README.md

---
language: gor
language_name: Gorontalo
language_family: austronesian_sulawesi
tags:
  - wikilangs
  - nlp
  - tokenizer
  - embeddings
  - n-gram
  - markov
  - wikipedia
  - feature-extraction
  - sentence-similarity
  - tokenization
  - n-grams
  - markov-chain
  - text-mining
  - fasttext
  - babelvec
  - vocabulous
  - vocabulary
  - monolingual
  - family-austronesian_sulawesi
license: mit
library_name: wikilangs
pipeline_tag: text-generation
datasets:
  - omarkamali/wikipedia-monthly
dataset_info:
  name: wikipedia-monthly
  description: Monthly snapshots of Wikipedia articles across 300+ languages
metrics:
  - name: best_compression_ratio
    type: compression
    value: 5.349
  - name: best_isotropy
    type: isotropy
    value: 0.7535
  - name: vocabulary_size
    type: vocab
    value: 0
generated: 2026-01-04
---

# Gorontalo - Wikilangs Models
## Comprehensive Research Report & Full Ablation Study

This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Gorontalo** Wikipedia data.
We analyze tokenizers, n-gram models, Markov chains, vocabulary statistics, and word embeddings.

## 📋 Repository Contents

### Models & Assets

- Tokenizers (8k, 16k, 32k, 64k)
- N-gram models (2, 3, 4, 5-gram)
- Markov chains (context of 1, 2, 3, 4 and 5)
- Subword N-gram and Markov chains
- Embeddings in various sizes and dimensions (aligned and unaligned)
- Language Vocabulary
- Language Statistics

![Performance Dashboard](visualizations/performance_dashboard.png)

### Analysis and Evaluation

- [1. Tokenizer Evaluation](#1-tokenizer-evaluation)
- [2. N-gram Model Evaluation](#2-n-gram-model-evaluation)
- [3. Markov Chain Evaluation](#3-markov-chain-evaluation)
- [4. Vocabulary Analysis](#4-vocabulary-analysis)
- [5. Word Embeddings Evaluation](#5-word-embeddings-evaluation)
- [6. Morphological Analysis (Experimental)](#6--morphological-analysis-experimental)
- [7. Summary & Recommendations](#7-summary--recommendations)
- [Metrics Glossary](#appendix-metrics-glossary--interpretation-guide)
- [Visualizations Index](#visualizations-index)

---
## 1. Tokenizer Evaluation

![Tokenizer Compression](visualizations/tokenizer_compression.png)

![Tokenizer Fertility](visualizations/tokenizer_fertility.png)

![Tokenizer OOV](visualizations/tokenizer_oov.png)

![Total Tokens](visualizations/tokenizer_total_tokens.png)

### Results

| Vocab Size | Compression | Avg Token Len | UNK Rate | Total Tokens |
|------------|-------------|---------------|----------|--------------|
| **8k** | 4.601x | 4.62 | 0.2718% | 69,912 |
| **16k** | 4.911x | 4.93 | 0.2900% | 65,506 |
| **32k** | 5.139x | 5.16 | 0.3035% | 62,598 |
| **64k** | 5.349x 🏆 | 5.37 | 0.3159% | 60,137 |

### Tokenization Examples

Below are sample sentences tokenized with each vocabulary size:

**Sample 1:** `Lajer yito tala tuwawu lo desa to Kecamatan Ambal, Kabupaten Kebumen, Provinsi J...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁la jer ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁ambal ... (+17 more)` | 27 |
| 16k | `▁la jer ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁ambal ... (+17 more)` | 27 |
| 32k | `▁lajer ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁ambal , ... (+16 more)` | 26 |
| 64k | `▁lajer ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁ambal , ... (+16 more)` | 26 |

**Sample 2:** `Sidomulyo yito tala tuwawu lo desa to Kecamatan Semen, Kabupaten Kediri, Provins...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁sidomulyo ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁semen , ... (+16 more)` | 26 |
| 16k | `▁sidomulyo ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁semen , ... (+16 more)` | 26 |
| 32k | `▁sidomulyo ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁semen , ... (+16 more)` | 26 |
| 64k | `▁sidomulyo ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁semen , ... (+16 more)` | 26 |

