{{annotation_task

{#- The labeling engine's request on this item. Blind: the server already withheld every machine answer; the note says why the AI help is gone. Shown: the machine's answer is pre-filled and the note says so. -#} {% if er %}

{%- set many = (annotation_schemes | default([]) | length) > 1 %} {%- if er.visibility == 'blind' %} {% if many %}Answer “{{ er.schema_label | default(er.schema) }}” yourself{% else %}Label this one yourself{% endif %}. Suggestions are off for this item. {%- elif er.shown %} {% if many %}For “{{ er.schema_label | default(er.schema) }}”, a model's answer is filled in{% else %}A model's answer is filled in{% endif %}. Keep it if it is right; change it if it is not. {%- elif er.purpose == 'adjudicate' %} People disagreed on this item. Your answer settles it. {%- else %} {% if many %}For “{{ er.schema_label | default(er.schema) }}”, the models could not settle this item{% else %}The models could not settle this item{% endif %}, so your answer is the label. {%- endif %}

{% endif %} {% if annotation_instructions | default('') %} {% endif %}

{% if has_instance_display | default(false) %} {{ display_html | safe }} {% elif has_video_annotation | default(false) %} {% if instance_text_is_media_path | default(false) %} {% else %}

{% endif %} {% elif has_audio_annotation | default(false) %} {% if instance_text_is_media_path | default(false) %} {% else %}

{% endif %} {% elif has_image_annotation | default(false) %} {% else %} {# Training shows its practice question here too. Gated on a separate flag defaulting to is_annotation_page so every non-training render is unchanged. #} {% if show_instance_text | default(is_annotation_page) %} {# The heading names what the item is. Items that embed a

{% if triage_info is defined and triage_info %}

Prioritized for review · {{ triage_info.reason }}

{% endif %} {% if judge_prediction is defined and judge_prediction %} {% set jp = judge_prediction %} {% set jconf = (jp.confidence * 100)|int %} {% set jqual = 'Low confidence' if jp.confidence < 0.5 else ('Moderate confidence' if jp.confidence < 0.8 else 'High confidence') %}

Judge suggests {{ jp.label }} for {{ jp.schema }} {{ jqual }} · {{ jconf }}%

{% if jp.running and jp.running.n %} {{ (jp.running.agreement_rate * 100)|int }}% agree (n={{ jp.running.n }}){% if jp.running.kappa is not none %} · κ {{ '%.2f'|format(jp.running.kappa) }}{% endif %} {% endif %}

{% if jp.reasoning %}

Judge reasoning

{% endif %}

{% endif %} {# Training chrome: practice progress and answer feedback. Behind a default(false) flag so every other phase renders identically. #} {% if is_training_page | default(false) %}

{{ ui_lang.training_question|default('Question') }} {{ training_current_question }} / {{ training_total_questions }} {{ ui_lang.training_correct|default('Correct') }}: {{ training_correct_count }} · {{ ui_lang.training_mistakes|default('Mistakes') }}: {{ training_mistake_count }}

{% if training_show_feedback and training_feedback %}

{% endif %} {% endif %}

{{ ui_lang.error_heading }}

Review

Curate (admin)

Pending proposals