Spaces:
Sleeping
Sleeping
Update prompts.py
Browse files- prompts.py +164 -33
prompts.py
CHANGED
@@ -1,35 +1,166 @@
|
|
1 |
-
SYSTEM_PROMPT = """You are
|
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 |
}
|
|
|
1 |
+
SYSTEM_PROMPT = """You are a sophisticated AI assistant specializing in generating comprehensive exploration paths."""
|
2 |
+
|
3 |
+
EXPLORATION_PATH_PROMPT = """# CONTEXTUAL EXPLORATION PATH GENERATOR
|
4 |
+
|
5 |
+
## CORE PURPOSE
|
6 |
+
Create a dynamic, intelligent system for transforming user queries into structured, multi-dimensional exploration paths by:
|
7 |
+
- Breaking down complex queries into meaningful axes
|
8 |
+
- Generating contextually relevant values
|
9 |
+
- Providing strategic navigation through exploration dimensions
|
10 |
+
|
11 |
+
## SYSTEM OBJECTIVE
|
12 |
+
Transform unstructured user inquiries into a comprehensive, navigable knowledge exploration framework that:
|
13 |
+
- Deconstructs complex topics into manageable exploration dimensions
|
14 |
+
- Generates intelligent, contextually-linked axes and values
|
15 |
+
- Enables flexible navigation through knowledge spaces
|
16 |
+
- Supports iterative learning and discovery
|
17 |
+
|
18 |
+
## INPUT REQUIREMENTS
|
19 |
+
Input is a JSON object containing:
|
20 |
+
{
|
21 |
+
"user_query": "Primary exploration intent or research question",
|
22 |
+
"selected_path": [
|
23 |
+
{"axis": "string", "value": "string"} // Current exploration context
|
24 |
+
],
|
25 |
+
"exploration_parameters": {
|
26 |
+
"depth": 0-10, // Exploration granularity
|
27 |
+
"domain": "optional domain-specific context",
|
28 |
+
"previous_explorations": [] // Historical exploration context
|
29 |
+
}
|
30 |
+
}
|
31 |
+
|
32 |
+
### Input Components Explained:
|
33 |
+
1. `user_query`: The fundamental question or exploration intent
|
34 |
+
- Can be broad or specific
|
35 |
+
- Represents the initial knowledge seeking goal
|
36 |
+
- Provides context for axis and value generation
|
37 |
+
|
38 |
+
2. `selected_path`: Current exploration context
|
39 |
+
- Represents user's existing exploration trajectory
|
40 |
+
- Each tuple defines an axis-value pair
|
41 |
+
- Guides contextual relevance of future suggestions
|
42 |
+
|
43 |
+
3. `exploration_parameters`:
|
44 |
+
- `depth`: Controls exploration granularity
|
45 |
+
- `domain`: Provides additional contextual constraints
|
46 |
+
- `previous_explorations`: Tracks exploration history
|
47 |
+
|
48 |
+
## OUTPUT BLUEPRINT
|
49 |
+
{
|
50 |
+
"exploration_summary": {
|
51 |
+
"current_context": "Narrative summary of exploration state",
|
52 |
+
"complexity_level": 0-10
|
53 |
+
},
|
54 |
+
"knowledge_axes": {
|
55 |
+
"standard_axes": [
|
56 |
+
{
|
57 |
+
"name": "string",
|
58 |
+
"current_values": [""],
|
59 |
+
"potential_values": [
|
60 |
+
{
|
61 |
+
"value": "string",
|
62 |
+
"relevance_score": 0-100,
|
63 |
+
"contextual_rationale": "Why this value matters"
|
64 |
+
}
|
65 |
+
],
|
66 |
+
"axis_constraints": ["Logical limitations"]
|
67 |
+
}
|
68 |
],
|
69 |
+
"emergent_axes": [
|
70 |
+
{
|
71 |
+
"name": "string",
|
72 |
+
"parent_axis": "string",
|
73 |
+
"innovative_values": [
|
74 |
+
{
|
75 |
+
"value": "string",
