Services: Admin-Lehrer, Backend-Lehrer, Studio v2, Website, Klausur-Service, School-Service, Voice-Service, Geo-Service, BreakPilot Drive, Agent-Core Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
504 lines
19 KiB
TypeScript
504 lines
19 KiB
TypeScript
'use client'
|
|
|
|
/**
|
|
* LLM Comparison Tool
|
|
*
|
|
* Vergleicht Antworten von verschiedenen LLM-Providern:
|
|
* - OpenAI/ChatGPT
|
|
* - Claude
|
|
* - Self-hosted + Tavily
|
|
* - Self-hosted + EduSearch
|
|
*/
|
|
|
|
import { useState, useEffect, useCallback } from 'react'
|
|
import { PagePurpose } from '@/components/common/PagePurpose'
|
|
import { AIToolsSidebarResponsive } from '@/components/ai/AIToolsSidebar'
|
|
|
|
interface LLMResponse {
|
|
provider: string
|
|
model: string
|
|
response: string
|
|
latency_ms: number
|
|
tokens_used?: number
|
|
search_results?: Array<{
|
|
title: string
|
|
url: string
|
|
content: string
|
|
score?: number
|
|
}>
|
|
error?: string
|
|
timestamp: string
|
|
}
|
|
|
|
interface ComparisonResult {
|
|
comparison_id: string
|
|
prompt: string
|
|
system_prompt?: string
|
|
responses: LLMResponse[]
|
|
created_at: string
|
|
}
|
|
|
|
const providerColors: Record<string, { bg: string; border: string; text: string }> = {
|
|
openai: { bg: 'bg-emerald-50', border: 'border-emerald-300', text: 'text-emerald-700' },
|
|
claude: { bg: 'bg-orange-50', border: 'border-orange-300', text: 'text-orange-700' },
|
|
selfhosted_tavily: { bg: 'bg-blue-50', border: 'border-blue-300', text: 'text-blue-700' },
|
|
selfhosted_edusearch: { bg: 'bg-purple-50', border: 'border-purple-300', text: 'text-purple-700' },
|
|
}
|
|
|
|
const providerLabels: Record<string, string> = {
|
|
openai: 'OpenAI GPT-4o-mini',
|
|
claude: 'Claude 3.5 Sonnet',
|
|
selfhosted_tavily: 'Self-hosted + Tavily',
|
|
selfhosted_edusearch: 'Self-hosted + EduSearch',
|
|
}
|
|
|
|
export default function LLMComparePage() {
|
|
// State
|
|
const [prompt, setPrompt] = useState('')
|
|
const [systemPrompt, setSystemPrompt] = useState('Du bist ein hilfreicher Assistent fuer Lehrkraefte in Deutschland.')
|
|
|
|
// Provider toggles
|
|
const [enableOpenAI, setEnableOpenAI] = useState(true)
|
|
const [enableClaude, setEnableClaude] = useState(true)
|
|
const [enableTavily, setEnableTavily] = useState(true)
|
|
const [enableEduSearch, setEnableEduSearch] = useState(true)
|
|
|
|
// Parameters
|
|
const [model, setModel] = useState('llama3.2:3b')
|
|
const [temperature, setTemperature] = useState(0.7)
|
|
const [maxTokens, setMaxTokens] = useState(2048)
|
|
|
|
// Results
|
|
const [isLoading, setIsLoading] = useState(false)
|
|
const [result, setResult] = useState<ComparisonResult | null>(null)
|
|
const [history, setHistory] = useState<ComparisonResult[]>([])
|
|
const [error, setError] = useState<string | null>(null)
|
|
|
|
// UI State
|
|
const [showSettings, setShowSettings] = useState(false)
|
|
const [showHistory, setShowHistory] = useState(false)
|
|
|
|
// API Base URL
|
|
const API_URL = process.env.NEXT_PUBLIC_LLM_GATEWAY_URL || 'http://localhost:8082'
|
|
const API_KEY = process.env.NEXT_PUBLIC_LLM_API_KEY || 'dev-key'
|
|
|
|
// Load history
|
|
const loadHistory = useCallback(async () => {
|
|
try {
|
|
const response = await fetch(`${API_URL}/v1/comparison/history?limit=20`, {
|
|
headers: { Authorization: `Bearer ${API_KEY}` },
|
|
})
|
|
if (response.ok) {
|
|
const data = await response.json()
|
|
setHistory(data.comparisons || [])
