daveadmin/dobetternorge-tools
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Simplify advocate engine options: remove GPU/dbn_legal_v3, fix time estimates

Browse Source 
GPU and DBN Legal Agent v3 are unsuitable for advocate synthesis (4-6K token
structured JSON output at 20-30 tok/s = 3-5 min on RTX 3060, plus both fall
through to Sonnet when Bedrock is enabled anyway). Reduce to two honest options:
Haiku (~2-4 min) and Sonnet (~3-5 min), with accurate description of why
time is dominated by multi-pass question answering, not synthesis.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
davegilligan
2026-05-25 21:27:34 +02:00
parent a925415ef7
commit c2735fa919
1 changed files with 3 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
advocate.php
+3 -6
View File
@@ -52,13 +52,10 @@ require_once __DIR__ . '/includes/layout.php';
<div class="control-row" id="advEngineControl"> <div class="control-row" id="advEngineControl">
<span class="control-label">Engine</span> <span class="control-label">Engine</span>
<label><input type="radio" name="advEngine" value="azure_mini" checked> &#x2601;&#xFE0F; Claude Haiku 4.5 (fast) <small class="control-hint">(~20-50s)</small></label> <label><input type="radio" name="advEngine" value="azure_mini" checked> &#x2601;&#xFE0F; Claude Haiku 4.5 <small class="control-hint">(fast · ~2-4 min)</small></label>
<label><input type="radio" name="advEngine" value="azure_full"> Azure gpt-4o <small class="control-hint">(best · ~60-180s)</small></label> <label><input type="radio" name="advEngine" value="claude_sonnet"> &#x2601;&#xFE0F; Claude Sonnet 4.6 &#9733;&#9733; <small class="control-hint">(thorough · ~3-5 min)</small></label>
<label><input type="radio" name="advEngine" value="gpu"> GPU (cuttlefish) <small class="control-hint">(local · ~30-90s)</small></label>
<label><input type="radio" name="advEngine" value="dbn_legal_v3"> &#x1F1F3;&#x1F1F4;&#9876;&#65039; DBN Legal Agent &#9733; <small class="control-hint">(dbn-legal-agent-v3 fine-tune · ~20-60s)</small></label>
<label><input type="radio" name="advEngine" value="claude_sonnet"> &#x2601;&#xFE0F; Claude Sonnet 4.6 (thorough) &#9733;&#9733; <small class="control-hint">(best · ~60-120s)</small></label>
</div> </div>
<p class="upload-hint">Claude Haiku 4.5 is fast and accurate for most cases. Claude Sonnet 4.6 produces the most thorough advocate brief — superior at multi-party legal reasoning, ECHR precedent weighting, and long-form argumentation. Norwegian specialist v3 is a Qwen2.5 fine-tune trained on barnevernsloven, ECHR, and forvaltningsloven — highest precision for § 4-25, Strand Lobben, and procedural red flags.</p> <p class="upload-hint">Both engines run on AWS Bedrock via Claude. Most of the time is spent on multiple question-answering passes — 6–10 sub-questions each requiring a full retrieval and answer cycle. Haiku is faster and handles most cases well. Sonnet produces a more thorough brief with deeper ECHR precedent analysis and stronger multi-party argumentation.</p>
<div class="dr-slice-section"> <div class="dr-slice-section">
<p class="control-label">Corpus slices</p> <p class="control-label">Corpus slices</p>