diff --git a/aperag/migration/sql/model_configs_init.sql b/aperag/migration/sql/model_configs_init.sql index 2ba0461a..21140a43 100644 --- a/aperag/migration/sql/model_configs_init.sql +++ b/aperag/migration/sql/model_configs_init.sql @@ -1,5 +1,5 @@ -- Model configuration initialization SQL script --- Generated directly from configuration data on 2026-03-03 14:09:57 +-- Generated directly from configuration data on 2026-03-03 14:16:57 -- This script populates llm_provider and llm_provider_models tables BEGIN; @@ -10236,6 +10236,6 @@ ON CONFLICT (provider_name, api, model) DO UPDATE SET COMMIT; --- Script completed. Generated on 2026-03-03 14:09:57 +-- Script completed. Generated on 2026-03-03 14:16:57 -- Total providers: 9 -- Total models: 670 \ No newline at end of file diff --git a/models/generate_model_configs.py b/models/generate_model_configs.py index 8e37c661..f3a192eb 100644 --- a/models/generate_model_configs.py +++ b/models/generate_model_configs.py @@ -27,6 +27,7 @@ # limitations under the License. import json +import logging import os from datetime import datetime from typing import Dict, List, Any, Optional @@ -35,6 +36,10 @@ import requests from litellm import model_cost +# Suppress litellm's verbose warnings and debug output +litellm.suppress_debug_info = True +logging.getLogger("LiteLLM").setLevel(logging.ERROR) + # --- Manual Override for Context Windows --- # For models where litellm's data might be ambiguous or incorrect, # we define the correct context window size here. @@ -289,10 +294,19 @@ def generate_model_specs(models, provider, mode, blocklist=None, tag_rules=None) def create_openai_config(): provider = "openai" - # Define blocklists - completion_blocklist = [] - embedding_blocklist = [] - rerank_blocklist = [] + # Define blocklists - deprecated/instruct models not in litellm's model_prices_and_context_window.json + completion_blocklist = [ + "babbage-002", "davinci-002", "ft:babbage-002", "ft:davinci-002", + "gpt-3.5-turbo-instruct", "gpt-3.5-turbo-instruct-0914", + ] + embedding_blocklist = [ + "babbage-002", "davinci-002", "ft:babbage-002", "ft:davinci-002", + "gpt-3.5-turbo-instruct", "gpt-3.5-turbo-instruct-0914", + ] + rerank_blocklist = [ + "babbage-002", "davinci-002", "ft:babbage-002", "ft:davinci-002", + "gpt-3.5-turbo-instruct", "gpt-3.5-turbo-instruct-0914", + ] # Define tag rules completion_tag_rules = { @@ -903,7 +917,13 @@ def create_openrouter_config(): if data is None: try: if os.path.exists(openrouter_file): + file_mtime = os.path.getmtime(openrouter_file) + file_age_days = (datetime.now().timestamp() - file_mtime) / 86400 + file_date = datetime.fromtimestamp(file_mtime).strftime('%Y-%m-%d') print(f"📁 Reading OpenRouter models from local file: {openrouter_file}") + print(f" (cached on {file_date}, {file_age_days:.0f} days ago)") + if file_age_days > 30: + print(f"⚠️ Warning: local cache is {file_age_days:.0f} days old — consider refreshing via VPN or proxy") with open(openrouter_file, 'r', encoding='utf-8') as f: data = json.load(f) print("✅ Successfully loaded OpenRouter models from local file") @@ -991,13 +1011,6 @@ def create_provider_config(): No need to pass parameters - block lists and tag rules are defined in each provider function. """ - print("\n📋 Block List Usage:") - print("- Block lists and tag rules are now defined internally in each provider function") - print("- To modify block lists or tag rules, edit the provider functions directly") - print("- Each provider supports completion_blocklist, embedding_blocklist, and rerank_blocklist") - print("- Tag rules map tag names to model name patterns") - print() - # Generate provider configurations result = [ create_openai_config(), @@ -1164,33 +1177,28 @@ def main(): """Main function to generate model configuration and SQL script""" try: print("Generating model configuration data...") - print("\n📋 Block List Usage:") - print("- Block lists and tag rules are now defined internally in each provider function") - print("- To modify block lists or tag rules, edit the provider functions directly") - print("- Each provider supports completion_blocklist, embedding_blocklist, and rerank_blocklist") - print("- Tag rules map tag names to model name patterns") + print(" (Block lists and tag rules are defined internally in each provider function)") print() providers_data = create_provider_config() - print("Generating SQL script...") + print("\n📊 Summary:") + total_models = 0 + for p in providers_data: + completion = len(p.get('completion', [])) + embedding = len(p.get('embedding', [])) + rerank = len(p.get('rerank', [])) + subtotal = completion + embedding + rerank + total_models += subtotal + print(f" {p['label']:<20} completion={completion:>4} embedding={embedding:>3} rerank={rerank:>3}") + print(f" {'TOTAL':<20} {total_models} models across {len(providers_data)} providers") + + print("\nGenerating SQL