|
|
}
|
|
} catch (e) {
|
|
console.error('Failed to load history:', e)
|
|
}
|
|
}, [API_URL, API_KEY])
|
|
|
|
useEffect(() => {
|
|
loadHistory()
|
|
}, [loadHistory])
|
|
|
|
const runComparison = async () => {
|
|
if (!prompt.trim()) {
|
|
setError('Bitte geben Sie einen Prompt ein')
|
|
return
|
|
}
|
|
|
|
setIsLoading(true)
|
|
setError(null)
|
|
setResult(null)
|
|
|
|
try {
|
|
const response = await fetch(`${API_URL}/v1/comparison/run`, {
|
|
method: 'POST',
|
|
headers: {
|
|
'Content-Type': 'application/json',
|
|
Authorization: `Bearer ${API_KEY}`,
|
|
},
|
|
body: JSON.stringify({
|
|
prompt,
|
|
system_prompt: systemPrompt || undefined,
|
|
enable_openai: enableOpenAI,
|
|
enable_claude: enableClaude,
|
|
enable_selfhosted_tavily: enableTavily,
|
|
enable_selfhosted_edusearch: enableEduSearch,
|
|
selfhosted_model: model,
|
|
temperature,
|
|
max_tokens: maxTokens,
|
|
}),
|
|
})
|
|
|
|
if (!response.ok) {
|
|
throw new Error(`API Error: ${response.status}`)
|
|
}
|
|
|
|
const data = await response.json()
|
|
setResult(data)
|
|
loadHistory()
|
|
} catch (e) {
|
|
setError(e instanceof Error ? e.message : 'Unbekannter Fehler')
|
|
} finally {
|
|
setIsLoading(false)
|
|
}
|
|
}
|
|
|
|
const ResponseCard = ({ response }: { response: LLMResponse }) => {
|
|
const colors = providerColors[response.provider] || {
|
|
bg: 'bg-slate-50',
|
|
border: 'border-slate-300',
|
|
text: 'text-slate-700',
|
|
}
|
|
const label = providerLabels[response.provider] || response.provider
|
|
|
|
return (
|
|
<div className={`rounded-xl border-2 ${colors.border} ${colors.bg} overflow-hidden`}>
|
|
<div className={`px-4 py-3 border-b ${colors.border} flex items-center justify-between`}>
|
|
<div>
|
|
<h3 className={`font-semibold ${colors.text}`}>{label}</h3>
|
|
<p className="text-xs text-slate-500">{response.model}</p>
|
|
</div>
|
|
<div className="text-right text-xs text-slate-500">
|
|
<div>{response.latency_ms}ms</div>
|
|
{response.tokens_used && <div>{response.tokens_used} tokens</div>}
|
|
</div>
|
|
</div>
|
|
|
|
<div className="p-4">
|
|
{response.error ? (
|
|
<div className="text-red-600 text-sm">
|
|
<strong>Fehler:</strong> {response.error}
|
|
</div>
|
|
) : (
|
|
<pre className="whitespace-pre-wrap text-sm text-slate-700 font-sans">
|
|
{response.response}
|
|
</pre>
|
|
)}
|
|
</div>
|
|
|
|
{response.search_results && response.search_results.length > 0 && (
|
|
<div className="px-4 pb-4">
|
|
<details className="text-xs">
|
|
<summary className="cursor-pointer text-slate-500 hover:text-slate-700">
|
|
{response.search_results.length} Suchergebnisse anzeigen
|
|
</summary>
|
|
<ul className="mt-2 space-y-2">
|
|
{response.search_results.map((sr, idx) => (
|
|
<li key={idx} className="bg-white rounded p-2 border border-slate-200">
|
|
<a
|
|
href={sr.url}
|
|
target="_blank"
|
|
rel="noopener noreferrer"
|
|
className="text-blue-600 hover:underline font-medium"
|
|
>
|
|
{sr.title || 'Untitled'}
|
|
</a>
|
|
<p className="text-slate-500 truncate">{sr.content}</p>
|
|
</li>
|
|
))}
|
|
</ul>
|
|
</details>
|
|
</div>
|
|
)}
|
|
</div>
|
|
)
|
|
}
|
|
|
|
return (
|
|
<div>
|
|
{/* Page Purpose */}
|
|
<PagePurpose
|
|
title="LLM Vergleich"
|
|
purpose="Vergleichen Sie Antworten verschiedener KI-Provider (OpenAI, Claude, Self-hosted) fuer Qualitaetssicherung. Optimieren Sie Parameter und System Prompts fuer beste Ergebnisse. Standalone-Werkzeug ohne direkten Datenfluss zur KI-Pipeline."
