File size: 9,530 Bytes
6835659 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 | """
Dynamic Council System with Modality Weighting and Priority Selection.
Implements:
- Dynamic modality weighting based on prompt analysis
- Leader-follower model for modality priority
- Content fusion agent for holistic multimodal outputs
"""
from __future__ import annotations
from dataclasses import dataclass
from typing import Dict, List, Optional, Tuple
from src.embeddings.aligned_embeddings import AlignedEmbedder
from src.planner.council import CouncilResult, Planner, SemanticPlanningCouncil
from src.planner.schema import SemanticPlan
from src.planner.merge_logic import merge_council_plans, MergeReport
@dataclass
class ModalityWeights:
"""Weights for different modalities based on prompt analysis."""
text_weight: float = 1.0
image_weight: float = 1.0
audio_weight: float = 1.0
@property
def total(self) -> float:
return self.text_weight + self.image_weight + self.audio_weight
def normalize(self) -> ModalityWeights:
"""Normalize weights so they sum to 3.0 (equal importance baseline)."""
total = self.total
if total == 0:
return ModalityWeights(1.0, 1.0, 1.0)
scale = 3.0 / total
return ModalityWeights(
text_weight=self.text_weight * scale,
image_weight=self.image_weight * scale,
audio_weight=self.audio_weight * scale,
)
@dataclass
class ModalityPriority:
"""Priority ranking for modalities."""
primary: str # "text", "image", or "audio"
secondary: str
tertiary: str
weights: ModalityWeights
class PromptAnalyzer:
"""Analyzes prompts to determine modality importance."""
def __init__(self):
# Keywords that suggest visual emphasis
self.visual_keywords = {
"see", "view", "look", "appear", "visible", "visual", "image", "photo",
"picture", "scene", "landscape", "color", "bright", "dark", "shade",
"shape", "form", "design", "style", "pattern", "texture",
}
# Keywords that suggest audio emphasis
self.audio_keywords = {
"hear", "sound", "listen", "audio", "music", "noise", "quiet", "loud",
"silence", "voice", "speak", "whisper", "shout", "echo", "resonance",
"tone", "pitch", "melody", "rhythm", "beat", "harmony",
}
# Keywords that suggest narrative/text emphasis
self.text_keywords = {
"story", "narrative", "tale", "describe", "tell", "explain", "detail",
"character", "plot", "scene", "moment", "event", "happen", "occur",
}
def analyze(self, prompt: str) -> ModalityPriority:
"""Analyze prompt and determine modality priority."""
prompt_lower = prompt.lower()
words = set(prompt_lower.split())
# Count keyword matches
visual_score = len(words & self.visual_keywords)
audio_score = len(words & self.audio_keywords)
text_score = len(words & self.text_keywords)
# Boost scores based on prompt length and structure
# Longer prompts with more descriptive words favor text/narrative
word_count = len(words)
if word_count > 15:
text_score += 1
if word_count > 25:
text_score += 1
# Create weights (add 1.0 base to avoid zero weights)
weights = ModalityWeights(
text_weight=1.0 + text_score * 0.5,
image_weight=1.0 + visual_score * 0.5,
audio_weight=1.0 + audio_score * 0.5,
)
# Normalize weights
weights = weights.normalize()
# Determine priority order
scores = [
("text", text_score),
("image", visual_score),
("audio", audio_score),
]
scores.sort(key=lambda x: x[1], reverse=True)
primary = scores[0][0]
secondary = scores[1][0]
tertiary = scores[2][0]
return ModalityPriority(
primary=primary,
secondary=secondary,
tertiary=tertiary,
weights=weights,
)
class ContentFusionAgent:
"""Agent that combines information from multiple modalities for holistic outputs."""
def __init__(self, embedder: Optional[AlignedEmbedder] = None):
self.embedder = embedder or AlignedEmbedder()
def fuse(
self,
plans: List[SemanticPlan],
weights: ModalityWeights,
priority: ModalityPriority,
) -> Tuple[SemanticPlan, Dict[str, any]]:
"""
Fuse multiple plans using weights and priority.
