Genderize v2: fused two-network model (v2.2), script-aware routing, country prior — honest model card; v1 preserved in tag v1

COMMERCIAL_USE.md

@@ -1,12 +1,13 @@
 # Licensing of the genderize weights
 
-**Weights** (`genderize_cora.pt`, `genderize_ultra.pt`, their `.config.json`, `.calibration.json` and `maps.json`)
+**Weights** (the `genderize_fuso_v2/` directory: the two internal networks `v1/` and `v2/`, `fuso.json`,
+`country_prior.json`, `dom113_rimedio.json` and `SHA256SUMS`)
 are released under **Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)** — see `LICENSE`.
 
 In plain words:
 
 - You may **download, run, study, modify and redistribute** the weights and derivatives **for non-commercial
-  purposes**, provided you give attribution (*dbtool.it, genderize cora/ultra, 2026*) and indicate changes.
+  purposes**, provided you give attribution (*dbtool.it, genderize v2, 2026*) and indicate changes.
 - Research, teaching, personal projects, non-profit and public-sector use are non-commercial uses.
 - **Commercial use is not granted by this licence.** Commercial use means, without limitation: offering the
   model or a derivative as a paid or ad-supported service or API, embedding it in a product that is sold,
@@ -15,7 +16,7 @@ In plain words:
   on-premise licence, both offered by the rights holder. A separate written commercial licence can be
   requested from the rights holder, Domenico Gigante (d.gigante@tech-time.it).
 
-**Inference code** (`genderize_infer.py`) is released under the **MIT License** (see `LICENSE-CODE`) so that it
+**Inference code** (`inference.py`, `inference_fuso.py`) is released under the **MIT License** (see `LICENSE-CODE`) so that it
 can be freely reused; it carries no model knowledge.
 
 No training data, training scripts or dictionary tables are part of this release.

README.md

@@ -8,209 +8,234 @@ tags:
 - gender-detection
 - nationality
 - name-analysis
-- byte-level
+- character-level
 - cpu
 pipeline_tag: text-classification
 ---
 
-# genderize — cora / ultra
+# Genderize v2
 
-Two models that take a personal name and return a **gender** (M/F) and a
-**country** (226 ISO-3166 alpha-2 codes), with probabilities. Byte-level,
-CPU-only, no tokenizer and no vocabulary file: you feed them a string.
+Genderize v2 estimates a **binary gender** (M/F) and a **likely country** (226
+ISO-3166 alpha-2 codes) from a personal name, with probabilities. It is a
+character-level classifier designed for **CPU inference** — no tokenizer, no
+vocabulary file. It supersedes `genderize v1 (cora/ultra)`, which remains
+available in this repository under the git tag **`v1`**.
 
-## Package contents
-
-| File | What it is |
-|---|---|
-| `genderize_cora.pt` | weights, cora (3.1 MB) — **CC BY-NC 4.0** |
-| `genderize_ultra.pt` | weights, ultra (12.9 MB) — **CC BY-NC 4.0** |
-| `genderize_cora.config.json` | `{"ch": 160}` — CC BY-NC 4.0 |
-| `genderize_ultra.config.json` | `{"ch": 384}` — CC BY-NC 4.0 |
-| `genderize_cora.calibration.json` | per-head temperature — CC BY-NC 4.0 |
-| `genderize_ultra.calibration.json` | per-head temperature (the file additionally records a calibration-error metric) — CC BY-NC 4.0 |
-| `maps.json` | class ids: gender `M/F`, 226 countries — CC BY-NC 4.0 |
-| `genderize_infer.py` | standalone inference code (needs `torch` and `numpy`) — **MIT** |
-| `LICENSE` | full text of CC BY-NC 4.0 (weights) |
-| `LICENSE-CODE` | full text of the MIT License (inference code) |
-| `COMMERCIAL_USE.md` | what counts as non-commercial use, and how to obtain a commercial licence |
-
-## What the models do
-
-| Variant | Gender | Country | Size |
-|---|---|---|---|
-| `genderize-cora` | 2 classes (M/F) | 226 classes | ~0.77M parameters, ch=160 |
-| `genderize-ultra` | 2 classes (M/F) | 226 classes | ~3.21M parameters, ch=384 |
-
-`ultra` is the larger, more accurate variant; `cora` is the lighter one. Both run
-on CPU: on an Intel i3-6100T (2 cores, 2 threads used) the network alone processes
-about **850 names/s** (`cora`) and **160 names/s** (`ultra`) in batches of 100.
+> **Read the limitations before using this model.** It is measurably **weaker
+> than v1 in Togo, Chad and Samoa**, and it does not improve Arabic and Hebrew.
+> It is intended for **aggregate statistical analyses**, not for decisions
+> about individual people.
 
-## How it works
+## Package contents
 
-**1. Normalisation.** The input name is normalised before anything else:
-Unicode **NFC**, **lowercase**, **whitespace collapsed** (internal runs and
-leading/trailing spaces become a single space). The normalised form is what the
-weights saw, so this step must not be changed or skipped:
-
-    "  MARÍA   GARCÍA  "  ->  "maría garcía"
-
-**2. Encoding.** The normalised text is encoded as **UTF-8 and truncated to 48
-bytes** (not 48 characters: an accented or non-Latin character takes more than
-one byte, so the effective character budget is smaller). Each byte becomes an
-integer in 0–255; byte value 0 is the padding value. A name shorter than 48 bytes
-is padded with zeros.
-
-**3. Network.** Byte-level, dual-head 1-D convolutional classifier:
-
-- one embedding table, 256 entries (one per byte value) of width 64, padding
-  index 0;
-- a 1×1 convolution projecting 64 → `ch` channels;
-- four residual 1-D convolution blocks, kernel sizes **3, 5, 7, 3**, each with
-  convolution + BatchNorm + GELU around a residual connection, padding to keep
-  the length;
-- the masked positions are pooled twice — mean-pooling and max-pooling — and the
-  two vectors are concatenated;
-- a shared layer of 512 units (GELU, plus a dropout layer that is inactive at
-  inference);
-- two linear heads read out from it: **gender**, 2 classes, and **country**,
-  226 classes.
-
-`cora` uses `ch = 160`, `ultra` `ch = 384`; those are the only differences
-between the two configs.
-
-**4. Probabilities.** Because it is a *convolutional* stack, the model sees all
-byte positions at once: character order matters through the convolution kernels,
-not through a recurrent state. Logits are divided by a **per-head temperature**
-before softmax — `temp_gender` and `temp_country` from the calibration file:
-
-| Variant | temp_gender | temp_country |
+| File | What it is | Licence |
 |---|---|---|
-| cora | 1.0096479654312134 | 1.0 |
-| ultra | 1.0460342168807983 | 0.939997673034668 |
-
-**5. Decision rule.** Gender is reported as `male` when P(M) ≥ 0.5, otherwise
-`female`; the reported probability is the probability of the reported class.
-Countries are returned as the **top-k** codes (default 5, `--top`) sorted by
-probability.
-
-**6. Output.** One record per input name:
-
-    {"name": "...", "gender": "male|female", "probability": 0.0-1.0,
-     "countries": [{"code": "IT", "probability": 0.0-1.0}, ...], "model": "cora|ultra"}
-
-## Loading and using the weights
-
-All files sit flat in this repository. Download them (for example with
-`huggingface_hub.snapshot_download("textpie/genderize")`) and run the reference
-driver from that directory:
-
-    pip install torch numpy
-    python genderize_infer.py --variant ultra --top 3 "Andrea Rossi"
-
-In Python:
+| `genderize_fuso_v2/fuso.json` | fusion configuration (thresholds, weights) | CC BY-NC 4.0 |
+| `genderize_fuso_v2/v1/` | internal network A: the v1 `ultra` weights, configs, calibration, class map | CC BY-NC 4.0 |
+| `genderize_fuso_v2/v2/` | internal network B (codepoint-level, distilled lineage), weights, configs, calibration, class map | CC BY-NC 4.0 |
+| `genderize_fuso_v2/country_prior.json` | empirical country prior used by the country head | CC BY-NC 4.0 |
+| `genderize_fuso_v2/dom113_rimedio.json` | configuration of the Latin-script country-trait rule (see below) | CC BY-NC 4.0 |
+| `genderize_fuso_v2/SHA256SUMS` | SHA-256 of every file in the model directory | CC BY-NC 4.0 |
+| `inference.py` | inference module for network A (byte-level model) | MIT |
+| `inference_fuso.py` | the Genderize v2 class `GenderizeFuso` (imports `inference.py`) | MIT |
+| `examples.txt` | example outputs reproduced with exactly these files | — |
+| `LICENSE` | CC BY-NC 4.0 (weights and model artifacts) | — |
+| `LICENSE-CODE` | MIT (inference code) | — |
+| `COMMERCIAL_USE.md` | what counts as non-commercial use, and how to obtain a commercial licence | — |
+
+Training data is **not** distributed with this package (see *Data provenance*).
+
+## Quickstart
+
+Requires Python 3.10+, `torch`, and `anyascii` (ISC licence; only invoked for
+non-Latin scripts that benefit from transliteration).
 
 ```python
-from genderize_infer import Genderize
+import sys
+from huggingface_hub import snapshot_download
 
-model = Genderize("ultra")              # or Genderize("cora"); pass models_dir=... if the files live elsewhere
-model.predict(["Andrea Rossi"], top=3)
-```
+path = snapshot_download("textpie/genderize")
+sys.path.insert(0, path)                     # inference_fuso.py + inference.py live at repo root
 
-Output for the five names published with the examples (`--variant ultra`),
-reproduced locally against these exact weight files:
+from inference_fuso import GenderizeFuso     # the class documented in the file header
 
-| Input | Gender | P | Country top-3 |
-|---|---|---|---|
-| Andrea Rossi | male | 0.849 | IT 0.84, FR 0.05, US 0.02 |
-| Yuki Tanaka | female | 0.589 | JP 0.98, US 0.01, ID 0.00 |
-| María García | female | 0.995 | ES 0.42, MX 0.13, AR 0.09 |
-| Chen Wei | male | 0.692 | CN 0.56, TW 0.13, SG 0.09 |
-| Fatima Al Sayed | female | 0.996 | AE 0.35, OM 0.19, SA 0.14 |
-
-## Benchmarks
+model = GenderizeFuso(f"{path}/genderize_fuso_v2")
+print(model.predict(["Andrea Rossi", "Yuki Tanaka"]))
+# -> [{'gender': 'M', 'p_male': 0.7937, 'country': 'IT', 'p_country': 0.8262, 'top5': ['IT', 'FR', 'US', 'CH', 'DE']},
+#     {'gender': 'F', 'p_male': 0.5037, 'country': 'JP', 'p_country': 0.9809, 'top5': ['JP', 'US', 'ID', 'DE', 'CN']}]
+```
 
-Source: **dbtool.it/benchmark**. These are **not** new measurements taken for
-this card and no re-measurement was performed here.
+`predict()` returns one dict per name: `gender` (`'M'` if p(M) ≥ 0.60, else
+`'F'`), `p_male`, `country` (argmax of the adjusted country distribution),
+`p_country`, and `top5` (top-5 country codes).
 
-- **Bench**: 25,000 names never seen in training, scored on the **network alone**.
-- The dictionary layer used by the hosted API does not contribute on unseen
-  names, so these figures describe the released weights.
+## How it works
 
-| Metric | cora | ultra |
-|---|---|---|
-| Gender accuracy | 97.9 % | 98.2 % |
-| Country top-1 accuracy | 82.6 % | 83.7 % |
-
-Per-country figures for the full hosted system are on https://dbtool.it/academic; an
-independent open bench on public WGND 2.0 names (network alone, losses included) is
-on https://dbtool.it/benchmark.
-
-## What this release does NOT include
-
-- **The dictionary layer of the API.** The hosted API combines the network with
-  a proprietary frequency layer (exact-name and per-country M/F frequencies)
-  derived from production data. That layer is **not part of this release**:
-  you get the network alone. Consequence: **the public API can answer
-  differently from the weights you download**, especially on very frequent
-  names. Documented example: `Yuki Tanaka` scores *female 0.59* with the network
-  alone (the value reproduced above), while the API with its dictionary layer
-  answers *male 0.58*.
-- **Everything about how the models were trained**: no recipe, no epochs, no
-  optimiser, no data split, no training hyperparameters, no data sources and no
-  per-country counts. This release ships a usable model, not the process behind
-  it, and no training example.
-
-## Training data — nature only
-
-The models were trained on **name–gender–country pairs covering 226 countries,
-tens of millions of examples**. That is the whole description this release
-provides: the corpus is **not redistributed**, and its sources, composition and
-per-country sizes are not part of the package.
-
-## Limitations
-
-- **East Asian names**: accuracy drops to roughly **82–88 %**; Chinese, Korean
-  and Japanese names are frequently confused with one another.
-- **Names ambiguous across countries**: many names are plausible in several
-  countries, so a top-1 of ~83 % means roughly one name in six gets the wrong
-  country. `Andrea` is the classic case: male in Italy, female elsewhere. Treat
-  the country score as a prior, and never as a single-country verdict.
-- **48-byte truncation**: only the first 48 UTF-8 bytes reach the model. Long
-  names and long compounds are silently cut, and multi-byte characters consume
-  more of the budget than plain ASCII.
-- **Empty or near-empty input**: an empty or fully-trimmed name is encoded as 48
-  padding bytes and still returns a prediction. Validate your input.
-- **Binary gender**: the gender head has two classes; names that do not fit them
-  are forced into the closer one.
-- **Transliteration**: results depend on how the name was romanised upstream;
-  different romanisations of the same name can disagree.
-- **Dictionary gap**: the released weights are the network alone, so for very
-  frequent names they can differ from the hosted API (see above).
+- **Two internal networks, one fused output.** Network A is the v1 `ultra`
+  byte-level CNN (UTF-8 bytes, 48-byte truncation). Network B reads the name
+  as Unicode codepoints (32 max, its own 226-country map). Gender probability
+  is the plain average of the two calibrated p(M); the country distribution is
+  0.7 × A + 0.3 × B.
+- **Calibration and threshold.** Each network ships per-head temperature
+  calibration files. The operating threshold M ≥ **0.60** was chosen on three
+  frozen benches and then verified — without retuning — on six fresh,
+  never-touched sets.
+- **Script-aware routing (T1).** The Unicode script of the name is detected
+  from character blocks; for measured scripts the model classifies the native
+  form, an `anyascii` transliteration, or the average of both (rule fixed on a
+  validation half only, no learned parameters). Latin and unmapped scripts
+  stay native. The country head always reads the original form.
+- **Country prior.** The country distribution is adjusted with a logit term
+  `log p(c) − 0.75 · log π(c)`, where `π` is the empirical prior in
+  `country_prior.json` (coefficient fixed on validation only).
+- **Per-country gender selector + Latin-script trait rule.** For a small set
+  of countries where v2.2 measured regressions (AM, GE, MM, TG, TD, WS), part
+  of the gender decision falls back to the v1 branch under rules fixed on old
+  benches and on a frozen validation bench — never on the 226-v2 test bench
+  (configuration in `dom113_rimedio.json`).
+
+## Results (gender accuracy, measured)
+
+v1 = `genderize-ultra` (calibrated, threshold 0.50). v2 = this release
+(threshold 0.60). All numbers are point measurements on frozen benches; they
+do not generalize beyond the named benches.
+
+### Independent bench: 226-countries v2 (50,258 rows, Latin script)
+
+Built with explicit exclusions against the training corpora and all earlier
+benches; **no overlap with training** (full = residual by construction).
+Paired bootstrap, 2,000 resamples, seed 20260927.
+
+| Group | n | v2 | v1 |
+|---|---:|---:|---:|
+| All | 50,258 | **94.08 %** | 93.71 % |
+| Women (F) | 20,648 | **92.98 %** | 90.30 % |
+| Men (M) | 29,610 | 94.86 % | **96.09 %** |
+
+Per country (countries with ≥ 400 bench rows): **14 countries significantly
+above v1, 94 statistically par, 1 below**. With the pre-registered criterion
+(≥ 200 rows): **14 above, 106 par, 3 below**. Countries above v1: AZ, DZ,
+EG, ET, HK, JM, LK, MY, NG, NP, SG, SY, UY, ZA. "Par" means the confidence
+interval includes zero — it is not proof of equivalence; no multiple-comparison
+correction is applied.
+
+### Where v2 is **worse than v1** (all measured deltas, IC95 in percentage points)
+
+| Country | n | Δ (v2 − v1) | IC95 |
+|---|---:|---:|---|
+| **Togo (TG)** | 400 | **−2.00** | [−3.50, −0.50] |
+| **Chad (TD)** | 207 | **−2.42** | [−4.83, −0.48] |
+| **Samoa (WS)** | 210 | **−4.76** | [−9.05, −0.95] |
+| Armenia (AM) | 400 | −0.25 | [−0.75, 0.00] |
+| Georgia (GE) | 400 | −0.25 | [−0.75, 0.00] |
+| Myanmar (MM) | 400 | −1.25 | [−3.50, +0.75] |
+
+TG is the only country below v1 at the ≥ 400-row quota; TD and WS are also
+below at the pre-registered ≥ 200-row criterion; AM, GE and MM measure
+statistically par. Without the Latin-script trait rule, deltas down to
+−7.3 pp were measured on the same bench. **If your
+analysis focuses on these countries, use v1.**
+
+### Other frozen benches
+
+| Bench | n | v2 | v1 | Notes |
+|---|---:|---:|---:|---|
+| Native scripts (with T1 routing) | 16,945 | **88.70 %** | 83.84 % | release path (v2.2 verdict) |
+| Native scripts + routing (T1) | 6,072 | **85.82 %** | 74.31 % | half-test split; largest gains: Georgian 50.9→92.3, Bengali 62.5→87.7, Armenian 67.5→90.8, Kazakh 93.1→97.5, Hindi 83.8→87.3 |
+| East-Asian | 5,199 | **80.96 %** | 79.75 % | women 79.3 vs 69.3 |
+| Old 226 bench (Wikidata) | 39,164 | 92.63 % | 92.14 % | **not independent: 34 % of this bench is inside v1's training data** (v1 residual: 90.41 %); kept for continuity with v1's history |
+
+Six fresh, never-touched sets (v2 / v1): Olympians 96.20 / 96.07; Vietnamese
+80.32 / 76.58; Tajik-Latin 79.09 / 76.13; Tajik-Cyrillic 72.12 / 67.65;
+Persian-Latin 77.57 / 74.07; Persian-script 67.29 / 64.59.
+
+### Country (nationality) output
+
+The country head is a **coarse signal**: on the old 226 bench, top-1 ≈ 28.4 %
+(v1: 28.1 %; v1 top-5 ≈ 56.3 %). Use `top5` and the probability, not the
+argmax alone.
+
+### Persian dedicated branch: tested and **rejected**
+
+A dedicated Persian transliteration branch (beam over k readings, < 5 MB) was
+integrated and measured on the only available native-Persian bench (n = 1,000;
+**not fresh** — already used by earlier experiments): +2.5 pp over the direct
+branch, IC95 [−0.4, +5.6] — the interval crosses zero, so the gain is **not
+confirmed** and the branch is **not part of this release**. It will be
+reconsidered only if an independent, fresh native-Persian bench becomes
+available (none found as of 2026-09-28).
+
+### Speed
+
+198.8 names/s with script routing (189.6 without), single CPU, 8 threads,
+batch 256, measured back-to-back on the bench machine. Throughput varies with
+hardware and load; no latency promise is made.
+
+## Limitations and honest notes
+
+- **Weaker than v1 in Togo, Chad, Samoa** (deltas above) — the headline
+  limitation of this release. Armenia, Georgia and Myanmar measure slightly
+  below v1 (statistically par).
+- **Women with non-Latin-script names remain weak**: on Persian-script and
+  Cyrillic names female accuracy stays around 29–35 %, as low as 19.9 % on
+  Persian-script names in the fresh sets. The aggregate gains do not fix this.
+- **Arabic and Hebrew: no gain.** Transliteration does not help these scripts
+  (measured: routing leaves them at the native branch); v2 is at best par with
+  v1 there. Thai likewise stays native.
+- **The model does not condition gender on country**: e.g. "Andrea Costa" is
+  classified F with p(M) = 0.44 (Andrea is a male name in Italian, female in
+  Spanish).
+- **No built-in abstention**: the model always answers. Probabilities are
+  temperature-calibrated per network; the 0.60 threshold was fixed before the
+  fresh-set verification. For analyses, use `p_male` (and a margin you choose)
+  rather than the hard label.
+- **Binary output only**: the model returns M/F. It cannot represent
+  non-binary identities, and name-inferred gender is a *perceived-attribute
+  proxy*, not a statement about any person's identity.
+- Bench numbers are specific to the named frozen benches; several are
+  Wikidata-derived and may share source characteristics with the training
+  corpora (the independent 226-v2 bench was built with explicit exclusions;
+  the old 226 bench overlaps v1 training by 34 %, as declared above).
+
+## Data provenance
+
+- Trained on **aggregated public name records with gender and country signals**
+  (Wikidata-derived corpora, Olympian records, per-country name lists),
+  internally reviewed. Training corpora, splits and recipes are **private and
+  not redistributed** with this package.
+- **No LLM-generated labels were used in the training of this model.** An
+  LLM second-opinion cascade was explored in research and measured, but it is
+  **not part of this release**.
+- A small fraction of the network-B corpus (**8,708 rows, ≈ 1.21 %**) comes
+  from a GFDL-1.3-licensed public dataset; this is disclosed for transparency
+  about obligations.
+- WGND and IPUMS were never used in training.
+- Bench sources: Wikidata (CC0), Olympian and public record compilations;
+  bench row-level data is not redistributed.
 
 ## Licence
 
-Two licences, one per artefact:
+- **Weights and model artifacts** (everything under `genderize_fuso_v2/`):
+  **CC BY-NC 4.0** — see `LICENSE`; commercial use requires a separate
+  licence (see `COMMERCIAL_USE.md`).
+- **Inference code** (`inference.py`, `inference_fuso.py`): **MIT** — see
+  `LICENSE-CODE`.
+- Runtime dependency `anyascii`: ISC licence (third-party).
 
-- **Weights** — `genderize_cora.pt`, `genderize_ultra.pt`, their `.config.json`
-  and `.calibration.json`, and `maps.json` — are released under **Creative
-  Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)**, full
-  text in `LICENSE`. You may download, run, study, modify and redistribute them
-  and their derivatives **for non-commercial purposes**, with attribution
-  (*dbtool.it, genderize cora/ultra, 2026*) and an indication of changes.
-- **Inference code** — `genderize_infer.py` — is released under the **MIT License**,
-  full text in `LICENSE-CODE`. It can be reused freely; it carries no model
-  knowledge.
+## Intended use
 
-**Commercial use of the weights is not granted by this licence.** The commercial
-channel is **dbtool.it** (the hosted API and the on-premise licence), or a
-separate written licence from the rights holder. What counts as commercial use,
-and how to request a licence, is set out in **`COMMERCIAL_USE.md`**.
+Aggregate, statistical analyses: bibliometrics and authorship studies,
+name-collection demography, dataset documentation. **Not** intended for
+decisions about individual people (hiring, credit, identity verification,
+content moderation), nor as ground truth about anyone's gender.
 
-The training data is **not** part of this release, and neither is the dictionary
-layer of the hosted API (see above).
+## Version history
 
-## Contact
+- **v2** (this release): fused two-network model, script-aware routing,
+  country prior, per-country selector. 14 above, 94 par, 1 below at ≥ 400 rows
+  (3 below at ≥ 200) on the independent 226-v2 bench (see above).
+- **v1** (`cora` / `ultra`): preserved in this repository at git tag **`v1`**.