|
|
audience={['Entwickler', 'Data Scientists', 'QA']}
|
|
architecture={{
|
|
services: ['llm-gateway (Python)', 'Ollama', 'OpenAI API', 'Claude API'],
|
|
databases: ['PostgreSQL (History)', 'Qdrant (RAG)'],
|
|
}}
|
|
relatedPages={[
|
|
{ name: 'Test Quality (BQAS)', href: '/ai/test-quality', description: 'Golden Suite & Synthetic Tests' },
|
|
{ name: 'GPU Infrastruktur', href: '/ai/gpu', description: 'GPU-Ressourcen verwalten' },
|
|
{ name: 'Agent Management', href: '/ai/agents', description: 'Multi-Agent System' },
|
|
]}
|
|
collapsible={true}
|
|
defaultCollapsed={true}
|
|
/>
|
|
|
|
{/* KI-Werkzeuge Sidebar */}
|
|
<AIToolsSidebarResponsive currentTool="llm-compare" />
|
|
|
|
<div className="grid grid-cols-1 lg:grid-cols-3 gap-6">
|
|
{/* Left Column: Input & Settings */}
|
|
<div className="lg:col-span-1 space-y-4">
|
|
{/* Prompt Input */}
|
|
<div className="bg-white rounded-xl border border-slate-200 p-4">
|
|
<h2 className="font-semibold text-slate-900 mb-3">Prompt</h2>
|
|
|
|
{/* System Prompt */}
|
|
<div className="mb-3">
|
|
<label className="block text-sm text-slate-600 mb-1">System Prompt</label>
|
|
<textarea
|
|
value={systemPrompt}
|
|
onChange={(e) => setSystemPrompt(e.target.value)}
|
|
rows={3}
|
|
className="w-full px-3 py-2 border border-slate-300 rounded-lg text-sm resize-none"
|
|
placeholder="System Prompt (optional)"
|
|
/>
|
|
</div>
|
|
|
|
{/* User Prompt */}
|
|
<div className="mb-3">
|
|
<label className="block text-sm text-slate-600 mb-1">User Prompt</label>
|
|
<textarea
|
|
value={prompt}
|
|
onChange={(e) => setPrompt(e.target.value)}
|
|
rows={4}
|
|
className="w-full px-3 py-2 border border-slate-300 rounded-lg text-sm resize-none"
|
|
placeholder="z.B.: Erstelle ein Arbeitsblatt zum Thema Bruchrechnung fuer Klasse 6..."
|
|
/>
|
|
</div>
|
|
|
|
{/* Provider Toggles */}
|
|
<div className="mb-4">
|
|
<label className="block text-sm text-slate-600 mb-2">Provider</label>
|
|
<div className="grid grid-cols-2 gap-2">
|
|
<label className="flex items-center gap-2 text-sm">
|
|
<input
|
|
type="checkbox"
|
|
checked={enableOpenAI}
|
|
onChange={(e) => setEnableOpenAI(e.target.checked)}
|
|
className="rounded"
|
|
/>
|
|
OpenAI
|
|
</label>
|
|
<label className="flex items-center gap-2 text-sm">
|
|
<input
|
|
type="checkbox"
|
|
checked={enableClaude}
|
|
onChange={(e) => setEnableClaude(e.target.checked)}
|
|
className="rounded"
|
|
/>
|
|
Claude
|
|
</label>
|
|
<label className="flex items-center gap-2 text-sm">
|
|
<input
|
|
type="checkbox"
|
|
checked={enableTavily}
|
|
onChange={(e) => setEnableTavily(e.target.checked)}
|
|
className="rounded"
|
|
/>
|
|
Self + Tavily
|
|
</label>
|
|
<label className="flex items-center gap-2 text-sm">
|
|
<input
|
|
type="checkbox"
|
|
checked={enableEduSearch}
|
|
onChange={(e) => setEnableEduSearch(e.target.checked)}
|
|
className="rounded"
|
|
/>
|
|
Self + EduSearch
|
|
</label>
|
|
</div>
|
|
</div>
|
|
|
|
{/* Run Button */}
|
|
<button
|
|
onClick={runComparison}
|
|
disabled={isLoading || !prompt.trim()}
|
|
className="w-full py-3 bg-teal-600 text-white rounded-lg font-medium hover:bg-teal-700 disabled:opacity-50 disabled:cursor-not-allowed"
|
|
>
|
|
{isLoading ? (
|
|
<span className="flex items-center justify-center gap-2">
|
|
<svg className="animate-spin w-5 h-5" fill="none" viewBox="0 0 24 24">
|
|
<circle className="opacity-25" cx="12" cy="12" r="10" stroke="currentColor" strokeWidth="4" />
|
|
<path className="opacity-75" fill="currentColor" d="M4 12a8 8 0 018-8V0C5.373 0 0 5.373 0 12h4z" />
|
|
</svg>
|
|
Vergleiche...