Returns:
- Fused semantic plan
- Fusion metadata (confidence, conflicts, etc.)
"""
if not plans:
raise ValueError("Cannot fuse empty list of plans")
if len(plans) == 1:
return plans[0], {"fusion_method": "single_plan", "confidence": 1.0}
# Use merge logic but with weighted consideration
if len(plans) == 3:
plan_a, plan_b, plan_c = plans
merged, merge_report = merge_council_plans(plan_a, plan_b, plan_c)
# Apply weights to merge report confidence
fusion_metadata = {
"fusion_method": "weighted_council_merge",
"weights": {
"text": weights.text_weight,
"image": weights.image_weight,
"audio": weights.audio_weight,
},
"priority": {
"primary": priority.primary,
"secondary": priority.secondary,
"tertiary": priority.tertiary,
},
"merge_report": merge_report.__dict__ if hasattr(merge_report, "__dict__") else {},
"confidence": merge_report.agreement_score if hasattr(merge_report, "agreement_score") else 0.5,
}
return merged, fusion_metadata
# Fallback: use first plan
return plans[0], {"fusion_method": "first_plan_fallback", "confidence": 0.5}
class DynamicSemanticCouncil(SemanticPlanningCouncil):
"""
Enhanced council with dynamic modality weighting and priority selection.
Features:
- Analyzes prompts to determine modality importance
- Applies weighted merging based on analysis
- Supports leader-follower generation strategies
"""
def __init__(
self,
planner_a: Planner,
planner_b: Planner,
planner_c: Planner,
enable_dynamic_weighting: bool = True,
embedder: Optional[AlignedEmbedder] = None,
):
super().__init__(planner_a, planner_b, planner_c)
self.enable_dynamic_weighting = enable_dynamic_weighting
self.analyzer = PromptAnalyzer()
self.fusion_agent = ContentFusionAgent(embedder=embedder)
def run(self, user_prompt: str) -> CouncilResult:
"""Run council with dynamic weighting."""
# Generate plans from all planners
plan_a = self.planner_a.plan(user_prompt)
plan_b = self.planner_b.plan(user_prompt)
plan_c = self.planner_c.plan(user_prompt)
if self.enable_dynamic_weighting:
# Analyze prompt to determine modality priority
priority = self.analyzer.analyze(user_prompt)
# Use fusion agent for weighted merging
plans = [plan_a, plan_b, plan_c]
merged, fusion_metadata = self.fusion_agent.fuse(plans, priority.weights, priority)
# Create standard merge report and enhance with dynamic weighting info
standard_merged, standard_report = merge_council_plans(plan_a, plan_b, plan_c)
# Enhance notes with dynamic weighting info
enhanced_notes = standard_report.notes
if enhanced_notes:
enhanced_notes += " | "
enhanced_notes += (
f"Dynamic weighting: Primary={priority.primary}, "
f"weights: T={priority.weights.text_weight:.2f}, "
f"I={priority.weights.image_weight:.2f}, "
f"A={priority.weights.audio_weight:.2f}"
)
# Use merged plan from fusion agent if available
final_merged = merged if merged else standard_merged
enhanced_report = MergeReport(
agreement_score=standard_report.agreement_score,
per_section_agreement=standard_report.per_section_agreement,
conflicts=standard_report.conflicts,
notes=enhanced_notes,
)
return CouncilResult(
plan_a=plan_a,
plan_b=plan_b,
plan_c=plan_c,
merged_plan=final_merged,
merge_report=enhanced_report,
)
else:
# Fall back to standard merge
merged, report = merge_council_plans(plan_a, plan_b, plan_c)
return CouncilResult(
plan_a=plan_a,
plan_b=plan_b,
plan_c=plan_c,
merged_plan=merged,
merge_report=report,
)
def get_modality_priority(self, user_prompt: str) -> ModalityPriority:
"""Get modality priority for a prompt without running full council."""
return self.analyzer.analyze(user_prompt)
|