+
+---
 
-dbtool — https://dbtool.it — open-weights page: https://dbtool.it/open-models.html
-Domenico Gigante, d.gigante@tech-time.it
+Support this work: https://github.com/sponsors/sheppard94g

examples.txt

@@ -0,0 +1,10 @@
+# Genderize v2 - example outputs, reproduced with exactly the packaged files
+# (CPU inference, deterministic; regenerate with the README quickstart)
+
+Andrea Rossi           M  p_male 0.7937  IT 0.8262  top5 IT FR US CH DE
+Yuki Tanaka            F  p_male 0.5037  JP 0.9809  top5 JP US ID DE CN
+María García           F  p_male 0.0041  ES 0.4072  top5 ES MX CO PE AR
+Chen Wei               M  p_male 0.7927  CN 0.4280  top5 CN TW SG MY HK
+Fatima Al Sayed        F  p_male 0.0089  AE 0.2588  top5 AE OM SA EG SY
+ნინო ალექსიძე          F  p_male 0.5304  GE 0.9764  top5 GE RU FR GR RS
+Andrea Costa           F  p_male 0.4408  IT 0.3496  top5 IT US BR FR AR   # known error: Andrea is a male name in Italian (no country conditioning)

genderize_fuso_v2/SHA256SUMS

@@ -0,0 +1,11 @@
+66ca7529702dab8241a9b275c30b3d24bf79c3fea81a8e6df4e01365c9987876  v1/genderize_ultra.calibration.json
+9105ba4cbe101b66215d68d6cb990872e5aada69fa244d3fe93ef93bbb68121a  v1/genderize_ultra.config.json
+e0b61d9a547aa1b11acad39af573ad63897578e35aa609c26b3de33156303b83  v1/genderize_ultra.pt
+79c5ae965c32da45899aa7aef1ce31192aa3daef16c8d2c46762a3550cdbc5e2  v1/maps.json
+a77c0a9a459b11c2dc7558eb55e359f145ecc4afdcf1b1f30f9908b78d3835f2  v2/genderize_ultra.calibration.json
+2daac90aba1db7980562bb683f1daa11d120bf6b78c92f44f956a93f0c2afbdd  v2/genderize_ultra.config.json
+079f974d050ac27c7ffb3e15e7bd0dd2a6eb450d6d6a759f02acaf5d67ab33c8  v2/genderize_ultra.pt
+2d4992a18e70435622e2f480814c0aaf28e9a32c555655a62f5c217fd18a4320  v2/maps.json
+fcb8bafbf58f1c49367c58b943641c2a426a8d486b4ab654672dec0fc7b24c7c  fuso.json
+120adde865c64296bf17f77f69419beb7d0fa647f1ce4fe80cefb00795930137  country_prior.json
+b68edafb32abcc05b3ec68c498722ef2ba281953bc00175cdafa976876e0516e  dom113_rimedio.json

genderize_fuso_v2/country_prior.json

@@ -0,0 +1,297 @@
+{
+ "beta": 0.75,
+ "fonte_prior": "conteggio empirico delle righe train (split=0) del corpus D2 precomputed G2D1; floor(counts,1); normalizzato a somma 1; identico a DOM-105 (reports/trainer/dom105/protocol.json)",
+ "input_sha256": {
+  "data/nomi_v2/banchi_overlap_wikidata.manifest.json": "567b447133681b75b18dd0f5da1f7e50478281759d3c7bbfb1e88504653b89d8",
+  "data/nomi_v2/banchi_overlap_wikidata_226.txt": "45eb3f3647e3563550cabab315244cffa53bbd3d8a0691605c1a87cb6427195c",
+  "data/nomi_v2/banchi_overlap_wikidata_nativo.txt": "2ca053efb5533afaf64d154380728d6306f61b426b04ec8c08cc2f60c1bdc751",
+  "data/nomi_v2/precomputed_v2_G2D1/manifest.json": "efa8485962b2b35bd8186a38b03cb4872bd7200d7c7a8c9b9612c927a147d8d3",
+  "data/nomi_v2/test_226_paesi.FROZEN.json": "b8dad3c5ba75d2acc36d93d0d846212d8af48833bc0ec36b83182ba6cdbc329a",
+  "data/nomi_v2/test_226_paesi.tsv": "0841f58d64ee224c60cb87a94d33131450a397e872e5cff9e70ee0d14eab12c6",
+  "data/nomi_v2/test_nativo_v1.FROZEN.json": "1782cdc371774d0d32be970e4fec89127ceef9ce044c5162960f8387885a338b",
+  "data/nomi_v2/test_nativo_v1.tsv": "a3c07f812446111ce9ba9dd35e562b3805b9dec9d930c25527edfe1b88154ce2",
+  "models/genderize_fuso_v2/v2/maps.json": "2d4992a18e70435622e2f480814c0aaf28e9a32c555655a62f5c217fd18a4320",
+  "release/inference.py": "35707c40dca9f4fe99da0ca1131d00808693b868b1a52c15aeb588ba2d1c2374",
+  "release/inference_fuso.py": "e50de64632785dc36f9afb65a04cda51edf82ce51ff7aa3bdc0e58f9dddf733e",
+  "reports/trainer/dom105/protocol.json": "61b14fb59526799ec127ff4ea8875c3f16890efa8888115d51dfb998cfd6a9dd",
+  "reports/trainer/genderize_D2.json": "7d2ef64c00f40d4c938780d653c61d7c85c44dfa5bff0133683b2db6bf6971ae",
+  "scripts/dom112_prepara.py": "74d78bc2199afffa8dedab5c20f6359c3a40c68f7e2254944834249fa8431f63"
+ },
+ "logit_adjustment": "log(p_c fusa 0,7v1+0,3v2) - beta*log(prior) in float64; argmax per il paese; p_c restituita rinormalizzata",
+ "prior": {
+  "AD": 1.6018768417489172e-05,
+  "AE": 0.009082800189765346,
+  "AF": 0.004839083969052669,
+  "AG": 6.022887861457011e-06,
+  "AI": 3.1699409797142164e-08,
+  "AL": 0.004687127564955102,
+  "AM": 0.00015520031036680802,
+  "AO": 0.005032661698213884,
+  "AR": 0.003951648425311742,
+  "AT": 0.012980168659034449,
+  "AU": 0.0009517325132794639,
+  "AX": 1.056646993238072e-08,
+  "AZ": 0.0010330097999993364,
+  "BA": 0.00013108762598111522,
+  "BB": 1.7857334185723418e-05,
+  "BD": 0.00588676002931815,
+  "BE": 0.007814792098469863,
+  "BF": 9.130486668570181e-05,
+  "BG": 0.0036473657906689744,
+  "BH": 0.009432265049838974,
+  "BI": 0.00018035907527580653,
+  "BJ": 4.985260514097224e-05,
+  "BM": 3.8461950553865825e-06,
+  "BN": 0.0021134947494048594,
+  "BO": 0.0056805553685877405,
+  "BR": 0.024410806768364993,
+  "BS": 1.3229220355340662e-05,
+  "BT": 1.5638375499923466e-06,
+  "BW": 0.002517968651946461,
+  "BY": 0.00022349140553978462,
+  "BZ": 9.795117627316928e-06,
+  "CA": 0.0095114713084521,
+  "CC": 1.056646993238072e-08,
+  "CD": 2.7620752403243204e-05,
+  "CF": 2.5031967269809928e-05,
+  "CG": 0.00013044307131524,
+  "CH": 0.0054746994013650995,
+  "CI": 0.00012076418485717926,
+  "CK": 3.592599777009445e-07,
+  "CL": 0.009715626074015628,
+  "CM": 0.0022206070551094026,
+  "CN": 0.006705333888509752,
+  "CO": 0.014075256469886522,
+  "CR": 0.0022267567406100484,
+  "CU": 0.00013013664368720097,
+  "CV": 3.5418807213340175e-05,
+  "CY": 0.001162163761982826,
+  "CZ": 0.012321317559340714,
+  "DE": 0.028080351579422038,
+  "DJ": 0.00015622525795024896,
+  "DK": 0.04178221013483809,
+  "DM": 1.7931299475250084e-05,
+  "DO": 8.341171364621342e-05,
+  "DZ": 0.01379913347761355,
+  "EC": 0.0030637373904238577,
+  "EE": 0.0014430944980750322,
+  "EG": 0.016905295244815915,
+  "EH": 1.056646993238072e-08,
+  "ER": 2.19782574593519e-06,
+  "ES": 0.02531284517355247,
+  "ET": 0.00018132062403965318,
+  "FI": 0.01249439633683311,
+  "FJ": 8.314755189790389e-05,
+  "FK": 2.113293986476144e-08,
+  "FM": 1.584970489857108e-07,
+  "FR": 0.13121066195700787,
+  "GA": 3.0980889841740275e-05,
+  "GB": 0.020088793005453378,
+  "GD": 1.9230975276932913e-06,
+  "GE": 0.0010607362171019034,
+  "GG": 2.3246233851237586e-07,
+  "GH": 0.008657510341456955,
+  "GI": 6.022887861457011e-07,
+  "GM": 1.5849704898571083e-05,
+  "GN": 6.439206776792811e-05,
+  "GP": 2.113293986476144e-08,
+  "GQ": 7.5127601219226924e-06,
+  "GR": 0.006941209196810287,
+  "GT": 0.012013262694932086,
+  "GW": 2.8318139418780333e-06,
+  "GY": 2.5718787815414674e-05,
+  "HK": 0.004499319128376967,
+  "HN": 0.00020840248647634497,
+  "HR": 0.006172308312770807,
+  "HT": 0.0002759116628743254,
+  "HU": 0.004415062097136164,
+  "ID": 0.002526897319039323,
+  "IE": 0.009992869112101434,
+  "IL": 0.00652543973791097,
+  "IM": 1.32080874154759e-06,
+  "IN": 0.009942234588185465,
+  "IO": 1.056646993238072e-08,
+  "IQ": 0.008264268596453475,
+  "IR": 0.005549827002584326,
+  "IS": 0.0004136350319729757,
+  "IT": 0.026147839327018994,
+  "JE": 3.2756056790380236e-07,
+  "JM": 0.0034164884226464557,
+  "JO": 0.004354262629145245,
+  "JP": 0.007213306364039023,
+  "KE": 9.528842585020934e-05,
+  "KG": 8.074896322325347e-05,
+  "KH": 0.00015076239299520813,
+  "KI": 5.705893763485589e-07,
+  "KM": 6.931604275641753e-05,
+  "KN": 7.92485244928554e-07,
+  "KP": 3.280888914004214e-05,
+  "KR": 0.0013601160096960465,
+  "KW": 0.0054166260826167345,
+  "KY": 3.2439062692408815e-06,
+  "KZ": 0.0036071603725762657,
+  "LA": 7.856170394725067e-05,
+  "LB": 0.0018794157409928293,
+  "LC": 2.6733168928923223e-06,
+  "LI": 1.3662445622568272e-05,
+  "LK": 0.00022036373043979993,
+  "LR": 7.924852449285541e-06,
+  "LS": 5.093038507407508e-06,
+  "LT": 0.0024641430541109135,
+  "LU": 0.0019066666669484392,
+  "LV": 0.00021158299392599157,
+  "LY": 0.0036326889639328977,
+  "MA": 0.009109797520442579,
+  "MC": 4.66932306311904e-05,
+  "MD": 0.0005427995604263976,
+  "ME": 4.7369484706862775e-05,
+  "MF": 1.056646993238072e-08,
+  "MG": 0.0002446877442241404,
+  "MH": 2.958611581066602e-07,
+  "MK": 8.37498406840496e-05,
+  "ML": 9.601751227554361e-05,
+  "MM": 1.935777291612148e-05,
+  "MN": 7.022475917060227e-05,
+  "MO": 0.002936073300700834,
+  "MQ": 2.113293986476144e-08,
+  "MR": 2.0741980477263354e-05,
+  "MS": 4.226587972952288e-08,
+  "MT": 0.0009922654919400763,
+  "MU": 0.006129260514266288,
+  "MV": 0.0007628251638283614,
+  "MW": 6.635743117535093e-06,
+  "MX": 0.015855970815240983,
+  "MY": 0.016463384339303888,
+  "MZ": 2.8814763505602226e-05,
+  "NA": 0.00392767310503517,
+  "NC": 6.656876057399854e-07,
+  "NE": 2.0192524040779557e-05,
+  "NF": 1.056646993238072e-08,
+  "NG": 0.012092912745282373,
+  "NI": 2.9004959964385078e-05,
+  "NL": 0.031615416927649666,
+  "NO": 0.0064510517895870105,
+  "NP": 0.00010499901171806722,
+  "NR": 5.610795534094163e-06,
+  "NU": 7.079534854695083e-07,
+  "NZ": 0.0003920899997808514,
+  "OM": 0.00832257437754035,
+  "PA": 0.010601687976665346,
+  "PE": 0.009714622259372052,
+  "PG": 1.306015683642257e-05,
+  "PH": 0.009393602336356393,
+  "PK": 0.00033100523710175847,
+  "PL": 0.012986719870392524,
+  "PN": 3.1699409797142164e-08,
+  "PR": 0.0011925740624482177,
+  "PS": 0.0049637894471946265,
+  "PT": 0.011482868170206304,
+  "PW": 3.6559985966037293e-06,
+  "PY": 5.9753387467612976e-05,
+  "QA": 0.003826562554252219,
+  "RO": 0.0009560858988916048,
+  "RS": 0.0020051461667182275,
+  "RU": 0.01639605479289476,
+  "RW": 2.434514672420518e-05,
+  "SA": 0.015217618667216068,
+  "SB": 3.3918368482942115e-06,
+  "SC": 3.4658021378208765e-06,
+  "SD": 0.009402319674050608,
+  "SE": 0.010882100955730865,
+  "SG": 0.005591406061768244,
+  "SI": 0.002685711362123005,
+  "SK": 0.0005211382970650172,
+  "SL": 6.477246068549382e-06,
+  "SM": 1.860755355092245e-05,
+  "SN": 0.00017695667195757994,
+  "SO": 4.733778529706563e-06,
+  "SR": 4.397764785856856e-05,
+  "SS": 3.1910739195789777e-06,
+  "ST": 2.620484543230419e-06,
+  "SV": 7.73359934350945e-05,
+  "SX": 1.056646993238072e-08,
+  "SY": 0.006802143886030224,
+  "SZ": 1.4687393206009202e-06,
+  "TC": 4.226587972952288e-08,
+  "TD": 1.4634560856347299e-05,
+  "TG": 4.246664265823812e-05,
+  "TH": 0.0002867211616151509,
+  "TJ": 1.5194583762763476e-05,
+  "TK": 1.056646993238072e-08,
+  "TL": 1.628293016579869e-05,
+  "TM": 0.0001901330599632587,
+  "TN": 0.008768162414588845,
+  "TO": 3.328438028699927e-06,
+  "TR": 0.011724364840510865,
+  "TT": 4.0680909239665774e-05,
+  "TV": 4.015258574304674e-07,
+  "TW": 0.004903127343312829,
+  "TZ": 2.9121191133641267e-05,
+  "UA": 0.0015944697463263185,
+  "UG": 2.5602556646158488e-05,
+  "US": 0.07691355653366785,
+  "UY": 0.010733251093793418,
+  "UZ": 8.81877580556495e-05,
+  "VA": 2.64161748309518e-07,
+  "VC": 1.1411787526971178e-06,
+  "VE": 0.00023895015105085762,
+  "VG": 8.347511246580769e-07,
+  "VN": 0.0003747926885015442,
+  "VU": 9.192828841171227e-07,
+  "WF": 6.339881959428433e-08,
+  "WS": 5.219836146596077e-06,
+  "XK": 3.2428496222476435e-05,
+  "YE": 0.004594174328959949,
+  "ZA": 0.01396672825721104,
+  "ZM": 2.154503219212429e-05,
+  "ZW": 2.4408545543799467e-05
+ },
+ "selettore": {
+  "griglia": [
+   "v1@0.50",
+   "fuso@0.50",
+   "fuso@0.55",
+   "fuso@0.60",
+   "fuso@0.65",
+   "fuso@0.70"
+  ],
+  "insieme_S": [
+   "AM",
+   "GE",
+   "MM",
+   "TG",
+   "TD",
+   "WS"
+  ],
+  "min_righe_validazione": 30,
+  "regola": "massima accuratezza genere sulle righe a paese vero c dei banchi vecchi (residuo 226 v1 + nativo); n<30 -> v1@0,5; parimerito: prima v1@0,5 poi fuso a soglia crescente",
+  "scelte": {
+   "AM": {
+    "ramo": "v1",
+    "soglia": 0.5
+   },
+   "GE": {
+    "ramo": "fuso",
+    "soglia": 0.55
+   },
+   "MM": {
+    "ramo": "fuso",
+    "soglia": 0.7
+   },
+   "TD": {
+    "ramo": "v1",
+    "soglia": 0.5
+   },
+   "TG": {
+    "ramo": "fuso",
+    "soglia": 0.7
+   },
+   "WS": {
+    "ramo": "v1",
+    "soglia": 0.5
+   }
+  },
+  "validazione": "banchi VECCHI: residuo 226 v1 (39.164 righe meno 13.425 overlap v1 training) + nativo v1 (16.945 righe); il banco 226 v2 non è mai stato letto in questa fase"
+ },
+ "versione": "2.1"
+}

genderize_fuso_v2/dom113_rimedio.json

@@ -0,0 +1,176 @@
+{
+ "ambito": "SOLO nomi di scritture latine (script_t1 = None): la validazione che ha scelto il rimedio è latina; sulle scritture native il ramo v1 leggerebbe byte fuori distribuzione e il trigger non è mai stato misurato",
+ "conseguenza": "se decisione v2.1 (taglio 0,60) != v1@0,50 e il trigger scatta: decisione e punteggio dal ramo v1, rimappati come il selettore DOM-112 (p_v1 + 0,60 - 0,50)",
+ "esclusi": {
+  "AM": "già pari col selettore DOM-112, non toccato",
+  "WS": "v1 peggiore del fuso sulla validazione (84,62 contro 85,71)"
+ },
+ "fonti": {
+  "residuo": "data/nomi_v2/test_226_paesi.tsv meno banchi_overlap_wikidata_226.txt",
+  "validazione": "data/nomi_v2/validazione_dom113_ge_mm_tg_td_ws.tsv (FROZEN, DOM-113)"
+ },
+ "lessico": {
+  "GE": [
+   "Abramidze",
+   "Abuladze",
+   "Aleksandre",
+   "Andria",
+   "Badri",
+   "Bakradze",
+   "Beka",
+   "Beridze",
+   "Berishvili",
+   "Beruashvili",
+   "Buba",
+   "Chagelishvili",
+   "Chechelashvili",
+   "Chkareuli",
+   "Davit",
+   "Elene",
+   "Emzar",
+   "Gagnidze",
+   "Gelashvili",
+   "Gocha",
+   "Gogi",
+   "Iashvili",
+   "Imnadze",
+   "Jgerenaia",
+   "Jordania",
+   "Kakhaber",
+   "Kavtaradze",
+   "Keso",
+   "Khaindrava",
+   "Khubua",
+   "Kiladze",
+   "Lasha",
+   "Leonid",
+   "Levan",
+   "Lizi",
+   "Losaberidze",
+   "Mamatsashvili",
+   "Mamuka",
+   "Mariami",
+   "Markoishvili",
+   "Mate",
+   "Meladze",
+   "Merab",
+   "Merabi",
+   "Mikheil",
+   "Mushkudiani",
+   "Natali",
+   "Nika",
+   "Nikoloz",
+   "Nikolozi",
+   "Nini",
+   "Nodar",
+   "Nozadze",
+   "Nugzar",
+   "Nutsubidze",
+   "Ormotsadze",
+   "Pitskhelauri",
+   "Ramaz",
+   "Revaz",
+   "Shalva",
+   "Shengelia",
+   "Shota",
+   "Tarba",
+   "Tato",
+   "Teimuraz",
+   "Tekla",
+   "Tornike",
+   "Tsertsvadze",
+   "Ugrekhelidze",
+   "Vakhtang",
+   "Vano",
+   "Vasadze",
+   "Vladimer",
+   "Vyacheslav",
+   "Zaza",
+   "Zodelava",
+   "Zurab"
+  ],
+  "MM": [
+   "Aike",
+   "Bai",
+   "Bay",
+   "Chit",
+   "Hein",
+   "Hlaing",
+   "Htan",
+   "Htet",
+   "Htike",
+   "Htin",
+   "Htoo",
+   "Htun",
+   "Htut",
+   "Kan",
+   "Khaing",
+   "Kham",
+   "Khat",
+   "Khun",
+   "Kyaing",
+   "Kyauk",
+   "Kyu",
+   "Latt",
+   "Law",
+   "Lay",
+   "Linn",
+   "Mann",
+   "Maw",
+   "Moon",
+   "Nyan",
+   "Nyein",
+   "Nyi",
+   "Nyo",
+   "Nyunt",
+   "Ohn",
+   "Phyo",
+   "Pyae",
+   "Ral",
+   "Saw",
+   "Sein",
+   "Shwe",
+   "Swe",
+   "Tha",
+   "Thant",
+   "Thar",
+   "Thaung",
+   "Thaw",
+   "Thura",
+   "Tint",
+   "Wunna",
+   "Yone"
+  ],
+  "TD": [
+   "Allasra",
+   "Timothee"
+  ],
+  "TG": [
+   "Akakpo",
+   "Coubageat",
+   "Dadzie",
+   "Eninful",
+   "Kodjovi",
+   "Lawson",
+   "Segbefia"
+  ]
+ },
+ "onorifici_mm": [
+  "Maung",
+  "Mg",
+  "Sao",
+  "Saw",
+  "U"
+ ],
+ "paesi_attivi": [
+  "GE",
+  "MM",
+  "TG",
+  "TD"
+ ],
+ "regola_lessico": "token esatto, qualsiasi posizione, supporto>=2 in validazione, lunghezza>=3, non più frequente negli altri paesi S, <=3 occorrenze nel residuo 226 v1",
+ "regola_massa": "somma delle p_paese AGGIUSTATE (logit β=0,75) su {GE MM TG TD WS} >= tau e paese S con p massima tra quelli attivi",
+ "regola_onorifici": "primo token onorifico birmano e nome di almeno 2 token (onorifici presenti come primo token nella validazione MM)",
+ "tau_massa": 0.15,
+ "versione": "dom113"
+}

genderize_fuso_v2/fuso.json

@@ -0,0 +1 @@
+{"nome":"genderize v2 fuso","v1":"release/pesi ultra","v2":"models/genderize_D2","soglia_M":0.60,"peso_paese_v1":0.7,"decisione":"DECISIONI 25/09"}

genderize_fuso_v2/v1/genderize_ultra.calibration.json

@@ -0,0 +1 @@
+{"temp_gender": 1.0460342168807983, "temp_country": 0.939997673034668, "ece_gender_pre": 0.0012399105666638377, "ece_gender_post": 0.0007778512481309008}

genderize_fuso_v2/v1/genderize_ultra.config.json

@@ -0,0 +1 @@
+{"ch": 384}

genderize_fuso_v2/v1/genderize_ultra.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0b61d9a547aa1b11acad39af573ad63897578e35aa609c26b3de33156303b83
+size 12868705

genderize_fuso_v2/v1/maps.json

@@ -0,0 +1 @@
+{"gender": {"M": 0, "F": 1}, "country": {"AD": 0, "AE": 1, "AF": 2, "AG": 3, "AI": 4, "AL": 5, "AM": 6, "AO": 7, "AR": 8, "AT": 9, "AU": 10, "AX": 11, "AZ": 12, "BA": 13, "BB": 14, "BD": 15, "BE": 16, "BF": 17, "BG": 18, "BH": 19, "BI": 20, "BJ": 21, "BM": 22, "BN": 23, "BO": 24, "BR": 25, "BS": 26, "BT": 27, "BW": 28, "BY": 29, "BZ": 30, "CA": 31, "CC": 32, "CD": 33, "CF": 34, "CG": 35, "CH": 36, "CI": 37, "CK": 38, "CL": 39, "CM": 40, "CN": 41, "CO": 42, "CR": 43, "CU": 44, "CV": 45, "CY": 46, "CZ": 47, "DE": 48, "DJ": 49, "DK": 50, "DM": 51, "DO": 52, "DZ": 53, "EC": 54, "EE": 55, "EG": 56, "EH": 57, "ER": 58, "ES": 59, "ET": 60, "FI": 61, "FJ": 62, "FK": 63, "FM": 64, "FR": 65, "GA": 66, "GB": 67, "GD": 68, "GE": 69, "GG": 70, "GH": 71, "GI": 72, "GM": 73, "GN": 74, "GP": 75, "GQ": 76, "GR": 77, "GT": 78, "GW": 79, "GY": 80, "HK": 81, "HN": 82, "HR": 83, "HT": 84, "HU": 85, "ID": 86, "IE": 87, "IL": 88, "IM": 89, "IN": 90, "IO": 91, "IQ": 92, "IR": 93, "IS": 94, "IT": 95, "JE": 96, "JM": 97, "JO": 98, "JP": 99, "KE": 100, "KG": 101, "KH": 102, "KI": 103, "KM": 104, "KN": 105, "KP": 106, "KR": 107, "KW": 108, "KY": 109, "KZ": 110, "LA": 111, "LB": 112, "LC": 113, "LI": 114, "LK": 115, "LR": 116, "LS": 117, "LT": 118, "LU": 119, "LV": 120, "LY": 121, "MA": 122, "MC": 123, "MD": 124, "ME": 125, "MF": 126, "MG": 127, "MH": 128, "MK": 129, "ML": 130, "MM": 131, "MN": 132, "MO": 133, "MQ": 134, "MR": 135, "MS": 136, "MT": 137, "MU": 138, "MV": 139, "MW": 140, "MX": 141, "MY": 142, "MZ": 143, "NA": 144, "NC": 145, "NE": 146, "NF": 147, "NG": 148, "NI": 149, "NL": 150, "NO": 151, "NP": 152, "NR": 153, "NU": 154, "NZ": 155, "OM": 156, "PA": 157, "PE": 158, "PG": 159, "PH": 160, "PK": 161, "PL": 162, "PN": 163, "PR": 164, "PS": 165, "PT": 166, "PW": 167, "PY": 168, "QA": 169, "RO": 170, "RS": 171, "RU": 172, "RW": 173, "SA": 174, "SB": 175, "SC": 176, "SD": 177, "SE": 178, "SG": 179, "SI": 180, "SK": 181, "SL": 182, "SM": 183, "SN": 184, "SO": 185, "SR": 186, "SS": 187, "ST": 188, "SV": 189, "SX": 190, "SY": 191, "SZ": 192, "TC": 193, "TD": 194, "TG": 195, "TH": 196, "TJ": 197, "TK": 198, "TL": 199, "TM": 200, "TN": 201, "TO": 202, "TR": 203, "TT": 204, "TV": 205, "TW": 206, "TZ": 207, "UA": 208, "UG": 209, "US": 210, "UY": 211, "UZ": 212, "VA": 213, "VC": 214, "VE": 215, "VG": 216, "VN": 217, "VU": 218, "WF": 219, "WS": 220, "XK": 221, "YE": 222, "ZA": 223, "ZM": 224, "ZW": 225}}

genderize_fuso_v2/v2/genderize_ultra.calibration.json

@@ -0,0 +1,4 @@
+{
+  "temp_gender": 1.0,
+  "temp_country": 1.0
+}

genderize_fuso_v2/v2/genderize_ultra.config.json

@@ -0,0 +1,4 @@
+{
+  "ch": 384,
+  "maxlen": 32
+}

genderize_fuso_v2/v2/genderize_ultra.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:079f974d050ac27c7ffb3e15e7bd0dd2a6eb450d6d6a759f02acaf5d67ab33c8
+size 14099084

genderize_fuso_v2/v2/maps.json

@@ -0,0 +1,5298 @@
+{
+  "gender": {
+    "M": 0,
+    "F": 1
+  },
+  "country": {
+    "AD": 0,
+    "AE": 1,
+    "AF": 2,
+    "AG": 3,
+    "AI": 4,
+    "AL": 5,
+    "AM": 6,
+    "AO": 7,
+    "AR": 8,
+    "AT": 9,
+    "AU": 10,
+    "AX": 11,
+    "AZ": 12,
+    "BA": 13,
+    "BB": 14,
+    "BD": 15,
+    "BE": 16,
+    "BF": 17,
+    "BG": 18,
+    "BH": 19,
+    "BI": 20,
+    "BJ": 21,
+    "BM": 22,
+    "BN": 23,
+    "BO": 24,
+    "BR": 25,
+    "BS": 26,
+    "BT": 27,
+    "BW": 28,
+    "BY": 29,
+    "BZ": 30,
+    "CA": 31,
+    "CC": 32,
+    "CD": 33,
+    "CF": 34,
+    "CG": 35,
+    "CH": 36,
+    "CI": 37,
+    "CK": 38,
+    "CL": 39,
+    "CM": 40,
+    "CN": 41,
+    "CO": 42,
+    "CR": 43,
+    "CU": 44,
+    "CV": 45,
+    "CY": 46,
+    "CZ": 47,
+    "DE": 48,
+    "DJ": 49,
+    "DK": 50,
+    "DM": 51,
+    "DO": 52,
+    "DZ": 53,
+    "EC": 54,
+    "EE": 55,
+    "EG": 56,
+    "EH": 57,
+    "ER": 58,
+    "ES": 59,
+    "ET": 60,
+    "FI": 61,
+    "FJ": 62,
+    "FK": 63,
+    "FM": 64,
+    "FR": 65,
+    "GA": 66,
+    "GB": 67,
+    "GD": 68,
+    "GE": 69,
+    "GG": 70,
+    "GH": 71,
+    "GI": 72,
+    "GM": 73,
+    "GN": 74,
+    "GP": 75,
+    "GQ": 76,
+    "GR": 77,
+    "GT": 78,
+    "GW": 79,
+    "GY": 80,
+    "HK": 81,
+    "HN": 82,
+    "HR": 83,
+    "HT": 84,
+    "HU": 85,
+    "ID": 86,
+    "IE": 87,
+    "IL": 88,
+    "IM": 89,
+    "IN": 90,
+    "IO": 91,
+    "IQ": 92,
+    "IR": 93,
+    "IS": 94,
+    "IT": 95,
+    "JE": 96,
+    "JM": 97,
+    "JO": 98,
+    "JP": 99,
+    "KE": 100,
+    "KG": 101,
+    "KH": 102,
+    "KI": 103,
+    "KM": 104,
+    "KN": 105,
+    "KP": 106,
+    "KR": 107,
+    "KW": 108,
+    "KY": 109,
+    "KZ": 110,
+    "LA": 111,
+    "LB": 112,
+    "LC": 113,
+    "LI": 114,
+    "LK": 115,
+    "LR": 116,
+    "LS": 117,
+    "LT": 118,
+    "LU": 119,
+    "LV": 120,
+    "LY": 121,
+    "MA": 122,
+    "MC": 123,
+    "MD": 124,
+    "ME": 125,
+    "MF": 126,
+    "MG": 127,
+    "MH": 128,
+    "MK": 129,
+    "ML": 130,
+    "MM": 131,
+    "MN": 132,
+    "MO": 133,
+    "MQ": 134,
+    "MR": 135,
+    "MS": 136,
+    "MT": 137,
+    "MU": 138,
+    "MV": 139,
+    "MW": 140,
+    "MX": 141,
+    "MY": 142,
+    "MZ": 143,
+    "NA": 144,
+    "NC": 145,
+    "NE": 146,
+    "NF": 147,
+    "NG": 148,
+    "NI": 149,
+    "NL": 150,
+    "NO": 151,
+    "NP": 152,
+    "NR": 153,
+    "NU": 154,
+    "NZ": 155,
+    "OM": 156,
+    "PA": 157,
+    "PE": 158,
+    "PG": 159,
+    "PH": 160,
+    "PK": 161,
+    "PL": 162,
+    "PN": 163,
+    "PR": 164,
+    "PS": 165,
+    "PT": 166,
+    "PW": 167,
+    "PY": 168,
+    "QA": 169,
+    "RO": 170,
+    "RS": 171,
+    "RU": 172,
+    "RW": 173,
+    "SA": 174,
+    "SB": 175,
+    "SC": 176,
+    "SD": 177,
+    "SE": 178,
+    "SG": 179,
+    "SI": 180,
+    "SK": 181,
+    "SL": 182,
+    "SM": 183,
+    "SN": 184,
+    "SO": 185,
+    "SR": 186,
+    "SS": 187,
+    "ST": 188,
+    "SV": 189,
+    "SX": 190,
+    "SY": 191,
+    "SZ": 192,
+    "TC": 193,
+    "TD": 194,
+    "TG": 195,
+    "TH": 196,
+    "TJ": 197,
+    "TK": 198,
+    "TL": 199,
+    "TM": 200,
+    "TN": 201,
+    "TO": 202,
+    "TR": 203,
+    "TT": 204,
+    "TV": 205,
+    "TW": 206,
+    "TZ": 207,
+    "UA": 208,
+    "UG": 209,
+    "US": 210,
+    "UY": 211,
+    "UZ": 212,
+    "VA": 213,
+    "VC": 214,
+    "VE": 215,
+    "VG": 216,
+    "VN": 217,
+    "VU": 218,
+    "WF": 219,
+    "WS": 220,
+    "XK": 221,
+    "YE": 222,
+    "ZA": 223,
+    "ZM": 224,
+    "ZW": 225
+  },
+  "vocab_codepoint": {
+    " ": 2,
+    "!": 3,
+    "\"": 4,
+    "#": 5,
+    "$": 6,
+    "%": 7,
+    "&": 8,
+    "'": 9,
+    "(": 10,
+    ")": 11,
+    "*": 12,
+    "+": 13,
+    ",": 14,
+    "-": 15,
+    ".": 16,
+    "/": 17,
+    "0": 18,
+    "1": 19,
+    "2": 20,
+    "3": 21,
+    "4": 22,
+    "5": 23,
+    "6": 24,
+    "7": 25,
+    "8": 26,
+    "9": 27,
+    ":": 28,
+    ";": 29,
+    "<": 30,
+    "=": 31,
+    ">": 32,
+    "?": 33,
+    "@": 34,
+    "[": 35,
+    "\\": 36,
+    "]": 37,
+    "^": 38,
+    "_": 39,
+    "`": 40,
+    "a": 41,
+    "b": 42,
+    "c": 43,
+    "d": 44,
+    "e": 45,
+    "f": 46,
+    "g": 47,
+    "h": 48,
+    "i": 49,
+    "j": 50,
+    "k": 51,
+    "l": 52,
+    "m": 53,
+    "n": 54,
+    "o": 55,
+    "p": 56,
+    "q": 57,
+    "r": 58,
+    "s": 59,
+    "t": 60,
+    "u": 61,
+    "v": 62,
+    "w": 63,
+    "x": 64,
+    "y": 65,
+    "z": 66,
+    "{": 67,
+    "|": 68,
+    "}": 69,
+    "": 70,
+    "¡": 71,
+    "¢": 72,
+    "£": 73,
+    "¥": 74,
+    "©": 75,
+    "ª": 76,
+    "«": 77,
+    "¬": 78,
+    "®": 79,
+    "°": 80,
+    "´": 81,
+    "µ": 82,
+    "·": 83,
+    "º": 84,
+    "»": 85,
+    "¼": 86,
+    "½": 87,
+    "¿": 88,
+    "×": 89,
+    "ß": 90,
+    "à": 91,
+    "á": 92,
+    "â": 93,
+    "ã": 94,
+    "ä": 95,
+    "å": 96,
+    "æ": 97,
+    "ç": 98,
+    "è": 99,
+    "é": 100,
+    "ê": 101,
+    "ë": 102,
+    "ì": 103,
+    "í": 104,
+    "î": 105,
+    "ï": 106,
+    "ð": 107,
+    "ñ": 108,
+    "ò": 109,
+    "ó": 110,
+    "ô": 111,
+    "õ": 112,
+    "ö": 113,
+    "÷": 114,
+    "ø": 115,
+    "ù": 116,
+    "ú": 117,
+    "û": 118,
+    "ü": 119,
+    "ý": 120,
+    "þ": 121,
+    "ÿ": 122,
+    "ā": 123,
+    "ă": 124,
+    "ą": 125,
+    "ć": 126,
+    "ĉ": 127,
+    "ċ": 128,
+    "č": 129,
+    "ď": 130,
+    "đ": 131,
+    "ē": 132,
+    "ĕ": 133,
+    "ė": 134,
+    "ę": 135,
+    "ě": 136,
+    "ĝ": 137,
+    "ğ": 138,
+    "ġ": 139,
+    "ģ": 140,
+    "ĥ": 141,
+    "ħ": 142,
+    "ĩ": 143,
+    "ī": 144,
+    "ĭ": 145,
+    "į": 146,
+    "ı": 147,
+    "ij": 148,
+    "ĵ": 149,
+    "ķ": 150,
+    "ĸ": 151,
+    "ĺ": 152,
+    "ļ": 153,
+    "ľ": 154,
+    "ŀ": 155,
+    "ł": 156,
+    "ń": 157,
+    "ņ": 158,
+    "ň": 159,
+    "ʼn": 160,
+    "ŋ": 161,
+    "ō": 162,
+    "ŏ": 163,
+    "ő": 164,
+    "œ": 165,
+    "ŕ": 166,
+    "ŗ": 167,
+    "ř": 168,
+    "ś": 169,
+    "ŝ": 170,
+    "ş": 171,
+    "š": 172,
+    "ţ": 173,
+    "ť": 174,
+    "ŧ": 175,
+    "ũ": 176,
+    "ū": 177,
+    "ŭ": 178,
+    "ů": 179,
+    "ű": 180,
+    "ų": 181,
+    "ŵ": 182,
+    "ŷ": 183,
+    "ź": 184,
+    "ż": 185,
+    "ž": 186,
+    "ƀ": 187,
+    "ƅ": 188,
+    "ƌ": 189,
+    "ƒ": 190,
+    "ƛ": 191,
+    "ƞ": 192,
+    "ơ": 193,
+    "ƥ": 194,
+    "ƨ": 195,
+    "ƪ": 196,
+    "ƭ": 197,
+    "ư": 198,
+    "ƴ": 199,
+    "ƶ": 200,
+    "ƺ": 201,
+    "ƽ": 202,
+    "ƿ": 203,
+    "dž": 204,
+    "ǎ": 205,
+    "ǐ": 206,
+    "ǒ": 207,
+    "ǔ": 208,
+    "ǖ": 209,
+    "ǘ": 210,
+    "ǚ": 211,
+    "ǜ": 212,
+    "ǝ": 213,
+    "ǟ": 214,
+    "ǡ": 215,
+    "ǥ": 216,
+    "ǧ": 217,
+    "ǩ": 218,
+    "ǯ": 219,
+    "ǰ": 220,
+    "ǵ": 221,
+    "ǹ": 222,
+    "ǻ": 223,
+    "ǿ": 224,
+    "ȁ": 225,
+    "ȃ": 226,
+    "ȅ": 227,
+    "ȇ": 228,
+    "ȉ": 229,
+    "ȋ": 230,
+    "ȍ": 231,
+    "ȑ": 232,
+    "ȓ": 233,
+    "ș": 234,
+    "ț": 235,
+    "ȝ": 236,
+    "ȟ": 237,
+    "ȡ": 238,
+    "ȧ": 239,
+    "ȩ": 240,
+    "ȭ": 241,
+    "ȳ": 242,
+    "ɇ": 243,
+    "ɐ": 244,
+    "ɑ": 245,
+    "ɓ": 246,
+    "ɖ": 247,
+    "ɗ": 248,
+    "ɘ": 249,
+    "ə": 250,
+    "ɛ": 251,
+    "ɞ": 252,
+    "ɣ": 253,
+    "ɥ": 254,
+    "ɩ": 255,
+    "ɪ": 256,
+    "ɭ": 257,
+    "ɯ": 258,
+    "ɱ": 259,
+    "ɲ": 260,
+    "ɳ": 261,
+    "ɽ": 262,
+    "ɾ": 263,
+    "ʀ": 264,
+    "ʆ": 265,
+    "ʇ": 266,
+    "ʉ": 267,
+    "ʋ": 268,
+    "ʌ": 269,
+    "ʑ": 270,
+    "ʚ": 271,
+    "ʛ": 272,
+    "ʜ": 273,
+    "ʞ": 274,
+    "ʟ": 275,
+    "ʬ": 276,
+    "ʹ": 277,
+    "ʺ": 278,
+    "ʻ": 279,
+    "ʼ": 280,
+    "ʾ": 281,
+    "ʿ": 282,
+    "˚": 283,
+    "̀": 284,
+    "́": 285,
+    "̂": 286,
+    "̃": 287,
+    "̄": 288,
+    "̅": 289,
+    "̆": 290,
+    "̇": 291,
+    "̈": 292,
+    "̊": 293,
+    "̋": 294,
+    "̐": 295,
+    "̑": 296,
+    "̕": 297,
+    "̛": 298,
+    "̜": 299,
+    "̝": 300,
+    "̠": 301,
+    "̣": 302,
+    "̤": 303,
+    "̥": 304,
+    "̦": 305,
+    "̧": 306,
+    "̨": 307,
+    "̩": 308,
+    "̫": 309,
+    "̭": 310,
+    "̮": 311,
+    "̯": 312,
+    "̰": 313,
+    "̱": 314,
+    "̲": 315,
+    "̳": 316,
+    "̴": 317,
+    "̵": 318,
+    "̶": 319,
+    "̷": 320,
+    "̸": 321,
+    "̺": 322,
+    "̽": 323,
+    "̾": 324,
+    "̿": 325,
+    "͂": 326,
+    "͆": 327,
+    "͇": 328,
+    "͋": 329,
+    "͍": 330,
+    "͎": 331,
+    "͏": 332,
+    "͒": 333,
+    "͓": 334,
+    "͙": 335,
+    "͜": 336,
+    "͝": 337,
+    "͞": 338,
+    "͟": 339,
+    "͠": 340,
+    "͡": 341,
+    "͢": 342,
+    "ͦ": 343,
+    "ͫ": 344,
+    "΄": 345,
+    "ΐ": 346,
+    "ά": 347,
+    "έ": 348,
+    "ή": 349,
+    "ί": 350,
+    "α": 351,
+    "β": 352,
+    "γ": 353,
+    "δ": 354,
+    "ε": 355,
+    "ζ": 356,
+    "η": 357,
+    "θ": 358,
+    "ι": 359,
+    "κ": 360,
+    "λ": 361,
+    "μ": 362,
+    "ν": 363,
+    "ξ": 364,
+    "ο": 365,
+    "π": 366,
+    "ρ": 367,
+    "ς": 368,
+    "σ": 369,
+    "τ": 370,
+    "υ": 371,
+    "φ": 372,
+    "χ": 373,
+    "ψ": 374,
+    "ω": 375,
+    "ϊ": 376,
+    "ό": 377,
+    "ύ": 378,
+    "ώ": 379,
+    "а": 380,
+    "б": 381,
+    "в": 382,
+    "г": 383,
+    "д": 384,
+    "е": 385,
+    "ж": 386,
+    "з": 387,
+    "и": 388,
+    "й": 389,
+    "к": 390,
+    "л": 391,
+    "м": 392,
+    "н": 393,
+    "о": 394,
+    "п": 395,
+    "р": 396,
+    "с": 397,
+    "т": 398,
+    "у": 399,
+    "ф": 400,
+    "х": 401,
+    "ц": 402,
+    "ч": 403,
+    "ш": 404,
+    "щ": 405,
+    "ъ": 406,
+    "ы": 407,
+    "ь": 408,
+    "э": 409,
+    "ю": 410,
+    "я": 411,
+    "ё": 412,
+    "ђ": 413,
+    "ѓ": 414,
+    "є": 415,
+    "ѕ": 416,
+    "і": 417,
+    "ї": 418,
+    "ј": 419,
+    "љ": 420,
+    "њ": 421,
+    "ћ": 422,
+    "ќ": 423,
+    "ѝ": 424,
+    "ў": 425,
+    "џ": 426,
+    "ѣ": 427,
+    "ѧ": 428,
+    "ѽ": 429,
+    "҈": 430,
+    "҉": 431,
+    "ґ": 432,
+    "ғ": 433,
+    "қ": 434,
+    "ң": 435,
+    "ҩ": 436,
+    "ү": 437,
+    "ұ": 438,
+    "ҳ": 439,
+    "ә": 440,
+    "ө": 441,
+    "ԁ": 442,
+    "ՙ": 443,
+    "ա": 444,
+    "բ": 445,
+    "գ": 446,
+    "դ": 447,
+    "ե": 448,
+    "զ": 449,
+    "է": 450,
+    "ը": 451,
+    "թ": 452,
+    "ժ": 453,
+    "ի": 454,
+    "լ": 455,
+    "խ": 456,
+    "ծ": 457,
+    "կ": 458,
+    "հ": 459,
+    "ձ": 460,
+    "ղ": 461,
+    "ճ": 462,
+    "մ": 463,
+    "յ": 464,
+    "ն": 465,
+    "շ": 466,
+    "ո": 467,
+    "չ": 468,
+    "պ": 469,
+    "ջ": 470,
+    "ռ": 471,
+    "ս": 472,
+    "վ": 473,
+    "տ": 474,
+    "ր": 475,
+    "ց": 476,
+    "ւ": 477,
+    "փ": 478,
+    "ք": 479,
+    "օ": 480,
+    "ֆ": 481,
+    "և": 482,
+    "ְ": 483,
+    "ִ": 484,
+    "ַ": 485,
+    "ָ": 486,
+    "ֹ": 487,
+    "ּ": 488,
+    "ׁ": 489,
+    "א": 490,
+    "ב": 491,
+    "ג": 492,
+    "ד": 493,
+    "ה": 494,
+    "ו": 495,
+    "ז": 496,
+    "ח": 497,
+    "ט": 498,
+    "י": 499,
+    "ך": 500,
+    "כ": 501,
+    "ל": 502,
+    "ם": 503,
+    "מ": 504,
+    "ן": 505,
+    "נ": 506,
+    "ס": 507,
+    "ע": 508,
+    "ף": 509,
+    "פ": 510,
+    "ץ": 511,
+    "צ": 512,
+    "ק": 513,
+    "ר": 514,
+    "ש": 515,
+    "ת": 516,
+    "׳": 517,
+    "״": 518,
+    "،": 519,
+    "ء": 520,
+    "آ": 521,
+    "أ": 522,
+    "ؤ": 523,
+    "إ": 524,
+    "ئ": 525,
+    "ا": 526,
+    "ب": 527,
+    "ة": 528,
+    "ت": 529,
+    "ث": 530,
+    "ج": 531,
+    "ح": 532,
+    "خ": 533,
+    "د": 534,
+    "ذ": 535,
+    "ر": 536,
+    "ز": 537,
+    "س": 538,
+    "ش": 539,
+    "ص": 540,
+    "ض": 541,
+    "ط": 542,
+    "ظ": 543,
+    "ع": 544,
+    "غ": 545,
+    "ـ": 546,
+    "ف": 547,
+    "ق": 548,
+    "ك": 549,
+    "ل": 550,
+    "م": 551,
+    "ن": 552,
+    "ه": 553,
+    "و": 554,
+    "ى": 555,
+    "ي": 556,
+    "ً": 557,
+    "ٌ": 558,
+    "ٍ": 559,
+    "َ": 560,
+    "ُ": 561,
+    "ِ": 562,
+    "ّ": 563,
+    "ْ": 564,
+    "ٓ": 565,
+    "ٕ": 566,
+    "ٖ": 567,
+    "ٗ": 568,
+    "٘": 569,
+    "ٛ": 570,
+    "٭": 571,
+    "ٰ": 572,
+    "ٱ": 573,
+    "ٲ": 574,
+    "ٳ": 575,
+    "ٴ": 576,
+    "ٵ": 577,
+    "ٷ": 578,
+    "ٹ": 579,
+    "ٺ": 580,
+    "ٻ": 581,
+    "ټ": 582,
+    "پ": 583,
+    "ٿ": 584,
+    "ڀ": 585,
+    "ځ": 586,
+    "ڂ": 587,
+    "ڃ": 588,
+    "څ": 589,
+    "چ": 590,
+    "ڇ": 591,
+    "ڈ": 592,
+    "ډ": 593,
+    "ڊ": 594,
+    "ڍ": 595,
+    "ڑ": 596,
+    "ڒ": 597,
+    "ړ": 598,
+    "ڔ": 599,
+    "ڕ": 600,
+    "ږ": 601,
+    "ڗ": 602,
+    "ژ": 603,
+    "ښ": 604,
+    "ڛ": 605,
+    "ڜ": 606,
+    "ڝ": 607,
+    "ڠ": 608,
+    "ڣ": 609,
+    "ڤ": 610,
+    "ڦ": 611,
+    "ڨ": 612,
+    "ک": 613,
+    "ڪ": 614,
+    "ګ": 615,
+    "گ": 616,
+    "ڰ": 617,
+    "ڱ": 618,
+    "ڲ": 619,
+    "ڳ": 620,
+    "ڴ": 621,
+    "ڵ": 622,
+    "ڶ": 623,
+    "ڷ": 624,
+    "ڸ": 625,
+    "ڹ": 626,
+    "ں": 627,
+    "ڻ": 628,
+    "ڼ": 629,
+    "ھ": 630,
+    "ہ": 631,
+    "ۂ": 632,
+    "ۃ": 633,
+    "ۄ": 634,
+    "ۅ": 635,
+    "ۆ": 636,
+    "ۇ": 637,
+    "ۈ": 638,
+    "ۉ": 639,
+    "ۋ": 640,
+    "ی": 641,
+    "ۍ": 642,
+    "ێ": 643,
+    "ې": 644,
+    "ۑ": 645,
+    "ے": 646,
+    "ۓ": 647,
+    "۔": 648,
+    "ە": 649,
+    "ۖ": 650,
+    "ۗ": 651,
+    "ۘ": 652,
+    "ۙ": 653,
+    "ۚ": 654,
+    "ۛ": 655,
+    "ۜ": 656,
+    "ۣ": 657,
+    "ۥ": 658,
+    "ۦ": 659,
+    "ۧ": 660,
+    "ۨ": 661,
+    "۪": 662,
+    "۫": 663,
+    "۬": 664,
+    "ۭ": 665,
+    "ۺ": 666,
+    "ۼ": 667,
+    "ۿ": 668,
+    "ܐ": 669,
+    "ܓ": 670,
+    "ܔ": 671,
+    "ܛ": 672,
+    "ܟ": 673,
+    "ݐ": 674,
+    "ݛ": 675,
+    "ݩ": 676,
+    "ݪ": 677,
+    "ހ": 678,
+    "ނ": 679,
+    "ރ": 680,
+    "ބ": 681,
+    "އ": 682,
+    "މ": 683,
+    "ފ": 684,
+    "ދ": 685,
+    "ތ": 686,
+    "ލ": 687,
+    "ސ": 688,
+    "ޒ": 689,
+    "ޗ": 690,
+    "ޙ": 691,
+    "ޝ": 692,
+    "ޢ": 693,
+    "ަ": 694,
+    "ާ": 695,
+    "ި": 696,
+    "ީ": 697,
+    "ު": 698,
+    "ް": 699,
+    "ࣰ": 700,
+    "ँ": 701,
+    "ं": 702,
+    "अ": 703,
+    "आ": 704,
+    "इ": 705,
+    "ई": 706,
+    "उ": 707,
+    "ऊ": 708,
+    "ऋ": 709,
+    "ए": 710,
+    "ऐ": 711,
+    "ओ": 712,
+    "क": 713,
+    "ख": 714,
+    "ग": 715,
+    "घ": 716,
+    "ङ": 717,
+    "च": 718,
+    "छ": 719,
+    "ज": 720,
+    "झ": 721,
+    "ञ": 722,
+    "ट": 723,
+    "ठ": 724,
+    "ड": 725,
+    "ढ": 726,
+    "ण": 727,
+    "त": 728,
+    "थ": 729,
+    "द": 730,
+    "ध": 731,
+    "न": 732,
+    "प": 733,
+    "फ": 734,
+    "ब": 735,
+    "भ": 736,
+    "म": 737,
+    "य": 738,
+    "र": 739,
+    "ल": 740,
+    "ळ": 741,
+    "व": 742,
+    "श": 743,
+    "ष": 744,
+    "स": 745,
+    "ह": 746,
+    "़": 747,
+    "ा": 748,
+    "ि": 749,
+    "ी": 750,
+    "ु": 751,
+    "ू": 752,
+    "ृ": 753,
+    "े": 754,
+    "ै": 755,
+    "ॉ": 756,
+    "ो": 757,
+    "ौ": 758,
+    "्": 759,
+    "।": 760,
+    "॰": 761,
+    "ঁ": 762,
+    "ং": 763,
+    "ঃ": 764,
+    "অ": 765,
+    "আ": 766,
+    "ই": 767,
+    "ঈ": 768,
+    "উ": 769,
+    "ঊ": 770,
+    "ঋ": 771,
+    "এ": 772,