script...") sql_script = generate_sql_script(providers_data) save_sql_to_file(sql_script) print("✅ Model configuration SQL script generated successfully!") - print("\nTo execute the script:") - print(" psql -h -U -d -f aperag/sql/model_configs_init.sql") - print("\nOr copy the contents and run in your PostgreSQL client.") - - print("\n🔧 Usage Examples:") - print("1. To customize block lists or tag rules:") - print(" Edit the block list and tag rule variables in each provider function") - print("2. To add/remove models from block lists:") - print(" Modify the completion_blocklist, embedding_blocklist, or rerank_blocklist in provider functions") - print("3. To add/modify model tags:") - print(" Update the tag_rules dictionaries in provider functions") - print(" Example: completion_tag_rules = {'enable_for_collection': ['model1', 'model2'], 'free': ['model3']}") except Exception as e: print(f"❌ Error generating SQL script: {e}") diff --git a/web/src/app/workspace/menu-footer.tsx b/web/src/app/workspace/menu-footer.tsx index b7b8a22b..2539a6af 100644 --- a/web/src/app/workspace/menu-footer.tsx +++ b/web/src/app/workspace/menu-footer.tsx @@ -23,6 +23,7 @@ import { ChevronRight, Key, Logs, + MessageSquareText, Package, Settings, } from 'lucide-react'; @@ -79,6 +80,12 @@ export const MenuFooter = () => { {sidebar_workspace('quotas')} + + + + {sidebar_workspace('prompts')} + + diff --git a/web/src/app/workspace/prompts/page.tsx b/web/src/app/workspace/prompts/page.tsx new file mode 100644 index 00000000..bae90b64 --- /dev/null +++ b/web/src/app/workspace/prompts/page.tsx @@ -0,0 +1,29 @@ +import { + PageContainer, + PageContent, + PageDescription, + PageHeader, + PageTitle, +} from '@/components/page-container'; +import { getServerApi } from '@/lib/api/server'; +import { toJson } from '@/lib/utils'; +import { getTranslations } from 'next-intl/server'; +import { PromptSettings } from './prompt-settings'; + +export default async function Page() { + const serverApi = await getServerApi(); + const res = await serverApi.defaultApi.promptsUserGet(); + const data = res.data; + const page_prompts = await getTranslations('page_prompts'); + + return ( + + + + {page_prompts('metadata.title')} + {page_prompts('metadata.description')} + + + + ); +} diff --git a/web/src/app/workspace/prompts/prompt-settings.tsx b/web/src/app/workspace/prompts/prompt-settings.tsx new file mode 100644 index 00000000..1d84af1e --- /dev/null +++ b/web/src/app/workspace/prompts/prompt-settings.tsx @@ -0,0 +1,156 @@ +'use client'; + +import { PromptDetail, UserPromptsResponse } from '@/api'; +import { Badge } from '@/components/ui/badge'; +import { Button } from '@/components/ui/button'; +import { + Card, + CardContent, + CardDescription, + CardFooter, + CardHeader, + CardTitle, +} from '@/components/ui/card'; +import { + Dialog, + DialogClose, + DialogContent, + DialogDescription, + DialogFooter, + DialogHeader, + DialogTitle, + DialogTrigger, +} from '@/components/ui/dialog'; +import { Textarea } from '@/components/ui/textarea'; +import { apiClient } from '@/lib/api/client'; +import { useTranslations } from 'next-intl'; +import { useRouter } from 'next/navigation'; +import { useCallback, useEffect, useState } from 'react'; +import { toast } from 'sonner'; + +type PromptType = 'agent_system' | 'agent_query'; + +const PROMPT_TYPES: PromptType[] = ['agent_system', 'agent_query']; + +interface PromptCardProps { + promptType: PromptType; + detail: PromptDetail | undefined; + onSaved: () => void; +} + +const PromptCard = ({ promptType, detail, onSaved }: PromptCardProps) => { + const page_prompts = useTranslations('page_prompts'); + const common_action = useTranslations('common.action'); + const [content, setContent] = useState(detail?.content ?? ''); + const [saving, setSaving] = useState(false); + const [resetOpen, setResetOpen] = useState(false); + + useEffect(() => { + setContent(detail?.content ?? ''); + }, [detail?.content]); + + const handleSave = useCallback(async () => { + setSaving(true); + try { + await apiClient.defaultApi.promptsUserPut({ + updateUserPromptsRequest: { + prompts: { [promptType]: content }, + }, + }); + toast.success(page_prompts('toast.save_success')); + onSaved(); + } finally { + setSaving(false); + } + }, [content, promptType, page_prompts, onSaved]); + + const handleReset = useCallback(async () => { + await apiClient.defaultApi.promptsUserPromptTypeDelete({ + promptType: promptType as never, + }); + toast.success(page_prompts('toast.reset_success')); + setResetOpen(false); + onSaved(); + }, [promptType, page_prompts, onSaved]); + + const isCustomized = detail?.customized === true; + + return ( + + +
+ {page_prompts(`${promptType}.title` as never)} + + {isCustomized + ? page_prompts('status.customized') + : page_prompts('status.default')} + +
+ + {page_prompts(`${promptType}.description` as never)} + + + + +