completion-kit 0.5.42 → 0.5.43

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (25) hide show
  1. checksums.yaml +4 -4
  2. data/app/assets/javascripts/completion_kit/application.js +17 -0
  3. data/app/assets/stylesheets/completion_kit/application.css +505 -39
  4. data/app/controllers/completion_kit/metrics_controller.rb +35 -24
  5. data/app/jobs/completion_kit/judge_review_job.rb +11 -0
  6. data/app/models/completion_kit/judge_version.rb +32 -1
  7. data/app/services/completion_kit/judge_variant_generator.rb +8 -6
  8. data/app/services/completion_kit/metric_calibration_stats.rb +16 -4
  9. data/app/views/completion_kit/api_reference/_body.html.erb +1 -1
  10. data/app/views/completion_kit/calibrations/_buttons.html.erb +43 -6
  11. data/app/views/completion_kit/calibrations/_trust_panel.html.erb +27 -28
  12. data/app/views/completion_kit/metrics/_form.html.erb +90 -4
  13. data/app/views/completion_kit/metrics/_rubric_diff.html.erb +25 -0
  14. data/app/views/completion_kit/metrics/_rubric_hint.html.erb +4 -0
  15. data/app/views/completion_kit/metrics/_starter_card.html.erb +13 -9
  16. data/app/views/completion_kit/metrics/edit.html.erb +5 -1
  17. data/app/views/completion_kit/metrics/index.html.erb +5 -3
  18. data/app/views/completion_kit/metrics/show.html.erb +132 -126
  19. data/app/views/completion_kit/metrics/starter_preview.html.erb +6 -6
  20. data/app/views/completion_kit/responses/show.html.erb +1 -1
  21. data/app/views/completion_kit/runs/_status_panel.html.erb +2 -2
  22. data/config/routes.rb +2 -1
  23. data/db/migrate/20260525000001_add_version_number_and_published_at_to_judge_versions.rb +24 -0
  24. data/lib/completion_kit/version.rb +1 -1
  25. metadata +4 -1
@@ -30,15 +30,17 @@
30
30
  <%= link_to metric_path(metric), class: "ck-record-name" do %><strong><%= metric.name %></strong><% end %>
31
31
  <% if CompletionKit.config.judge_calibration_enabled %>
32
32
  <% s = CompletionKit::MetricCalibrationStats.for(metric) %>
33
- <p class="ck-metrics-table__trust">
33
+ <p class="ck-metrics-table__trust" title="Calibration: how often this metric's scores match the humans who reviewed them.">
34
+ <%= heroicon_tag "adjustments-horizontal", variant: :outline, "aria-hidden": "true", class: "ck-trust-icon" %>
35
+ <span class="ck-metrics-table__trust-label">Calibration</span>
34
36
  <% if s.counter_only? %>
35
37
  <% if s.sample_size.zero? %>
36
- No verdicts yet
38
+ <span class="ck-metrics-table__trust-state">Not measured yet</span>
37
39
  <% else %>
38
40
  <%= s.sample_size %> / <%= CompletionKit::MetricCalibrationStats::PROVISIONAL_MIN %> verdicts
39
41
  <% end %>
40
42
  <% else %>
41
- <span class="ck-metrics-table__trust-rate" title="<%= s.firm? ? 'Settled read.' : 'Early read keep giving verdicts.' %>">~<%= (s.agreement_point * 100).round %>%</span>
43
+ <span class="ck-metrics-table__trust-rate" title="<%= s.firm? ? 'Settled read.' : 'Early read. Keep giving verdicts.' %>">~<%= (s.agreement_point * 100).round %>%</span>
42
44
  ±<%= (s.margin * 100).round %> pt · <%= s.firm? ? "settled" : "early" %>
43
45
  <% end %>
44
46
  </p>
@@ -8,70 +8,40 @@
8
8
  <h1 class="ck-title"><%= @metric.name %></h1>
9
9
  <% if CompletionKit.config.judge_calibration_enabled %>
