---
language: din
language_name: Dinka
language_family: african_nilotic
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-african_nilotic
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: 4.248
  - name: best_isotropy
    type: isotropy
    value: 0.2108
  - name: vocabulary_size
    type: vocab
    value: 0
generated: 2026-01-04
---

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

This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Dinka** 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** | 3.696x | 3.70 | 1.0395% | 137,657 |
| **16k** | 3.984x | 3.99 | 1.1206% | 127,694 |
| **32k** | 4.248x 🏆 | 4.25 | 1.1949% | 119,761 |

### Tokenization Examples

Below are sample sentences tokenized with each vocabulary size:

**Sample 1:** `Ukraine ee paan en Yurop Penëdhiäk ee Volodymyr Zelensky. Genamaatnhomde ayee cɔ...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁ukraine ▁ee ▁paan ▁en ▁yurop ▁penëdhiäk ▁ee ▁v ol od ... (+15 more)` | 25 |
| 16k | `▁ukraine ▁ee ▁paan ▁en ▁yurop ▁penëdhiäk ▁ee ▁v olodymyr ▁zelensky ... (+8 more)` | 18 |
| 32k | `▁ukraine ▁ee ▁paan ▁en ▁yurop ▁penëdhiäk ▁ee ▁volodymyr ▁zelensky . ... (+5 more)` | 15 |

**Sample 2:** `Monteaguila ee gendït Chile. Cinëkɔcde aa tëcit ruonic`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁mon te agu ila ▁ee ▁gendït ▁ch ile . ▁cinëkɔcde ... (+3 more)` | 13 |
| 16k | `▁mon te agu ila ▁ee ▁gendït ▁chile . ▁cinëkɔcde ▁aa ... (+2 more)` | 12 |
| 32k | `▁monteaguila ▁ee ▁gendït ▁chile . ▁cinëkɔcde ▁aa ▁tëcit ▁ruonic` | 9 |

**Sample 3:** `Dhambia ee Apirïka. Genamaatnhomde ayee cɔl Lusaka.`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁dhambia ▁ee ▁apirïka . ▁genamaatnhomde ▁ayee ▁cɔl ▁lu sak a ... (+1 more)` | 11 |
| 16k | `▁dhambia ▁ee ▁apirïka . ▁genamaatnhomde ▁ayee ▁cɔl ▁lusaka .` | 9 |
| 32k | `▁dhambia ▁ee ▁apirïka . ▁genamaatnhomde ▁ayee ▁cɔl ▁lusaka .` | 9 |


### Key Findings

- **Best Compression:** 32k achieves 4.248x compression
- **Lowest UNK Rate:** 8k with 1.0395% 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 | 846 | 9.72 | 1,522 | 38.9% | 86.3% |
| **2-gram** | Subword | 328 | 8.36 | 1,563 | 62.0% | 99.1% |
| **3-gram** | Word | 240 | 7.90 | 785 | 62.9% | 100.0% |
| **3-gram** | Subword | 2,240 | 11.13 | 9,446 | 25.3% | 71.0% |
| **4-gram** | Word | 166 | 7.38 | 882 | 69.6% | 100.0% |
| **4-gram** | Subword | 8,823 | 13.11 | 31,591 | 13.0% | 43.0% |
| **5-gram** | Word | 59 🏆 | 5.89 | 373 | 86.5% | 100.0% |
| **5-gram** | Subword | 18,719 | 14.19 | 51,151 | 8.6% | 31.8% |

### Top 5 N-grams by Size

**2-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `glossary derived` | 167 |
| 2 | `derived from` | 167 |
| 3 | `from sil` | 167 |
| 4 | `sil internationals` | 167 |
| 5 | `internationals draft` | 167 |

**3-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `internationals draft dinka` | 167 |
| 2 | `from sil internationals` | 167 |
| 3 | `derived from sil` | 167 |
| 4 | `dinka glossary derived` | 167 |
| 5 | `educational foundation sil` | 167 |

**4-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `english to dinka glossary` | 167 |
| 2 | `to dinka glossary derived` | 167 |
| 3 | `dinka glossary derived from` | 167 |
| 4 | `glossary derived from sil` | 167 |
| 5 | `from sil internationals draft` | 167 |

**5-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `dinka glossary derived from sil` | 167 |
| 2 | `williamson educational foundation sil international` | 167 |
| 3 | `kay williamson educational foundation sil` | 167 |
| 4 | `dictionary kay williamson educational foundation` | 167 |
| 5 | `english dictionary kay williamson educational` | 167 |

**2-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ k` | 14,243 |
| 2 | `e _` | 10,060 |
| 3 | `_ a` | 9,948 |
| 4 | `ë _` | 8,555 |
| 5 | `n _` | 7,924 |

**3-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ k u` | 4,510 |
| 2 | `n ë _` | 3,923 |
| 3 | `k u _` | 3,559 |
| 4 | `_ k e` | 3,459 |
| 5 | `_ t h` | 3,193 |

**4-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ k u _` | 3,514 |
| 2 | `_ n ë _` | 2,762 |
| 3 | `_ d e _` | 2,147 |
| 4 | `_ k e _` | 1,756 |
| 5 | `_ y e _` | 1,452 |

**5-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ k ɔ c _` | 1,091 |
| 2 | `, _ k u _` | 836 |
| 3 | `_ y e n _` | 729 |
| 4 | `a t i o n` | 718 |
| 5 | `t i o n a` | 686 |


### Key Findings

- **Best Perplexity:** 5-gram (word) with 59
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
- **Coverage:** Top-1000 patterns cover ~32% 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 | 0.6343 | 1.552 | 3.69 | 17,365 | 36.6% |
| **1** | Subword | 1.5315 | 2.891 | 11.78 | 318 | 0.0% |
| **2** | Word | 0.1750 | 1.129 | 1.30 | 63,845 | 82.5% |
| **2** | Subword | 1.1046 | 2.150 | 5.58 | 3,744 | 0.0% |
| **3** | Word | 0.0333 | 1.023 | 1.04 | 83,004 | 96.7% |
| **3** | Subword | 0.7588 | 1.692 | 3.12 | 20,888 | 24.1% |
| **4** | Word | 0.0076 🏆 | 1.005 | 1.01 | 86,340 | 99.2% |
| **4** | Subword | 0.5088 | 1.423 | 2.08 | 65,173 | 49.1% |

### Generated Text Samples (Word-based)

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

**Context Size 1:**

1. `ku gɛɛth puɔɔth ben jam ë kɔcnhiaardiɛtë acik gam ke panmäcalëi french indochina bï ya kë`
2. `në bɛ̈ɛ̈i tënë tïmëtïm 57 ku tiem thidhic ku kek aa kï alëk dɛl miɲ kaːl`
3. `de spain ku aye raan döŋ acï giit en kɛ̈ɛ̈cë anyak atɔ̈ thïn rin keloirɔt wët`

**Context Size 2:**

1. `english dictionary kay williamson educational foundation sil international dikconari thudän`
2. `english to dinka glossary derived from sil internationals draft dinka english dictionary kay william...`
3. `to dinka glossary derived from sil internationals draft dinka english dictionary kay williamson educ...`

**Context Size 3:**

1. `and roger blench english to dinka glossary derived from sil internationals draft dinka english dicti...`
2. `internationals draft dinka english dictionary kay williamson educational foundation sil internationa...`
3. `roger blench english to dinka glossary derived from sil internationals draft dinka english dictionar...`

**Context Size 4:**

1. `internationals draft dinka english dictionary kay williamson educational foundation sil internationa...`
2. `to dinka glossary derived from sil internationals draft dinka english dictionary kay williamson educ...`
3. `derived from sil internationals draft dinka english dictionary kay williamson educational foundation...`


### Generated Text Samples (Subword-based)

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

**Context Size 1:**

1. `_adde_cïnapae_lu`
2. `a_piic_ciän_anya`
3. `kuɛ̈c_arabo_san_k`

**Context Size 2:**

1. `_ku_acï_raŋdec_bï`
2. `e_bïk_ëk_cök_de_y`
3. `_aŋrɛn,_juäi_adhi`

**Context Size 3:**

1. `_ku_yiic,_thudän._`
2. `në_2._“tx2_awɛ̈ɛ̈rde`
3. `ku_puses)._ë_makut`

**Context Size 4:**

1. `_ku_cɔl_muɔɔr_aacë_`
2. `_në_keye,_ee_noŋic_`
3. `_de_joŋlei_paguot_k`


### Key Findings

- **Best Predictability:** Context-4 (word) with 99.2% predictability
- **Branching Factor:** Decreases with context size (more deterministic)
- **Memory Trade-off:** Larger contexts require more storage (65,173 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 | 5,848 |
| Total Tokens | 81,189 |
| Mean Frequency | 13.88 |
| Median Frequency | 3 |
| Frequency Std Dev | 86.66 |

### Most Common Words

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | ku | 3,546 |
| 2 | në | 2,775 |
| 3 | de | 2,158 |
| 4 | ë | 1,890 |
| 5 | ke | 1,776 |
| 6 | ye | 1,484 |
| 7 | ee | 1,173 |
| 8 | kɔc | 1,137 |
| 9 | cï | 883 |
| 10 | yen | 747 |

### Least Common Words (from vocabulary)

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | mayall | 2 |
| 2 | cream | 2 |
| 3 | puɔ̈k | 2 |
| 4 | layla | 2 |
| 5 | adëgëk | 2 |
| 6 | skobarkä | 2 |
| 7 | pïlïbït | 2 |
| 8 | tïgër | 2 |
| 9 | rësärwë | 2 |
| 10 | terai | 2 |

### Zipf's Law Analysis

| Metric | Value |
|--------|-------|
| Zipf Coefficient | 1.0295 |
| R² (Goodness of Fit) | 0.989261 |
| Adherence Quality | **excellent** |

### Coverage Analysis

| Top N Words | Coverage |
|-------------|----------|
| Top 100 | 47.4% |
| Top 1,000 | 78.6% |
| Top 5,000 | 97.9% |
| Top 10,000 | 0.0% |

### Key Findings

- **Zipf Compliance:** R²=0.9893 indicates excellent adherence to Zipf's law
- **High Frequency Dominance:** Top 100 words cover 47.4% of corpus
- **Long Tail:** -4,152 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 | 0.2108 🏆 | 0.6155 | N/A | N/A |
| **mono_64d** | 64 | 0.0418 | 0.6059 | N/A | N/A |
| **mono_128d** | 128 | 0.0088 | 0.6443 | N/A | N/A |
| **aligned_32d** | 32 | 0.2108 | 0.5998 | 0.0070 | 0.0607 |
| **aligned_64d** | 64 | 0.0418 | 0.5881 | 0.0187 | 0.1028 |
| **aligned_128d** | 128 | 0.0088 | 0.6544 | 0.0164 | 0.0911 |

### Key Findings

- **Best Isotropy:** mono_32d with 0.2108 (more uniform distribution)
- **Semantic Density:** Average pairwise similarity of 0.6180. Lower values indicate better semantic separation.
- **Alignment Quality:** Aligned models achieve up to 1.9% 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 | **1.232** | High morphological productivity | Reliable analysis |
| Idiomaticity Gap | **2.143** | 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 |
|--------|----------|
| `-th` | thiεkde, thɔ̈r, thiɛɛr |

#### Productive Suffixes
| Suffix | Examples |
|--------|----------|
| `-ic` | tocdïtic, nyinic, ciaryic |

### 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 |
|------|----------|------------------|----------|
| `thiä` | 1.36x | 12 contexts | thiär, thiäŋ, thiäi |

### 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 |
|--------|--------|-----------|----------|
| `-th` | `-ic` | 10 words | thändïtic, thudänic |

### 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 |
|------|-----------------|------------|------|
| kathɛɛric | **`kathɛɛr-ic`** | 4.5 | `kathɛɛr` |
| wëlëmiiric | **`wëlëmiir-ic`** | 4.5 | `wëlëmiir` |
| ruɔ̈ɔ̈nic | **`ruɔ̈ɔ̈n-ic`** | 4.5 | `ruɔ̈ɔ̈n` |
| pïïrdenic | **`pïïrden-ic`** | 4.5 | `pïïrden` |
| manywëëthic | **`manywëëth-ic`** | 4.5 | `manywëëth` |
| pinynhomic | **`pinynhom-ic`** | 4.5 | `pinynhom` |
| krïthmathic | **`krïthmath-ic`** | 4.5 | `krïthmath` |
| käcïpuric | **`käcïpur-ic`** | 4.5 | `käcïpur` |
| abëkruöönic | **`abëkruöön-ic`** | 4.5 | `abëkruöön` |
| thändïtic | **`th-ändït-ic`** | 3.0 | `ändït` |
| thiɛ̈ɛ̈ric | **`th-iɛ̈ɛ̈r-ic`** | 3.0 | `iɛ̈ɛ̈r` |
| wëljamiic | **`wëljami-ic`** | 1.5 | `wëljami` |
| pabakciɛlic | **`pabakciɛl-ic`** | 1.5 | `pabakciɛl` |
| thanypiny | **`th-anypiny`** | 1.5 | `anypiny` |
| lëkthɛɛric | **`lëkthɛɛr-ic`** | 1.5 | `lëkthɛɛr` |

### 6.6 Linguistic Interpretation

> **Automated Insight:**
The language Dinka shows moderate morphological complexity. There is a balanced trade-off between whole-word memorization and subword composition.

> **Note on Idiomaticity:** The high Idiomaticity Gap suggests a large number of frequent multi-word expressions or formulaic sequences that are statistically distinct from their component parts.

---
## 7. Summary & Recommendations

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

### Production Recommendations

| Component | Recommended | Rationale |
|-----------|-------------|-----------|
| Tokenizer | **32k BPE** | Best compression (4.25x) |
| N-gram | **5-gram** | Lowest perplexity (59) |
| Markov | **Context-4** | Highest predictability (99.2%) |
| 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 02:12:14*