enact-context 0.0.2

Context window management and compaction for Enact
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
default_preset: gpt4_128k
budget:
  total_tokens: 128000
  output_reserve: 4096
  warning_threshold: 80
  critical_threshold: 95
condenser:
  target_tokens: 1500
  max_tokens: 2000
  max_steps: 10
  max_decisions: 5
calibrator:
  max_tokens: 8000
  response_reserve: 2000
  max_history_messages: 20
  max_rag_chunks: 5