
+ "Come to think of it,",
108
+ "Actually,",
109
+ "Now that I think about it,",
110
+ "Good question,",
111
+ "Let me think,",
112
+ "Let's see now,",
113
+ "Interesting,",
114
+ "Now then,"
115
+ ]
116
  ```