+    "ঐ": 773,
+    "ও": 774,
+    "ক": 775,
+    "খ": 776,
+    "গ": 777,
+    "ঘ": 778,
+    "ঙ": 779,
+    "চ": 780,
+    "ছ": 781,
+    "জ": 782,
+    "ঝ": 783,
+    "ঞ": 784,
+    "ট": 785,
+    "ঠ": 786,
+    "ড": 787,
+    "ঢ": 788,
+    "ণ": 789,
+    "ত": 790,
+    "থ": 791,
+    "দ": 792,
+    "ধ": 793,
+    "ন": 794,
+    "প": 795,
+    "ফ": 796,
+    "ব": 797,
+    "ভ": 798,
+    "ম": 799,
+    "য": 800,
+    "র": 801,
+    "ল": 802,
+    "শ": 803,
+    "ষ": 804,
+    "স": 805,
+    "হ": 806,
+    "়": 807,
+    "া": 808,
+    "ি": 809,
+    "ী": 810,
+    "ু": 811,
+    "ূ": 812,
+    "ৃ": 813,
+    "ে": 814,
+    "ৈ": 815,
+    "ো": 816,
+    "ৌ": 817,
+    "্": 818,
+    "ৎ": 819,
+    "ৰ": 820,
+    "ৱ": 821,
+    "ਂ": 822,
+    "ਅ": 823,
+    "ਆ": 824,
+    "ਕ": 825,
+    "ਖ": 826,
+    "ਗ": 827,
+    "ਘ": 828,
+    "ਚ": 829,
+    "ਜ": 830,
+    "ਟ": 831,
+    "ਡ": 832,
+    "ਢ": 833,
+    "ਣ": 834,
+    "ਤ": 835,
+    "ਦ": 836,
+    "ਧ": 837,
+    "ਨ": 838,
+    "ਪ": 839,
+    "ਫ": 840,
+    "ਬ": 841,
+    "ਭ": 842,
+    "ਮ": 843,
+    "ਯ": 844,
+    "ਰ": 845,
+    "ਲ": 846,
+    "ਵ": 847,
+    "ਸ": 848,
+    "ਹ": 849,
+    "਼": 850,
+    "ਾ": 851,
+    "ਿ": 852,
+    "ੀ": 853,
+    "ੁ": 854,
+    "ੂ": 855,
+    "ੇ": 856,
+    "ੈ": 857,
+    "ੋ": 858,
+    "ੌ": 859,
+    "੍": 860,
+    "ੜ": 861,
+    "ੰ": 862,
+    "ੱ": 863,
+    "ં": 864,
+    "ઈ": 865,
+    "ક": 866,
+    "ગ": 867,
+    "જ": 868,
+    "ટ": 869,
+    "ઠ": 870,
+    "ડ": 871,
+    "ણ": 872,
+    "ત": 873,
+    "દ": 874,
+    "ન": 875,
+    "પ": 876,
+    "બ": 877,
+    "ભ": 878,
+    "મ": 879,
+    "ય": 880,
+    "ર": 881,
+    "લ": 882,
+    "વ": 883,
+    "શ": 884,
+    "સ": 885,
+    "હ": 886,
+    "ા": 887,
+    "િ": 888,
+    "ી": 889,
+    "ુ": 890,
+    "ે": 891,
+    "ો": 892,
+    "્": 893,
+    "ା": 894,
+    "୍": 895,
+    "அ": 896,
+    "ஆ": 897,
+    "இ": 898,
+    "உ": 899,
+    "எ": 900,
+    "ஒ": 901,
+    "க": 902,
+    "ங": 903,
+    "ச": 904,
+    "ஜ": 905,
+    "ஞ": 906,
+    "ட": 907,
+    "ண": 908,
+    "த": 909,
+    "ந": 910,
+    "ன": 911,
+    "ப": 912,
+    "ம": 913,
+    "ய": 914,
+    "ர": 915,
+    "ற": 916,
+    "ல": 917,
+    "ள": 918,
+    "ழ": 919,
+    "வ": 920,
+    "ஷ": 921,
+    "ஸ": 922,
+    "ஹ": 923,
+    "ா": 924,
+    "ி": 925,
+    "ீ": 926,
+    "ு": 927,
+    "ூ": 928,
+    "ெ": 929,
+    "ே": 930,
+    "ை": 931,
+    "ொ": 932,
+    "ோ": 933,
+    "்": 934,
+    "ం": 935,
+    "అ": 936,
+    "ఆ": 937,
+    "క": 938,
+    "గ": 939,
+    "చ": 940,
+    "జ": 941,
+    "ట": 942,
+    "డ": 943,
+    "ణ": 944,
+    "త": 945,
+    "ద": 946,
+    "ధ": 947,
+    "న": 948,
+    "ప": 949,
+    "బ": 950,
+    "భ": 951,
+    "మ": 952,
+    "య": 953,
+    "ర": 954,
+    "ల": 955,
+    "వ": 956,
+    "శ": 957,
+    "ష": 958,
+    "స": 959,
+    "హ": 960,
+    "ా": 961,
+    "ి": 962,
+    "ీ": 963,
+    "ు": 964,
+    "ూ": 965,
+    "ె": 966,
+    "ే": 967,
+    "ో": 968,
+    "్": 969,
+    "ಂ": 970,
+    "ಌ": 971,
+    "ಕ": 972,
+    "ಗ": 973,
+    "ಜ": 974,
+    "ಟ": 975,
+    "ಡ": 976,
+    "ತ": 977,
+    "ದ": 978,
+    "ನ": 979,
+    "ಪ": 980,
+    "ಬ": 981,
+    "ಮ": 982,
+    "ಯ": 983,
+    "ರ": 984,
+    "ಲ": 985,
+    "ವ": 986,
+    "ಶ": 987,
+    "ಸ": 988,
+    "ಹ": 989,
+    "ಾ": 990,
+    "ಿ": 991,
+    "ೀ": 992,
+    "ು": 993,
+    "ೆ": 994,
+    "ೇ": 995,
+    "ೋ": 996,
+    "್": 997,
+    "ം": 998,
+    "അ": 999,
+    "ആ": 1000,
+    "ഇ": 1001,
+    "ഉ": 1002,
+    "എ": 1003,
+    "ഒ": 1004,
+    "ക": 1005,
+    "ഖ": 1006,
+    "ഗ": 1007,
+    "ങ": 1008,
+    "ച": 1009,
+    "ജ": 1010,
+    "ഞ": 1011,
+    "ട": 1012,
+    "ഡ": 1013,
+    "ണ": 1014,
+    "ത": 1015,
+    "ഥ": 1016,
+    "ദ": 1017,
+    "ധ": 1018,
+    "ന": 1019,
+    "പ": 1020,
+    "ഫ": 1021,
+    "ബ": 1022,
+    "ഭ": 1023,
+    "മ": 1024,
+    "യ": 1025,
+    "ര": 1026,
+    "റ": 1027,
+    "ല": 1028,
+    "ള": 1029,
+    "ഴ": 1030,
+    "വ": 1031,
+    "ശ": 1032,
+    "ഷ": 1033,
+    "സ": 1034,
+    "ഹ": 1035,
+    "ാ": 1036,
+    "ി": 1037,
+    "ീ": 1038,
+    "ു": 1039,
+    "ൂ": 1040,
+    "ൃ": 1041,
+    "െ": 1042,
+    "േ": 1043,
+    "ൈ": 1044,
+    "ൊ": 1045,
+    "ോ": 1046,
+    "്": 1047,
+    "ൺ": 1048,
+    "ൻ": 1049,
+    "ർ": 1050,
+    "ൽ": 1051,
+    "ൾ": 1052,
+    "ං": 1053,
+    "අ": 1054,
+    "ආ": 1055,
+    "ක": 1056,
+    "ග": 1057,
+    "ච": 1058,
+    "ජ": 1059,
+    "ට": 1060,
+    "ඩ": 1061,
+    "ණ": 1062,
+    "ත": 1063,
+    "ද": 1064,
+    "න": 1065,
+    "ප": 1066,
+    "බ": 1067,
+    "ම": 1068,
+    "ය": 1069,
+    "ර": 1070,
+    "ල": 1071,
+    "ව": 1072,
+    "ශ": 1073,
+    "ෂ": 1074,
+    "ස": 1075,
+    "හ": 1076,
+    "්": 1077,
+    "ා": 1078,
+    "ැ": 1079,
+    "ි": 1080,
+    "ී": 1081,
+    "ු": 1082,
+    "ෙ": 1083,
+    "ේ": 1084,
+    "ො": 1085,
+    "ෝ": 1086,
+    "ก": 1087,
+    "ข": 1088,
+    "ค": 1089,
+    "ฆ": 1090,
+    "ง": 1091,
+    "จ": 1092,
+    "ฉ": 1093,
+    "ช": 1094,
+    "ซ": 1095,
+    "ญ": 1096,
+    "ฎ": 1097,
+    "ฏ": 1098,
+    "ฐ": 1099,
+    "ฑ": 1100,
+    "ฒ": 1101,
+    "ณ": 1102,
+    "ด": 1103,
+    "ต": 1104,
+    "ถ": 1105,
+    "ท": 1106,
+    "ธ": 1107,
+    "น": 1108,
+    "บ": 1109,
+    "ป": 1110,
+    "ผ": 1111,
+    "ฝ": 1112,
+    "พ": 1113,
+    "ฟ": 1114,
+    "ภ": 1115,
+    "ม": 1116,
+    "ย": 1117,
+    "ร": 1118,
+    "ฤ": 1119,
+    "ล": 1120,
+    "ว": 1121,
+    "ศ": 1122,
+    "ษ": 1123,
+    "ส": 1124,
+    "ห": 1125,
+    "ฬ": 1126,
+    "อ": 1127,
+    "ฮ": 1128,
+    "ะ": 1129,
+    "ั": 1130,
+    "า": 1131,
+    "ำ": 1132,
+    "ิ": 1133,
+    "ี": 1134,
+    "ึ": 1135,
+    "ื": 1136,
+    "ุ": 1137,
+    "ู": 1138,
+    "ฺ": 1139,
+    "เ": 1140,
+    "แ": 1141,
+    "โ": 1142,
+    "ใ": 1143,
+    "ไ": 1144,
+    "ๆ": 1145,
+    "็": 1146,
+    "่": 1147,
+    "้": 1148,
+    "๊": 1149,
+    "๋": 1150,
+    "์": 1151,
+    "ํ": 1152,
+    "ກ": 1153,
+    "ຄ": 1154,
+    "ງ": 1155,
+    "ຈ": 1156,
+    "ຊ": 1157,
+    "ຍ": 1158,
+    "ດ": 1159,
+    "ຕ": 1160,
+    "ທ": 1161,
+    "ນ": 1162,
+    "ບ": 1163,
+    "ພ": 1164,
+    "ມ": 1165,
+    "ລ": 1166,
+    "ວ": 1167,
+    "ສ": 1168,
+    "ຫ": 1169,
+    "ອ": 1170,
+    "ະ": 1171,
+    "ັ": 1172,
+    "າ": 1173,
+    "ຳ": 1174,
+    "ິ": 1175,
+    "ີ": 1176,
+    "ຸ": 1177,
+    "ົ": 1178,
+    "ເ": 1179,
+    "ແ": 1180,
+    "ໂ": 1181,
+    "ໄ": 1182,
+    "່": 1183,
+    "້": 1184,
+    "ໍ": 1185,
+    "་": 1186,
+    "།": 1187,
+    "ག": 1188,
+    "ང": 1189,
+    "ད": 1190,
+    "ན": 1191,
+    "བ": 1192,
+    "མ": 1193,
+    "འ": 1194,
+    "ར": 1195,
+    "ལ": 1196,
+    "ས": 1197,
+    "ི": 1198,
+    "ུ": 1199,
+    "ེ": 1200,
+    "ོ": 1201,
+    "ྱ": 1202,
+    "က": 1203,
+    "ခ": 1204,
+    "ဂ": 1205,
+    "င": 1206,
+    "စ": 1207,
+    "ဆ": 1208,
+    "ဇ": 1209,
+    "ည": 1210,
+    "တ": 1211,
+    "ထ": 1212,
+    "ဒ": 1213,
+    "န": 1214,
+    "ပ": 1215,
+    "ဖ": 1216,
+    "ဗ": 1217,
+    "ဘ": 1218,
+    "မ": 1219,
+    "ယ": 1220,
+    "ရ": 1221,
+    "လ": 1222,
+    "ဝ": 1223,
+    "သ": 1224,
+    "ဟ": 1225,
+    "အ": 1226,
+    "ဥ": 1227,
+    "ဦ": 1228,
+    "ါ": 1229,
+    "ာ": 1230,
+    "ိ": 1231,
+    "ီ": 1232,
+    "ု": 1233,
+    "ူ": 1234,
+    "ေ": 1235,
+    "ဲ": 1236,
+    "ဳ": 1237,
+    "ဴ": 1238,
+    "ံ": 1239,
+    "့": 1240,
+    "း": 1241,
+    "္": 1242,
+    "်": 1243,
+    "ျ": 1244,
+    "ြ": 1245,
+    "ွ": 1246,
+    "ၥ": 1247,
+    "ၾ": 1248,
+    "ႀ": 1249,
+    "ႊ": 1250,
+    "ႏ": 1251,
+    "ა": 1252,
+    "ბ": 1253,
+    "გ": 1254,
+    "დ": 1255,
+    "ე": 1256,
+    "ვ": 1257,
+    "ზ": 1258,
+    "თ": 1259,
+    "ი": 1260,
+    "კ": 1261,
+    "ლ": 1262,
+    "მ": 1263,
+    "ნ": 1264,
+    "ო": 1265,
+    "პ": 1266,
+    "ჟ": 1267,
+    "რ": 1268,
+    "ს": 1269,
+    "ტ": 1270,
+    "უ": 1271,
+    "ფ": 1272,
+    "ქ": 1273,
+    "ღ": 1274,
+    "ყ": 1275,
+    "შ": 1276,
+    "ჩ": 1277,
+    "ც": 1278,
+    "ძ": 1279,
+    "წ": 1280,
+    "ჭ": 1281,
+    "ხ": 1282,
+    "ჯ": 1283,
+    "ჰ": 1284,
+    "ሀ": 1285,
+    "ሁ": 1286,
+    "ሂ": 1287,
+    "ሃ": 1288,
+    "ሄ": 1289,
+    "ህ": 1290,
+    "ሆ": 1291,
+    "ለ": 1292,
+    "ሉ": 1293,
+    "ሊ": 1294,
+    "ላ": 1295,
+    "ሌ": 1296,
+    "ል": 1297,
+    "ሎ": 1298,
+    "ሏ": 1299,
+    "ሐ": 1300,
+    "ሓ": 1301,
+    "ሔ": 1302,
+    "ሕ": 1303,
+    "መ": 1304,
+    "ሙ": 1305,
+    "ሚ": 1306,
+    "ማ": 1307,
+    "ሜ": 1308,
+    "ም": 1309,
+    "ሞ": 1310,
+    "ሠ": 1311,
+    "ሡ": 1312,
+    "ሣ": 1313,
+    "ሥ": 1314,
+    "ረ": 1315,
+    "ሩ": 1316,
+    "ሪ": 1317,
+    "ራ": 1318,
+    "ሬ": 1319,
+    "ር": 1320,
+    "ሮ": 1321,
+    "ሰ": 1322,
+    "ሱ": 1323,
+    "ሲ": 1324,
+    "ሳ": 1325,
+    "ሴ": 1326,
+    "ስ": 1327,
+    "ሶ": 1328,
+    "ሸ": 1329,
+    "ሺ": 1330,
+    "ሻ": 1331,
+    "ሽ": 1332,
+    "ቀ": 1333,
+    "ቁ": 1334,
+    "ቂ": 1335,
+    "ቃ": 1336,
+    "ቄ": 1337,
+    "ቅ": 1338,
+    "ቆ": 1339,
+    "ቋ": 1340,
+    "በ": 1341,
+    "ቡ": 1342,
+    "ቢ": 1343,
+    "ባ": 1344,
+    "ቤ": 1345,
+    "ብ": 1346,
+    "ቦ": 1347,
+    "ተ": 1348,
+    "ቱ": 1349,
+    "ቲ": 1350,
+    "ታ": 1351,
+    "ቴ": 1352,
+    "ቶ": 1353,
+    "ቷ": 1354,
+    "ቸ": 1355,
+    "ቻ": 1356,
+    "ች": 1357,
+    "ነ": 1358,
+    "ኑ": 1359,
+    "ኒ": 1360,
+    "ና": 1361,
+    "ኔ": 1362,
+    "ን": 1363,
+    "ኖ": 1364,
+    "ኛ": 1365,
+    "ኝ": 1366,
+    "አ": 1367,
+    "ኡ": 1368,
+    "ኢ": 1369,
+    "ኣ": 1370,
+    "ኤ": 1371,
+    "እ": 1372,
+    "ኦ": 1373,
+    "ከ": 1374,
+    "ኩ": 1375,
+    "ኪ": 1376,
+    "ካ": 1377,
+    "ኬ": 1378,
+    "ክ": 1379,
+    "ኮ": 1380,
+    "ኸ": 1381,
+    "ኽ": 1382,
+    "ወ": 1383,
+    "ዉ": 1384,
+    "ዊ": 1385,
+    "ዋ": 1386,
+    "ው": 1387,
+    "ዎ": 1388,
+    "ዑ": 1389,
+    "ዓ": 1390,
+    "ዕ": 1391,
+    "ዘ": 1392,
+    "ዙ": 1393,
+    "ዚ": 1394,
+    "ዛ": 1395,
+    "ዜ": 1396,
+    "ዝ": 1397,
+    "ዞ": 1398,
+    "የ": 1399,
+    "ዩ": 1400,
+    "ያ": 1401,
+    "ዬ": 1402,
+    "ይ": 1403,
+    "ዮ": 1404,
+    "ደ": 1405,
+    "ዱ": 1406,
+    "ዲ": 1407,
+    "ዳ": 1408,
+    "ዴ": 1409,
+    "ድ": 1410,
+    "ዶ": 1411,
+    "ጀ": 1412,
+    "ጂ": 1413,
+    "ጃ": 1414,
+    "ጄ": 1415,
+    "ጅ": 1416,
+    "ጆ": 1417,
+    "ገ": 1418,
+    "ጉ": 1419,
+    "ጊ": 1420,
+    "ጋ": 1421,
+    "ጌ": 1422,
+    "ግ": 1423,
+    "ጎ": 1424,
+    "ጠ": 1425,
+    "ጡ": 1426,
+    "ጣ": 1427,
+    "ጤ": 1428,
+    "ጥ": 1429,
+    "ጦ": 1430,
+    "ጨ": 1431,
+    "ጭ": 1432,
+    "ጸ": 1433,
+    "ጽ": 1434,
+    "ፀ": 1435,
+    "ፅ": 1436,
+    "ፈ": 1437,
+    "ፉ": 1438,
+    "ፊ": 1439,
+    "ፋ": 1440,
+    "ፌ": 1441,
+    "ፍ": 1442,
+    "ፎ": 1443,
+    "፡": 1444,
+    "።": 1445,
+    "ᏸ": 1446,
+    "ᐝ": 1447,
+    "ᐸ": 1448,
+    "ᓲ": 1449,
+    "ᕮ": 1450,
+    "ᕱ": 1451,
+    "ᕼ": 1452,
+    "ᖆ": 1453,
+    "ᖇ": 1454,
+    "ᖱ": 1455,
+    "ᖶ": 1456,
+    "ᖺ": 1457,
+    "ᗅ": 1458,
+    "ᗗ": 1459,
+    "ᗩ": 1460,
+    "ᗰ": 1461,
+    "ᘉ": 1462,
+    "ᘎ": 1463,
+    "ᘙ": 1464,
+    "ᘜ": 1465,
+    "ᙢ": 1466,
+    "ᛁ": 1467,
+    "ᛟ": 1468,
+    "ᜒ": 1469,
+    "᜔": 1470,
+    "ក": 1471,
+    "ខ": 1472,
+    "គ": 1473,
+    "ង": 1474,
+    "ច": 1475,
+    "ឆ": 1476,
+    "ជ": 1477,
+    "ញ": 1478,
+    "ដ": 1479,
+    "ណ": 1480,
+    "ត": 1481,
+    "ថ": 1482,
+    "ទ": 1483,
+    "ធ": 1484,
+    "ន": 1485,
+    "ប": 1486,
+    "ផ": 1487,
+    "ព": 1488,
+    "ភ": 1489,
+    "ម": 1490,
+    "យ": 1491,
+    "រ": 1492,
+    "ល": 1493,
+    "វ": 1494,
+    "ស": 1495,
+    "ហ": 1496,
+    "ឡ": 1497,
+    "អ": 1498,
+    "ា": 1499,
+    "ិ": 1500,
+    "ី": 1501,
+    "ឹ": 1502,
+    "ុ": 1503,
+    "ូ": 1504,
+    "ួ": 1505,
+    "ើ": 1506,
+    "ឿ": 1507,
+    "ៀ": 1508,
+    "េ": 1509,
+    "ែ": 1510,
+    "ោ": 1511,
+    "ៅ": 1512,
+    "ំ": 1513,
+    "ះ": 1514,
+    "៉": 1515,
+    "៊": 1516,
+    "់": 1517,
+    "័": 1518,
+    "្": 1519,
+    "ḁ": 1520,
+    "ḃ": 1521,
+    "ḅ": 1522,
+    "ḇ": 1523,
+    "ḉ": 1524,
+    "ḋ": 1525,
+    "ḍ": 1526,
+    "ḏ": 1527,
+    "ḑ": 1528,
+    "ḓ": 1529,
+    "ḕ": 1530,
+    "ḗ": 1531,
+    "ḛ": 1532,
+    "ḝ": 1533,
+    "ḟ": 1534,
+    "ḡ": 1535,
+    "ḣ": 1536,
+    "ḥ": 1537,
+    "ḧ": 1538,
+    "ḩ": 1539,
+    "ḫ": 1540,
+    "ḭ": 1541,
+    "ḯ": 1542,
+    "ḱ": 1543,
+    "ḳ": 1544,
+    "ḵ": 1545,
+    "ḷ": 1546,
+    "ḻ": 1547,
+    "ḿ": 1548,
+    "ṁ": 1549,
+    "ṃ": 1550,
+    "ṅ": 1551,
+    "ṇ": 1552,
+    "ṉ": 1553,
+    "ṍ": 1554,
+    "ṏ": 1555,
+    "ṓ": 1556,
+    "ṕ": 1557,
+    "ṗ": 1558,
+    "ṙ": 1559,
+    "ṛ": 1560,
+    "ṝ": 1561,
+    "ṟ": 1562,
+    "ṡ": 1563,
+    "ṣ": 1564,
+    "ṥ": 1565,
+    "ṧ": 1566,
+    "ṩ": 1567,
+    "ṫ": 1568,
+    "ṭ": 1569,
+    "ṯ": 1570,
+    "ṱ": 1571,
+    "ṳ": 1572,
+    "ṽ": 1573,
+    "ṿ": 1574,
+    "ẁ": 1575,
+    "ẃ": 1576,
+    "ẅ": 1577,
+    "ẋ": 1578,
+    "ẍ": 1579,
+    "ẏ": 1580,
+    "ẑ": 1581,
+    "ẓ": 1582,
+    "ẕ": 1583,
+    "ẖ": 1584,
+    "ẗ": 1585,
+    "ẙ": 1586,
+    "ạ": 1587,
+    "ả": 1588,
+    "ấ": 1589,
+    "ầ": 1590,
+    "ẩ": 1591,
+    "ẫ": 1592,
+    "ậ": 1593,
+    "ắ": 1594,
+    "ằ": 1595,
+    "ẳ": 1596,
+    "ẵ": 1597,
+    "ặ": 1598,
+    "ẹ": 1599,
+    "ẻ": 1600,
+    "ẽ": 1601,
+    "ế": 1602,
+    "ề": 1603,
+    "ể": 1604,
+    "ễ": 1605,
+    "ệ": 1606,
+    "ỉ": 1607,
+    "ị": 1608,
+    "ọ": 1609,
+    "ỏ": 1610,
+    "���": 1611,
+    "ồ": 1612,
+    "ổ": 1613,
+    "ỗ": 1614,
+    "ộ": 1615,
+    "ớ": 1616,
+    "ờ": 1617,
+    "ở": 1618,
+    "ỡ": 1619,
+    "ợ": 1620,
+    "ụ": 1621,
+    "ủ": 1622,
+    "ứ": 1623,
+    "ừ": 1624,
+    "ử": 1625,
+    "ữ": 1626,
+    "ự": 1627,
+    "ỳ": 1628,
+    "ỵ": 1629,
+    "ỷ": 1630,
+    "ỹ": 1631,
+    "‎": 1632,
+    "‏": 1633,
+    "‐": 1634,
+    "‑": 1635,
+    "–": 1636,
+    "‘": 1637,
+    "’": 1638,
+    "“": 1639,
+    "”": 1640,
+    "…": 1641,
+    "′": 1642,
+    "⃜": 1643,
+    "⃟": 1644,
+    "⃣": 1645,
+    "─": 1646,
+    "│": 1647,
+    "┐": 1648,
+    "┘": 1649,
+    "├": 1650,
+    "┤": 1651,
+    "┼": 1652,
+    "║": 1653,
+    "╗": 1654,
+    "╝": 1655,
+    "╣": 1656,
+    "░": 1657,
+    "▒": 1658,
+    "▓": 1659,
+    "ⱥ": 1660,
+    "ⴰ": 1661,
+    "ⴱ": 1662,
+    "ⴷ": 1663,
+    "ⴻ": 1664,
+    "ⵀ": 1665,
+    "ⵉ": 1666,
+    "ⵍ": 1667,
+    "ⵎ": 1668,
+    "ⵏ": 1669,
+    "ⵓ": 1670,
+    "ⵔ": 1671,
+    "ⵙ": 1672,
+    "ⵣ": 1673,
+    "々": 1674,
+    "ぁ": 1675,
+    "あ": 1676,
+    "ぃ": 1677,
+    "い": 1678,
+    "う": 1679,
+    "ぇ": 1680,
+    "え": 1681,
+    "お": 1682,
+    "か": 1683,
+    "が": 1684,
+    "き": 1685,
+    "ぎ": 1686,
+    "く": 1687,
+    "ぐ": 1688,
+    "け": 1689,
+    "げ": 1690,
+    "こ": 1691,
+    "ご": 1692,
+    "さ": 1693,
+    "ざ": 1694,
+    "し": 1695,
+    "じ": 1696,
+    "す": 1697,
+    "ず": 1698,
+    "せ": 1699,
+    "ぜ": 1700,
+    "そ": 1701,
+    "ぞ": 1702,
+    "た": 1703,
+    "だ": 1704,
+    "ち": 1705,
+    "っ": 1706,
+    "つ": 1707,
+    "づ": 1708,
+    "て": 1709,
+    "で": 1710,
+    "と": 1711,
+    "ど": 1712,
+    "な": 1713,
+    "に": 1714,
+    "ぬ": 1715,
+    "ね": 1716,
+    "の": 1717,
+    "は": 1718,
+    "ば": 1719,
+    "ぱ": 1720,
+    "ひ": 1721,
+    "び": 1722,
+    "ぴ": 1723,
+    "ふ": 1724,
+    "ぶ": 1725,
+    "ぷ": 1726,
+    "へ": 1727,
+    "べ": 1728,
+    "ぺ": 1729,
+    "ほ": 1730,
+    "ぼ": 1731,
+    "ぽ": 1732,
+    "ま": 1733,
+    "み": 1734,
+    "む": 1735,
+    "め": 1736,
+    "も": 1737,
+    "ゃ": 1738,
+    "や": 1739,
+    "ゅ": 1740,
+    "ゆ": 1741,
+    "ょ": 1742,
+    "よ": 1743,
+    "ら": 1744,
+    "り": 1745,
+    "る": 1746,
+    "れ": 1747,
+    "ろ": 1748,
+    "わ": 1749,
+    "を": 1750,
+    "ん": 1751,
+    "ァ": 1752,
+    "ア": 1753,
+    "ィ": 1754,
+    "イ": 1755,
+    "ゥ": 1756,
+    "ウ": 1757,
+    "ェ": 1758,
+    "エ": 1759,
+    "ォ": 1760,
+    "オ": 1761,
+    "カ": 1762,
+    "ガ": 1763,
+    "キ": 1764,
+    "ギ": 1765,
+    "ク": 1766,
+    "グ": 1767,
+    "ケ": 1768,
+    "ゲ": 1769,
+    "コ": 1770,
+    "ゴ": 1771,
+    "サ": 1772,
+    "ザ": 1773,
+    "シ": 1774,
+    "ジ": 1775,
+    "ス": 1776,
+    "ズ": 1777,
+    "セ": 1778,
+    "ゼ": 1779,
+    "ソ": 1780,
+    "ゾ": 1781,
+    "タ": 1782,
+    "ダ": 1783,
+    "チ": 1784,
+    "ヂ": 1785,
+    "ッ": 1786,
+    "ツ": 1787,
+    "ヅ": 1788,
+    "テ": 1789,
+    "デ": 1790,
+    "ト": 1791,
+    "ド": 1792,
+    "ナ": 1793,
+    "ニ": 1794,
+    "ヌ": 1795,
+    "ネ": 1796,
+    "ノ": 1797,
+    "ハ": 1798,
+    "バ": 1799,
+    "パ": 1800,
+    "ヒ": 1801,
+    "ビ": 1802,
+    "ピ": 1803,
+    "フ": 1804,
+    "ブ": 1805,
+    "プ": 1806,
+    "ヘ": 1807,
+    "ベ": 1808,
+    "ペ": 1809,
+    "ホ": 1810,
+    "ボ": 1811,
+    "ポ": 1812,
+    "マ": 1813,
+    "ミ": 1814,
+    "ム": 1815,
+    "メ": 1816,
+    "モ": 1817,
+    "ャ": 1818,
+    "ヤ": 1819,
+    "ュ": 1820,
+    "ユ": 1821,
+    "ョ": 1822,
+    "ヨ": 1823,
+    "ラ": 1824,
+    "リ": 1825,
+    "ル": 1826,
+    "レ": 1827,
+    "ロ": 1828,
+    "ワ": 1829,
+    "ン": 1830,
+    "ヴ": 1831,
+    "ヶ": 1832,
+    "・": 1833,
+    "ー": 1834,
+    "一": 1835,
+    "丁": 1836,
+    "七": 1837,
+    "万": 1838,
+    "丈": 1839,
+    "三": 1840,
+    "上": 1841,
+    "下": 1842,
+    "不": 1843,
+    "与": 1844,
+    "丕": 1845,
+    "世": 1846,
+    "丘": 1847,
+    "丙": 1848,
+    "业": 1849,
+    "丛": 1850,
+    "东": 1851,
+    "丝": 1852,
+    "丞": 1853,
+    "両": 1854,
+    "严": 1855,
+    "並": 1856,
+    "丫": 1857,
+    "中": 1858,