10
10
  <%= render "completion_kit/calibrations/trust_panel",
11
- stats: CompletionKit::MetricCalibrationStats.for(@metric) %>
12
- <% if @edit_draft %>
13
- <% pub_instr = @published_judge_version&.instruction.to_s %>
14
- <% draft_instr = @edit_draft.instruction.to_s %>
15
- <% instruction_changed = pub_instr != draft_instr %>
16
- <% rubric_changed = @published_judge_version && @published_judge_version.rubric_bands != @edit_draft.rubric_bands %>
17
- <section class="ck-card ck-card--spaced ck-draft-pending">
18
- <div class="ck-prompt-preview__header">
19
- <p class="ck-kicker">Draft pending</p>
20
- <%= button_to "Publish this version", publish_draft_metric_path(@metric, draft_id: @edit_draft.id),
21
- method: :post, form_class: "inline-block",
22
- class: ck_button_classes(:dark) %>
23
- </div>
24
- <p class="ck-meta-copy">A draft of this metric is saved. Publishing it replaces the live instruction<%= ", rubric," if rubric_changed %> for future runs. Here's what changes.</p>
25
-
26
- <% if instruction_changed %>
27
- <div class="ck-suggest-diff">
28
- <div class="ck-suggest-diff__pane">
29
- <div class="ck-suggest-diff__header">
30
- <span class="ck-suggest-diff__label ck-suggest-diff__label--before">Currently live</span>
31
- </div>
32
- <pre class="ck-suggest-diff__code"><%= ck_word_diff_old(pub_instr, draft_instr) %></pre>
33
- </div>
34
- <div class="ck-suggest-diff__pane">
35
- <div class="ck-suggest-diff__header">
36
- <span class="ck-suggest-diff__label ck-suggest-diff__label--after">Draft</span>
37
- </div>
38
- <pre class="ck-suggest-diff__code"><%= ck_word_diff_new(pub_instr, draft_instr) %></pre>
39
- </div>
40
- </div>
41
- <% else %>
42
- <p class="ck-meta-copy">The instruction is unchanged.</p>
43
- <% end %>
44
-
45
- <% if rubric_changed %>
46
- <p class="ck-meta-copy"><strong>Rubric also changed.</strong> Edit the metric to inspect each band, or publish to apply the new wording.</p>
47
- <% end %>
48
- </section>
49
- <% end %>
11
+ stats: CompletionKit::MetricCalibrationStats.for(@metric),
12
+ metric: @metric %>
13
+ <% end %>
14
+ <% if @metric.tags.any? %>
15
+ <div class="tag-marks-row tag-marks-row--header">
16
+ <%= render "completion_kit/tags/marks", tags: @metric.tags %>
17
+ </div>
50
18
  <% end %>
51
19
  </div>
52
20
  <div class="ck-actions">
53
21
  <% if CompletionKit.config.judge_calibration_enabled %>
54
- <% if @improve_disagreement_count.positive? %>
22
+ <% if @suggestion_draft %>
23
+ <%= link_to "Review improvements →", edit_metric_path(@metric),
24
+ class: ck_button_classes(:dark),
25
+ title: "The model proposed improvements based on your disagreements. Review and apply what you want." %>
26
+ <% elsif @edit_draft %>
27
+ <%= link_to "Review draft →", edit_metric_path(@metric),
28
+ class: ck_button_classes(:dark),
29
+ title: "An unpublished draft of this metric is saved. Review and publish, or keep editing." %>
30
+ <% elsif @improve_disagreement_count.positive? %>
55
31
  <%= button_to "Improve the metric", suggest_variants_metric_path(@metric),
56
32
  method: :post, form_class: "inline-block",
57
33
  class: ck_button_classes(:light, variant: :outline),
58
- title: "Rewrite this metric based on the disagreements collected so far.",