|
|
</span>
|
|
) : (
|
|
'Vergleich starten'
|
|
)}
|
|
</button>
|
|
|
|
{error && (
|
|
<div className="mt-3 p-3 bg-red-50 border border-red-200 rounded-lg text-red-700 text-sm">
|
|
{error}
|
|
</div>
|
|
)}
|
|
</div>
|
|
|
|
{/* Settings Panel */}
|
|
<div className="bg-white rounded-xl border border-slate-200 overflow-hidden">
|
|
<button
|
|
onClick={() => setShowSettings(!showSettings)}
|
|
className="w-full px-4 py-3 flex items-center justify-between hover:bg-slate-50"
|
|
>
|
|
<span className="font-semibold text-slate-900">Parameter</span>
|
|
<svg
|
|
className={`w-5 h-5 transition-transform ${showSettings ? 'rotate-180' : ''}`}
|
|
fill="none"
|
|
stroke="currentColor"
|
|
viewBox="0 0 24 24"
|
|
>
|
|
<path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M19 9l-7 7-7-7" />
|
|
</svg>
|
|
</button>
|
|
|
|
{showSettings && (
|
|
<div className="p-4 border-t border-slate-200 space-y-4">
|
|
<div>
|
|
<label className="block text-sm text-slate-600 mb-1">Self-hosted Modell</label>
|
|
<select
|
|
value={model}
|
|
onChange={(e) => setModel(e.target.value)}
|
|
className="w-full px-3 py-2 border border-slate-300 rounded-lg text-sm"
|
|
>
|
|
<option value="llama3.2:3b">Llama 3.2 3B</option>
|
|
<option value="llama3.1:8b">Llama 3.1 8B</option>
|
|
<option value="mistral:7b">Mistral 7B</option>
|
|
<option value="qwen2.5:7b">Qwen 2.5 7B</option>
|
|
</select>
|
|
</div>
|
|
|
|
<div>
|
|
<label className="block text-sm text-slate-600 mb-1">
|
|
Temperature: {temperature.toFixed(2)}
|
|
</label>
|
|
<input
|
|
type="range"
|
|
min="0"
|
|
max="2"
|
|
step="0.1"
|
|
value={temperature}
|
|
onChange={(e) => setTemperature(parseFloat(e.target.value))}
|
|
className="w-full"
|
|
/>
|
|
</div>
|
|
|
|
<div>
|
|
<label className="block text-sm text-slate-600 mb-1">Max Tokens: {maxTokens}</label>
|
|
<input
|
|
type="range"
|
|
min="256"
|
|
max="4096"
|
|
step="256"
|
|
value={maxTokens}
|
|
onChange={(e) => setMaxTokens(parseInt(e.target.value))}
|
|
className="w-full"
|
|
/>
|
|
</div>
|
|
</div>
|
|
)}
|
|
</div>
|
|
|
|
{/* History Panel */}
|
|
<div className="bg-white rounded-xl border border-slate-200 overflow-hidden">
|
|
<button
|
|
onClick={() => setShowHistory(!showHistory)}
|
|
className="w-full px-4 py-3 flex items-center justify-between hover:bg-slate-50"
|
|
>
|
|
<span className="font-semibold text-slate-900">Verlauf ({history.length})</span>
|
|
<svg
|
|
className={`w-5 h-5 transition-transform ${showHistory ? 'rotate-180' : ''}`}
|
|
fill="none"
|
|
stroke="currentColor"
|
|
viewBox="0 0 24 24"
|
|
>
|
|
<path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M19 9l-7 7-7-7" />
|
|
</svg>
|
|
</button>
|
|
|
|
{showHistory && history.length > 0 && (
|
|
<div className="border-t border-slate-200 max-h-64 overflow-y-auto">
|
|
{history.map((h) => (
|
|
<button
|
|
key={h.comparison_id}
|
|
onClick={() => {
|
|
setResult(h)
|
|
