completion-kit 0.28.20 → 0.28.21

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 6a05f4f94c9f7663a486ca16dcf6d7a60215b3c8017be6576760dd929a6df49d
-  data.tar.gz: d5e26b475bb73c1402d68856f833d91241a6feac0402fdfd0e1ab605a658578f
+  metadata.gz: 8242f60832e3c23d204132839a832e082eb5115c26afd46eb2082ca4e342f908
+  data.tar.gz: 5798c4e0f853e3aa5a2ae20291a7909cc5f70dd7baaa5e9bd2765cb5cbbe5f4f
 SHA512:
-  metadata.gz: b0153fca04674849516b94f8d12021be77dedec325e8e716e40037699194002102ac28fb80ffccc2f01c1cd08a7ddca498055f352ac6bdb5e00721e3c2a39732
-  data.tar.gz: 6b11f2fbd4e155b191f27c561624b3e2ba7ee0358ffc2f97e6430749ee251c6a0ad628696d9b1f48c5827ebce91001d583ab7a2d07c796dd2c28b695db16a7aa
+  metadata.gz: ec13a779260638ccd8968c9c3be4bd250780fdda06f01ea13b9963553756329cb0c092603a7ee66118e854227bcdb68b2b360e66a1e7830c2cfeffc879bf35d3
+  data.tar.gz: 92dd2c71fd38a9cc31638c77927a21a9a53f6ba8a4f4b81dc876f652dbf91c3d601722b402dd71f108d7530ab96442d1d2c113aadf60684ab767902c1e3ec1c9

data/app/controllers/completion_kit/runs_controller.rb

@@ -6,8 +6,9 @@ module CompletionKit
     before_action :load_form_collections, only: [:new, :edit, :create, :update]
 
     def index
-      scope = Run.includes(:prompt, :dataset, :tags, responses: :reviews).order(created_at: :desc).display_scoped
-      @runs = apply_tag_filter(scope)
+      scope = Run.includes(:prompt, :dataset, :tags).order(created_at: :desc).display_scoped
+      @runs = apply_tag_filter(scope).load
+      Run.preload_summaries(@runs)
     end
 
     RESPONSES_PER_PAGE = 100

data/app/models/completion_kit/run.rb

@@ -14,6 +14,8 @@ module CompletionKit
     has_many :suggestions, dependent: :destroy
     has_many :dashboard_dismissals, as: :dismissable, dependent: :destroy
 
+    attr_writer :avg_score, :check_pass_rate, :metric_averages, :response_count
+
     validates :name, presence: true
     validates :status, inclusion: { in: STATUSES }
     validate :dataset_supplies_prompt_variables
@@ -34,6 +36,58 @@ module CompletionKit
       display_scoped.select(:id)
     end
 
+    # Batch-compute the list-view summaries (response count, avg score, check
+    # pass rate, per-metric averages) for a set of runs in a constant number of
+    # grouped queries, injecting the results so the index never loads a single
+    # response or review object. Mirrors the per-run reader methods exactly.
+    def self.preload_summaries(runs)
+      runs = runs.to_a
+      return runs if runs.empty?
+
+      run_ids = runs.map(&:id)
+      counts = Response.where(run_id: run_ids).group(:run_id).count
+
+      run_col = Arel.sql("completion_kit_responses.run_id")
+      base = Review.joins(:response).where(completion_kit_responses: {run_id: run_ids})
+
+      run_rows = base.group(run_col).pluck(
+        run_col,
+        Arel.sql("AVG(ai_score)"),
+        Arel.sql("COUNT(passed)"),
+        Arel.sql("SUM(CASE WHEN passed THEN 1 ELSE 0 END)")
+      )
+      run_stats = run_rows.each_with_object({}) do |(rid, avg, resolved, passed), h|
+        h[rid] = {avg: avg, resolved: resolved.to_i, passed: passed.to_i}
+      end
+
+      metric_rows = base.group(run_col, :metric_name).pluck(
+        run_col,
+        :metric_name,
+        Arel.sql("AVG(ai_score)"),
+        Arel.sql("COUNT(ai_score)"),
+        Arel.sql("COUNT(passed)"),
+        Arel.sql("SUM(CASE WHEN passed THEN 1 ELSE 0 END)")
+      )
+      metrics_by_run = metric_rows.group_by(&:first)
+
+      runs.each do |run|
+        run.response_count = counts.fetch(run.id, 0)
+
+        stats = run_stats[run.id]
+        run.avg_score = stats && stats[:avg] ? stats[:avg].to_f.round(2) : nil
+        run.check_pass_rate = stats && stats[:resolved] > 0 ? (stats[:passed].to_f / stats[:resolved]).round(2) : nil
+
+        run.metric_averages = (metrics_by_run[run.id] || []).filter_map do |(_rid, name, avg, scored, resolved, passed)|
+          if scored.to_i > 0
+            {name: name, avg: avg.to_f.round(1)}
+          elsif resolved.to_i > 0
+            {name: name, kind: "check", pass_rate: (passed.to_i.to_f / resolved.to_i).round(2)}
+          end
+        end
+      end
+      runs
+    end
+
     # A scoring-only run grades a pre-existing column on the dataset instead of
     # generating new outputs. No prompt is attached; the response text is read
     # from row[output_column]; no LLM generation happens.
@@ -126,7 +180,15 @@ module CompletionKit
         end
     end
 
+    def response_count
+      return @response_count if defined?(@response_count)
+
+      responses.size
+    end
+
     def avg_score
+      return @avg_score if defined?(@avg_score)
+
       scores = reviews_for_summary.map(&:ai_score).compact.map(&:to_f)
       return nil if scores.empty?
 
@@ -134,6 +196,8 @@ module CompletionKit
     end
 
     def metric_averages
+      return @metric_averages if defined?(@metric_averages)
+
       reviews_for_summary.group_by(&:metric_name).filter_map do |name, reviews|
         scored = reviews.select { |r| r.ai_score.present? }
         if scored.any?
@@ -150,6 +214,8 @@ module CompletionKit
     end
 
     def check_pass_rate
+      return @check_pass_rate if defined?(@check_pass_rate)
+
       resolved = reviews_for_summary.reject { |r| r.passed.nil? }
       return nil if resolved.empty?
 

data/app/views/completion_kit/runs/_row.html.erb

@@ -27,7 +27,7 @@
   </td>
   <td>
     <span class="ck-runs-table__count">
-      <%= run.responses.size %><% if run.dataset %><span class="ck-runs-table__count-of">/<%= run.dataset.row_count %></span><% end %>
+      <%= run.response_count %><% if run.dataset %><span class="ck-runs-table__count-of">/<%= run.dataset.row_count %></span><% end %>
     </span>
   </td>
   <td>

data/lib/completion_kit/version.rb

@@ -1,3 +1,3 @@
 module CompletionKit
-  VERSION = "0.28.20"
+  VERSION = "0.28.21"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: completion-kit
 version: !ruby/object:Gem::Version
-  version: 0.28.20
+  version: 0.28.21
 platform: ruby
 authors:
 - Damien Bastin