**Sample 3:** `Pengejaran yito tala tuwawu lo desa to Kecamatan Kintamani, Kabupaten Bangli, Pr...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁penge jaran ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁kintamani ... (+15 more)` | 25 |
| 16k | `▁penge jaran ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁kintamani ... (+15 more)` | 25 |
| 32k | `▁pengejaran ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁kintamani , ... (+14 more)` | 24 |
| 64k | `▁pengejaran ▁yito ▁tala ▁tuwawu ▁lo ▁desa ▁to ▁kecamatan ▁kintamani , ... (+14 more)` | 24 |


### Key Findings

- **Best Compression:** 64k achieves 5.349x compression
- **Lowest UNK Rate:** 8k with 0.2718% unknown tokens
- **Trade-off:** Larger vocabularies improve compression but increase model size
- **Recommendation:** 32k vocabulary provides optimal balance for production use

---
## 2. N-gram Model Evaluation

![N-gram Perplexity](visualizations/ngram_perplexity.png)

![N-gram Unique](visualizations/ngram_unique.png)

![N-gram Coverage](visualizations/ngram_coverage.png)

### Results

| N-gram | Variant | Perplexity | Entropy | Unique N-grams | Top-100 Coverage | Top-1000 Coverage |
|--------|---------|------------|---------|----------------|------------------|-------------------|
| **2-gram** | Word | 993 | 9.96 | 11,670 | 50.6% | 74.0% |
| **2-gram** | Subword | 193 🏆 | 7.59 | 1,918 | 76.2% | 99.8% |
| **3-gram** | Word | 1,301 | 10.35 | 13,510 | 43.5% | 73.2% |
| **3-gram** | Subword | 1,161 | 10.18 | 14,643 | 40.9% | 82.7% |
| **4-gram** | Word | 2,359 | 11.20 | 22,939 | 34.9% | 65.1% |
| **4-gram** | Subword | 4,301 | 12.07 | 70,956 | 30.9% | 59.4% |
| **5-gram** | Word | 3,036 | 11.57 | 19,983 | 29.7% | 60.3% |
| **5-gram** | Subword | 9,570 | 13.22 | 162,371 | 27.8% | 49.7% |

### Top 5 N-grams by Size

**2-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `tala tuwawu` | 13,812 |
| 2 | `tuwawu lo` | 13,680 |
| 3 | `yito tala` | 13,639 |
| 4 | `to kabupaten` | 13,061 |
| 5 | `to kecamatan` | 12,421 |

**3-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `yito tala tuwawu` | 13,623 |
| 2 | `tala tuwawu lo` | 13,621 |
| 3 | `tuwawu lo desa` | 10,100 |
| 4 | `lo desa to` | 9,932 |
| 5 | `desa to kecamatan` | 9,896 |

**4-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `yito tala tuwawu lo` | 13,541 |
| 2 | `tala tuwawu lo desa` | 10,097 |
| 3 | `tuwawu lo desa to` | 9,928 |
| 4 | `lo desa to kecamatan` | 9,892 |
| 5 | `indonesia referensi to kabupaten` | 5,671 |

**5-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `yito tala tuwawu lo desa` | 10,097 |
| 2 | `tala tuwawu lo desa to` | 9,928 |
| 3 | `tuwawu lo desa to kecamatan` | 9,892 |
| 4 | `provinsi jawa timur indonesia referensi` | 2,631 |
| 5 | `jawa timur indonesia referensi to` | 2,240 |

**2-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ t` | 132,254 |
| 2 | `o _` | 131,021 |
| 3 | `a n` | 117,480 |
| 4 | `a _` | 103,041 |
| 5 | `t a` | 92,939 |

**3-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `t o _` | 73,949 |
| 2 | `_ t o` | 55,873 |
| 3 | `a n _` | 52,432 |
| 4 | `a _ t` | 42,455 |
| 5 | `s i _` | 41,839 |

**4-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ t o _` | 52,046 |
| 2 | `t o _ k` | 28,897 |
| 3 | `p a t e` | 26,517 |
| 4 | `_ k a b` | 26,515 |
| 5 | `n s i _` | 26,402 |

**5-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ t o _ k` | 28,543 |
| 2 | `b u p a t` | 26,199 |
| 3 | `p a t e n` | 25,982 |
| 4 | `k a b u p` | 25,981 |
| 5 | `a b u p a` | 25,976 |


### Key Findings

- **Best Perplexity:** 2-gram (subword) with 193
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
- **Coverage:** Top-1000 patterns cover ~50% of corpus
- **Recommendation:** 4-gram or 5-gram for best predictive performance

---
## 3. Markov Chain Evaluation

![Markov Entropy](visualizations/markov_entropy.png)

![Markov Contexts](visualizations/markov_contexts.png)

![Markov Branching](visualizations/markov_branching.png)

### Results

| Context | Variant | Avg Entropy | Perplexity | Branching Factor | Unique Contexts | Predictability |
|---------|---------|-------------|------------|------------------|-----------------|----------------|
| **1** | Word | 0.7343 | 1.664 | 4.45 | 55,977 | 26.6% |
| **1** | Subword | 0.7592 | 1.693 | 5.36 | 1,113 | 24.1% |
| **2** | Word | 0.2089 | 1.156 | 1.42 | 248,155 | 79.1% |
| **2** | Subword | 0.8081 | 1.751 | 4.92 | 5,961 | 19.2% |
| **3** | Word | 0.0606 | 1.043 | 1.10 | 350,996 | 93.9% |
| **3** | Subword | 0.8459 | 1.797 | 4.10 | 29,344 | 15.4% |
| **4** | Word | 0.0227 🏆 | 1.016 | 1.04 | 384,496 | 97.7% |
| **4** | Subword | 0.6476 | 1.567 | 2.71 | 120,194 | 35.2% |

### Generated Text Samples (Word-based)

Below are text samples generated from each word-based Markov chain model:

**Context Size 1:**

1. `to as dan kemudian lulusan sma katolik atau minang andala yito tala tuwawu masjid agung dari`
2. `kabupaten parigi barat ntb 31 juni september 22 23 april sambe sma negeri chu penyanyi asal`
3. `lo desa to kabupaten kebumen to sulawesi selatan parigi moutong to jawa timur to kabupaten indramayu`

**Context Size 2:**

1. `tala tuwawu lo desa to kecamatan somba opu kabupaten gowa to kabupaten konawe provinsi sulawesi teng...`
2. `tuwawu lo desa to kecamatan ayah kabupaten kebumen to jawa timur indonesia referensi to kota surabay...`
3. `yito tala tuwawu lo desa to kecamatan sekaran kabupaten lamongan to jawa timur to indonesia hari bel...`

**Context Size 3:**

1. `yito tala tuwawu lo kelurahan to kecamatan tompobulu kabupaten bantaeng provinsi sulawesi selatan in...`
2. `tala tuwawu lo kelurahan to kecamatan enrekang kabupaten enrekang provinsi sulawesi selatan indonesi...`
3. `tuwawu lo desa to kecamatan tabanan kabupaten tabanan provinsi bali indonesia referensi to kabupaten...`

**Context Size 4:**

1. `yito tala tuwawu lo desa to kecamatan sahu kabupaten halmahera barat provinsi maluku utara indonesia...`
2. `tala tuwawu lo desa to kecamatan pasarwajo kabupaten buton provinsi sulawesi tenggara indonesia refe...`
3. `tuwawu lo desa to kecamatan prajekan kabupaten bondowoso provinsi jawa timur indonesia referensi to ...`


### Generated Text Samples (Subword-based)

Below are text samples generated from each subword-based Markov chain model:

**Context Size 1:**

1. `_tawahi_bu_lo_t_`
2. `andresema_tengov`
3. `nolo_inyot_tando`

**Context Size 2:**

1. `_to_kotan_bontin_`
2. `o_to_dioneng_bang`
3. `an_timus_kecamasa`

**Context Size 3:**

1. `to_kutara,_to_nusa`
2. `_to_ngodiyo_lendra`
3. `an_desa_to_sikorbe`

**Context Size 4:**

1. `_to_kecamatan_to_ba`
2. `to_kecamatan_parigi`
3. `paten_hongajara_pem`


### Key Findings

- **Best Predictability:** Context-4 (word) with 97.7% predictability
- **Branching Factor:** Decreases with context size (more deterministic)
- **Memory Trade-off:** Larger contexts require more storage (120,194 contexts)
- **Recommendation:** Context-3 or Context-4 for text generation

