README.md

---
language: crh
language_name: Crimean Tatar
language_family: turkic_kipchak
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-turkic_kipchak
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: 4.779
  - name: best_isotropy
    type: isotropy
    value: 0.7031
  - name: vocabulary_size
    type: vocab
    value: 0
generated: 2026-01-03
---

# Crimean Tatar - Wikilangs Models
## Comprehensive Research Report & Full Ablation Study

This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Crimean Tatar** 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** | 3.646x | 3.65 | 0.2038% | 212,471 |
| **16k** | 4.078x | 4.08 | 0.2279% | 189,960 |
| **32k** | 4.457x | 4.46 | 0.2492% | 173,772 |
| **64k** | 4.779x 🏆 | 4.79 | 0.2672% | 162,079 |

### Tokenization Examples

Below are sample sentences tokenized with each vocabulary size:

**Sample 1:** `İslanovo () - Rusiyede, Başqırtistan Cumhuriyetiniñ Kuşnarenko rayonında bir köy...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁İs lan ovo ▁() ▁- ▁rusiyede , ▁başqırtistan ