mirror of
https://github.com/onestardao/WFGY.git
synced 2026-04-28 11:40:07 +00:00
9.9 KiB
9.9 KiB
Language & Multilingual — Global Fix Map
A compact hub to stabilize cross-lingual retrieval and reasoning. Use this folder when your corpus or queries span CJK, RTL, Indic, Cyrillic, accented Latin, or when users code-switch.
Quick routes to per-page guides
- Tokenizer mismatch → tokenizer_mismatch.md
- Script mixing in one query → script_mixing.md
- Locale drift and normalization → locale_drift.md
- End-to-end overview and recipes → multilingual_guide.md
When to use this folder
- High similarity yet wrong meaning on bilingual or mixed-script corpora
- Citations point to the wrong section after translating the question
- Hybrid retrievers underperform a single retriever across languages
- Index “looks” healthy while coverage remains low for non-Latin scripts
- Names flip between native, transliteration, and English aliases
- zh-Hans and zh-Hant never co-retrieve, Thai recall collapses without clear reason
Acceptance targets
- ΔS(question, retrieved) ≤ 0.45 across language variants
- Coverage of the target section ≥ 0.70 after repair
- λ remains convergent across three paraphrases and two seeds
- E_resonance stays flat on long windows that mix scripts
Open these first
- Visual map and recovery → rag-architecture-and-recovery.md
- End-to-end retrieval knobs → retrieval-playbook.md
- Why this snippet, how to cite → retrieval-traceability.md
- Snippet schema fence → data-contracts.md
- Embedding vs meaning → embedding-vs-semantic.md
- Chunk boundary sanity → chunking-checklist.md
Map symptoms → structural fixes (Problem Map)
| Symptom | Likely cause | Open this |
|---|---|---|
| Wrong-meaning hits despite high similarity | Embedding not multilingual or pre-norm mismatch | embedding-vs-semantic.md |
| Citations jump sections after translation | Snippet schema too loose | data-contracts.md, retrieval-traceability.md |
| zh-Hans/zh-Hant never co-retrieve | Variant mapping missing | locale_drift.md |
| Thai/CJK recall collapses | Tokenizer mismatch, missing segmenter | tokenizer_mismatch.md |
| Mixed Latin + CJK query under-recalls | Analyzer split across scripts | script_mixing.md |
| Hybrid retriever worse than single | Query parsing split, mis-weighted rerank | patterns/pattern_query_parsing_split.md, rerankers.md |
Fix in 60 seconds
-
Measure ΔS across variants Run question in source language, target language, and a code-switched version. If ΔS improves only after transliteration or segmentation, language handling is the failing layer.
-
Probe λ_observe Hold citations fixed and flip only script or segmentation. If λ flips, lock schema and unify analyzers.
-
Apply the smallest change
- Unify index and search analyzers, add Thai/CJK segmenters
- Map zh-Hans ↔ zh-Hant at ingest or store in variant subfields
- Add alias fields for proper nouns:
native,romanized,normalized - Normalize digits, punctuation, and diacritics for RTL/Indic consistently
- Use a multilingual embedding; otherwise add a lexical sidecar and deterministic rerank
- Verify Coverage ≥ 0.70 and ΔS ≤ 0.45 across two languages and one mixed query. λ convergent on two seeds.
Store-agnostic quick recipes
- Elasticsearch/OpenSearch: ICU normalization, width fold, bigram analyzers for CJK, Thai segmenter, keyword subfields for product names.
- Vector stores: pre-normalize identically for corpus and queries, validate with a bilingual gold set, rerank deterministically if embeddings are monolingual.
🔗 Quick-Start Downloads (60 sec)
| Tool | Link | 3-Step Setup |
|---|---|---|
| WFGY 1.0 PDF | Engine Paper | 1️⃣ Download · 2️⃣ Upload to your LLM · 3️⃣ Ask “Answer using WFGY + <your question>” |
| TXT OS (plain-text OS) | TXTOS.txt | 1️⃣ Download · 2️⃣ Paste into any LLM chat · 3️⃣ Type “hello world” — OS boots instantly |
🧭 Explore More
| Module | Description | Link |
|---|---|---|
| WFGY Core | WFGY 2.0 engine is live: full symbolic reasoning architecture and math stack | View → |
| Problem Map 1.0 | Initial 16-mode diagnostic and symbolic fix framework | View → |
| Problem Map 2.0 | RAG-focused failure tree, modular fixes, and pipelines | View → |
| Semantic Clinic Index | Expanded failure catalog: prompt injection, memory bugs, logic drift | View → |
| Semantic Blueprint | Layer-based symbolic reasoning & semantic modulations | View → |
| Benchmark vs GPT-5 | Stress test GPT-5 with full WFGY reasoning suite | View → |
| 🧙♂️ Starter Village 🏡 | New here? Lost in symbols? Click here and let the wizard guide you through | Start → |
👑 Early Stargazers: See the Hall of Fame —
Engineers, hackers, and open source builders who supported WFGY from day one.
⭐ WFGY Engine 2.0 is already unlocked. ⭐ Star the repo to help others discover it and unlock more on the Unlock Board.