---
## 4. Vocabulary Analysis

![Zipf's Law](visualizations/zipf_law.png)

![Top Words](visualizations/top20_words.png)

![Coverage Curve](visualizations/vocab_coverage.png)

### Statistics

| Metric | Value |
|--------|-------|
| Vocabulary Size | 23,411 |
| Total Tokens | 697,935 |
| Mean Frequency | 29.81 |
| Median Frequency | 4 |
| Frequency Std Dev | 517.90 |

### Most Common Words

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | to | 52,619 |
| 2 | kabupaten | 25,927 |
| 3 | lo | 21,709 |
| 4 | indonesia | 17,695 |
| 5 | yito | 15,438 |
| 6 | kecamatan | 15,418 |
| 7 | provinsi | 14,413 |
| 8 | tuwawu | 14,287 |
| 9 | tala | 13,963 |
| 10 | referensi | 12,443 |

### Least Common Words (from vocabulary)

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | kiblati | 2 |
| 2 | modelowa | 2 |
| 3 | lienchiang | 2 |
| 4 | sekitarliyo | 2 |
| 5 | lopotanda | 2 |
| 6 | hemoklaim | 2 |
| 7 | wangchuck | 2 |
| 8 | bodhisatva | 2 |
| 9 | rekontruksi | 2 |
| 10 | ཐིམ | 2 |

### Zipf's Law Analysis

| Metric | Value |
|--------|-------|
| Zipf Coefficient | 1.1171 |
| R² (Goodness of Fit) | 0.996028 |
| Adherence Quality | **excellent** |

### Coverage Analysis

| Top N Words | Coverage |
|-------------|----------|
| Top 100 | 54.1% |
| Top 1,000 | 76.9% |
| Top 5,000 | 90.3% |
| Top 10,000 | 95.2% |

### Key Findings

- **Zipf Compliance:** R²=0.9960 indicates excellent adherence to Zipf's law
- **High Frequency Dominance:** Top 100 words cover 54.1% of corpus
- **Long Tail:** 13,411 words needed for remaining 4.8% coverage

---
## 5. Word Embeddings Evaluation

![Embedding Isotropy](visualizations/embedding_isotropy.png)

![Similarity Matrix](visualizations/embedding_similarity.png)

![t-SNE Words](visualizations/tsne_words.png)

![t-SNE Sentences](visualizations/tsne_sentences.png)


### 5.1 Cross-Lingual Alignment

![Alignment Quality](visualizations/embedding_alignment_quality.png)

![Multilingual t-SNE](visualizations/embedding_tsne_multilingual.png)


### 5.2 Model Comparison

| Model | Dimension | Isotropy | Semantic Density | Alignment R@1 | Alignment R@10 |
|-------|-----------|----------|------------------|---------------|----------------|
| **mono_32d** | 32 | 0.7535 🏆 | 0.3948 | N/A | N/A |
| **mono_64d** | 64 | 0.3983 | 0.3683 | N/A | N/A |
| **mono_128d** | 128 | 0.0933 | 0.3561 | N/A | N/A |
| **aligned_32d** | 32 | 0.7535 | 0.3841 | 0.0360 | 0.2160 |
| **aligned_64d** | 64 | 0.3983 | 0.3682 | 0.0340 | 0.2280 |
| **aligned_128d** | 128 | 0.0933 | 0.3466 | 0.0840 | 0.2920 |

### Key Findings

- **Best Isotropy:** mono_32d with 0.7535 (more uniform distribution)
- **Semantic Density:** Average pairwise similarity of 0.3697. Lower values indicate better semantic separation.
- **Alignment Quality:** Aligned models achieve up to 8.4% R@1 in cross-lingual retrieval.
- **Recommendation:** 128d aligned for best cross-lingual performance

---
## 6.  Morphological Analysis (Experimental)

This section presents an automated morphological analysis derived from the statistical divergence between word-level and subword-level models. By analyzing where subword predictability spikes and where word-level coverage fails, we can infer linguistic structures without supervised data.

