Module: Legion::Extensions::Agentic::Affect::Reward::Runners::Reward

Includes:
Helpers::Lex
Included in:
Client
Defined in:
lib/legion/extensions/agentic/affect/reward/runners/reward.rb

Instance Method Summary collapse

Instance Method Details

#compute_reward(tick_results: {}) ⇒ Object 



13
14
15
16
17
18
19
20
 
# File 'lib/legion/extensions/agentic/affect/reward/runners/reward.rb', line 13

def compute_reward(tick_results: {}, **)
  result = reward_store.process_tick(tick_results)

  log.debug("[reward] reward=#{result[:reward]} rpe=#{result[:rpe]} " \
            "class=#{result[:rpe_class]} learning=#{result[:learning_signal]}")

  result
end

#domain_rewards ⇒ Object 



49
50
51
52
53
54
55
56
57
58
59
 
# File 'lib/legion/extensions/agentic/affect/reward/runners/reward.rb', line 49

def domain_rewards(**)
  averages = reward_store.all_domain_averages
  log.debug("[reward] domains: #{averages.size} tracked")

  {
    domains:      averages,
    domain_count: averages.size,
    best_domain:  averages.max_by { |_, v| v }&.first,
    worst_domain: averages.min_by { |_, v| v }&.first
  }
end

#reward_for(domain:) ⇒ Object 



32
33
34
35
36
 
# File 'lib/legion/extensions/agentic/affect/reward/runners/reward.rb', line 32

def reward_for(domain:, **)
  report = reward_store.domain_report(domain)
  log.debug("[reward] domain=#{domain} avg=#{report[:average].round(3)} trend=#{report[:trend]}")
  report
end

#reward_history(limit: 20) ⇒ Object 



38
39
40
41
42
43
44
45
46
47
 
# File 'lib/legion/extensions/agentic/affect/reward/runners/reward.rb', line 38

def reward_history(limit: 20, **)
  recent = reward_store.signal.recent_rewards(limit)
  log.debug("[reward] history: #{recent.size} entries")

  {
    history:           recent,
    total:             reward_store.signal.history.size,
    discounted_return: reward_store.signal.discounted_return(limit).round(4)
  }
end

#reward_stats ⇒ Object 



61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
 
# File 'lib/legion/extensions/agentic/affect/reward/runners/reward.rb', line 61

def reward_stats(**)
  sig = reward_store.signal
  health = reward_store.health_assessment

  log.debug('[reward] stats')

  {
    running_average:   sig.running_average.round(4),
    predicted_reward:  sig.predicted_reward.round(4),
    volatility:        sig.reward_volatility.round(4),
    tick_count:        sig.tick_count,
    health:            health,
    domains_tracked:   sig.domain_history.keys.size,
    history_size:      sig.history.size,
    discounted_return: sig.discounted_return.round(4),
    anhedonic:         sig.anhedonic?,
    euphoric:          sig.euphoric?
  }
end

#reward_status ⇒ Object 



22
23
24
25
26
27
28
29
30
 
# File 'lib/legion/extensions/agentic/affect/reward/runners/reward.rb', line 22

def reward_status(**)
  sig = reward_store.signal
  health = reward_store.health_assessment

  log.debug("[reward] status: avg=#{sig.running_average.round(3)} " \
            "predicted=#{sig.predicted_reward.round(3)} health=#{health[:status]}")

  sig.to_h.merge(health: health)
end