34
+ title: "Have the model rewrite this metric's instruction and rubric based on the disagreements collected so far.",
59
35
  data: { turbo_confirm: "Rewrite this metric based on the disagreements collected so far?" } %>
60
36
  <% else %>
61
37
  <button type="button" class="<%= ck_button_classes(:light, variant: :outline) %>" disabled
62
- title="Mark at least one row as Disagree before the model can suggest a change.">Improve the metric</button>
38
+ title="Mark at least one case as Disagree before the model can suggest a change.">Improve the metric</button>
63
39
  <% end %>
64
40
  <% end %>
65
41
  <%= link_to "Edit", edit_metric_path(@metric), class: ck_button_classes(:light, variant: :outline) %>
66
42
  </div>
67
43
  </section>
68
44
 
69
- <% if @metric.tags.any? %>
70
- <div class="tag-marks-row tag-marks-row--header">
71
- <%= render "completion_kit/tags/marks", tags: @metric.tags %>
72
- </div>
73
- <% end %>
74
-
75
45
  <% if @metric.instruction.present? %>
76
46
  <section class="ck-card">
77
47
  <p class="ck-kicker">Instruction</p>
@@ -80,7 +50,7 @@
80
50
  <% end %>
81
51
 
82
52
  <section class="ck-card ck-card--spaced">
83
- <p class="ck-kicker">Rubric</p>
53
+ <p class="ck-kicker">Rubric<%= render "completion_kit/metrics/rubric_hint" %></p>
84
54
  <div class="ck-rubric-display">
85
55
  <% @metric.rubric_bands_for_form.each do |band| %>
86
56
  <div class="ck-rubric-row ck-rubric-row--display">
@@ -97,64 +67,121 @@
97
67
  </div>
98
68
  </section>
99
69
 
100
- <% if CompletionKit.config.judge_calibration_enabled && @suggestion_draft %>
101
- <% sd_current_instr = @published_judge_version&.instruction.to_s %>
102
- <% sd_draft_instr = @suggestion_draft.instruction.to_s %>
103
- <% sd_current_rubric = @published_judge_version&.rubric_bands || [] %>
104
- <% sd_rubric_changed = @suggestion_draft.rubric_bands != sd_current_rubric %>
105
- <section class="ck-card ck-card--spaced ck-draft-pending">
106
- <div class="ck-prompt-preview__header">
107
- <p class="ck-kicker">Suggested change</p>
108
- <time class="ck-meta-copy" data-relative-time datetime="<%= @suggestion_draft.created_at.utc.iso8601 %>"><%= time_ago_in_words(@suggestion_draft.created_at) %> ago</time>
109
- </div>
110
- <p class="ck-meta-copy">Based on your disagreements, the model proposed this rewrite. Use it to replace the live version, or discard.</p>
70
+ <% if CompletionKit.config.judge_calibration_enabled && @versions.size > 1 %>
71
+ <% predecessor_of = @versions.index_with { |v| @versions.detect { |o| o.version_number < v.version_number } } %>
72
+ <% version_changed = ->(v, pred) { pred && (pred.instruction.to_s != v.instruction.to_s || pred.rubric_bands != v.rubric_bands) } %>
73
+ <section class="ck-card ck-card--spaced">
74
+ <p class="ck-kicker">Versions</p>
75
+ <table class="ck-results-table ck-metric-versions-table">
76
+ <thead>
77
+ <tr>
78
+ <th scope="col">Version</th>
79
+ <th scope="col">Source</th>
80
+ <th scope="col">Created</th>
81
+ </tr>
82
+ </thead>
83
+ <tbody>
84
+ <% @versions.each do |v| %>
85
+ <% pred = predecessor_of[v] %>
86
+ <tr>
87
+ <td>
88
+ <div class="ck-version-cell">
89
+ <div class="ck-version-cell__label">
90
+ <strong><%= v.version_label %></strong>
91
+ <% if v.current? %>
92
+ <span class="ck-chip">Published</span>
93
+ <% elsif v.draft? %>
94
+ <%= button_to "Publish", publish_draft_metric_path(@metric, draft_id: v.id),
95
+ method: :post, form_class: "inline-block",
96
+ class: "ck-chip ck-chip--publish" %>
97
+ <% else %>
98
+ <%= button_to "Make current", publish_draft_metric_path(@metric, draft_id: v.id),
99
+ method: :post, form_class: "inline-block",
100
+ class: "ck-chip ck-chip--publish",
101
+ data: { turbo_confirm: "Roll the live judge back to #{v.version_label}? Calibration verdicts collected against the current version stay tied to it." } %>
102
+ <% end %>
103
+ </div>
104
+ <% if version_changed.call(v, pred) %>
105
+ <button type="button" class="ck-cell-link ck-cell-link--delta"
106
+ title="What changed from #{pred.version_label}"
107
+ onclick="document.getElementById('ck-mvdiff-<%= v.id %>').showModal()">&Delta;</button>
108
+ <% end %>
109
+ </div>
110
+ </td>
111
+ <td>
112
+ <% source_label, source_class = case v.source
113
+ when "suggestion" then ["AI suggestion", "ck-source-chip ck-source-chip--ai"]
114
+ when "edit" then ["Manual edit", "ck-source-chip ck-source-chip--manual"]
115
+ else ["Original", "ck-source-chip ck-source-chip--initial"]
116
+ end %>
117
+ <span class="<%= source_class %>"><%= source_label %></span>
118
+ </td>
119
+ <td class="ck-meta-copy">
120
+ <time datetime="<%= v.created_at.utc.iso8601 %>" data-relative-time><%= time_ago_in_words(v.created_at) %> ago</time>
121
+ </td>
122
+ </tr>
123
+ <% end %>
124
+ </tbody>
125
+ </table>
126
+ </section>
111
127
 
112
- <div class="ck-suggest-diff">
113
- <div class="ck-suggest-diff__pane">
114
- <div class="ck-suggest-diff__header">
115
- <span class="ck-suggest-diff__label ck-suggest-diff__label--before">Currently live</span>
116
- </div>
117
- <pre class="ck-suggest-diff__code"><%= ck_word_diff_old(sd_current_instr, sd_draft_instr) %></pre>
118
- </div>
119
- <div class="ck-suggest-diff__pane">
120
- <div class="ck-suggest-diff__header">
121
- <span class="ck-suggest-diff__label ck-suggest-diff__label--after">Proposed</span>
128
+ <% @versions.each do |v| %>
129
+ <% pred = predecessor_of[v] %>
130
+ <% next unless version_changed.call(v, pred) %>
131
+ <dialog id="ck-mvdiff-<%= v.id %>" class="ck-modal" onclick="if(event.target===this)this.close()">
132
+ <article class="ck-modal__panel" tabindex="-1" onclick="event.stopPropagation()">
133
+ <header class="ck-modal__header">
134
+ <div class="ck-modal__heading">
135
+ <h2 class="ck-modal__title"><%= pred.version_label %> &rarr; <%= v.version_label %></h2>
136
+ <span class="ck-modal__meta">What changed in <%= v.version_label %><% if v.current? %> (live)<% elsif v.draft? %> (draft)<% end %></span>
137
+ </div>
138
+ <button type="button" class="ck-modal__close" aria-label="Close" onclick="this.closest('dialog').close()">&times;</button>
139
+ </header>
140
+ <div class="ck-modal__body">
141
+ <% if pred.instruction.to_s != v.instruction.to_s %>
142
+ <div class="ck-suggest-diff">
143
+ <div class="ck-suggest-diff__pane">
144
+ <div class="ck-suggest-diff__header"><span class="ck-suggest-diff__label ck-suggest-diff__label--before"><%= pred.version_label %> instruction</span></div>
145
+ <pre class="ck-suggest-diff__code"><%= ck_word_diff_old(pred.instruction.to_s, v.instruction.to_s) %></pre>
146
+ </div>
147
+ <div class="ck-suggest-diff__pane">
148
+ <div class="ck-suggest-diff__header"><span class="ck-suggest-diff__label ck-suggest-diff__label--after"><%= v.version_label %> instruction</span></div>
149
+ <pre class="ck-suggest-diff__code"><%= ck_word_diff_new(pred.instruction.to_s, v.instruction.to_s) %></pre>
150
+ </div>
151
+ </div>
152
+ <% end %>
153
+ <% if pred.rubric_bands != v.rubric_bands %>
154
+ <p class="ck-kicker ck-kicker--inset">Rubric changes</p>
155
+ <%= render "completion_kit/metrics/rubric_diff",
156
+ current_bands: pred.rubric_bands || [],
157
+ draft_bands: v.rubric_bands || [] %>
158
+ <% end %>
122
159
  </div>
123
- <pre class="ck-suggest-diff__code"><%= ck_word_diff_new(sd_current_instr, sd_draft_instr) %></pre>
124
- </div>
125
- </div>
126
-
127
- <% if sd_rubric_changed %>
128
- <p class="ck-meta-copy"><strong>Rubric also changed.</strong> Publishing applies the new rubric too.</p>
129
- <% end %>
130
-
131
- <div class="ck-actions">
132
- <%= button_to "Discard", dismiss_suggestion_metric_path(@metric, draft_id: @suggestion_draft.id),
133
- method: :delete, form_class: "inline-block",
134
- class: ck_button_classes(:light, variant: :outline),
135
- data: { turbo_confirm: "Drop this suggestion?" } %>
136
- <%= button_to "Use this version", publish_draft_metric_path(@metric, draft_id: @suggestion_draft.id),
137
- method: :post, form_class: "inline-block",
138
- class: ck_button_classes(:dark) %>
139
- </div>
140
- </section>
160
+ </article>
161
+ </dialog>
162
+ <% end %>
141
163
  <% end %>
142
164
 
143
165
  <% if CompletionKit.config.judge_calibration_enabled && @disagreements.any? %>
144
166
  <section class="ck-card ck-card--spaced">
145
167
  <div class="ck-prompt-preview__header">
146
168
  <p class="ck-kicker">Cases to learn from</p>
147
- <span class="ck-chip"><%= pluralize(@disagreements.size, "row") %></span>
169
+ <span class="ck-chip"><%= pluralize(@disagreements.size, "case") %></span>
148
170
  </div>
149
- <p class="ck-meta-copy">Rows where a reviewer's score didn't match the judge's. Pin useful ones with <strong>Remember this</strong> the judge sees them next time it grades.</p>
171
+ <p class="ck-meta-copy">Cases where a reviewer's score didn't match the judge's. Pin useful ones with <strong>Remember this</strong> so the judge sees them next time it grades.</p>
150
172
  <% existing_ids = Array(@metric.few_shot_examples).map { |fs| fs["calibration_id"] } %>
151
173
  <ul class="ck-disagreement-list">
152
174
  <% @disagreements.each do |cal| %>
153
175
  <% review = cal.response.reviews.find { |r| r.metric_id == @metric.id } %>
154
176
  <% already = existing_ids.include?(cal.id) %>
155
- <li class="ck-disagreement">
177
+ <% cal_version = cal.judge_version %>
178
+ <% on_current = cal_version&.id == @published_judge_version.id %>
179
+ <li class="ck-disagreement<%= " ck-disagreement--remembered" if already %><%= " ck-disagreement--stale" unless on_current %>">
156
180
  <div class="ck-disagreement__head">
157
181
  <div class="ck-disagreement__scores">
182
+ <% if cal_version %>
183
+ <span class="ck-source-chip <%= on_current ? "ck-source-chip--current" : "ck-source-chip--past" %>" title="<%= on_current ? "Verdict on the live judge version." : "Verdict on a superseded judge version." %>"><%= cal_version.version_label %></span>
184
+ <% end %>
158
185
  <span class="ck-disagreement__scores-label">Judge</span>
159
186
  <% if review&.ai_score %>
160
187
  <span class="<%= ck_badge_classes(ck_score_kind(review.ai_score.to_f)) %>"><%= review.ai_score %></span>
@@ -171,14 +198,21 @@
171
198
  </div>
172
199
  <div class="ck-disagreement__action">
173
200
  <% if already %>
174
- <span class="ck-chip ck-chip--done">Remembered</span>
201
+ <%= button_to "Forget",
202
+ remove_few_shot_metric_path(@metric, calibration_id: cal.id),
203
+ method: :delete,
204
+ form_class: "inline-block",
205
+ class: ck_button_classes(:light, variant: :outline),
206
+ title: "Stop showing this case to the judge.",
207
+ data: { turbo_confirm: "Stop showing this case to the judge?" } %>
208
+ <span class="ck-chip ck-chip--done" title="The judge sees this row when it grades for this metric.">Remembered</span>
175
209
  <% else %>
176
210
  <%= button_to "Remember this",
177
211
  add_few_shot_metric_path(@metric, calibration_id: cal.id),
178
212
  method: :post,
179
213
  form_class: "inline-block",
180
214
  class: ck_button_classes(:light, variant: :outline),
181
- title: "Pin this row so the judge sees it next time it grades for this metric." %>
215
+ title: "Pin this case so the judge sees it next time it grades for this metric." %>
182
216
  <% end %>
183
217
  </div>
184
218
  </div>
@@ -186,42 +220,14 @@
186
220
  <p class="ck-disagreement__note"><%= cal.note %></p>
187
221
  <% end %>
188
222
  <p class="ck-disagreement__source ck-meta-copy">
189
- <%= link_to cal.response.run.name.to_s.truncate(50), ck_run_path(cal.response.run), class: "ck-link" %>
190
- ·
191
- <%= link_to "row ##{cal.response.id}", run_response_path(cal.response.run, cal.response), class: "ck-link" %>
223
+ <%= link_to run_response_path(cal.response.run, cal.response, anchor: @metric.name.parameterize),
224
+ class: "ck-disagreement__source-link" do %>
225
+ <% case_display = cal.response.row_index.nil? ? "##{cal.response.id}" : (cal.response.row_index + 1).to_s %>
226
+ View case <%= case_display %> in <%= cal.response.run.name.to_s.truncate(50) %> →
227
+ <% end %>
192
228
  </p>
193
229
  </li>
194
230
  <% end %>
195
231
  </ul>
196
232
  </section>
197
-
198
-
199
- <% if Array(@metric.few_shot_examples).any? %>
200
- <section class="ck-card ck-card--spaced">
201
- <div class="ck-prompt-preview__header">
202
- <p class="ck-kicker">What the judge remembers</p>
203
- <span class="ck-chip"><%= pluralize(Array(@metric.few_shot_examples).size, "case") %></span>
204
- </div>
205
- <p class="ck-meta-copy">Rows you've pinned so the judge sees them next time it grades. Each one shows what the judge gave and what a human said it should have been.</p>
206
- <ol class="ck-few-shot-list">
207
- <% Array(@metric.few_shot_examples).each do |fs| %>
208
- <li class="ck-few-shot-item">
209
- <div class="ck-few-shot-item__scores">
210
- <span class="ck-meta-copy">judge said</span>
211
- <% if fs["judge_score"] %>
212
- <span class="<%= ck_badge_classes(ck_score_kind(fs["judge_score"].to_f)) %>"><%= fs["judge_score"] %></span>
213
- <% end %>
214
- <span class="ck-meta-copy">human said</span>
215
- <% if fs["human_score"] %>
216
- <span class="<%= ck_badge_classes(ck_score_kind(fs["human_score"].to_f)) %>"><%= fs["human_score"] %></span>
217
- <% end %>
218
- </div>
219
- <% if fs["human_note"].to_s.present? %>
220
- <p class="ck-copy"><%= fs["human_note"] %></p>
221
- <% end %>
222
- </li>
223
- <% end %>
224
- </ol>
225
- </section>
226
- <% end %>
227
233
  <% end %>
@@ -2,7 +2,6 @@
2
2
  <div>
3
3
  <p class="ck-kicker">Starter metric</p>
4
4
  <h1 class="ck-title"><%= @starter.name %></h1>
5
- <p class="ck-lead"><%= @starter.description %></p>
6
5
  </div>
7
6
  <div class="ck-actions">
8
7
  <%= link_to "← Back to metrics", metrics_path, class: ck_button_classes(:light, variant: :outline) %>
@@ -10,17 +9,18 @@
10
9
  </section>
11
10
 
12
11
  <section class="ck-card ck-card--spaced">
13
- <p class="ck-kicker">What this catches</p>
12
+ <p class="ck-kicker">Why use this</p>
13
+ <p class="ck-copy"><strong><%= @starter.description %></strong></p>
14
14
  <p class="ck-copy"><%= @starter.catches %></p>
15
15
  </section>
16
16
 
17
17
  <section class="ck-card ck-card--spaced">
18
- <p class="ck-kicker">Instruction the judge will see</p>
18
+ <p class="ck-kicker">Judge instruction</p>
19
19
  <p class="ck-copy"><%= @starter.instruction %></p>
20
20
  </section>
21
21
 
22
22
  <section class="ck-card ck-card--spaced">
23
- <p class="ck-kicker">Rubric</p>
23
+ <p class="ck-kicker">Rubric<%= render "completion_kit/metrics/rubric_hint" %></p>
24
24
  <div class="ck-rubric-display">
25
25
  <% @starter.rubric_bands.sort_by { |b| -b["stars"] }.each do |band| %>
26
26
  <div class="ck-rubric-row ck-rubric-row--display">
@@ -37,9 +37,9 @@
37
37
  </div>
38
38
  </section>
39
39
 
40
- <div class="ck-actions ck-actions--right">
40
+ <div class="ck-starter-actions">
41
41
  <%= link_to "Cancel", metrics_path, class: ck_button_classes(:light, variant: :outline) %>
42
- <%= button_to "Add this metric", adopt_starter_metrics_path(key: @starter.key),
42
+ <%= button_to "Add #{@starter.name} to my metrics", adopt_starter_metrics_path(key: @starter.key),
43
43
  method: :post, form_class: "inline-block",
44
44
  class: ck_button_classes(:dark) %>
45
45
  </div>
@@ -98,7 +98,7 @@
98
98
 
99
99
  <div class="ck-review-list">
100
100
  <% @reviews.each do |review| %>
101
- <div class="ck-review-card">
101
+ <div class="ck-review-card" id="<%= review.metric&.name&.parameterize || "review-#{review.id}" %>">
102
102
  <div class="ck-review-card__header">
103
103
  <span class="ck-review-card__metric"><% if review.metric %><%= link_to review.metric_name, metric_path(review.metric), class: "ck-link" %><% else %><%= review.metric_name %><% end %></span>
104
104
  <div class="ck-inline">
@@ -46,7 +46,7 @@
46
46
  <% else %>
47
47
  <span class="ck-metric-pip ck-metric-pip--pending">
48
48
  <span class="ck-metric-pip__bar"></span>
49
- <span class="ck-metric-pip__label"><%= metric.name %> <em>pending</em></span>
49
+ <span class="ck-metric-pip__label"><%= metric.name %> <span class="ck-metric-pip__sub">pending</span></span>
50
50
  </span>
51
51
  <% end %>
52
52
  <% end %>
@@ -72,7 +72,7 @@
72
72
 
73
73
  <% if failed_count > 0 %>
74
74
  <%= button_to retry_failures_run_path(run), method: :post, class: "ck-run-status__retry", form_class: "ck-run-status__action" do %>
75
- Retry <%= failed_count %> failed <%= "row".pluralize(failed_count) %>
75
+ Retry <%= failed_count %> failed <%= "case".pluralize(failed_count) %>
76
76
  <% end %>
77
77
  <% end %>
78
78
  </section>
data/config/routes.rb CHANGED
@@ -19,7 +19,8 @@ CompletionKit::Engine.routes.draw do
19
19
  post "starters/:key/dismiss", to: "metrics#dismiss_starter", as: :dismiss_starter
20
20
  end
21
21
  member do
22
- post :add_few_shot
22
+ post :add_few_shot
23
+ delete :remove_few_shot
23
24
  post :publish_draft
24
25
  post :suggest_variants
25
26
  delete :dismiss_suggestion
@@ -0,0 +1,24 @@
1
+ class AddVersionNumberAndPublishedAtToJudgeVersions < ActiveRecord::Migration[8.1]
2
+ def change
3
+ add_column :completion_kit_judge_versions, :version_number, :integer
4
+ add_column :completion_kit_judge_versions, :published_at, :datetime
5
+
6
+ reversible do |dir|
7
+ dir.up do
8
+ jv = Class.new(ActiveRecord::Base) { self.table_name = "completion_kit_judge_versions" }
9
+ jv.distinct.pluck(:metric_id).each do |metric_id|
10
+ jv.where(metric_id: metric_id).order(:created_at, :id).each_with_index do |row, i|
11
+ updates = { version_number: i + 1 }
12
+ updates[:published_at] = row.created_at if row[:state] == "published"
13
+ jv.where(id: row.id).update_all(updates)
14
+ end
15
+ end
16
+ end
17
+ end
18
+
19
+ change_column_null :completion_kit_judge_versions, :version_number, false
20
+ add_index :completion_kit_judge_versions,
21
+ [:metric_id, :version_number],
22
+ name: "index_ck_judge_versions_on_metric_version"
23
+ end
24
+ end
@@ -1,3 +1,3 @@
1
1
  module CompletionKit
2
- VERSION = "0.5.42"
2
+ VERSION = "0.5.43"
3
3
  end
metadata CHANGED
@@ -1,7 +1,7 @@
1
1
  --- !ruby/object:Gem::Specification
2
2
  name: completion-kit
3
3
  version: !ruby/object:Gem::Version
4
- version: 0.5.42
4
+ version: 0.5.43
5
5
  platform: ruby
6
6
  authors:
7
7
  - Damien Bastin
@@ -347,6 +347,8 @@ files:
347
347
  - app/views/completion_kit/metric_groups/new.html.erb
348
348
  - app/views/completion_kit/metric_groups/show.html.erb
349
349
  - app/views/completion_kit/metrics/_form.html.erb
350
+ - app/views/completion_kit/metrics/_rubric_diff.html.erb
351
+ - app/views/completion_kit/metrics/_rubric_hint.html.erb
350
352
  - app/views/completion_kit/metrics/_starter_card.html.erb
351
353
  - app/views/completion_kit/metrics/edit.html.erb
352
354
  - app/views/completion_kit/metrics/index.html.erb
@@ -419,6 +421,7 @@ files:
419
421
  - db/migrate/20260523000001_add_few_shot_examples_to_completion_kit_metrics.rb
420
422
  - db/migrate/20260523000002_add_state_to_completion_kit_judge_versions.rb
421
423
  - db/migrate/20260524000001_create_completion_kit_starter_metric_dismissals.rb
424
+ - db/migrate/20260525000001_add_version_number_and_published_at_to_judge_versions.rb
422
425
  - lib/completion-kit.rb
423
426
  - lib/completion_kit.rb
424
427
  - lib/completion_kit/concurrency_check.rb