+    "丰": 1859,
+    "串": 1860,
+    "临": 1861,
+    "丸": 1862,
+    "丹": 1863,
+    "为": 1864,
+    "主": 1865,
+    "丽": 1866,
+    "举": 1867,
+    "乃": 1868,
+    "久": 1869,
+    "么": 1870,
+    "义": 1871,
+    "之": 1872,
+    "乌": 1873,
+    "乐": 1874,
+    "乔": 1875,
+    "乗": 1876,
+    "乘": 1877,
+    "乙": 1878,
+    "九": 1879,
+    "也": 1880,
+    "习": 1881,
+    "书": 1882,
+    "买": 1883,
+    "乾": 1884,
+    "亀": 1885,
+    "了": 1886,
+    "予": 1887,
+    "争": 1888,
+    "事": 1889,
+    "二": 1890,
+    "于": 1891,
+    "云": 1892,
+    "五": 1893,
+    "井": 1894,
+    "亘": 1895,
+    "亚": 1896,
+    "亜": 1897,
+    "亞": 1898,
+    "亠": 1899,
+    "交": 1900,
+    "亦": 1901,
+    "亨": 1902,
+    "享": 1903,
+    "京": 1904,
+    "亭": 1905,
+    "亮": 1906,
+    "人": 1907,
+    "亿": 1908,
+    "什": 1909,
+    "仁": 1910,
+    "仇": 1911,
+    "今": 1912,
+    "介": 1913,
+    "从": 1914,
+    "仔": 1915,
+    "仕": 1916,
+    "付": 1917,
+    "仙": 1918,
+    "仟": 1919,
+    "代": 1920,
+    "令": 1921,
+    "以": 1922,
+    "仪": 1923,
+    "仰": 1924,
+    "仲": 1925,
+    "任": 1926,
+    "企": 1927,
+    "伃": 1928,
+    "伊": 1929,
+    "伍": 1930,
+    "伏": 1931,
+    "众": 1932,
+    "优": 1933,
+    "伙": 1934,
+    "会": 1935,
+    "伝": 1936,
+    "伟": 1937,
+    "传": 1938,
+    "伦": 1939,
+    "伯": 1940,
+    "伴": 1941,
+    "伶": 1942,
+    "伸": 1943,
+    "似": 1944,
+    "伽": 1945,
+    "位": 1946,
+    "住": 1947,
+    "佐": 1948,
+    "佑": 1949,
+    "体": 1950,
+    "何": 1951,
+    "余": 1952,
+    "佛": 1953,
+    "作": 1954,
+    "佟": 1955,
+    "佩": 1956,
+    "佰": 1957,
+    "佳": 1958,
+    "佶": 1959,
+    "佾": 1960,
+    "侃": 1961,
+    "來": 1962,
+    "侍": 1963,
+    "侑": 1964,
+    "侖": 1965,
+    "依": 1966,
+    "侠": 1967,
+    "侨": 1968,
+    "侯": 1969,
+    "俊": 1970,
+    "俏": 1971,
+    "俐": 1972,
+    "保": 1973,
+    "俞": 1974,
+    "俠": 1975,
+    "信": 1976,
+    "俣": 1977,
+    "俪": 1978,
+    "俭": 1979,
+    "修": 1980,
+    "俳": 1981,
+    "俵": 1982,
+    "倉": 1983,
+    "個": 1984,
+    "倍": 1985,
+    "倖": 1986,
+    "倚": 1987,
+    "倢": 1988,
+    "倩": 1989,
+    "倪": 1990,
+    "倫": 1991,
+    "倬": 1992,
+    "假": 1993,
+    "偉": 1994,
+    "偌": 1995,
+    "健": 1996,
+    "偲": 1997,
+    "側": 1998,
+    "偵": 1999,
+    "傅": 2000,
+    "傑": 2001,
+    "備": 2002,
+    "傲": 2003,
+    "傳": 2004,
+    "像": 2005,
+    "僑": 2006,
+    "僖": 2007,
+    "僚": 2008,
+    "儀": 2009,
+    "億": 2010,
+    "儉": 2011,
+    "儒": 2012,
+    "優": 2013,
+    "儷": 2014,
+    "儿": 2015,
+    "允": 2016,
+    "元": 2017,
+    "兄": 2018,
+    "充": 2019,
+    "兆": 2020,
+    "先": 2021,
+    "光": 2022,
+    "克": 2023,
+    "児": 2024,
+    "兒": 2025,
+    "兔": 2026,
+    "兜": 2027,
+    "入": 2028,
+    "全": 2029,
+    "兩": 2030,
+    "八": 2031,
+    "公": 2032,
+    "六": 2033,
+    "兰": 2034,
+    "共": 2035,
+    "关": 2036,
+    "兴": 2037,
+    "兵": 2038,
+    "其": 2039,
+    "具": 2040,
+    "典": 2041,
+    "养": 2042,
+    "兼": 2043,
+    "冀": 2044,
+    "内": 2045,
+    "円": 2046,
+    "冉": 2047,
+    "再": 2048,
+    "军": 2049,
+    "农": 2050,
+    "冠": 2051,
+    "冨": 2052,
+    "冬": 2053,
+    "冯": 2054,
+    "冰": 2055,
+    "冲": 2056,
+    "冴": 2057,
+    "冶": 2058,
+    "冷": 2059,
+    "准": 2060,
+    "凉": 2061,
+    "凌": 2062,
+    "凛": 2063,
+    "凜": 2064,
+    "凝": 2065,
+    "凡": 2066,
+    "凤": 2067,
+    "凪": 2068,
+    "凯": 2069,
+    "凰": 2070,
+    "凱": 2071,
+    "出": 2072,
+    "函": 2073,
+    "刀": 2074,
+    "分": 2075,
+    "切": 2076,
+    "刈": 2077,
+    "列": 2078,
+    "刘": 2079,
+    "则": 2080,
+    "刚": 2081,
+    "创": 2082,
+    "初": 2083,
+    "別": 2084,
+    "利": 2085,
+    "到": 2086,
+    "則": 2087,
+    "前": 2088,
+    "剑": 2089,
+    "剛": 2090,
+    "剣": 2091,
+    "副": 2092,
+    "創": 2093,
+    "劉": 2094,
+    "劍": 2095,
+    "力": 2096,
+    "功": 2097,
+    "加": 2098,
+    "动": 2099,
+    "助": 2100,
+    "努": 2101,
+    "劭": 2102,
+    "励": 2103,
+    "劲": 2104,
+    "劼": 2105,
+    "勁": 2106,
+    "勃": 2107,
+    "勇": 2108,
+    "勉": 2109,
+    "勋": 2110,
+    "勒": 2111,
+    "動": 2112,
+    "勘": 2113,
+    "務": 2114,
+    "勛": 2115,
+    "勝": 2116,
+    "勢": 2117,
+    "勤": 2118,
+    "勲": 2119,
+    "勳": 2120,
+    "勵": 2121,
+    "勻": 2122,
+    "包": 2123,
+    "化": 2124,
+    "北": 2125,
+    "匠": 2126,
+    "匡": 2127,
+    "匯": 2128,
+    "區": 2129,
+    "十": 2130,
+    "千": 2131,
+    "升": 2132,
+    "午": 2133,
+    "卉": 2134,
+    "半": 2135,
+    "华": 2136,
+    "协": 2137,
+    "卓": 2138,
+    "協": 2139,
+    "单": 2140,
+    "南": 2141,
+    "博": 2142,
+    "卜": 2143,
+    "占": 2144,
+    "卡": 2145,
+    "卢": 2146,
+    "卫": 2147,
+    "卯": 2148,
+    "印": 2149,
+    "卷": 2150,
+    "卿": 2151,
+    "历": 2152,
+    "厚": 2153,
+    "原": 2154,
+    "又": 2155,
+    "及": 2156,
+    "友": 2157,
+    "双": 2158,
+    "反": 2159,
+    "収": 2160,
+    "发": 2161,
+    "叔": 2162,
+    "取": 2163,
+    "叡": 2164,
+    "口": 2165,
+    "古": 2166,
+    "只": 2167,
+    "召": 2168,
+    "叮": 2169,
+    "可": 2170,
+    "台": 2171,
+    "史": 2172,
+    "右": 2173,
+    "叶": 2174,
+    "号": 2175,
+    "司": 2176,
+    "各": 2177,
+    "合": 2178,
+    "吉": 2179,
+    "同": 2180,
+    "名": 2181,
+    "后": 2182,
+    "吏": 2183,
+    "吐": 2184,
+    "向": 2185,
+    "吕": 2186,
+    "君": 2187,
+    "吟": 2188,
+    "含": 2189,
+    "启": 2190,
+    "吳": 2191,
+    "吴": 2192,
+    "吹": 2193,
+    "吾": 2194,
+    "呂": 2195,
+    "呆": 2196,
+    "呈": 2197,
+    "呉": 2198,
+    "告": 2199,
+    "员": 2200,
+    "周": 2201,
+    "味": 2202,
+    "呼": 2203,
+    "命": 2204,
+    "和": 2205,
+    "咏": 2206,
+    "咖": 2207,
+    "咨": 2208,
+    "咪": 2209,
+    "咲": 2210,
+    "咸": 2211,
+    "品": 2212,
+    "哈": 2213,
+    "哉": 2214,
+    "員": 2215,
+    "哥": 2216,
+    "哲": 2217,
+    "唐": 2218,
+    "唯": 2219,
+    "商": 2220,
+    "問": 2221,
+    "啓": 2222,
+    "啟": 2223,
+    "啦": 2224,
+    "啸": 2225,
+    "善": 2226,
+    "喆": 2227,
+    "喜": 2228,
+    "喬": 2229,
+    "單": 2230,
+    "喵": 2231,
+    "喻": 2232,
+    "嗣": 2233,
+    "嘉": 2234,
+    "嘎": 2235,
+    "嘯": 2236,
+    "嘻": 2237,
+    "器": 2238,
+    "噹": 2239,
+    "嚴": 2240,
+    "四": 2241,
+    "回": 2242,
+    "因": 2243,
+    "团": 2244,
+    "団": 2245,
+    "园": 2246,
+    "固": 2247,
+    "国": 2248,
+    "图": 2249,
+    "囿": 2250,
+    "圃": 2251,
+    "圆": 2252,
+    "圈": 2253,
+    "國": 2254,
+    "園": 2255,
+    "圓": 2256,
+    "圖": 2257,
+    "團": 2258,
+    "土": 2259,
+    "圣": 2260,
+    "在": 2261,
+    "圭": 2262,
+    "地": 2263,
+    "圳": 2264,
+    "圻": 2265,
+    "坂": 2266,
+    "均": 2267,
+    "坊": 2268,
+    "坚": 2269,
+    "坡": 2270,
+    "坤": 2271,
+    "坦": 2272,
+    "坪": 2273,
+    "垂": 2274,
+    "垚": 2275,
+    "垣": 2276,
+    "城": 2277,
+    "埜": 2278,
+    "域": 2279,
+    "執": 2280,
+    "培": 2281,
+    "基": 2282,
+    "堀": 2283,
+    "堂": 2284,
+    "堃": 2285,
+    "堅": 2286,
+    "堆": 2287,
+    "堉": 2288,
+    "堡": 2289,
+    "堤": 2290,
+    "堯": 2291,
+    "場": 2292,
+    "堺": 2293,
+    "塁": 2294,
+    "塏": 2295,
+    "塔": 2296,
+    "塗": 2297,
+    "塚": 2298,
+    "塩": 2299,
+    "塵": 2300,
+    "境": 2301,
+    "墉": 2302,
+    "増": 2303,
+    "增": 2304,
+    "墨": 2305,
+    "壁": 2306,
+    "壕": 2307,
+    "士": 2308,
+    "壬": 2309,
+    "壮": 2310,
+    "壯": 2311,
+    "声": 2312,
+    "壱": 2313,
+    "壹": 2314,
+    "壽": 2315,
+    "夆": 2316,
+    "复": 2317,
+    "夏": 2318,
+    "夕": 2319,
+    "外": 2320,
+    "夙": 2321,
+    "多": 2322,
+    "夜": 2323,
+    "夢": 2324,
+    "大": 2325,
+    "天": 2326,
+    "太": 2327,
+    "夫": 2328,
+    "央": 2329,
+    "夷": 2330,
+    "奇": 2331,
+    "奈": 2332,
+    "奉": 2333,
+    "奋": 2334,
+    "奎": 2335,
+    "奏": 2336,
+    "奐": 2337,
+    "奔": 2338,
+    "奕": 2339,
+    "奥": 2340,
+    "奧": 2341,
+    "奨": 2342,
+    "女": 2343,
+    "奴": 2344,
+    "好": 2345,
+    "如": 2346,
+    "妃": 2347,
+    "妍": 2348,
+    "妏": 2349,
+    "妗": 2350,
+    "妘": 2351,
+    "妙": 2352,
+    "妡": 2353,
+    "妤": 2354,
+    "妮": 2355,
+    "妹": 2356,
+    "妺": 2357,
+    "妻": 2358,
+    "姆": 2359,
+    "始": 2360,
+    "姍": 2361,
+    "姐": 2362,
+    "委": 2363,
+    "姗": 2364,
+    "姚": 2365,
+    "姜": 2366,
+    "姝": 2367,
+    "姣": 2368,
+    "姫": 2369,
+    "姬": 2370,
+    "姮": 2371,
+    "姵": 2372,
+    "姸": 2373,
+    "姿": 2374,
+    "威": 2375,
+    "娃": 2376,
+    "娅": 2377,
+    "娇": 2378,
+    "娉": 2379,
+    "娘": 2380,
+    "娜": 2381,
+    "娟": 2382,
+    "娣": 2383,
+    "娥": 2384,
+    "娴": 2385,
+    "娸": 2386,
+    "婉": 2387,
+    "婕": 2388,
+    "婧": 2389,
+    "婭": 2390,
+    "婵": 2391,
+    "婷": 2392,
+    "媄": 2393,
+    "媖": 2394,
+    "媗": 2395,
+    "媚": 2396,
+    "媛": 2397,
+    "媜": 2398,
+    "媺": 2399,
+    "嫚": 2400,
+    "嫣": 2401,
+    "嫦": 2402,
+    "嫺": 2403,
+    "嫻": 2404,
+    "嬅": 2405,
+    "嬋": 2406,
+    "嬌": 2407,
+    "嬡": 2408,
+    "嬪": 2409,
+    "嬿": 2410,
+    "子": 2411,
+    "孔": 2412,
+    "字": 2413,
+    "存": 2414,
+    "孙": 2415,
+    "孚": 2416,
+    "孜": 2417,
+    "孝": 2418,
+    "孟": 2419,
+    "季": 2420,
+    "学": 2421,
+    "孫": 2422,
+    "學": 2423,
+    "宁": 2424,
+    "宅": 2425,
+    "宇": 2426,
+    "守": 2427,
+    "安": 2428,
+    "宋": 2429,
+    "完": 2430,
+    "宍": 2431,
+    "宏": 2432,
+    "宓": 2433,
+    "宗": 2434,
+    "官": 2435,
+    "宙": 2436,
+    "定": 2437,
+    "宛": 2438,
+    "宜": 2439,
+    "宝": 2440,
+    "实": 2441,
+    "実": 2442,
+    "客": 2443,
+    "宣": 2444,
+    "室": 2445,
+    "宥": 2446,
+    "宪": 2447,
+    "宫": 2448,
+    "宮": 2449,
+    "宰": 2450,
+    "宴": 2451,
+    "宵": 2452,
+    "家": 2453,
+    "宸": 2454,
+    "容": 2455,
+    "宽": 2456,
+    "宾": 2457,
+    "宿": 2458,
+    "寂": 2459,
+    "寄": 2460,
+    "寅": 2461,
+    "密": 2462,
+    "富": 2463,
+    "寒": 2464,
+    "寓": 2465,
+    "寛": 2466,
+    "實": 2467,
+    "寧": 2468,
+    "寬": 2469,
+    "寰": 2470,
+    "寳": 2471,
+    "寶": 2472,
+    "寺": 2473,
+    "寻": 2474,
+    "寿": 2475,
+    "封": 2476,
+    "専": 2477,
+    "将": 2478,
+    "將": 2479,
+    "專": 2480,
+    "尉": 2481,
+    "尊": 2482,
+    "尋": 2483,
+    "對": 2484,
+    "導": 2485,
+    "小": 2486,
+    "少": 2487,
+    "尔": 2488,
+    "尘": 2489,
+    "尚": 2490,
+    "尤": 2491,
+    "尧": 2492,
+    "就": 2493,
+    "尹": 2494,
+    "尻": 2495,
+    "尼": 2496,
+    "尾": 2497,
+    "局": 2498,
+    "居": 2499,
+    "屋": 2500,
+    "屏": 2501,
+    "展": 2502,
+    "屠": 2503,
+    "山": 2504,
+    "屹": 2505,
+    "岐": 2506,
+    "岑": 2507,
+    "岗": 2508,
+    "岚": 2509,
+    "岡": 2510,
+    "岩": 2511,
+    "岫": 2512,
+    "岬": 2513,
+    "岭": 2514,
+    "岱": 2515,
+    "岳": 2516,
+    "岷": 2517,
+    "岸": 2518,
+    "峥": 2519,
+    "峨": 2520,
+    "峯": 2521,
+    "峰": 2522,
+    "島": 2523,
+    "峻": 2524,
+    "崇": 2525,
+    "崎": 2526,
+    "崑": 2527,
+    "崔": 2528,
+    "崙": 2529,
+    "崚": 2530,
+    "崧": 2531,
+    "崴": 2532,
+    "嵐": 2533,
+    "嵘": 2534,
+    "嵜": 2535,
+    "嵩": 2536,
+    "嵯": 2537,
+    "嶋": 2538,
+    "嶌": 2539,
+    "嶸": 2540,
+    "嶺": 2541,
+    "嶽": 2542,
+    "巌": 2543,
+    "巍": 2544,
+    "巖": 2545,
+    "川": 2546,
+    "州": 2547,
+    "工": 2548,
+    "左": 2549,
+    "巧": 2550,
+    "巨": 2551,
+    "巫": 2552,
+    "己": 2553,
+    "已": 2554,
+    "巳": 2555,
+    "巴": 2556,
+    "巻": 2557,
+    "巽": 2558,
+    "市": 2559,
+    "布": 2560,
+    "帅": 2561,
+    "帆": 2562,
+    "师": 2563,
+    "希": 2564,
+    "帕": 2565,
+    "帛": 2566,
+    "帝": 2567,
+    "帥": 2568,
+    "带": 2569,
+    "師": 2570,
+    "席": 2571,
+    "帮": 2572,
+    "帶": 2573,
+    "帷": 2574,
+    "常": 2575,
+    "幃": 2576,
+    "幕": 2577,
+    "幡": 2578,
+    "干": 2579,
+    "平": 2580,
+    "年": 2581,
+    "幸": 2582,
+    "幹": 2583,
+    "幻": 2584,
+    "幼": 2585,
+    "幽": 2586,
+    "幾": 2587,
+    "广": 2588,
+    "広": 2589,
+    "庄": 2590,
+    "庆": 2591,
+    "床": 2592,
+    "序": 2593,
+    "应": 2594,
+    "庚": 2595,
+    "府": 2596,
+    "庠": 2597,
+    "度": 2598,
+    "座": 2599,
+    "庫": 2600,
+    "庭": 2601,
+    "庵": 2602,
+    "康": 2603,
+    "庸": 2604,
+    "廉": 2605,
+    "廖": 2606,
+    "廣": 2607,
+    "延": 2608,
+    "廷": 2609,
+    "廸": 2610,
+    "建": 2611,
+    "开": 2612,
+    "弈": 2613,
+    "弋": 2614,
+    "式": 2615,
+    "弓": 2616,
+    "引": 2617,
+    "弘": 2618,
+    "弟": 2619,
+    "张": 2620,
+    "弥": 2621,
+    "弦": 2622,
+    "張": 2623,
+    "強": 2624,
+    "强": 2625,
+    "弼": 2626,
+    "彌": 2627,
+    "当": 2628,
+    "彗": 2629,
+    "形": 2630,
+    "彣": 2631,
+    "彤": 2632,
+    "彥": 2633,
+    "彦": 2634,
+    "彧": 2635,
+    "彩": 2636,
+    "彪": 2637,
+    "彬": 2638,
+    "彭": 2639,
+    "彰": 2640,
+    "影": 2641,
+    "彼": 2642,
+    "征": 2643,
+    "律": 2644,
+    "後": 2645,
+    "徐": 2646,
+    "徒": 2647,
+    "得": 2648,
+    "從": 2649,
+    "御": 2650,
+    "復": 2651,
+    "循": 2652,
+    "微": 2653,
+    "徳": 2654,
+    "徵": 2655,
+    "德": 2656,
+    "徹": 2657,
+    "徽": 2658,
+    "心": 2659,
+    "必": 2660,
+    "忆": 2661,
+    "忍": 2662,
+    "志": 2663,
+    "忠": 2664,
+    "快": 2665,
+    "念": 2666,
+    "忻": 2667,
+    "怀": 2668,
+    "怜": 2669,
+    "思": 2670,
+    "怡": 2671,
+    "性": 2672,
+    "怪": 2673,
+    "恆": 2674,
+    "恋": 2675,
+    "恒": 2676,
+    "恕": 2677,
+    "恩": 2678,
+    "恬": 2679,
+    "恭": 2680,
+    "恵": 2681,
+    "恺": 2682,
+    "悅": 2683,
+    "悌": 2684,
+    "悟": 2685,
+    "悠": 2686,
+    "悦": 2687,
+    "情": 2688,
+    "惇": 2689,
+    "惜": 2690,
+    "惟": 2691,
+    "惠": 2692,
+    "惣": 2693,
+    "想": 2694,
+    "愉": 2695,
+    "意": 2696,
+    "愚": 2697,
+    "愛": 2698,
+    "愷": 2699,
+    "愿": 2700,
+    "慈": 2701,
+    "慎": 2702,
+    "慕": 2703,
+    "慧": 2704,
+    "慰": 2705,
+    "慶": 2706,
+    "憂": 2707,
+    "憐": 2708,
+    "憬": 2709,
+    "憲": 2710,
+    "憶": 2711,
+    "應": 2712,
+    "懋": 2713,
+    "懷": 2714,
+    "懿": 2715,
+    "戀": 2716,
+    "戈": 2717,
+    "戎": 2718,
+    "成": 2719,
+    "我": 2720,
+    "战": 2721,
+    "戚": 2722,
+    "戰": 2723,
+    "戴": 2724,
+    "戸": 2725,
+    "房": 2726,
+    "所": 2727,
+    "扇": 2728,
+    "手": 2729,
+    "才": 2730,
+    "扎": 2731,
+    "打": 2732,
+    "托": 2733,
+    "扣": 2734,
+    "扬": 2735,
+    "扶": 2736,
+    "承": 2737,
+    "技": 2738,
+    "抒": 2739,
+    "折": 2740,
+    "押": 2741,
+    "拉": 2742,
+    "拓": 2743,
+    "拔": 2744,
+    "招": 2745,
+    "拜": 2746,
+    "拳": 2747,
+    "持": 2748,
+    "指": 2749,
+    "振": 2750,
+    "挺": 2751,
+    "捷": 2752,
+    "授": 2753,
+    "掌": 2754,
+    "掛": 2755,
+    "措": 2756,
+    "提": 2757,
+    "揚": 2758,
+    "揮": 2759,
+    "摘": 2760,
+    "摩": 2761,
+    "播": 2762,
+    "擇": 2763,
+    "操": 2764,
+    "擎": 2765,
+    "攀": 2766,
+    "支": 2767,
+    "收": 2768,
+    "攸": 2769,
+    "改": 2770,
+    "放": 2771,
+    "政": 2772,
+    "效": 2773,
+    "敏": 2774,
+    "敖": 2775,
+    "教": 2776,
+    "敢": 2777,
+    "敦": 2778,
+    "敬": 2779,
+    "数": 2780,
+    "整": 2781,
+    "敷": 2782,
+    "數": 2783,
+    "文": 2784,
+    "斉": 2785,
+    "斌": 2786,
+    "斎": 2787,
+    "斐": 2788,
+    "斗": 2789,
+    "斯": 2790,
+    "新": 2791,
+    "方": 2792,
+    "於": 2793,
+    "施": 2794,
+    "旅": 2795,
+    "旋": 2796,
+    "旎": 2797,
+    "族": 2798,
+    "旖": 2799,
+    "旗": 2800,
+    "日": 2801,
+    "旦": 2802,
+    "旨": 2803,
+    "早": 2804,
+    "旬": 2805,
+    "旭": 2806,
+    "时": 2807,
+    "旸": 2808,
+    "旺": 2809,
+    "旻": 2810,
+    "昀": 2811,
+    "昂": 2812,
+    "昆": 2813,
+    "昇": 2814,
+    "昊": 2815,
+    "昌": 2816,
+    "明": 2817,
+    "易": 2818,
+    "昕": 2819,
+    "星": 2820,
+    "映": 2821,
+    "春": 2822,
+    "昭": 2823,
+    "是": 2824,
+    "昱": 2825,
+    "昶": 2826,
+    "显": 2827,
+    "晁": 2828,
+    "時": 2829,
+    "晃": 2830,
+    "晉": 2831,
+    "晋": 2832,
+    "晏": 2833,
+    "晓": 2834,
+    "晔": 2835,
+    "晖": 2836,
+    "晗": 2837,
+    "晚": 2838,
+    "晞": 2839,
+    "晟": 2840,
+    "晧": 2841,
+    "晨": 2842,
+    "普": 2843,
+    "景": 2844,
+    "晰": 2845,
+    "晴": 2846,
+    "晶": 2847,
+    "智": 2848,
+    "暁": 2849,
+    "暄": 2850,
+    "暉": 2851,
+    "暐": 2852,
+    "暖": 2853,
+    "暘": 2854,
+    "暢": 2855,
+    "暮": 2856,
+    "曄": 2857,
+    "曈": 2858,
+    "曉": 2859,
+    "曙": 2860,
+    "曜": 2861,
+    "曦": 2862,
+    "曰": 2863,
+    "曲": 2864,
+    "更": 2865,
+    "書": 2866,
+    "曹": 2867,
+    "曼": 2868,
+    "曽": 2869,
+    "曾": 2870,
+    "最": 2871,
+    "會": 2872,
+    "月": 2873,
+    "有": 2874,
+    "朋": 2875,
+    "服": 2876,
+    "朔": 2877,
+    "朕": 2878,
+    "朗": 2879,
+    "望": 2880,
+    "朝": 2881,
+    "期": 2882,
+    "木": 2883,
+    "未": 2884,
+    "末": 2885,
+    "本": 2886,
+    "术": 2887,
+    "朱": 2888,
+    "朴": 2889,
+    "朵": 2890,
+    "权": 2891,
+    "杉": 2892,
+    "李": 2893,
+    "杏": 2894,
+    "材": 2895,
+    "村": 2896,
+    "杜": 2897,
+    "杞": 2898,
+    "束": 2899,
+    "条": 2900,
+    "来": 2901,
+    "杨": 2902,
+    "杭": 2903,
+    "杯": 2904,
+    "杰": 2905,
+    "東": 2906,
+    "杵": 2907,
+    "松": 2908,
+    "板": 2909,
+    "林": 2910,
+    "枚": 2911,
+    "果": 2912,
+    "枝": 2913,
+    "枫": 2914,
+    "柄": 2915,
+    "柊": 2916,
+    "柏": 2917,
+    "柑": 2918,
+    "染": 2919,
+    "柔": 2920,
+    "柘": 2921,
+    "柚": 2922,
+    "柠": 2923,
+    "查": 2924,
+    "柯": 2925,
+    "柱": 2926,
+    "柳": 2927,
+    "柴": 2928,
+    "柾": 2929,
