README.md

---
language: dz
language_name: Dzongkha
language_family: tibetoburman_tibetic
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-tibetoburman_tibetic
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: 5.510
  - name: best_isotropy
    type: isotropy
    value: 0.6999
  - name: vocabulary_size
    type: vocab
    value: 0
generated: 2026-01-04
---

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

This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Dzongkha** 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** | 4.484x | 4.49 | 0.0965% | 813,691 |
| **16k** | 4.768x | 4.77 | 0.1026% | 765,197 |
| **32k** | 5.092x | 5.09 | 0.1096% | 716,539 |
| **64k** | 5.510x 🏆 | 5.51 | 0.1185% | 662,175 |

### Tokenization Examples

Below are sample sentences tokenized with each vocabulary size:

**Sample 1:** `རྒྱལ་ཁབ ཇཱ་པཱན། 日本 ཇ་པན་གྱི་རྒྱལ་ཁབ་འདི་ཤར་ཨེ་ཤི་ཡ་ལུ་ཆགས་ཏི་ཡོད་མི་མཚོ་གླིང་གྱི...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁རྒྱལ་ཁབ ▁ཇ ཱ་ པ ཱན། ▁ 日本 ▁ཇ་པ ན་ གྱི་རྒྱལ་ཁབ་ ... (+31 more)` | 41 |
| 16k | `▁རྒྱལ་ཁབ ▁ཇཱ་པཱན། ▁ 日本 ▁ཇ་པན་ གྱི་རྒྱལ་ཁབ་ འདི་ ཤར་ཨེ་ཤི་ཡ་ ལུ་ཆགས་ ཏི་ ... (+23 more)` | 33 |
| 32k | `▁རྒྱལ་ཁབ ▁ཇཱ་པཱན། ▁ 日本 ▁ཇ་པན་ གྱི་རྒྱལ་ཁབ་ འདི་ཤར་ཨེ་ཤི་ཡ་ ལུ་ཆགས་ཏི་ ཡོད་མི་ མཚོ་གླིང་གྱི་ ... (+12 more)` | 22 |
| 64k | `▁རྒྱལ་ཁབ ▁ཇཱ་པཱན། ▁ 日本 ▁ཇ་པན་ གྱི་རྒྱལ་ཁབ་ འདི་ཤར་ཨེ་ཤི་ཡ་ ལུ་ཆགས་ཏི་ ཡོད་མི་ མཚོ་གླིང་གྱི་ ... (+12 more)` | 22 |

**Sample 2:** `སེམས་ཅན བྱི་ལི ཁྱི ཉ སྟག​ བྱམོ དོམ ལུག རྟ བྱི་ཙི པར་རིས་བར་འཁྱམས། ཁུངས་གཏུག། ཕྱི...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁སེམས་ ཅན ▁བྱི་ ལི ▁ཁྱ ི ▁ཉ ▁སྟ ག ▁བྱ ... (+15 more)` | 25 |
| 16k | `▁སེམས་ཅན ▁བྱི་ལི ▁ཁྱ ི ▁ཉ ▁སྟ ག ▁བྱ མོ ▁ད ... (+13 more)` | 23 |
| 32k | `▁སེམས་ཅན ▁བྱི་ལི ▁ཁྱི ▁ཉ ▁སྟག ▁བྱམོ ▁དོམ ▁ལུག ▁རྟ ▁བྱི་ཙི ... (+5 more)` | 15 |
| 64k | `▁སེམས་ཅན ▁བྱི་ལི ▁ཁྱི ▁ཉ ▁སྟག ▁བྱམོ ▁དོམ ▁ལུག ▁རྟ ▁བྱི་ཙི ... (+5 more)` | 15 |

**Sample 3:** `ཞི་ཆོག་གི་སྐབས་ལུ་འཕུ་ནི་གི་ཆོས་ཆས། རྒྱ་མཚོ་ནང་གི་སེམས་ཅན་ཅིག་གི་ཕྱི་ཤུབས། དུང་ད...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁ཞི་ ཆོག་ གི་ སྐབས་ལུ་ འཕ ུ་ ནི་གི་ ཆོས་ ཆས། ▁རྒྱ་མཚོ་ ... (+15 more)` | 25 |
| 16k | `▁ཞི་ ཆོག་ གི་སྐབས་ལུ་ འཕ ུ་ ནི་གི་ ཆོས་ ཆས། ▁རྒྱ་མཚོ་ ནང་གི་ ... (+12 more)` | 22 |
| 32k | `▁ཞི་ཆོག་ གི་སྐབས་ལུ་ འཕུ་ནི་གི་ ཆོས་ཆས། ▁རྒྱ་མཚོ་ ནང་གི་སེམས་ཅན་ ཅིག་གི་ཕྱི་ཤུབས། ▁དུང་དཀར་གྱི་ མིང་གཞན་ ▁སྐྱེ་བ་ལྔ་པ་ ... (+1 more)` | 11 |
| 64k | `▁ཞི་ཆོག་ གི་སྐབས་ལུ་ འཕུ་ནི་གི་ ཆོས་ཆས། ▁རྒྱ་མཚོ་ ནང་གི་སེམས་ཅན་ ཅིག་གི་ཕྱི་ཤུབས། ▁དུང་དཀར་གྱི་ མིང་གཞན་ ▁སྐྱེ་བ་ལྔ་པ་ ... (+1 more)` | 11 |


### Key Findings

- **Best Compression:** 64k achieves 5.510x compression
- **Lowest UNK Rate:** 8k with 0.0965% 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 | 11,790 | 13.53 | 28,884 | 11.2% | 35.3% |
| **2-gram** | Subword | 488 🏆 | 8.93 | 5,527 | 57.6% | 90.8% |
| **3-gram** | Word | 34,131 | 15.06 | 59,067 | 5.7% | 18.6% |
| **3-gram** | Subword | 3,461 | 11.76 | 28,498 | 24.5% | 62.8% |
| **4-gram** | Word | 80,153 | 16.29 | 114,752 | 2.9% | 10.7% |
| **4-gram** | Subword | 15,479 | 13.92 | 106,273 | 12.4% | 37.5% |
| **5-gram** | Word | 77,316 | 16.24 | 96,422 | 2.3% | 8.9% |
| **5-gram** | Subword | 44,243 | 15.43 | 194,726 | 7.1% | 23.4% |

### Top 5 N-grams by Size

**2-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `ཡོདཔ ཨིན` | 3,325 |
| 2 | `རྒྱལ ཁབ` | 2,719 |
| 3 | `སྤྱི ལོ` | 1,933 |
| 4 | `ཨིན མས` | 1,872 |
| 5 | `ནང ལུ` | 1,628 |

**3-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `རིན པོ ཆེ` | 778 |
| 2 | `ཡོདཔ ཨིན མས` | 778 |
| 3 | `རྒྱལ ཁབ ནང` | 732 |
| 4 | `སྤྱི ལོ ལུ` | 688 |
| 5 | `འབྲུག རྒྱལ ཁབ` | 623 |

**4-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `རྒྱལ ཁབ ནང ལུ` | 309 |
| 2 | `འབྲུག རྒྱལ ཁབ ནང` | 288 |
| 3 | `དཔལ ལྡན འབྲུག པའི` | 272 |
| 4 | `གུ རུ རིན པོ` | 250 |
| 5 | `སྡེ སྲིད ཁྲི རབས` | 223 |

**5-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `གུ རུ རིན པོ ཆེ` | 184 |
| 2 | `གནམ ལོ མེད སྤྱི ལོ` | 162 |
| 3 | `ཞབས དྲུང རིན པོ ཆེ` | 150 |
| 4 | `རྒྱལ ཡོངས དགའ སྐྱིད དཔལ` | 127 |
| 5 | `ཡོངས དགའ སྐྱིད དཔལ འཛོམས` | 125 |

**2-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `ས ་` | 123,525 |
| 2 | `ང ་` | 91,851 |
| 3 | `ན ་` | 70,834 |
| 4 | `་ _` | 62,281 |
| 5 | `་ བ` | 59,589 |

**3-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `ག ས ་` | 25,075 |
| 2 | `ད ང ་` | 18,381 |
| 3 | `་ ད ང` | 17,725 |
| 4 | `། _ །` | 15,647 |
| 5 | `་ པ ་` | 15,536 |

**4-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `་ ད ང ་` | 17,384 |
| 2 | `་ པ འི ་` | 13,232 |
| 3 | `་ ལ ས ་` | 12,579 |
| 4 | `་ འ དི ་` | 8,184 |
| 5 | `་ ན ང ་` | 6,539 |

**5-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `་ ཡོ ད པ ་` | 5,559 |
| 2 | `་ ལ ས ་ _` | 4,930 |
| 3 | `་ ད ང ་ _` | 4,145 |
| 4 | `་ འ བ ད ་` | 3,971 |
| 5 | `ས ་ པ འི ་` | 3,925 |


### Key Findings

- **Best Perplexity:** 2-gram (subword) with 488
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
- **Coverage:** Top-1000 patterns cover ~23% 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 | 1.1820 | 2.269 | 14.12 | 12,061 | 0.0% |
| **1** | Subword | 0.8884 | 1.851 | 7.57 | 1,607 | 11.2% |
| **2** | Word | 0.5611 | 1.475 | 2.65 | 170,162 | 43.9% |
| **2** | Subword | 0.6433 | 1.562 | 5.02 | 12,152 | 35.7% |
| **3** | Word | 0.2267 | 1.170 | 1.41 | 449,950 | 77.3% |
| **3** | Subword | 0.5247 | 1.439 | 3.26 | 61,009 | 47.5% |
| **4** | Word | 0.0989 🏆 | 1.071 | 1.15 | 633,460 | 90.1% |
| **4** | Subword | 0.3500 | 1.275 | 2.11 | 199,035 | 65.0% |

### Generated Text Samples (Word-based)

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

**Context Size 1:**

1. `དང སྲས ཚུ གིས ས ཐག མ སླེབས ཚེ འདི ལེགས སོ བཅོམ ཡིད གསུམ གྱི`
2. `པ ཨིན པས དེ མ མི ཡུལ བྱིན ཅན ཁྱོད འདི གི ལཱ འབད ནི ཀ`
3. `ལུ བདག སྐྱོང ལེགས སོ དཀརཔོ ཅིག གཅིག པུར ལུ ༡༡ ག གིས པདྨ རིགས མ`

**Context Size 2:**

1. `ཡོདཔ ཨིན པས རྒྱབ རྟེན ༡ དྲག ཤོས ཀྱི གསོལ ར ༤ གློག འཕྲིན གྱི ཁྱབ བདག`
2. `རྒྱལ ཁབ ཀྱི སྐུ རིམ དང པོ ནས བློ གྲོས བཟང མོ གིས ཨ ལུ འདི ཆ`
3. `སྤྱི ལོ སྤྱི ཟླ ༤ པ ༡༡ པ ལས འཛིན ཟེར བཙུགས མི མཐོ ཚད ཀི ལོ`

**Context Size 3:**

1. `རིན པོ ཆེ སངས རྒྱས ཀུན གྱི སྐུ འཆང བ སངས རྒྱས ཀུན གྱི གསུང ཡང ཡིན རྡོ`
2. `ཡོདཔ ཨིན མས ཨོ རྒྱན ཆོས གླིང ལྷ ཁང འདི དུས རབས ༨ པའི ནང གུ རུ རིན`
3. `རྒྱལ ཁབ ནང ལུ ཡང དམངས གཙོའི རིང ལུགས ཀྱི རྒྱལ པོའི བརྟན བཞུགས གི རྩ ཚིག གསར`

**Context Size 4:**

1. `རྒྱལ ཁབ ནང ལུ དཔལ འབྱོར གྱི སྡེ ཚན ཅིག ཡང གཞི གཙུགས འབད དེ འདུག དེ ཡང སྔོན`
2. `འབྲུག རྒྱལ ཁབ ནང ཡོད པའི རྒྱལ ཁབ ཅིག ཨིན དེ ཡང གྷི རེཊ བིརི ཊེན ཟེར མི འདི`
3. `དཔལ ལྡན འབྲུག པའི གདུང བརྒྱུད ཅིག ཞུ ནིའི དོན ལུ ཚེས ཉེར དགུ ལུ བླ མ གུ རུ`


### Generated Text Samples (Subword-based)

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

**Context Size 1:**

1. `་_ཚར་ཏེ་གིས་ཡོདཔོན་འ`
2. `_རྫོང་སྟེངས་_ཞེང་བ་རུའི`
3. `སལཔ་ལུ་ག་བཏུབཟོཔ་ཡིག`

**Context Size 2:**

1. `ས་ལུང་ཞིན་པ་འབྲུག་འབྲུག`
2. `ང་ཁྲུང་ཁབ་སྦྲུལ་ཙ་ཝཊ་ཛ`
3. `ན་ནང་བླ་མཆོད་ཆོས་དཔ་`

**Context Size 3:**

1. `གས་རིག་པའི་ནུས་པ་སྦེ་ཐོན`
2. `དང་རའི་ཨཔ་ཟླཝ་ག་རང་འ`
3. `་དང་ཕྱི་མས།_།ཉི་ཟླ་_༢༩`

**Context Size 4:**

1. `་དང་གཅིག་ནང་_ཡན་ལག་ཁ`
2. `་པའི་བླ་མ་ཐུབ།_།དགེ་བ་སྟོ`
3. `་ལས་_འབྱུང་ཁུངས།_།དགའ་`


### Key Findings

- **Best Predictability:** Context-4 (word) with 90.1% predictability
- **Branching Factor:** Decreases with context size (more deterministic)
- **Memory Trade-off:** Larger contexts require more storage (199,035 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 | 6,761 |
| Total Tokens | 898,876 |
| Mean Frequency | 132.95 |
| Median Frequency | 6 |
| Frequency Std Dev | 709.47 |

### Most Common Words

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | དང | 18,802 |
| 2 | པ | 17,903 |
| 3 | ལུ | 15,384 |
| 4 | པའི | 14,560 |
| 5 | ལས | 14,391 |
| 6 | མི | 11,348 |
| 7 | དེ | 11,091 |
| 8 | མ | 10,372 |
| 9 | གི | 10,307 |
| 10 | འདི | 9,382 |

### Least Common Words (from vocabulary)

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | printer | 2 |
| 2 | fortress | 2 |
| 3 | gods | 2 |
| 4 | wordpress | 2 |
| 5 | phurdo | 2 |
| 6 | gonpa | 2 |
| 7 | assam | 2 |
| 8 | pelgen | 2 |
| 9 | anecdotes | 2 |
| 10 | kheng | 2 |

### Zipf's Law Analysis

| Metric | Value |
|--------|-------|
| Zipf Coefficient | 1.8277 |
| R² (Goodness of Fit) | 0.959592 |
| Adherence Quality | **excellent** |

### Coverage Analysis

| Top N Words | Coverage |
|-------------|----------|
| Top 100 | 49.0% |
| Top 1,000 | 92.3% |
| Top 5,000 | 99.6% |
| Top 10,000 | 0.0% |

### Key Findings

- **Zipf Compliance:** R²=0.9596 indicates excellent adherence to Zipf's law
- **High Frequency Dominance:** Top 100 words cover 49.0% of corpus
- **Long Tail:** -3,239 words needed for remaining 100.0% 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 | 0.6999 🏆 | 0.3567 | N/A | N/A |
| **mono_64d** | 64 | 0.4345 | 0.3403 | N/A | N/A |
| **mono_128d** | 128 | 0.1109 | 0.3305 | N/A | N/A |
| **aligned_32d** | 32 | 0.6999 | 0.3594 | 0.0547 | 0.2644 |
| **aligned_64d** | 64 | 0.4345 | 0.3388 | 0.1307 | 0.4103 |
| **aligned_128d** | 128 | 0.1109 | 0.3270 | 0.2340 | 0.4742 |

### Key Findings

- **Best Isotropy:** mono_32d with 0.6999 (more uniform distribution)
- **Semantic Density:** Average pairwise similarity of 0.3421. Lower values indicate better semantic separation.
- **Alignment Quality:** Aligned models achieve up to 23.4% 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 | **-0.621** | Low formulaic 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.

*No productive affixes detected.*


### 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.

*No significant bound stems detected.*


### 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.

*No significant affix co-occurrences detected.*


### 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`).

*Insufficient data for recursive segmentation.*


### 6.6 Linguistic Interpretation

> **Automated Insight:**
The language Dzongkha 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 (5.51x) |
| N-gram | **2-gram** | Lowest perplexity (488) |
| Markov | **Context-4** | Highest predictability (90.1%) |
| 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 03:00:40*