### 6.1 Productivity & Complexity

| Metric | Value | Interpretation | Recommendation |
|--------|-------|----------------|----------------|
| Productivity Index | **5.000** | High morphological productivity | Reliable analysis |
| Idiomaticity Gap | **0.214** | High formulaic/idiomatic content | - |

### 6.2 Affix Inventory (Productive Units)

These are the most productive prefixes and suffixes identified by sampling the vocabulary for global substitutability patterns. A unit is considered an affix if stripping it leaves a valid stem that appears in other contexts.

#### Productive Prefixes
| Prefix | Examples |
|--------|----------|
| `-me` | menghabiskan, menyerah, membina |
| `-pe` | perkembangbiakan, penyimpanan, pengepungan |
| `-mo` | molihuto, mopiohu, mototoheto |

#### Productive Suffixes
| Suffix | Examples |
|--------|----------|
| `-a` | imana, yilorusa, babaliya |
| `-n` | michigan, gurun, kekebalan |
| `-an` | michigan, kekebalan, menghabiskan |
| `-ng` | kucing, kindang, dipasang |
| `-yo` | potaliliyo, delomiyo, kuasaliyo |
| `-iyo` | potaliliyo, delomiyo, kuasaliyo |
| `-kan` | menghabiskan, perkembangbiakan, dibatalkan |

### 6.3 Bound Stems (Lexical Roots)

Bound stems are high-frequency subword units that are semantically cohesive but rarely appear as standalone words. These often correspond to the 'core' of a word that requires inflection or derivation to be valid.

| Stem | Cohesion | Substitutability | Examples |
|------|----------|------------------|----------|
| `anga` | 1.65x | 93 contexts | langa, banga, sanga |
| `rang` | 1.64x | 62 contexts | range, orang, orange |
| `angg` | 1.38x | 115 contexts | tanggu, wangga, anggia |
| `enga` | 1.72x | 38 contexts | engau, sengau, dengan |
| `mong` | 1.84x | 26 contexts | mongo, omong, among |
| `ngan` | 1.72x | 30 contexts | dengan, nganga, pangan |
| `aran` | 1.46x | 56 contexts | haran, saran, siaran |
| `ngga` | 1.34x | 77 contexts | jingga, wangga, mangga |
| `anta` | 1.46x | 53 contexts | banta, santa, panta |
| `arat` | 1.68x | 28 contexts | barat, marat, darat |
| `ahan` | 1.51x | 41 contexts | jahan, lahan, bahan |
| `owal` | 1.77x | 23 contexts | owala, owali, owalo |

### 6.4 Affix Compatibility (Co-occurrence)

This table shows which prefixes and suffixes most frequently co-occur on the same stems, revealing the 'stacking' rules of the language's morphology.

| Prefix | Suffix | Frequency | Examples |
|--------|--------|-----------|----------|
| `-pe` | `-n` | 171 words | persimpangan, persiapan |
| `-pe` | `-an` | 157 words | persimpangan, persiapan |
| `-me` | `-n` | 111 words | mengumpulkan, menyiapkan |
| `-me` | `-an` | 107 words | mengumpulkan, menyiapkan |
| `-me` | `-kan` | 103 words | mengumpulkan, menyiapkan |
| `-mo` | `-a` | 44 words | motita, modaha |
| `-pe` | `-a` | 35 words | pertamanya, penjara |
| `-me` | `-a` | 24 words | menggawa, meiliana |
| `-pe` | `-ng` | 13 words | performing, petambang |
| `-pe` | `-kan` | 11 words | percetakan, penunjukan |

### 6.5 Recursive Morpheme Segmentation

Using **Recursive Hierarchical Substitutability**, we decompose complex words into their constituent morphemes. This approach handles nested affixes (e.g., `prefix-prefix-root-suffix`).