+    "柿": 2930,
+    "栁": 2931,
+    "栃": 2932,
+    "栄": 2933,
+    "标": 2934,
+    "栋": 2935,
+    "树": 2936,
+    "栖": 2937,
+    "栗": 2938,
+    "栞": 2939,
+    "校": 2940,
+    "栢": 2941,
+    "栩": 2942,
+    "根": 2943,
+    "格": 2944,
+    "桀": 2945,
+    "桂": 2946,
+    "桃": 2947,
+    "案": 2948,
+    "桐": 2949,
+    "桑": 2950,
+    "桓": 2951,
+    "桔": 2952,
+    "桜": 2953,
+    "桝": 2954,
+    "桢": 2955,
+    "桥": 2956,
+    "桦": 2957,
+    "梁": 2958,
+    "梅": 2959,
+    "梓": 2960,
+    "條": 2961,
+    "梢": 2962,
+    "梦": 2963,
+    "梧": 2964,
+    "梨": 2965,
+    "梵": 2966,
+    "梶": 2967,
+    "棉": 2968,
+    "棋": 2969,
+    "棕": 2970,
+    "棚": 2971,
+    "棟": 2972,
+    "棠": 2973,
+    "棣": 2974,
+    "棨": 2975,
+    "森": 2976,
+    "棻": 2977,
+    "椋": 2978,
+    "植": 2979,
+    "椎": 2980,
+    "椰": 2981,
+    "椿": 2982,
+    "楊": 2983,
+    "楓": 2984,
+    "楚": 2985,
+    "楠": 2986,
+    "楢": 2987,
+    "楨": 2988,
+    "業": 2989,
+    "極": 2990,
+    "楷": 2991,
+    "楹": 2992,
+    "楼": 2993,
+    "楽": 2994,
+    "榆": 2995,
+    "榊": 2996,
+    "榎": 2997,
+    "榕": 2998,
+    "榛": 2999,
+    "榜": 3000,
+    "榤": 3001,
+    "榮": 3002,
+    "槇": 3003,
+    "槐": 3004,
+    "槙": 3005,
+    "槻": 3006,
+    "樂": 3007,
+    "樊": 3008,
+    "樋": 3009,
+    "樑": 3010,
+    "樓": 3011,
+    "標": 3012,
+    "樟": 3013,
+    "模": 3014,
+    "権": 3015,
+    "横": 3016,
+    "樫": 3017,
+    "樱": 3018,
+    "樵": 3019,
+    "樹": 3020,
+    "樺": 3021,
+    "樽": 3022,
+    "橋": 3023,
+    "橘": 3024,
+    "橙": 3025,
+    "機": 3026,
+    "檀": 3027,
+    "檜": 3028,
+    "檬": 3029,
+    "櫛": 3030,
+    "櫻": 3031,
+    "權": 3032,
+    "次": 3033,
+    "欢": 3034,
+    "欣": 3035,
+    "欧": 3036,
+    "欽": 3037,
+    "歆": 3038,
+    "歌": 3039,
+    "歐": 3040,
+    "歡": 3041,
+    "止": 3042,
+    "正": 3043,
+    "此": 3044,
+    "步": 3045,
+    "武": 3046,
+    "歩": 3047,
+    "歪": 3048,
+    "歳": 3049,
+    "歷": 3050,
+    "殊": 3051,
+    "段": 3052,
+    "殷": 3053,
+    "殿": 3054,
+    "毅": 3055,
+    "母": 3056,
+    "毓": 3057,
+    "比": 3058,
+    "毕": 3059,
+    "毛": 3060,
+    "氏": 3061,
+    "民": 3062,
+    "气": 3063,
+    "気": 3064,
+    "氣": 3065,
+    "水": 3066,
+    "氷": 3067,
+    "永": 3068,
+    "汀": 3069,
+    "求": 3070,
+    "汇": 3071,
+    "汉": 3072,
+    "汎": 3073,
+    "汐": 3074,
+    "汗": 3075,
+    "汝": 3076,
+    "江": 3077,
+    "池": 3078,
+    "汤": 3079,
+    "汪": 3080,
+    "汰": 3081,
+    "汶": 3082,
+    "沁": 3083,
+    "沂": 3084,
+    "沃": 3085,
+    "沅": 3086,
+    "沈": 3087,
+    "沉": 3088,
+    "沐": 3089,
+    "沖": 3090,
+    "沙": 3091,
+    "沛": 3092,
+    "沢": 3093,
+    "沫": 3094,
+    "河": 3095,
+    "油": 3096,
+    "治": 3097,
+    "沼": 3098,
+    "沿": 3099,
+    "泉": 3100,
+    "泊": 3101,
+    "泓": 3102,
+    "法": 3103,
+    "波": 3104,
+    "泥": 3105,
+    "泫": 3106,
+    "泯": 3107,
+    "泰": 3108,
+    "泱": 3109,
+    "泳": 3110,
+    "泽": 3111,
+    "洁": 3112,
+    "洋": 3113,
+    "洛": 3114,
+    "洞": 3115,
+    "津": 3116,
+    "洧": 3117,
+    "洪": 3118,
+    "洲": 3119,
+    "洳": 3120,
+    "洵": 3121,
+    "洸": 3122,
+    "洺": 3123,
+    "活": 3124,
+    "洽": 3125,
+    "派": 3126,
+    "流": 3127,
+    "浅": 3128,
+    "济": 3129,
+    "浚": 3130,
+    "浜": 3131,
+    "浠": 3132,
+    "浤": 3133,
+    "浦": 3134,
+    "浩": 3135,
+    "浪": 3136,
+    "浮": 3137,
+    "海": 3138,
+    "涂": 3139,
+    "涌": 3140,
+    "涓": 3141,
+    "涛": 3142,
+    "润": 3143,
+    "涯": 3144,
+    "涵": 3145,
+    "涼": 3146,
+    "淇": 3147,
+    "淋": 3148,
+    "淏": 3149,
+    "淑": 3150,
+    "淘": 3151,
+    "淙": 3152,
+    "淞": 3153,
+    "淡": 3154,
+    "淦": 3155,
+    "淨": 3156,
+    "淩": 3157,
+    "淮": 3158,
+    "淯": 3159,
+    "深": 3160,
+    "淳": 3161,
+    "淵": 3162,
+    "淺": 3163,
+    "添": 3164,
+    "淼": 3165,
+    "清": 3166,
+    "済": 3167,
+    "渉": 3168,
+    "渊": 3169,
+    "渋": 3170,
+    "渓": 3171,
+    "渕": 3172,
+    "渚": 3173,
+    "渝": 3174,
+    "渟": 3175,
+    "渠": 3176,
+    "渡": 3177,
+    "渥": 3178,
+    "温": 3179,
+    "渭": 3180,
+    "港": 3181,
+    "渲": 3182,
+    "游": 3183,
+    "渼": 3184,
+    "湄": 3185,
+    "湊": 3186,
+    "湖": 3187,
+    "湘": 3188,
+    "湛": 3189,
+    "湧": 3190,
+    "湯": 3191,
+    "満": 3192,
+    "源": 3193,
+    "準": 3194,
+    "溝": 3195,
+    "溢": 3196,
+    "溥": 3197,
+    "溪": 3198,
+    "溫": 3199,
+    "溱": 3200,
+    "溶": 3201,
+    "滄": 3202,
+    "滉": 3203,
+    "滋": 3204,
+    "滑": 3205,
+    "滔": 3206,
+    "滕": 3207,
+    "滝": 3208,
+    "满": 3209,
+    "滢": 3210,
+    "滨": 3211,
+    "滴": 3212,
+    "滿": 3213,
+    "漁": 3214,
+    "漆": 3215,
+    "演": 3216,
+    "漢": 3217,
+    "漩": 3218,
+    "漪": 3219,
+    "漫": 3220,
+    "漳": 3221,
+    "漾": 3222,
+    "潁": 3223,
+    "潇": 3224,
+    "潔": 3225,
+    "潘": 3226,
+    "潜": 3227,
+    "潞": 3228,
+    "潤": 3229,
+    "潭": 3230,
+    "潮": 3231,
+    "潼": 3232,
+    "澁": 3233,
+    "澄": 3234,
+    "澍": 3235,
+    "澎": 3236,
+    "澐": 3237,
+    "澔": 3238,
+    "澜": 3239,
+    "澤": 3240,
+    "澧": 3241,
+    "澪": 3242,
+    "澳": 3243,
+    "濂": 3244,
+    "濃": 3245,
+    "濟": 3246,
+    "濠": 3247,
+    "濤": 3248,
+    "濬": 3249,
+    "濰": 3250,
+    "濱": 3251,
+    "濵": 3252,
+    "瀅": 3253,
+    "瀚": 3254,
+    "瀛": 3255,
+    "瀞": 3256,
+    "瀟": 3257,
+    "瀧": 3258,
+    "瀨": 3259,
+    "瀬": 3260,
+    "瀾": 3261,
+    "灃": 3262,
+    "灏": 3263,
+    "灝": 3264,
+    "灣": 3265,
+    "火": 3266,
+    "灯": 3267,
+    "灰": 3268,
+    "灵": 3269,
+    "灶": 3270,
+    "灼": 3271,
+    "灿": 3272,
+    "炎": 3273,
+    "炘": 3274,
+    "炜": 3275,
+    "炤": 3276,
+    "炫": 3277,
+    "炬": 3278,
+    "炮": 3279,
+    "炯": 3280,
+    "炳": 3281,
+    "点": 3282,
+    "為": 3283,
+    "炼": 3284,
+    "炽": 3285,
+    "烁": 3286,
+    "烈": 3287,
+    "烏": 3288,
+    "烜": 3289,
+    "烟": 3290,
+    "烨": 3291,
+    "热": 3292,
+    "烱": 3293,
+    "烽": 3294,
+    "焕": 3295,
+    "焜": 3296,
+    "無": 3297,
+    "焦": 3298,
+    "焯": 3299,
+    "焰": 3300,
+    "焱": 3301,
+    "然": 3302,
+    "煇": 3303,
+    "煊": 3304,
+    "煌": 3305,
+    "煒": 3306,
+    "煙": 3307,
+    "煜": 3308,
+    "煥": 3309,
+    "煦": 3310,
+    "照": 3311,
+    "煬": 3312,
+    "熊": 3313,
+    "熏": 3314,
+    "熒": 3315,
+    "熙": 3316,
+    "熠": 3317,
+    "熱": 3318,
+    "熹": 3319,
+    "熾": 3320,
+    "燁": 3321,
+    "燃": 3322,
+    "燈": 3323,
+    "燊": 3324,
+    "燕": 3325,
+    "營": 3326,
+    "燦": 3327,
+    "燮": 3328,
+    "燿": 3329,
+    "爪": 3330,
+    "爱": 3331,
+    "爵": 3332,
+    "父": 3333,
+    "爺": 3334,
+    "爽": 3335,
+    "爾": 3336,
+    "片": 3337,
+    "牙": 3338,
+    "牛": 3339,
+    "牟": 3340,
+    "牧": 3341,
+    "物": 3342,
+    "特": 3343,
+    "犬": 3344,
+    "狄": 3345,
+    "狩": 3346,
+    "猛": 3347,
+    "猪": 3348,
+    "猫": 3349,
+    "献": 3350,
+    "猴": 3351,
+    "猷": 3352,
+    "猿": 3353,
+    "獅": 3354,
+    "獻": 3355,
+    "玄": 3356,
+    "玉": 3357,
+    "王": 3358,
+    "玖": 3359,
+    "玗": 3360,
+    "玛": 3361,
+    "玟": 3362,
+    "玠": 3363,
+    "玥": 3364,
+    "玨": 3365,
+    "玩": 3366,
+    "玫": 3367,
+    "玮": 3368,
+    "环": 3369,
+    "现": 3370,
+    "玲": 3371,
+    "玳": 3372,
+    "玹": 3373,
+    "玺": 3374,
+    "珀": 3375,
+    "珂": 3376,
+    "珅": 3377,
+    "珈": 3378,
+    "珉": 3379,
+    "珊": 3380,
+    "珍": 3381,
+    "珏": 3382,
+    "珑": 3383,
+    "珞": 3384,
+    "珠": 3385,
+    "珣": 3386,
+    "珩": 3387,
+    "班": 3388,
+    "珮": 3389,
+    "珺": 3390,
+    "珽": 3391,
+    "現": 3392,
+    "琁": 3393,
+    "球": 3394,
+    "理": 3395,
+    "琇": 3396,
+    "琉": 3397,
+    "琍": 3398,
+    "琛": 3399,
+    "琢": 3400,
+    "琥": 3401,
+    "琦": 3402,
+    "琨": 3403,
+    "琪": 3404,
+    "琬": 3405,
+    "琮": 3406,
+    "琰": 3407,
+    "琳": 3408,
+    "琴": 3409,
+    "琼": 3410,
+    "瑀": 3411,
+    "瑄": 3412,
+    "瑋": 3413,
+    "瑗": 3414,
+    "瑚": 3415,
+    "瑛": 3416,
+    "瑜": 3417,
+    "瑞": 3418,
+    "瑟": 3419,
+    "瑠": 3420,
+    "瑢": 3421,
+    "瑤": 3422,
+    "瑩": 3423,
+    "瑪": 3424,
+    "瑰": 3425,
+    "瑱": 3426,
+    "瑶": 3427,
+    "瑾": 3428,
+    "璁": 3429,
+    "璃": 3430,
+    "璇": 3431,
+    "璉": 3432,
+    "璋": 3433,
+    "璐": 3434,
+    "璘": 3435,
+    "璞": 3436,
+    "璟": 3437,
+    "璠": 3438,
+    "璦": 3439,
+    "璧": 3440,
+    "璨": 3441,
+    "環": 3442,
+    "璽": 3443,
+    "璿": 3444,
+    "瓊": 3445,
+    "瓏": 3446,
+    "瓔": 3447,
+    "瓜": 3448,
+    "瓦": 3449,
+    "瓶": 3450,
+    "甄": 3451,
+    "甘": 3452,
+    "甚": 3453,
+    "甜": 3454,
+    "生": 3455,
+    "用": 3456,
+    "甫": 3457,
+    "甯": 3458,
+    "田": 3459,
+    "由": 3460,
+    "甲": 3461,
+    "申": 3462,
+    "男": 3463,
+    "町": 3464,
+    "画": 3465,
+    "畅": 3466,
+    "畇": 3467,
+    "界": 3468,
+    "畑": 3469,
+    "畔": 3470,
+    "留": 3471,
+    "畠": 3472,
+    "番": 3473,
+    "畫": 3474,
+    "畯": 3475,
+    "異": 3476,
+    "當": 3477,
+    "癸": 3478,
+    "登": 3479,
+    "發": 3480,
+    "白": 3481,
+    "百": 3482,
+    "的": 3483,
+    "皆": 3484,
+    "皇": 3485,
+    "皓": 3486,
+    "皖": 3487,
+    "皮": 3488,
+    "盈": 3489,
+    "益": 3490,
+    "盛": 3491,
+    "盟": 3492,
+    "盤": 3493,
+    "盧": 3494,
+    "目": 3495,
+    "直": 3496,
+    "相": 3497,
+    "盼": 3498,
+    "省": 3499,
+    "眉": 3500,
+    "眞": 3501,
+    "真": 3502,
+    "眼": 3503,
+    "睛": 3504,
+    "督": 3505,
+    "睦": 3506,
+    "睿": 3507,
+    "瞬": 3508,
+    "瞳": 3509,
+    "矢": 3510,
+    "知": 3511,
+    "矩": 3512,
+    "石": 3513,
+    "砂": 3514,
+    "研": 3515,
+    "砚": 3516,
+    "破": 3517,
+    "硕": 3518,
+    "硬": 3519,
+    "硯": 3520,
+    "碓": 3521,
+    "碧": 3522,
+    "碩": 3523,
+    "碹": 3524,
+    "磊": 3525,
+    "磐": 3526,
+    "磨": 3527,
+    "磯": 3528,
+    "礒": 3529,
+    "示": 3530,
+    "礼": 3531,
+    "社": 3532,
+    "祁": 3533,
+    "祈": 3534,
+    "祉": 3535,
+    "祎": 3536,
+    "祐": 3537,
+    "祖": 3538,
+    "祚": 3539,
+    "祝": 3540,
+    "神": 3541,
+    "祥": 3542,
+    "祯": 3543,
+    "祺": 3544,
+    "祿": 3545,
+    "禄": 3546,
+    "禅": 3547,
+    "禎": 3548,
+    "福": 3549,
+    "禕": 3550,
+    "禧": 3551,
+    "禮": 3552,
+    "禹": 3553,
+    "禾": 3554,
+    "秀": 3555,
+    "私": 3556,
+    "秉": 3557,
+    "秋": 3558,
+    "科": 3559,
+    "秘": 3560,
+    "秝": 3561,
+    "秦": 3562,
+    "积": 3563,
+    "稀": 3564,
+    "程": 3565,
+    "稔": 3566,
+    "稚": 3567,
+    "稜": 3568,
+    "種": 3569,
+    "稲": 3570,
+    "稳": 3571,
+    "稼": 3572,
+    "穂": 3573,
+    "穆": 3574,
+    "積": 3575,
+    "穎": 3576,
+    "穗": 3577,
+    "穣": 3578,
+    "穩": 3579,
+    "穴": 3580,
+    "空": 3581,
+    "窪": 3582,
+    "立": 3583,
+    "竑": 3584,
+    "竜": 3585,
+    "竞": 3586,
+    "竟": 3587,
+    "章": 3588,
+    "竣": 3589,
+    "童": 3590,
+    "端": 3591,
+    "競": 3592,
+    "竹": 3593,
+    "竺": 3594,
+    "笑": 3595,
+    "笙": 3596,
+    "笛": 3597,
+    "笠": 3598,
+    "符": 3599,
+    "笹": 3600,
+    "筆": 3601,
+    "等": 3602,
+    "筌": 3603,
+    "筑": 3604,
+    "筒": 3605,
+    "策": 3606,
+    "筠": 3607,
+    "筧": 3608,
+    "筱": 3609,
+    "简": 3610,
+    "箕": 3611,
+    "管": 3612,
+    "箱": 3613,
+    "箴": 3614,
+    "節": 3615,
+    "範": 3616,
+    "築": 3617,
+    "篠": 3618,
+    "篤": 3619,
+    "簡": 3620,
+    "籠": 3621,
+    "米": 3622,
+    "籽": 3623,
+    "粉": 3624,
+    "粒": 3625,
+    "粟": 3626,
+    "粤": 3627,
+    "精": 3628,
+    "糸": 3629,
+    "紀": 3630,
+    "約": 3631,
+    "紅": 3632,
+    "紋": 3633,
+    "納": 3634,
+    "純": 3635,
+    "紗": 3636,
+    "紘": 3637,
+    "紙": 3638,
+    "紜": 3639,
+    "素": 3640,
+    "索": 3641,
+    "紫": 3642,
+    "細": 3643,
+    "紳": 3644,
+    "紹": 3645,
+    "紺": 3646,
+    "絃": 3647,
+    "経": 3648,
+    "結": 3649,
+    "絜": 3650,
+    "絢": 3651,
+    "絨": 3652,
+    "絮": 3653,
+    "統": 3654,
+    "絲": 3655,
+    "絵": 3656,
+    "絹": 3657,
+    "綉": 3658,
+    "經": 3659,
+    "継": 3660,
+    "綜": 3661,
+    "綠": 3662,
+    "維": 3663,
+    "綱": 3664,
+    "網": 3665,
+    "綵": 3666,
+    "綸": 3667,
+    "綺": 3668,
+    "綽": 3669,
+    "綾": 3670,
+    "綿": 3671,
+    "緁": 3672,
+    "緋": 3673,
+    "総": 3674,
+    "緑": 3675,
+    "緒": 3676,
+    "緗": 3677,
+    "緣": 3678,
+    "緯": 3679,
+    "練": 3680,
+    "緹": 3681,
+    "緻": 3682,
+    "縄": 3683,
+    "縈": 3684,
+    "縉": 3685,
+    "縣": 3686,
+    "總": 3687,
+    "繁": 3688,
+    "織": 3689,
+    "繡": 3690,
+    "繹": 3691,
+    "繼": 3692,
+    "红": 3693,
+    "约": 3694,
+    "纪": 3695,
+    "纬": 3696,
+    "纯": 3697,
+    "纳": 3698,
+    "纹": 3699,
+    "练": 3700,
+    "细": 3701,
+    "绍": 3702,
+    "经": 3703,
+    "结": 3704,
+    "统": 3705,
+    "绣": 3706,
+    "继": 3707,
+    "绪": 3708,
+    "绮": 3709,
+    "维": 3710,
+    "绵": 3711,
+    "绿": 3712,
+    "缘": 3713,
+    "罗": 3714,
+    "罡": 3715,
+    "置": 3716,
+    "羅": 3717,
+    "羊": 3718,
+    "美": 3719,
+    "羚": 3720,
+    "羡": 3721,
+    "群": 3722,
+    "義": 3723,
+    "羲": 3724,
+    "羽": 3725,
+    "羿": 3726,
+    "翀": 3727,
+    "翁": 3728,
+    "翃": 3729,
+    "翊": 3730,
+    "翌": 3731,
+    "翎": 3732,
+    "習": 3733,
+    "翔": 3734,
+    "翟": 3735,
+    "翠": 3736,
+    "翰": 3737,
+    "翹": 3738,
+    "翼": 3739,
+    "耀": 3740,
+    "老": 3741,
+    "考": 3742,
+    "者": 3743,
+    "而": 3744,
+    "耐": 3745,
+    "耕": 3746,
+    "耘": 3747,
+    "耳": 3748,
+    "耶": 3749,
+    "耿": 3750,
+    "聆": 3751,
+    "联": 3752,
+    "聖": 3753,
+    "聘": 3754,
+    "聚": 3755,
+    "聞": 3756,
+    "聡": 3757,
+    "聪": 3758,
+    "聯": 3759,
+    "聰": 3760,
+    "聲": 3761,
+    "聿": 3762,
+    "肇": 3763,
+    "肖": 3764,
+    "股": 3765,
+    "肥": 3766,
+    "肯": 3767,
+    "育": 3768,
+    "背": 3769,
+    "胖": 3770,
+    "胜": 3771,
+    "胡": 3772,
+    "胤": 3773,
+    "能": 3774,
+    "脇": 3775,
+    "脩": 3776,
+    "腰": 3777,
+    "腾": 3778,
+    "臣": 3779,
+    "臧": 3780,
+    "臨": 3781,
+    "自": 3782,
+    "至": 3783,
+    "致": 3784,
+    "臺": 3785,
+    "臻": 3786,
+    "臼": 3787,
+    "與": 3788,
+    "興": 3789,
+    "舉": 3790,
+    "舒": 3791,
+    "舘": 3792,
+    "舛": 3793,
+    "舜": 3794,
+    "舞": 3795,
+    "舟": 3796,
+    "舩": 3797,
+    "航": 3798,
+    "舫": 3799,
+    "船": 3800,
+    "艇": 3801,
+    "良": 3802,
+    "色": 3803,
+    "艳": 3804,
+    "艶": 3805,
+    "艷": 3806,
+    "艺": 3807,
+    "艾": 3808,
+    "芃": 3809,
+    "芊": 3810,
+    "芋": 3811,
+    "芙": 3812,
+    "芝": 3813,
+    "芥": 3814,
+    "芦": 3815,
+    "芩": 3816,
+    "芫": 3817,
+    "芬": 3818,
+    "芮": 3819,
+    "芯": 3820,
+    "花": 3821,
+    "芳": 3822,
+    "芷": 3823,
+    "芸": 3824,
+    "芹": 3825,
+    "芽": 3826,
+    "苅": 3827,
+    "苇": 3828,
+    "苏": 3829,
+    "苑": 3830,
+    "苓": 3831,
+    "苗": 3832,
+    "苡": 3833,
+    "若": 3834,
+    "苦": 3835,
+    "英": 3836,
+    "苹": 3837,
+    "茂": 3838,
+    "范": 3839,
+    "茅": 3840,
+    "茉": 3841,
+    "茗": 3842,
+    "茜": 3843,
+    "茨": 3844,
+    "茱": 3845,
+    "茲": 3846,
+    "茵": 3847,
+    "茶": 3848,
+    "茹": 3849,
+    "荃": 3850,
+    "草": 3851,
+    "荏": 3852,
+    "荒": 3853,
+    "荔": 3854,
+    "荘": 3855,
+    "荣": 3856,
+    "荧": 3857,
+    "荷": 3858,
+    "荻": 3859,
+    "莉": 3860,
+    "莊": 3861,
+    "莎": 3862,
+    "莓": 3863,
+    "莘": 3864,
+    "莛": 3865,
+    "莞": 3866,
+    "莫": 3867,
+    "莱": 3868,
+    "莲": 3869,
+    "莹": 3870,
+    "莺": 3871,
+    "菀": 3872,
+    "菁": 3873,
+    "菅": 3874,
+    "菇": 3875,
+    "菊": 3876,
+    "菖": 3877,
+    "菘": 3878,
+    "菜": 3879,
+    "菡": 3880,
+    "華": 3881,
+    "菱": 3882,
+    "菲": 3883,
+    "萊": 3884,
+    "萌": 3885,
+    "萍": 3886,
+    "营": 3887,
+    "萧": 3888,
+    "萨": 3889,
+    "萩": 3890,
+    "萬": 3891,
+    "萱": 3892,
+    "落": 3893,
+    "葆": 3894,
+    "葉": 3895,
+    "著": 3896,
+    "葛": 3897,
+    "董": 3898,
+    "葦": 3899,
+    "葳": 3900,
+    "葵": 3901,
+    "葶": 3902,
+    "蒂": 3903,
+    "蒋": 3904,
+    "蒔": 3905,
+    "蒙": 3906,
+    "蒨": 3907,
+    "蒲": 3908,
+    "蒼": 3909,
+    "蓁": 3910,
+    "蓉": 3911,
+    "蓑": 3912,
+    "蓓": 3913,
+    "蓝": 3914,
+    "蓬": 3915,
+    "蓮": 3916,
+    "蔓": 3917,
+    "蔚": 3918,
+    "蔡": 3919,
+    "蔣": 3920,
+    "蔭": 3921,
+    "蔵": 3922,
+    "蕊": 3923,
+    "蕎": 3924,
+    "蕓": 3925,
+    "蕙": 3926,
+    "蕭": 3927,
+    "蕴": 3928,
+    "蕾": 3929,
+    "薄": 3930,
+    "薇": 3931,
+    "薏": 3932,
+    "薗": 3933,
+    "薙": 3934,
+    "薛": 3935,
+    "薩": 3936,
+    "薪": 3937,
+    "薫": 3938,
+    "薬": 3939,
+    "薮": 3940,
+    "薰": 3941,
+    "藍": 3942,
+    "藏": 3943,
+    "藝": 3944,
+    "藤": 3945,
+    "藩": 3946,
+    "藪": 3947,
+    "蘇": 3948,
+    "蘊": 3949,
+    "蘋": 3950,
+    "蘭": 3951,
+    "蘿": 3952,
+    "虎": 3953,
+    "虔": 3954,
+    "虞": 3955,
+    "虹": 3956,
+    "蛋": 3957,
+    "蛯": 3958,
+    "蜀": 3959,
+    "蜂": 3960,
+    "蜜": 3961,
+    "蝦": 3962,
+    "蝶": 3963,
+    "融": 3964,
+    "螢": 3965,
+    "蟹": 3966,
+    "行": 3967,
+    "衍": 3968,
+    "衛": 3969,
+    "衞": 3970,
+    "衡": 3971,
+    "衣": 3972,
+    "表": 3973,
+    "衿": 3974,
+    "袁": 3975,
+    "袋": 3976,
+    "袖": 3977,
+    "裔": 3978,
+    "裕": 3979,
+    "裘": 3980,
+    "裴": 3981,
+    "襄": 3982,
+    "西": 3983,
+    "覀": 3984,
+    "要": 3985,
+    "見": 3986,
+    "規": 3987,
+    "視": 3988,
+    "覚": 3989,
+    "親": 3990,
+    "観": 3991,
+    "覺": 3992,
+    "觀": 3993,
+    "见": 3994,
+    "观": 3995,
+    "觉": 3996,
+    "角": 3997,
+    "解": 3998,
+    "言": 3999,
+    "計": 4000,
+    "訓": 4001,