setPrompt(h.prompt)
|
|
if (h.system_prompt) setSystemPrompt(h.system_prompt)
|
|
}}
|
|
className="w-full px-4 py-2 text-left hover:bg-slate-50 border-b border-slate-100 last:border-0"
|
|
>
|
|
<div className="text-sm text-slate-700 truncate">{h.prompt}</div>
|
|
<div className="text-xs text-slate-400">
|
|
{new Date(h.created_at).toLocaleString('de-DE')}
|
|
</div>
|
|
</button>
|
|
))}
|
|
</div>
|
|
)}
|
|
</div>
|
|
</div>
|
|
|
|
{/* Right Column: Results */}
|
|
<div className="lg:col-span-2">
|
|
{result ? (
|
|
<div className="space-y-4">
|
|
<div className="bg-white rounded-xl border border-slate-200 p-4">
|
|
<div className="flex items-center justify-between">
|
|
<div>
|
|
<h2 className="font-semibold text-slate-900">Ergebnisse</h2>
|
|
<p className="text-sm text-slate-500">ID: {result.comparison_id}</p>
|
|
</div>
|
|
<div className="text-sm text-slate-500">
|
|
{new Date(result.created_at).toLocaleString('de-DE')}
|
|
</div>
|
|
</div>
|
|
<div className="mt-2 p-3 bg-slate-50 rounded-lg">
|
|
<p className="text-sm text-slate-700">{result.prompt}</p>
|
|
</div>
|
|
</div>
|
|
|
|
<div className="grid grid-cols-1 xl:grid-cols-2 gap-4">
|
|
{result.responses.map((response, idx) => (
|
|
<ResponseCard key={`${response.provider}-${idx}`} response={response} />
|
|
))}
|
|
</div>
|
|
</div>
|
|
) : (
|
|
<div className="bg-white rounded-xl border border-slate-200 p-12 text-center">
|
|
<svg
|
|
className="w-16 h-16 mx-auto text-slate-300 mb-4"
|
|
fill="none"
|
|
stroke="currentColor"
|
|
viewBox="0 0 24 24"
|
|
>
|
|
<path
|
|
strokeLinecap="round"
|
|
strokeLinejoin="round"
|
|
strokeWidth={1.5}
|
|
d="M9 3v2m6-2v2M9 19v2m6-2v2M5 9H3m2 6H3m18-6h-2m2 6h-2M7 19h10a2 2 0 002-2V7a2 2 0 00-2-2H7a2 2 0 00-2 2v10a2 2 0 002 2zM9 9h6v6H9V9z"
|
|
/>
|
|
</svg>
|
|
<h3 className="text-lg font-medium text-slate-700 mb-2">LLM-Vergleich starten</h3>
|
|
<p className="text-slate-500 max-w-md mx-auto">
|
|
Geben Sie einen Prompt ein und klicken Sie auf "Vergleich starten", um
|
|
die Antworten verschiedener LLM-Provider zu vergleichen.
|
|
</p>
|
|
</div>
|
|
)}
|
|
</div>
|
|
</div>
|
|
|
|
{/* Info Box */}
|
|
<div className="mt-8 bg-teal-50 border border-teal-200 rounded-xl p-6">
|
|
<div className="flex items-start gap-4">
|
|
<svg className="w-6 h-6 text-teal-600 flex-shrink-0 mt-0.5" fill="none" stroke="currentColor" viewBox="0 0 24 24">
|
|
<path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M13 16h-1v-4h-1m1-4h.01M21 12a9 9 0 11-18 0 9 9 0 0118 0z" />
|
|
</svg>
|
|
<div>
|
|
<h3 className="font-semibold text-teal-900">Qualitaetssicherung</h3>
|
|
<p className="text-sm text-teal-800 mt-1">
|
|
Dieses Tool dient zur Qualitaetssicherung der KI-Antworten. Vergleichen Sie verschiedene Provider,
|
|
um die optimalen Parameter und System Prompts zu finden. Die Ergebnisse werden fuer Audits gespeichert.
|
|
</p>
|
|
</div>
|
|
</div>
|
|
</div>
|
|
</div>
|
|
)
|
|
}
|