| Word | Suggested Split | Confidence | Stem |
|------|-----------------|------------|------|
| limbangan | **`limba-ng-an`** | 6.0 | `limba` |
| pelepasan | **`pe-lepas-an`** | 6.0 | `lepas` |
| penerbangan | **`pe-nerba-ng-an`** | 4.5 | `nerba` |
| mempertimbangkan | **`me-mpertimba-ng-kan`** | 4.5 | `mpertimba` |
| tanggapan | **`tanggap-an`** | 4.5 | `tanggap` |
| bersamaan | **`bersama-an`** | 4.5 | `bersama` |
| memperjuangkan | **`me-mperjua-ng-kan`** | 4.5 | `mperjua` |
| molanggato | **`mo-langgato`** | 4.5 | `langgato` |
| mohulango | **`mo-hulango`** | 4.5 | `hulango` |
| motilango | **`mo-tilango`** | 4.5 | `tilango` |
| memerdekakan | **`me-me-rdeka-kan`** | 4.5 | `rdeka` |
| penggulingan | **`pe-ngguli-ng-an`** | 4.5 | `ngguli` |
| sampaikan | **`sampai-kan`** | 4.5 | `sampai` |
| pertandingan | **`pe-rtandi-ng-an`** | 4.5 | `rtandi` |
| pembangkangan | **`pe-mbangka-ng-an`** | 4.5 | `mbangka` |

### 6.6 Linguistic Interpretation

> **Automated Insight:**
The language Gorontalo shows high morphological productivity. The subword models are significantly more efficient than word models, suggesting a rich system of affixation or compounding.

---
## 7. Summary & Recommendations

![Performance Dashboard](visualizations/performance_dashboard.png)

### Production Recommendations

| Component | Recommended | Rationale |
|-----------|-------------|-----------|
| Tokenizer | **64k BPE** | Best compression (5.35x) |
| N-gram | **2-gram** | Lowest perplexity (193) |
| Markov | **Context-4** | Highest predictability (97.7%) |
| Embeddings | **100d** | Balanced semantic capture and isotropy |


---
## Appendix: Metrics Glossary & Interpretation Guide

This section provides definitions, intuitions, and guidance for interpreting the metrics used throughout this report.

### Tokenizer Metrics

**Compression Ratio**
> *Definition:* The ratio of characters to tokens (chars/token). Measures how efficiently the tokenizer represents text.
>
> *Intuition:* Higher compression means fewer tokens needed to represent the same text, reducing sequence lengths for downstream models. A 3x compression means ~3 characters per token on average.
>
> *What to seek:* Higher is generally better for efficiency, but extremely high compression may indicate overly aggressive merging that loses morphological information.

**Average Token Length (Fertility)**
> *Definition:* Mean number of characters per token produced by the tokenizer.
>
> *Intuition:* Reflects the granularity of tokenization. Longer tokens capture more context but may struggle with rare words; shorter tokens are more flexible but increase sequence length.
>
> *What to seek:* Balance between 2-5 characters for most languages. Arabic/morphologically-rich languages may benefit from slightly longer tokens.

**Unknown Token Rate (OOV Rate)**
> *Definition:* Percentage of tokens that map to the unknown/UNK token, indicating words the tokenizer cannot represent.
>
> *Intuition:* Lower OOV means better vocabulary coverage. High OOV indicates the tokenizer encounters many unseen character sequences.
>
> *What to seek:* Below 1% is excellent; below 5% is acceptable. BPE tokenizers typically achieve very low OOV due to subword fallback.

### N-gram Model Metrics

**Perplexity**
> *Definition:* Measures how "surprised" the model is by test data. Mathematically: 2^(cross-entropy). Lower values indicate better prediction.
>
> *Intuition:* If perplexity is 100, the model is as uncertain as if choosing uniformly among 100 options at each step. A perplexity of 10 means effectively choosing among 10 equally likely options.
>
> *What to seek:* Lower is better. Perplexity decreases with larger n-grams (more context). Values vary widely by language and corpus size.

**Entropy**
> *Definition:* Average information content (in bits) needed to encode the next token given the context. Related to perplexity: perplexity = 2^entropy.
>
> *Intuition:* High entropy means high uncertainty/randomness; low entropy means predictable patterns. Natural language typically has entropy between 1-4 bits per character.
>
> *What to seek:* Lower entropy indicates more predictable text patterns. Entropy should decrease as n-gram size increases.

**Coverage (Top-K)**
> *Definition:* Percentage of corpus occurrences explained by the top K most frequent n-grams.
>
> *Intuition:* High coverage with few patterns indicates repetitive/formulaic text; low coverage suggests diverse vocabulary usage.
>
> *What to seek:* Depends on use case. For language modeling, moderate coverage (40-60% with top-1000) is typical for natural text.

### Markov Chain Metrics

**Average Entropy**
> *Definition:* Mean entropy across all contexts, measuring average uncertainty in next-word prediction.
>
> *Intuition:* Lower entropy means the model is more confident about what comes next. Context-1 has high entropy (many possible next words); Context-4 has low entropy (few likely continuations).
>
> *What to seek:* Decreasing entropy with larger context sizes. Very low entropy (<0.1) indicates highly deterministic transitions.

**Branching Factor**
> *Definition:* Average number of unique next tokens observed for each context.
>
> *Intuition:* High branching = many possible continuations (flexible but uncertain); low branching = few options (predictable but potentially repetitive).
>
> *What to seek:* Branching factor should decrease with context size. Values near 1.0 indicate nearly deterministic chains.

**Predictability**
> *Definition:* Derived metric: (1 - normalized_entropy) × 100%. Indicates how deterministic the model's predictions are.
>
> *Intuition:* 100% predictability means the next word is always certain; 0% means completely random. Real text falls between these extremes.
>
> *What to seek:* Higher predictability for text generation quality, but too high (>98%) may produce repetitive output.

### Vocabulary & Zipf's Law Metrics

**Zipf's Coefficient**
> *Definition:* The slope of the log-log plot of word frequency vs. rank. Zipf's law predicts this should be approximately -1.
>
> *Intuition:* A coefficient near -1 indicates the corpus follows natural language patterns where a few words are very common and most words are rare.
>
> *What to seek:* Values between -0.8 and -1.2 indicate healthy natural language distribution. Deviations may suggest domain-specific or artificial text.

**R² (Coefficient of Determination)**
> *Definition:* Measures how well the linear fit explains the frequency-rank relationship. Ranges from 0 to 1.
>
> *Intuition:* R² near 1.0 means the data closely follows Zipf's law; lower values indicate deviation from expected word frequency patterns.
>
> *What to seek:* R² > 0.95 is excellent; > 0.99 indicates near-perfect Zipf adherence typical of large natural corpora.

**Vocabulary Coverage**
> *Definition:* Cumulative percentage of corpus tokens accounted for by the top N words.
>
> *Intuition:* Shows how concentrated word usage is. If top-100 words cover 50% of text, the corpus relies heavily on common words.
>
> *What to seek:* Top-100 covering 30-50% is typical. Higher coverage indicates more repetitive text; lower suggests richer vocabulary.

### Word Embedding Metrics

**Isotropy**
> *Definition:* Measures how uniformly distributed vectors are in the embedding space. Computed as the ratio of minimum to maximum singular values.
>
> *Intuition:* High isotropy (near 1.0) means vectors spread evenly in all directions; low isotropy means vectors cluster in certain directions, reducing expressiveness.
>
> *What to seek:* Higher isotropy generally indicates better-quality embeddings. Values > 0.1 are reasonable; > 0.3 is good. Lower-dimensional embeddings tend to have higher isotropy.

**Average Norm**
> *Definition:* Mean magnitude (L2 norm) of word vectors in the embedding space.
>
> *Intuition:* Indicates the typical "length" of vectors. Consistent norms suggest stable training; high variance may indicate some words are undertrained.
>
> *What to seek:* Relatively consistent norms across models. The absolute value matters less than consistency (low std deviation).

**Cosine Similarity**
> *Definition:* Measures angular similarity between vectors, ranging from -1 (opposite) to 1 (identical direction).
>
> *Intuition:* Words with similar meanings should have high cosine similarity. This is the standard metric for semantic relatedness in embeddings.
>
> *What to seek:* Semantically related words should score > 0.5; unrelated words should be near 0. Synonyms often score > 0.7.

**t-SNE Visualization**
> *Definition:* t-Distributed Stochastic Neighbor Embedding - a dimensionality reduction technique that preserves local structure for visualization.
>
> *Intuition:* Clusters in t-SNE plots indicate groups of semantically related words. Spread indicates vocabulary diversity; tight clusters suggest semantic coherence.
>
> *What to seek:* Meaningful clusters (e.g., numbers together, verbs together). Avoid over-interpreting distances - t-SNE preserves local, not global, structure.

### General Interpretation Guidelines

1. **Compare within model families:** Metrics are most meaningful when comparing models of the same type (e.g., 8k vs 64k tokenizer).
2. **Consider trade-offs:** Better performance on one metric often comes at the cost of another (e.g., compression vs. OOV rate).
3. **Context matters:** Optimal values depend on downstream tasks. Text generation may prioritize different metrics than classification.
4. **Corpus influence:** All metrics are influenced by corpus characteristics. Wikipedia text differs from social media or literature.
5. **Language-specific patterns:** Morphologically rich languages (like Arabic) may show different optimal ranges than analytic languages.


### Visualizations Index

| Visualization | Description |
|---------------|-------------|
| Tokenizer Compression | Compression ratios by vocabulary size |
| Tokenizer Fertility | Average token length by vocabulary |
| Tokenizer OOV | Unknown token rates |
| Tokenizer Total Tokens | Total tokens by vocabulary |
| N-gram Perplexity | Perplexity by n-gram size |
| N-gram Entropy | Entropy by n-gram size |
| N-gram Coverage | Top pattern coverage |
| N-gram Unique | Unique n-gram counts |
| Markov Entropy | Entropy by context size |
| Markov Branching | Branching factor by context |
| Markov Contexts | Unique context counts |
| Zipf's Law | Frequency-rank distribution with fit |
| Vocab Frequency | Word frequency distribution |
| Top 20 Words | Most frequent words |
| Vocab Coverage | Cumulative coverage curve |
| Embedding Isotropy | Vector space uniformity |
| Embedding Norms | Vector magnitude distribution |
| Embedding Similarity | Word similarity heatmap |
| Nearest Neighbors | Similar words for key terms |
| t-SNE Words | 2D word embedding visualization |
| t-SNE Sentences | 2D sentence embedding visualization |
| Position Encoding | Encoding method comparison |
| Model Sizes | Storage requirements |
| Performance Dashboard | Comprehensive performance overview |

---
## About This Project

### Data Source

Models trained on [wikipedia-monthly](https://huggingface.co/datasets/omarkamali/wikipedia-monthly) - a monthly snapshot of Wikipedia articles across 300+ languages.

### Project

A project by **[Wikilangs](https://wikilangs.org)** - Open-source NLP models for every Wikipedia language.

### Maintainer

[Omar Kamali](https://omarkamali.com) - [Omneity Labs](https://omneitylabs.com)

### Citation

If you use these models in your research, please cite:

```bibtex
@misc{wikilangs2025,
  author = {Kamali, Omar},
  title = {Wikilangs: Open NLP Models for Wikipedia Languages},
  year = {2025},
  doi = {10.5281/zenodo.18073153},
  publisher = {Zenodo},
  url = {https://huggingface.co/wikilangs}
  institution = {Omneity Labs}
}
```

### License

MIT License - Free for academic and commercial use.

### Links

- 🌐 Website: [wikilangs.org](https://wikilangs.org)
- 🤗 Models: [huggingface.co/wikilangs](https://huggingface.co/wikilangs)
- 📊 Data: [wikipedia-monthly](https://huggingface.co/datasets/omarkamali/wikipedia-monthly)
- 👤 Author: [Omar Kamali](https://huggingface.co/omarkamali)
- 🤝 Sponsor: [Featherless AI](https://featherless.ai)
---
*Generated by Wikilangs Models Pipeline*

*Report Date: 2026-01-04 15:24:59*