+    "記": 4002,
+    "訪": 4003,
+    "設": 4004,
+    "許": 4005,
+    "証": 4006,
+    "詒": 4007,
+    "詔": 4008,
+    "評": 4009,
+    "詞": 4010,
+    "詠": 4011,
+    "詣": 4012,
+    "詩": 4013,
+    "詮": 4014,
+    "詰": 4015,
+    "詳": 4016,
+    "詹": 4017,
+    "誉": 4018,
+    "誌": 4019,
+    "誓": 4020,
+    "語": 4021,
+    "誠": 4022,
+    "誼": 4023,
+    "調": 4024,
+    "諄": 4025,
+    "諏": 4026,
+    "諒": 4027,
+    "論": 4028,
+    "諠": 4029,
+    "諦": 4030,
+    "諭": 4031,
+    "諸": 4032,
+    "諺": 4033,
+    "諾": 4034,
+    "謀": 4035,
+    "謙": 4036,
+    "謝": 4037,
+    "謹": 4038,
+    "證": 4039,
+    "識": 4040,
+    "譚": 4041,
+    "譯": 4042,
+    "議": 4043,
+    "譲": 4044,
+    "護": 4045,
+    "譽": 4046,
+    "讚": 4047,
+    "计": 4048,
+    "训": 4049,
+    "记": 4050,
+    "许": 4051,
+    "译": 4052,
+    "诗": 4053,
+    "诚": 4054,
+    "语": 4055,
+    "诸": 4056,
+    "诺": 4057,
+    "谊": 4058,
+    "谢": 4059,
+    "谦": 4060,
+    "谨": 4061,
+    "谭": 4062,
+    "谷": 4063,
+    "豆": 4064,
+    "豈": 4065,
+    "豊": 4066,
+    "豐": 4067,
+    "豔": 4068,
+    "象": 4069,
+    "豪": 4070,
+    "豫": 4071,
+    "豬": 4072,
+    "貓": 4073,
+    "貝": 4074,
+    "貞": 4075,
+    "財": 4076,
+    "貢": 4077,
+    "貫": 4078,
+    "貳": 4079,
+    "貴": 4080,
+    "費": 4081,
+    "貽": 4082,
+    "貿": 4083,
+    "賀": 4084,
+    "資": 4085,
+    "賈": 4086,
+    "賓": 4087,
+    "賜": 4088,
+    "賞": 4089,
+    "賢": 4090,
+    "賦": 4091,
+    "質": 4092,
+    "賴": 4093,
+    "賽": 4094,
+    "贊": 4095,
+    "贏": 4096,
+    "贝": 4097,
+    "贞": 4098,
+    "财": 4099,
+    "贤": 4100,
+    "贵": 4101,
+    "费": 4102,
+    "贺": 4103,
+    "贻": 4104,
+    "贾": 4105,
+    "赐": 4106,
+    "赖": 4107,
+    "赛": 4108,
+    "赞": 4109,
+    "赟": 4110,
+    "赤": 4111,
+    "赫": 4112,
+    "走": 4113,
+    "赵": 4114,
+    "起": 4115,
+    "超": 4116,
+    "越": 4117,
+    "趙": 4118,
+    "足": 4119,
+    "跃": 4120,
+    "路": 4121,
+    "躍": 4122,
+    "身": 4123,
+    "車": 4124,
+    "軍": 4125,
+    "軒": 4126,
+    "軽": 4127,
+    "輔": 4128,
+    "輝": 4129,
+    "輪": 4130,
+    "輿": 4131,
+    "轉": 4132,
+    "轟": 4133,
+    "车": 4134,
+    "轩": 4135,
+    "轶": 4136,
+    "辉": 4137,
+    "辛": 4138,
+    "辰": 4139,
+    "農": 4140,
+    "辺": 4141,
+    "辻": 4142,
+    "込": 4143,
+    "达": 4144,
+    "迅": 4145,
+    "过": 4146,
+    "迎": 4147,
+    "运": 4148,
+    "近": 4149,
+    "进": 4150,
+    "远": 4151,
+    "连": 4152,
+    "迟": 4153,
+    "迦": 4154,
+    "迪": 4155,
+    "迫": 4156,
+    "述": 4157,
+    "迷": 4158,
+    "迺": 4159,
+    "追": 4160,
+    "送": 4161,
+    "选": 4162,
+    "逊": 4163,
+    "逍": 4164,
+    "透": 4165,
+    "通": 4166,
+    "速": 4167,
+    "造": 4168,
+    "逢": 4169,
+    "連": 4170,
+    "���": 4171,
+    "逵": 4172,
+    "逸": 4173,
+    "遂": 4174,
+    "遇": 4175,
+    "遊": 4176,
+    "運": 4177,
+    "過": 4178,
+    "道": 4179,
+    "達": 4180,
+    "遙": 4181,
+    "遠": 4182,
+    "遥": 4183,
+    "適": 4184,
+    "遵": 4185,
+    "選": 4186,
+    "遼": 4187,
+    "邉": 4188,
+    "邊": 4189,
+    "邑": 4190,
+    "邓": 4191,
+    "邢": 4192,
+    "那": 4193,
+    "邦": 4194,
+    "邨": 4195,
+    "邱": 4196,
+    "邵": 4197,
+    "邹": 4198,
+    "郁": 4199,
+    "郎": 4200,
+    "郑": 4201,
+    "郝": 4202,
+    "郡": 4203,
+    "部": 4204,
+    "郭": 4205,
+    "郷": 4206,
+    "都": 4207,
+    "鄉": 4208,
+    "鄒": 4209,
+    "鄧": 4210,
+    "鄭": 4211,
+    "酉": 4212,
+    "配": 4213,
+    "酒": 4214,
+    "酸": 4215,
+    "醇": 4216,
+    "醒": 4217,
+    "采": 4218,
+    "释": 4219,
+    "釋": 4220,
+    "里": 4221,
+    "重": 4222,
+    "野": 4223,
+    "量": 4224,
+    "金": 4225,
+    "釗": 4226,
+    "釘": 4227,
+    "釜": 4228,
+    "針": 4229,
+    "釧": 4230,
+    "鈞": 4231,
+    "鈴": 4232,
+    "鈺": 4233,
+    "鈿": 4234,
+    "鉄": 4235,
+    "鉅": 4236,
+    "鉉": 4237,
+    "鉦": 4238,
+    "鉴": 4239,
+    "銀": 4240,
+    "銅": 4241,
+    "銓": 4242,
+    "銘": 4243,
+    "銜": 4244,
+    "銨": 4245,
+    "銮": 4246,
+    "銳": 4247,
+    "鋆": 4248,
+    "鋐": 4249,
+    "鋒": 4250,
+    "鋭": 4251,
+    "鋼": 4252,
+    "錄": 4253,
+    "錕": 4254,
+    "錚": 4255,
+    "錞": 4256,
+    "錠": 4257,
+    "錡": 4258,
+    "錢": 4259,
+    "錦": 4260,
+    "錩": 4261,
+    "錫": 4262,
+    "鍇": 4263,
+    "鍋": 4264,
+    "鍛": 4265,
+    "鍬": 4266,
+    "鍵": 4267,
+    "鍾": 4268,
+    "鎂": 4269,
+    "鎌": 4270,
+    "鎔": 4271,
+    "鎧": 4272,
+    "鎮": 4273,
+    "鎰": 4274,
+    "鏑": 4275,
+    "鏗": 4276,
+    "鏞": 4277,
+    "鏡": 4278,
+    "鏵": 4279,
+    "鐘": 4280,
+    "鐵": 4281,
+    "鐸": 4282,
+    "鑑": 4283,
+    "鑫": 4284,
+    "鑽": 4285,
+    "鑾": 4286,
+    "钊": 4287,
+    "钟": 4288,
+    "钢": 4289,
+    "钦": 4290,
+    "钧": 4291,
+    "钰": 4292,
+    "钱": 4293,
+    "钿": 4294,
+    "铁": 4295,
+    "铃": 4296,
+    "铎": 4297,
+    "铖": 4298,
+    "铨": 4299,
+    "铭": 4300,
+    "铮": 4301,
+    "银": 4302,
+    "铿": 4303,
+    "锋": 4304,
+    "锐": 4305,
+    "锟": 4306,
+    "锡": 4307,
+    "锦": 4308,
+    "键": 4309,
+    "锴": 4310,
+    "镇": 4311,
+    "镜": 4312,
+    "長": 4313,
+    "长": 4314,
+    "門": 4315,
+    "開": 4316,
+    "閎": 4317,
+    "閑": 4318,
+    "閒": 4319,
+    "間": 4320,
+    "閔": 4321,
+    "関": 4322,
+    "閣": 4323,
+    "閩": 4324,
+    "閻": 4325,
+    "關": 4326,
+    "门": 4327,
+    "闫": 4328,
+    "问": 4329,
+    "闻": 4330,
+    "阎": 4331,
+    "阡": 4332,
+    "阪": 4333,
+    "阮": 4334,
+    "阳": 4335,
+    "阿": 4336,
+    "际": 4337,
+    "陆": 4338,
+    "陈": 4339,
+    "陌": 4340,
+    "降": 4341,
+    "陞": 4342,
+    "院": 4343,
+    "陣": 4344,
+    "陳": 4345,
+    "陵": 4346,
+    "陶": 4347,
+    "陸": 4348,
+    "陽": 4349,
+    "隅": 4350,
+    "隆": 4351,
+    "隈": 4352,
+    "階": 4353,
+    "際": 4354,
+    "隨": 4355,
+    "隱": 4356,
+    "隻": 4357,
+    "隼": 4358,
+    "隽": 4359,
+    "雀": 4360,
+    "雁": 4361,
+    "雄": 4362,
+    "雅": 4363,
+    "集": 4364,
+    "雋": 4365,
+    "雍": 4366,
+    "雙": 4367,
+    "雛": 4368,
+    "難": 4369,
+    "雨": 4370,
+    "雪": 4371,
+    "雯": 4372,
+    "雲": 4373,
+    "零": 4374,
+    "雷": 4375,
+    "電": 4376,
+    "霄": 4377,
+    "霆": 4378,
+    "震": 4379,
+    "霈": 4380,
+    "霊": 4381,
+    "霍": 4382,
+    "霏": 4383,
+    "霓": 4384,
+    "霖": 4385,
+    "霙": 4386,
+    "霜": 4387,
+    "霞": 4388,
+    "霧": 4389,
+    "露": 4390,
+    "霸": 4391,
+    "霽": 4392,
+    "靄": 4393,
+    "靈": 4394,
+    "青": 4395,
+    "靓": 4396,
+    "靖": 4397,
+    "静": 4398,
+    "靚": 4399,
+    "靜": 4400,
+    "非": 4401,
+    "面": 4402,
+    "革": 4403,
+    "鞠": 4404,
+    "韋": 4405,
+    "韓": 4406,
+    "韜": 4407,
+    "韦": 4408,
+    "韩": 4409,
+    "韬": 4410,
+    "音": 4411,
+    "韵": 4412,
+    "韶": 4413,
+    "韻": 4414,
+    "響": 4415,
+    "頂": 4416,
+    "順": 4417,
+    "須": 4418,
+    "頌": 4419,
+    "頎": 4420,
+    "頓": 4421,
+    "領": 4422,
+    "頡": 4423,
+    "頤": 4424,
+    "頭": 4425,
+    "頴": 4426,
+    "頻": 4427,
+    "頼": 4428,
+    "額": 4429,
+    "顏": 4430,
+    "顕": 4431,
+    "顗": 4432,
+    "願": 4433,
+    "類": 4434,
+    "顥": 4435,
+    "顧": 4436,
+    "顯": 4437,
+    "项": 4438,
+    "顺": 4439,
+    "顾": 4440,
+    "顿": 4441,
+    "颂": 4442,
+    "领": 4443,
+    "颐": 4444,
+    "颖": 4445,
+    "颜": 4446,
+    "颢": 4447,
+    "風": 4448,
+    "颯": 4449,
+    "飄": 4450,
+    "风": 4451,
+    "飘": 4452,
+    "飛": 4453,
+    "飞": 4454,
+    "飯": 4455,
+    "飼": 4456,
+    "餅": 4457,
+    "養": 4458,
+    "餘": 4459,
+    "館": 4460,
+    "饶": 4461,
+    "首": 4462,
+    "香": 4463,
+    "馥": 4464,
+    "馨": 4465,
+    "馬": 4466,
+    "馮": 4467,
+    "馳": 4468,
+    "駒": 4469,
+    "駱": 4470,
+    "駿": 4471,
+    "騎": 4472,
+    "騏": 4473,
+    "騫": 4474,
+    "騰": 4475,
+    "驊": 4476,
+    "驛": 4477,
+    "马": 4478,
+    "驰": 4479,
+    "驹": 4480,
+    "骁": 4481,
+    "骅": 4482,
+    "骆": 4483,
+    "骏": 4484,
+    "骥": 4485,
+    "骨": 4486,
+    "高": 4487,
+    "髙": 4488,
+    "鬼": 4489,
+    "魁": 4490,
+    "魂": 4491,
+    "魏": 4492,
+    "魚": 4493,
+    "魯": 4494,
+    "鮎": 4495,
+    "鮫": 4496,
+    "鮮": 4497,
+    "鯉": 4498,
+    "鯨": 4499,
+    "鲁": 4500,
+    "鲍": 4501,
+    "鲜": 4502,
+    "鳥": 4503,
+    "鳩": 4504,
+    "鳯": 4505,
+    "鳳": 4506,
+    "鳴": 4507,
+    "鴨": 4508,
+    "鴻": 4509,
+    "鵜": 4510,
+    "鵬": 4511,
+    "鶯": 4512,
+    "鶴": 4513,
+    "鷲": 4514,
+    "鷹": 4515,
+    "鸞": 4516,
+    "鸣": 4517,
+    "鸥": 4518,
+    "鸽": 4519,
+    "鸾": 4520,
+    "鸿": 4521,
+    "鹏": 4522,
+    "鹤": 4523,
+    "鹰": 4524,
+    "鹿": 4525,
+    "麒": 4526,
+    "麗": 4527,
+    "麟": 4528,
+    "麥": 4529,
+    "麦": 4530,
+    "麻": 4531,
+    "麿": 4532,
+    "黃": 4533,
+    "黄": 4534,
+    "黎": 4535,
+    "黑": 4536,
+    "黒": 4537,
+    "默": 4538,
+    "黛": 4539,
+    "點": 4540,
+    "鼎": 4541,
+    "鼻": 4542,
+    "齊": 4543,
+    "齋": 4544,
+    "齐": 4545,
+    "齡": 4546,
+    "龄": 4547,
+    "龍": 4548,
+    "龐": 4549,
+    "龔": 4550,
+    "龙": 4551,
+    "龚": 4552,
+    "ꙮ": 4553,
+    "ꧾ": 4554,
+    "ꭰ": 4555,
+    "ꭱ": 4556,
+    "ꭵ": 4557,
+    "ꭶ": 4558,
+    "ꭷ": 4559,
+    "ꭹ": 4560,
+    "ꭺ": 4561,
+    "ꭼ": 4562,
+    "ꭿ": 4563,
+    "ꮀ": 4564,
+    "ꮄ": 4565,
+    "ꮆ": 4566,
+    "ꮇ": 4567,
+    "ꮋ": 4568,
+    "ꮍ": 4569,
+    "ꮎ": 4570,
+    "ꮏ": 4571,
+    "ꮑ": 4572,
+    "ꮒ": 4573,
+    "ꮓ": 4574,
+    "ꮙ": 4575,
+    "ꮛ": 4576,
+    "ꮜ": 4577,
+    "ꮢ": 4578,
+    "ꮤ": 4579,
+    "ꮥ": 4580,
+    "ꮦ": 4581,
+    "ꮧ": 4582,
+    "ꮪ": 4583,
+    "ꮭ": 4584,
+    "ꮮ": 4585,
+    "ꮲ": 4586,
+    "ꮶ": 4587,
+    "ꮸ": 4588,
+    "ꮻ": 4589,
+    "ꮼ": 4590,
+    "가": 4591,
+    "각": 4592,
+    "감": 4593,
+    "갑": 4594,
+    "강": 4595,
+    "거": 4596,
+    "건": 4597,
+    "걸": 4598,
+    "결": 4599,
+    "겸": 4600,
+    "경": 4601,
+    "계": 4602,
+    "고": 4603,
+    "곤": 4604,
+    "공": 4605,
+    "곽": 4606,
+    "관": 4607,
+    "광": 4608,
+    "교": 4609,
+    "구": 4610,
+    "국": 4611,
+    "군": 4612,
+    "궁": 4613,
+    "권": 4614,
+    "귀": 4615,
+    "규": 4616,
+    "균": 4617,
+    "그": 4618,
+    "극": 4619,
+    "근": 4620,
+    "금": 4621,
+    "기": 4622,
+    "길": 4623,
+    "김": 4624,
+    "나": 4625,
+    "낙": 4626,
+    "난": 4627,
+    "남": 4628,
+    "내": 4629,
+    "네": 4630,
+    "녀": 4631,
+    "년": 4632,
+    "녕": 4633,
+    "노": 4634,
+    "농": 4635,
+    "누": 4636,
+    "늘": 4637,
+    "능": 4638,
+    "니": 4639,
+    "님": 4640,
+    "다": 4641,
+    "단": 4642,
+    "달": 4643,
+    "담": 4644,
+    "대": 4645,
+    "더": 4646,
+    "덕": 4647,
+    "도": 4648,
+    "돈": 4649,
+    "동": 4650,
+    "두": 4651,
+    "드": 4652,
+    "득": 4653,
+    "디": 4654,
+    "라": 4655,
+    "락": 4656,
+    "란": 4657,
+    "람": 4658,
+    "랑": 4659,
+    "래": 4660,
+    "량": 4661,
+    "레": 4662,
+    "련": 4663,
+    "렬": 4664,
+    "령": 4665,
+    "례": 4666,
+    "로": 4667,
+    "록": 4668,
+    "룡": 4669,
+    "루": 4670,
+    "류": 4671,
+    "륜": 4672,
+    "률": 4673,
+    "르": 4674,
+    "름": 4675,
+    "리": 4676,
+    "린": 4677,
+    "림": 4678,
+    "립": 4679,
+    "마": 4680,
+    "만": 4681,
+    "말": 4682,
+    "매": 4683,
+    "맹": 4684,
+    "면": 4685,
+    "명": 4686,
+    "모": 4687,
+    "목": 4688,
+    "몽": 4689,
+    "무": 4690,
+    "묵": 4691,
+    "문": 4692,
+    "미": 4693,
+    "민": 4694,
+    "바": 4695,
+    "박": 4696,
+    "반": 4697,
+    "방": 4698,
+    "배": 4699,
+    "백": 4700,
+    "범": 4701,
+    "법": 4702,
+    "변": 4703,
+    "별": 4704,
+    "병": 4705,
+    "보": 4706,
+    "복": 4707,
+    "본": 4708,
+    "봉": 4709,
+    "부": 4710,
+    "분": 4711,
+    "브": 4712,
+    "비": 4713,
+    "빈": 4714,
+    "사": 4715,
+    "산": 4716,
+    "삼": 4717,
+    "상": 4718,
+    "새": 4719,
+    "생": 4720,
+    "서": 4721,
+    "석": 4722,
+    "선": 4723,
+    "설": 4724,
+    "섭": 4725,
+    "성": 4726,
+    "세": 4727,
+    "소": 4728,
+    "손": 4729,
+    "솔": 4730,
+    "송": 4731,
+    "수": 4732,
+    "숙": 4733,
+    "순": 4734,
+    "술": 4735,
+    "숭": 4736,
+    "스": 4737,
+    "슬": 4738,
+    "승": 4739,
+    "시": 4740,
+    "식": 4741,
+    "신": 4742,
+    "실": 4743,
+    "심": 4744,
+    "아": 4745,
+    "악": 4746,
+    "안": 4747,
+    "알": 4748,
+    "암": 4749,
+    "애": 4750,
+    "야": 4751,
+    "양": 4752,
+    "어": 4753,
+    "억": 4754,
+    "언": 4755,
+    "엄": 4756,
+    "업": 4757,
+    "에": 4758,
+    "엘": 4759,
+    "여": 4760,
+    "연": 4761,
+    "열": 4762,
+    "염": 4763,
+    "엽": 4764,
+    "영": 4765,
+    "예": 4766,
+    "오": 4767,
+    "옥": 4768,
+    "온": 4769,
+    "완": 4770,
+    "왕": 4771,
+    "외": 4772,
+    "요": 4773,
+    "용": 4774,
+    "우": 4775,
+    "욱": 4776,
+    "운": 4777,
+    "울": 4778,
+    "웅": 4779,
+    "원": 4780,
+    "월": 4781,
+    "위": 4782,
+    "유": 4783,
+    "육": 4784,
+    "윤": 4785,
+    "율": 4786,
+    "은": 4787,
+    "을": 4788,
+    "음": 4789,
+    "응": 4790,
+    "의": 4791,
+    "이": 4792,
+    "익": 4793,
+    "인": 4794,
+    "일": 4795,
+    "임": 4796,
+    "자": 4797,
+    "작": 4798,
+    "장": 4799,
+    "재": 4800,
+    "전": 4801,
+    "점": 4802,
+    "정": 4803,
+    "제": 4804,
+    "조": 4805,
+    "종": 4806,
+    "좌": 4807,
+    "주": 4808,
+    "준": 4809,
+    "중": 4810,
+    "즈": 4811,
+    "증": 4812,
+    "지": 4813,
+    "직": 4814,
+    "진": 4815,
+    "집": 4816,
+    "차": 4817,
+    "찬": 4818,
+    "창": 4819,
+    "채": 4820,
+    "천": 4821,
+    "철": 4822,
+    "청": 4823,
+    "초": 4824,
+    "최": 4825,
+    "추": 4826,
+    "춘": 4827,
+    "출": 4828,
+    "충": 4829,
+    "치": 4830,
+    "칠": 4831,
+    "카": 4832,
+    "케": 4833,
+    "크": 4834,
+    "키": 4835,
+    "타": 4836,
+    "탁": 4837,
+    "태": 4838,
+    "택": 4839,
+    "터": 4840,
+    "트": 4841,
+    "티": 4842,
+    "파": 4843,
+    "판": 4844,
+    "팔": 4845,
+    "평": 4846,
+    "포": 4847,
+    "표": 4848,
+    "풍": 4849,
+    "피": 4850,
+    "필": 4851,
+    "하": 4852,
+    "학": 4853,
+    "한": 4854,
+    "함": 4855,
+    "항": 4856,
+    "해": 4857,
+    "행": 4858,
+    "향": 4859,
+    "허": 4860,
+    "헌": 4861,
+    "혁": 4862,
+    "현": 4863,
+    "협": 4864,
+    "형": 4865,
+    "혜": 4866,
+    "호": 4867,
+    "홍": 4868,
+    "화": 4869,
+    "환": 4870,
+    "황": 4871,
+    "회": 4872,
+    "효": 4873,
+    "후": 4874,
+    "훈": 4875,
+    "휘": 4876,
+    "휴": 4877,
+    "흔": 4878,
+    "흠": 4879,
+    "흥": 4880,
+    "희": 4881,
+    "히": 4882,
+    "﨑": 4883,
+    "ﭐ": 4884,
+    "ﭑ": 4885,
+    "ﭔ": 4886,
+    "ﭕ": 4887,
+    "ﭘ": 4888,
+    "ﭡ": 4889,
+    "ﭤ": 4890,
+    "ﭭ": 4891,
+    "ﭱ": 4892,
+    "ﭼ": 4893,
+    "ﭽ": 4894,
+    "ﮉ": 4895,
+    "ﮊ": 4896,
+    "ﮍ": 4897,
+    "ﮏ": 4898,
+    "ﮐ": 4899,
+    "ﮑ": 4900,
+    "ﮒ": 4901,
+    "ﮓ": 4902,
+    "ﮔ": 4903,
+    "ﮕ": 4904,
+    "ﮥ": 4905,
+    "ﮧ": 4906,
+    "ﮨ": 4907,
+    "ﮩ": 4908,
+    "ﮪ": 4909,
+    "ﮫ": 4910,
+    "ﮭ": 4911,
+    "ﮮ": 4912,
+    "ﮯ": 4913,
+    "ﮰ": 4914,
+    "ﮱ": 4915,
+    "ﯚ": 4916,
+    "ﯛ": 4917,
+    "ﯜ": 4918,
+    "ﯝ": 4919,
+    "ﯞ": 4920,
+    "ﯠ": 4921,
+    "ﯡ": 4922,
+    "ﯢ": 4923,
+    "ﯥ": 4924,
+    "ﯧ": 4925,
+    "ﯼ": 4926,
+    "ﯽ": 4927,
+    "ﯾ": 4928,
+    "ﯿ": 4929,
+    "ﱞ": 4930,
+    "ﱟ": 4931,
+    "﴾": 4932,
+    "﴿": 4933,
+    "ﷻ": 4934,
+    "︠": 4935,
+    "︡": 4936,
+    "ﹷ": 4937,
+    "ﹺ": 4938,
+    "ﹻ": 4939,
+    "ﹾ": 4940,
+    "ﺀ": 4941,
+    "ﺁ": 4942,
+    "ﺂ": 4943,
+    "ﺃ": 4944,
+    "ﺄ": 4945,
+    "ﺅ": 4946,
+    "ﺆ": 4947,
+    "ﺇ": 4948,
+    "ﺈ": 4949,
+    "ﺉ": 4950,
+    "ﺊ": 4951,
+    "ﺋ": 4952,
+    "ﺌ": 4953,
+    "ﺍ": 4954,
+    "ﺎ": 4955,
+    "ﺏ": 4956,
+    "ﺐ": 4957,
+    "ﺑ": 4958,
+    "ﺒ": 4959,
+    "ﺓ": 4960,
+    "ﺔ": 4961,
+    "ﺕ": 4962,
+    "ﺖ": 4963,
+    "ﺗ": 4964,
+    "ﺘ": 4965,
+    "ﺙ": 4966,
+    "ﺚ": 4967,
+    "ﺛ": 4968,
+    "ﺜ": 4969,
+    "ﺝ": 4970,
+    "ﺞ": 4971,
+    "ﺟ": 4972,
+    "ﺠ": 4973,
+    "ﺡ": 4974,
+    "ﺢ": 4975,
+    "ﺣ": 4976,
+    "ﺤ": 4977,
+    "ﺥ": 4978,
+    "ﺦ": 4979,
+    "ﺧ": 4980,
+    "ﺨ": 4981,
+    "ﺩ": 4982,
+    "ﺪ": 4983,
+    "ﺫ": 4984,
+    "ﺬ": 4985,
+    "ﺭ": 4986,
+    "ﺮ": 4987,
+    "ﺯ": 4988,
+    "ﺰ": 4989,
+    "ﺱ": 4990,
+    "ﺲ": 4991,
+    "ﺳ": 4992,
+    "ﺴ": 4993,
+    "ﺵ": 4994,
+    "ﺶ": 4995,
+    "ﺷ": 4996,
+    "ﺸ": 4997,
+    "ﺹ": 4998,
+    "ﺺ": 4999,
+    "ﺻ": 5000,
+    "ﺼ": 5001,
+    "ﺽ": 5002,
+    "ﺾ": 5003,
+    "ﺿ": 5004,
+    "ﻀ": 5005,
+    "ﻁ": 5006,
+    "ﻂ": 5007,
+    "ﻃ": 5008,
+    "ﻄ": 5009,
+    "ﻆ": 5010,
+    "ﻇ": 5011,
+    "ﻈ": 5012,
+    "ﻉ": 5013,
+    "ﻊ": 5014,
+    "ﻋ": 5015,
+    "ﻌ": 5016,
+    "ﻏ": 5017,
+    "ﻐ": 5018,
+    "ﻑ": 5019,
+    "ﻒ": 5020,
+    "ﻓ": 5021,
+    "ﻔ": 5022,
+    "ﻕ": 5023,
+    "ﻖ": 5024,
+    "ﻗ": 5025,
+    "ﻘ": 5026,
+    "ﻙ": 5027,
+    "ﻚ": 5028,
+    "ﻛ": 5029,
+    "ﻜ": 5030,
+    "ﻝ": 5031,
+    "ﻞ": 5032,
+    "ﻟ": 5033,
+    "ﻠ": 5034,
+    "ﻡ": 5035,
+    "ﻢ": 5036,
+    "ﻣ": 5037,
+    "ﻤ": 5038,
+    "ﻥ": 5039,
+    "ﻦ": 5040,
+    "ﻧ": 5041,
+    "ﻨ": 5042,
+    "ﻩ": 5043,
+    "ﻪ": 5044,
+    "ﻫ": 5045,
+    "ﻬ": 5046,
+    "ﻭ": 5047,
+    "ﻮ": 5048,
+    "ﻯ": 5049,
+    "ﻰ": 5050,
+    "ﻱ": 5051,
+    "ﻲ": 5052,
+    "ﻳ": 5053,
+    "ﻴ": 5054,
+    "ﻵ": 5055,
+    "ﻶ": 5056,
+    "ﻷ": 5057,
+    "ﻹ": 5058,
+    "ﻻ": 5059,
+    "ﻼ": 5060,
+    "ａ": 5061,
+    "ｉ": 5062,
+    "�": 5063
+  }
+}

inference.py

@@ -0,0 +1,159 @@
+#!/usr/bin/env python3
+"""genderize (cora / ultra) — standalone inference for the released weights.
