ResearchIT · HF comparison

{{ 'Synthetic demonstration — no real relevance result.' if report.get('synthetic') else 'Unjudged experiment — not evidence of improved relevance.' }}
{{ report.get('method','') }}
Up to {{ report.source_cap }} HF replacements, within the same assigned interest. Threshold and source cap are experimental.

{% for key,label in [('baseline','Current baseline'),('hf_only','HF candidates only'),('combined','Combined experiment')] %}

{% for p in report[key] %}

{{ p.origin }} · Interest {{ p.interest+1 }}

{{ loop.index }}. {{ p.title }}

Topic similarity {{ '%.3f'|format(p.similarity) }} · Score {{ '%.3f'|format(p.score) }}

{% if report.get("synthetic") %}Simulated ID: {{ p.arxiv_id }}{% else %}{{ p.arxiv_id }}{% endif %}

{% else %}

No eligible candidates. This is not a successful relevance result.

{% endfor %}

Exactly what changed

{% for c in report.changes %}

{{ c.removed }} → {{ c.added }} · Interest {{ c.interest+1 }}
{{ c.reason }}

{% else %}

No replacements. Source failure, stale data, missing embeddings or poor topic fit can all lead to the baseline remaining unchanged.

{% endfor %}

Rejected candidates and source health

{{ report.rejected|tojson(indent=2) }}

{{ report.get('store',{})|tojson(indent=2) }}

{{ report.get('profile','Reader') }}: what would change?

{{ label }}

{{ loop.index }}. {{ p.title }}