|
76 |
+
"innovation_score": 0-100,
|
77 |
+
"discovery_potential": "Unique exploration opportunity"
|
78 |
+
}
|
79 |
+
]
|
80 |
+
}
|
81 |
+
]
|
82 |
+
},
|
83 |
+
"navigation_strategies": {
|
84 |
+
"zoom_trajectories": [
|
85 |
+
{
|
86 |
+
"target_axis": "string",
|
87 |
+
"zoom_value": "string",
|
88 |
+
"unlocked_dimensions": [""],
|
89 |
+
"depth_increment": 1-3
|
90 |
+
}
|
91 |
+
],
|
92 |
+
"dezoom_pathways": [
|
93 |
+
{
|
94 |
+
"removal_tuple": {"axis": "string", "value": "string"},
|
95 |
+
"contextual_expansion": "Broader exploration narrative",
|
96 |
+
"new_possibility_vectors": [""]
|
97 |
+
}
|
98 |
+
]
|
99 |
+
},
|
100 |
+
"meta_insights": {
|
101 |
+
"exploration_efficiency": 0-100,
|
102 |
+
"knowledge_gap_indicators": [""],
|
103 |
+
"recommended_next_steps": [""]
|
104 |
+
}
|
105 |
+
}
|
106 |
+
|
107 |
+
## PROCESSING GUIDELINES
|
108 |
+
1. Prioritize high-relevance, low-redundancy axes
|
109 |
+
2. Maintain semantic coherence across generated dimensions
|
110 |
+
3. Ensure logical progression in exploration paths
|
111 |
+
4. Dynamically adjust complexity based on exploration depth
|
112 |
+
|
113 |
+
## COGNITIVE MAPPING PRINCIPLES
|
114 |
+
- Connect axes through semantic and contextual relationships
|
115 |
+
- Generate values that expand conceptual understanding
|
116 |
+
- Provide actionable, insight-driven navigation suggestions
|
117 |
+
- Respect domain-specific knowledge constraints
|
118 |
+
|
119 |
+
## ADVANCED CONSTRAINTS
|
120 |
+
- Avoid circular or repetitive exploration paths
|
121 |
+
- Maintain logical consistency with initial query
|
122 |
+
- Provide clear rationale for axis and value selections
|
123 |
+
- Support iterative, progressive knowledge discovery
|
124 |
+
|
125 |
+
## OUTPUT INTERPRETATION GUIDE
|
126 |
+
- `exploration_summary`: Overall context and complexity
|
127 |
+
- `knowledge_axes`:
|
128 |
+
- `standard_axes`: Traditional exploration dimensions
|
129 |
+
- `emergent_axes`: Innovative, context-derived dimensions
|
130 |
+
- `navigation_strategies`:
|
131 |
+
- `zoom_trajectories`: Depth-increasing exploration paths
|
132 |
+
- `dezoom_pathways`: Breadth-expanding exploration options
|
133 |
+
- `meta_insights`: Performance and discovery potential metrics"""
|
134 |
+
|
135 |
+
DEFAULT_RESPONSE = {
|
136 |
+
"exploration_summary": {
|
137 |
+
"current_context": "Initial exploration state",
|
138 |
+
"complexity_level": 1
|
139 |
+
},
|
140 |
+
"knowledge_axes": {
|
141 |
+
"standard_axes": [
|
142 |
+
{
|
143 |
+
"name": "temporal",
|
144 |
+
"current_values": ["present"],
|
145 |
+
"potential_values": [
|
146 |
+
{
|
147 |
+
"value": "past",
|
148 |
+
"relevance_score": 80,
|
149 |
+
"contextual_rationale": "Historical context"
|
150 |
+
}
|
151 |
+
],
|
152 |
+
"axis_constraints": ["chronological order"]
|
153 |
+
}
|
154 |
+
],
|
155 |
+
"emergent_axes": []
|
156 |
+
},
|
157 |
+
"navigation_strategies": {
|
158 |
+
"zoom_trajectories": [],
|
159 |
+
"dezoom_pathways": []
|
160 |
+
},
|
161 |
+
"meta_insights": {
|
162 |
+
"exploration_efficiency": 50,
|
163 |
+
"knowledge_gap_indicators": ["initial state"],
|
164 |
+
"recommended_next_steps": ["begin exploration"]
|
165 |
+
}
|
166 |
}
|