+
+What the model is
+-----------------
+A byte-level dual-head convolutional classifier. A personal name is normalised
+(Unicode NFC, lowercase, whitespace collapsed), UTF-8 encoded and truncated to 48 bytes for legacy weights, or to the configured
+``maxlen``. Each byte is embedded (256 x 64), projected to `ch` channels and passed
+through four residual 1-D convolution blocks (kernels 3, 5, 7, 3; BatchNorm +
+GELU). Masked mean-pooling and max-pooling over the byte positions are
+concatenated and fed to a shared 512-unit layer, from which two linear heads
+read out: gender (2 classes, M/F) and country (226 ISO-3166 alpha-2 codes).
+`cora` uses ch=160 (~0.77M parameters), `ultra` uses ch=384 (~3.2M).
+Logits are divided by a per-head temperature (calibration.json) before softmax.
+
+Files expected next to this script (or pass --models-dir):
+  genderize_<variant>.pt                state_dict (PyTorch)
+  genderize_<variant>.config.json       {"ch": 160 | 384}
+  genderize_<variant>.calibration.json  {"temp_gender": t, "temp_country": t}
+  maps.json                             {"gender": {"M":0,"F":1}, "country": {"AD":0, ...}}
+
+Usage
+-----
+  python genderize_infer.py --variant ultra "Andrea Rossi" "Yuki Tanaka" "María García"
+  python genderize_infer.py --variant cora --top 3 --json "Chen Wei"
+
+Requires only torch and numpy. CPU is enough.
+"""
+from __future__ import annotations
+
+import argparse
+import json
+import sys
+import unicodedata
+from pathlib import Path
+
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+MAXLEN = 48
+
+
+class _GBlock(nn.Module):
+    def __init__(self, ch: int, k: int):
+        super().__init__()
+        self.conv = nn.Conv1d(ch, ch, k, padding=k // 2)
+        self.norm = nn.BatchNorm1d(ch)
+
+    def forward(self, x):
+        return x + F.gelu(self.norm(self.conv(x)))
+
+
+class NameModel(nn.Module):
+    """Byte/codepoint classifier: gender (2) + country (n_countries)."""
+
+    def __init__(self, n_countries: int, emb: int = 64, ch: int = 160,
+                 vocab_size: int = 256):
+        super().__init__()
+        self.emb = nn.Embedding(vocab_size, emb, padding_idx=0)
+        self.proj = nn.Conv1d(emb, ch, 1)
+        self.blocks = nn.Sequential(_GBlock(ch, 3), _GBlock(ch, 5), _GBlock(ch, 7), _GBlock(ch, 3))
+        self.shared = nn.Sequential(nn.Linear(ch * 2, 512), nn.GELU(), nn.Dropout(0.15))
+        self.head_g = nn.Linear(512, 2)
+        self.head_c = nn.Linear(512, n_countries)
+
+    def forward(self, x):
+        mask = (x != 0).float().unsqueeze(1)
+        h = self.proj(self.emb(x.long()).transpose(1, 2))
+        h = self.blocks(h) * mask
+        mean = h.sum(-1) / mask.sum(-1).clamp(min=1)
+        mx = h.masked_fill(mask == 0, -1e9).max(-1).values
+        z = self.shared(torch.cat([mean, mx], -1))
+        return self.head_g(z), self.head_c(z)
+
+
+def normalise(name: str) -> str:
+    """NFC + lowercase + collapse whitespace. Must match what the weights saw."""
+    return " ".join(unicodedata.normalize("NFC", (name or "").lower()).split())
+
+
+def encode(texts: list[str], maxlen: int = MAXLEN,
+           vocab_codepoint: dict[str, int] | None = None) -> torch.Tensor:
+    """Codifica byte legacy oppure codepoint (0=pad, 1=unk)."""
+    dtype = np.uint8 if vocab_codepoint is None else np.int64
+    X = np.zeros((len(texts), maxlen), dtype=dtype)
+    for i, t in enumerate(texts):
+        if vocab_codepoint is None:
+            b = t.encode("utf-8")[:maxlen]
+            X[i, : len(b)] = np.frombuffer(b, dtype=np.uint8)
+        else:
+            ids = [vocab_codepoint.get(c, 1) for c in t[:maxlen]]
+            X[i, : len(ids)] = ids
+    return torch.from_numpy(X)
+
+
+class Genderize:
+    def __init__(self, variant: str = "ultra", models_dir: str | Path | None = None):
+        md = Path(models_dir) if models_dir else Path(__file__).resolve().parent
+        maps = json.loads((md / "maps.json").read_text())
+        self.vocab_codepoint = maps.get("vocab_codepoint")
+        self.inv_country = {v: k for k, v in maps["country"].items()}
+        self.idx_m = maps["gender"].get("M", 0)
+        cfg = json.loads((md / f"genderize_{variant}.config.json").read_text())
+        self.maxlen = int(cfg.get("maxlen", maps.get("maxlen", MAXLEN)))
+        cal_path = md / f"genderize_{variant}.calibration.json"
+        self.cal = json.loads(cal_path.read_text()) if cal_path.exists() else {}
+        vocab_size = len(self.vocab_codepoint) + 2 if self.vocab_codepoint is not None else 256
+        self.model = NameModel(len(maps["country"]), ch=int(cfg.get("ch", 160)),
+                               vocab_size=vocab_size)
+        self.model.load_state_dict(torch.load(md / f"genderize_{variant}.pt", map_location="cpu"))
+        self.model.eval()
+        self.variant = variant
+
+    @torch.no_grad()
+    def predict(self, names: list[str], top: int = 5) -> list[dict]:
+        texts = [normalise(n) for n in names]
+        lg, lc = self.model(encode(texts, maxlen=self.maxlen,
+                                   vocab_codepoint=self.vocab_codepoint))
+        pg = F.softmax(lg / self.cal.get("temp_gender", 1.0), dim=-1)
+        pc = F.softmax(lc / self.cal.get("temp_country", 1.0), dim=-1)
+        vals, idx = pc.topk(min(top, pc.shape[-1]), dim=-1)
+        out = []
+        for i, name in enumerate(names):
+            pm = float(pg[i, self.idx_m])
+            gender = "male" if pm >= 0.5 else "female"
+            out.append({
+                "name": name,
+                "gender": gender,
+                "probability": round(pm if gender == "male" else 1.0 - pm, 4),
+                "countries": [{"code": self.inv_country.get(int(j), "??"), "probability": round(float(v), 4)}
+                              for v, j in zip(vals[i].tolist(), idx[i].tolist())],
+                "model": self.variant,
+            })
+        return out
+
+
+def main() -> int:
+    ap = argparse.ArgumentParser(description="genderize cora/ultra — gender + nationality from a name")
+    ap.add_argument("names", nargs="+")
+    ap.add_argument("--variant", choices=["cora", "ultra"], default="ultra")
+    ap.add_argument("--models-dir", default=None)
+    ap.add_argument("--top", type=int, default=5)
+    ap.add_argument("--json", action="store_true", help="print JSON instead of a table")
+    a = ap.parse_args()
+    g = Genderize(a.variant, a.models_dir)
+    rows = g.predict(a.names, top=a.top)
+    if a.json:
+        print(json.dumps(rows, ensure_ascii=False, indent=1))
+    else:
+        for r in rows:
+            cs = ", ".join(f"{c['code']} {c['probability']:.2f}" for c in r["countries"])
+            print(f"{r['name']:<28} {r['gender']:<7} {r['probability']:.3f}   {cs}")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

inference_fuso.py

@@ -0,0 +1,250 @@
+"""genderize v2 «fuso» — un solo modello, due reti interne (v1 ultra + v2 distillata D2).
+
+Uso:
+    g = GenderizeFuso('models/genderize_fuso_v2')
+    g.predict(['Maria Rossi', 'Nguyen Van An'])   # -> [{'gender','p_male','country','p_country','top5'}]
+
+Regole (DECISIONI 25/09): p(M) = media delle p(M) calibrate delle due reti; «M» se p(M) >= soglia
+(0,60: scelta sui banchi, verificata su 6 insiemi freschi); paese = 0,7·v1 + 0,3·v2 sulle probabilità.
+Ogni rete riceve il nome con la propria codifica (v1 byte maxlen 48, v2 codepoint maxlen 32).
+
+T1 — instradamento per scrittura con traslitterazione (DOM-102, 27/09): la scrittura del nome
+viene riconosciuta dai blocchi Unicode e si applica la regola misurata in
+reports/trainer/g3_traslit_T1.json (scelta solo sulla metà di validazione, nessun parametro
+appreso; totale test 85,82 contro 78,11 fuso e 74,31 v1): «nativo» classifica il nome originale,
+«anyascii» la sua traslitterazione, «mix» la media delle probabilità delle due forme; le scritture
+latine e quelle non mappate restano nativo. Il paese è sempre calcolato sul nome originale.
+anyascii è una dipendenza runtime (licenza ISC) richiesta solo quando serve traslitterare.
+Disambiguazioni: kana → giapponese (senza kana il solo Han va in cinese); lettere persiane
+specifiche (پ چ ژ گ ک ی) distinguono il farsi dall'arabo; lettere kazake specifiche
+(ә ғ қ ң ө ұ ү һ) e ucraine (ї є ґ і) distinguono i cirillici, altrimenti russo.
+
+v2.1 (DOM-112, 28/09) — testa paese e selettore per paese:
+• logit adjustment β=0,75 della testa paese: log(p_c) − β·log(π), dove p_c è la probabilità
+  fusa (0,7·v1 + 0,3·v2) sul nome originale e π è la frequenza empirica del corpus v2 letta da
+  models/genderize_fuso_v2/country_prior.json (accanto ai pesi, SHA nel manifest); β scelto solo
+  sulla validation in DOM-105. La distribuzione restituita è rinormalizzata; l'argmax decide
+  il paese.
+• selettore per paese: se il paese predetto (dopo l'aggiustamento) è in S {AM GE MM TG TD WS},
+  la decisione di genere usa il ramo scelto solo sui banchi VECCHI (residuo 226 v1 + nativo;
+  mai il banco 226 v2): v1@0,5 oppure fuso a soglia diversa. Per quelle righe il p(M) restituito
+  è il punteggio del ramo rimappato in modo che il taglio 0,60 riproduca esattamente la
+  decisione del ramo (p_ramo + 0,60 − soglia_ramo, limitato a [0,1]); per tutte le altre righe
+  resta fuso@0,60. «gender» e «p_male» restano sempre coerenti con la stessa regola di taglio.
+
+v2.2 (DOM-113, 28/09) — rimedio «tratti + massa» per GE MM TG TD (WS escluso):
+• se la decisione v2.1 e v1@0,50 discordano E il nome ha tratti tipici di quei paesi, la
+  decisione di genere (e il punteggio, rimappato come il selettore: p_v1 + 0,60 − 0,50) viene
+  dal ramo v1. Ambito: SOLO nomi di scritture latine (la validazione che ha scelto il rimedio
+  è latina; sulle scritture native il ramo v1 leggerebbe byte fuori distribuzione e il trigger
+  non è mai stato misurato). Tratti scelti SOLO sul banco di validazione DOM-113 (Wikidata CC0,
+  congelato; mai il 226 v2) e vincolati a non danneggiare il residuo 226 v1 (−0,023 pp
+  misurati):
+  (i) token esatto del lessico del paese (supporto ≥ 2 in validazione, ovunque nel nome);
+  (ii) onorifico birmano iniziale {Maung, Mg, Saw, Sao, U} con almeno 2 token;
+  (iii) somma delle p_paese aggiustate su {AM GE MM TG TD WS} ≥ 0,15 con il paese a p massima
+  tra {GE MM TG TD WS} attivo. Configurazione in models/genderize_fuso_v2/dom113_rimedio.json
+  (SHA nel manifest del verdetto); se il file manca, il comportamento resta v2.1.
+"""
+from __future__ import annotations
+import json
+from pathlib import Path
+import torch, torch.nn.functional as F
+from inference import NameModel, encode, normalise
+
+try:
+    from anyascii import anyascii as _anyascii
+except ImportError:
+    _anyascii = None
+
+REGOLA_T1 = {
+    'zh': 'nativo', 'ja': 'nativo', 'ko': 'nativo', 'ar': 'nativo', 'he': 'nativo',
+    'th': 'nativo', 'el': 'nativo', 'uk': 'nativo',
+    'ka': 'anyascii', 'hy': 'anyascii', 'kk': 'anyascii', 'ru': 'anyascii',
+    'fa': 'mix', 'bn': 'mix', 'hi': 'mix',
+}
+
+_BLOCCHI = (
+    ('ka', 0x10A0, 0x10FF), ('hy', 0x0530, 0x058F), ('he', 0x0590, 0x05FF),
+    ('ar', 0x0600, 0x06FF), ('bn', 0x0980, 0x09FF), ('hi', 0x0900, 0x097F),
+    ('th', 0x0E00, 0x0E7F), ('el', 0x0370, 0x03FF), ('han', 0x3400, 0x9FFF),
+    ('hangul', 0xAC00, 0xD7AF), ('hiragana', 0x3040, 0x309F), ('katakana', 0x30A0, 0x30FF),
+    ('cyrillic', 0x0400, 0x052F), ('latino', 0x0041, 0x024F),
+)
+_PERSIANO = frozenset('پچژگکی')
+_KAZAKO = frozenset('әғқңөұүһӘҒҚҢӨҰҮҺ')
+_UCRAINO = frozenset('їєґіЇЄҐІ')
+
+
+def script_t1(testo: str):
+    conteggi: dict[str, int] = {}
+    for c in testo:
+        cp = ord(c)
+        for nome, a, b in _BLOCCHI:
+            if a <= cp <= b:
+                conteggi[nome] = conteggi.get(nome, 0) + 1
+                break
+    if not conteggi:
+        return None
+    if 'hiragana' in conteggi or 'katakana' in conteggi:
+        return 'ja'
+    if 'hangul' in conteggi:
+        return 'ko'
+    if 'han' in conteggi:
+        return 'zh'
+    dominante = max(conteggi, key=lambda k: (conteggi[k], k != 'latino'))
+    if dominante == 'latino':
+        return None
+    if dominante == 'ar':
+        return 'fa' if _PERSIANO & set(testo) else 'ar'
+    if dominante == 'cyrillic':
+        if _KAZAKO & set(testo):
+            return 'kk'
+        if _UCRAINO & set(testo):
+            return 'uk'
+        return 'ru'
+    return dominante
+
+
+def _cap(s: str) -> str:
+    return ' '.join(w[:1].upper() + w[1:] for w in s.split())
+
+
+def _traslittera(nome: str) -> str:
+    if _anyascii is None:
+        raise RuntimeError('anyascii è richiesto per le scritture traslitterate: pip install anyascii (licenza ISC)')
+    return _cap(_anyascii(nome)) or nome
+
+
+def varianti_t1(nome: str) -> list[str]:
+    regola = REGOLA_T1.get(script_t1(nome))
+    if regola in (None, 'nativo'):
+        return [nome]
+    tr = _traslittera(nome)
+    return [nome, tr] if regola == 'mix' else [tr]
+
+
+class GenderizeFuso:
+    def __init__(self, cartella: str, device: str = 'cpu', batch_size: int = 256):
+        if batch_size < 1:
+            raise ValueError('batch_size deve essere positivo')
+        self.batch_size = batch_size
+        base = Path(cartella); cfg = json.loads((base / 'fuso.json').read_text())
+        self.soglia, self.peso_paese_v1, self.dev = float(cfg['soglia_M']), float(cfg['peso_paese_v1']), device
+        self.reti = []
+        for sub in ('v1', 'v2'):
+            d = base / sub; c = json.loads((d / 'genderize_ultra.config.json').read_text())
+            cal = json.loads((d / 'genderize_ultra.calibration.json').read_text()); mp = json.loads((d / 'maps.json').read_text())
+            vc = mp.get('vocab_codepoint') if sub == 'v2' else None
+            m = NameModel(len(mp['country']), ch=int(c['ch']), vocab_size=len(vc) + 2 if vc else 256)
+            st = torch.load(d / 'genderize_ultra.pt', map_location='cpu', weights_only=True)
+            m.load_state_dict(st.get('state_dict', st)); m.eval().to(device)
+            paesi = mp['country'] if isinstance(mp['country'], list) else [k for k, _ in sorted(mp['country'].items(), key=lambda x: x[1])]
+            self.reti.append((m, cal, vc, int(c.get('maxlen', mp.get('maxlen', 48))), paesi))
+        self.paesi = self.reti[1][4]
+        self.perm = torch.tensor([self.reti[0][4].index(p) for p in self.paesi])   # v1 -> ordine v2
+        pri = json.loads((base / 'country_prior.json').read_text())
+        self.beta = float(pri['beta'])
+        pr = pri['prior']; assert len(pr) == len(self.paesi) and set(pr) == set(self.paesi)
+        self.log_prior = torch.tensor([float(pr[p]) for p in self.paesi], dtype=torch.float64)
+        sc = pri['selettore']['scelte']
+        self._sel_v1 = torch.zeros(len(self.paesi), dtype=torch.bool)
+        self._sel_off = torch.zeros(len(self.paesi))
+        for c, b in sc.items():
+            i = self.paesi.index(c)
+            self._sel_v1[i] = b['ramo'] == 'v1'
+            self._sel_off[i] = self.soglia - float(b['soglia'])
+        # v2.2 (DOM-113): rimedio «tratti + massa»; senza config resta v2.1
+        self._rim_lessico: frozenset[str] = frozenset()
+        self._rim_onorifici: frozenset[str] = frozenset()
+        self._rim_tau = 1.0
+        rp = base / 'dom113_rimedio.json'
+        if rp.exists():
+            rim = json.loads(rp.read_text())
+            lessico = set()
+            for c in rim['paesi_attivi']:
+                lessico.update(rim['lessico'][c])
+            self._rim_lessico = frozenset(lessico)
+            self._rim_onorifici = frozenset(rim['onorifici_mm'])
+            self._rim_tau = float(rim['tau_massa'])
+            s6 = ['AM', 'GE', 'MM', 'TG', 'TD', 'WS']   # somma di massa (come DOM-113)
+            s5 = ['GE', 'MM', 'TG', 'TD', 'WS']          # argmax di massa (senza AM)
+            self._rim_s6 = torch.tensor([self.paesi.index(c) for c in s6])
+            self._rim_s5 = torch.tensor([self.paesi.index(c) for c in s5])
+            self._rim_attivi = torch.zeros(len(self.paesi), dtype=torch.bool)
+            for c in rim['paesi_attivi']:
+                self._rim_attivi[self.paesi.index(c)] = True
+
+    @torch.no_grad()
+    def probabilita(self, nomi: list[str]):
+        if not nomi:
+            return torch.empty(0), torch.empty((0, len(self.paesi)))
+        # One normalized work list shared by both networks. Only selected rows
+        # add a transliteration; equal forms and repeated names reuse inference.
+        piatti, righe, indice = [], [], {}
+        def aggiungi(nome):
+            testo = normalise(nome)
+            if testo not in indice:
+                indice[testo] = len(piatti)
+                piatti.append(testo)
+            return indice[testo]
+        for n in nomi:
+            regola = REGOLA_T1.get(script_t1(n))
+            a = aggiungi(n)
+            b = aggiungi(_traslittera(n)) if regola in ('anyascii', 'mix') else None
+            righe.append((a, b, regola))
+        pm_reti, pc_reti = [], []
+        for m, cal, vc, ml, _ in self.reti:
+            lg_all, lc_all = [], []
+            for i in range(0, len(piatti), self.batch_size):
+                lg, lc = m(encode(piatti[i:i + self.batch_size], maxlen=ml, vocab_codepoint=vc).to(self.dev))
+                lg_all.append(lg / cal.get('temp_gender', 1.0)); lc_all.append(lc / cal.get('temp_country', 1.0))
+            pm_reti.append(F.softmax(torch.cat(lg_all), -1)[:, 0].cpu())
+            pc_reti.append(F.softmax(torch.cat(lc_all), -1).cpu())
+        a_t = torch.tensor([a for a, _, _ in righe])
+        b_t = torch.tensor([b if b is not None else 0 for _, b, _ in righe])
+        aa_m = torch.tensor([regola == 'anyascii' for _, _, regola in righe], dtype=torch.bool)
+        mix_m = torch.tensor([regola == 'mix' for _, _, regola in righe], dtype=torch.bool)
+        pn = (pm_reti[0][a_t] + pm_reti[1][a_t]) / 2
+        pt = (pm_reti[0][b_t] + pm_reti[1][b_t]) / 2
+        p_m = torch.where(aa_m, pt, torch.where(mix_m, (pn + pt) / 2, pn))
+        p_c = self.peso_paese_v1 * pc_reti[0][a_t][:, self.perm] + (1 - self.peso_paese_v1) * pc_reti[1][a_t]
+        # v2.1: logit adjustment della testa paese — log(p_c) − β·log(π), float64 come DOM-105;
+        # la distribuzione restituita è rinormalizzata e l'argmax decide il paese.
+        lg_c = torch.log(p_c.to(torch.float64)) - self.beta * self.log_prior
+        paese = lg_c.argmax(-1)
+        p_c = torch.softmax(lg_c, -1).to(p_m.dtype)
+        # v2.1: selettore per paese — per i paesi di S il punteggio del ramo scelto è rimappato
+        # perché il taglio self.soglia (0,60) riproduca esattamente la decisione del ramo.
+        base = torch.where(self._sel_v1[paese], pm_reti[0][a_t], p_m)
+        p_m = torch.clamp(base + self._sel_off[paese], 0.0, 1.0)
+        # v2.2: rimedio DOM-113 — se v2.1 e v1 discordano e il nome ha tratti tipici
+        # (lessico / onorifico birmano / massa del paese), decide il ramo v1
+        # (punteggio rimappato: p_v1 + 0,60 − 0,50), come il selettore DOM-112.
+        if self._rim_lessico:
+            s6 = p_c[:, self._rim_s6].sum(-1)
+            cS = self._rim_s5[p_c[:, self._rim_s5].argmax(-1)]
+            massa = (s6 >= self._rim_tau) & self._rim_attivi[cS]
+            # il rimedio vale solo per nomi latini: la validazione DOM-113 che lo ha
+            # scelto è latina; sulle scritture native il ramo v1 girerebbe sui byte
+            # originali (fuori distribuzione) e il trigger non è stato mai misurato
+            tratti = torch.tensor(
+                [script_t1(n) is None and (bool(massa[j]) or
+                 any(t in self._rim_lessico for t in n.split())
+                 or (len(n.split()) >= 2 and n.split()[0] in self._rim_onorifici))
+                 for j, n in enumerate(nomi)], dtype=torch.bool)
+            pv1 = pm_reti[0][a_t]
+            discorda = (p_m >= self.soglia) != (pv1 >= 0.5)
+            p_m = torch.where(tratti & discorda,
+                              torch.clamp(pv1 + self.soglia - 0.50, 0.0, 1.0), p_m)
+        return p_m, p_c
+
+    def predict(self, nomi: list[str]):
+        out = []
+        for i in range(0, len(nomi), self.batch_size):
+            p_m, p_c = self.probabilita(nomi[i:i + self.batch_size]); top = p_c.topk(5, -1)
+            for j in range(len(p_m)):
+                out.append({'gender': 'M' if p_m[j] >= self.soglia else 'F', 'p_male': round(float(p_m[j]), 4),
+                            'country': self.paesi[int(top.indices[j, 0])], 'p_country': round(float(top.values[j, 0]), 4),
+                            'top5': [self.paesi[int(k)] for k in top.indices[j]]})
+        return out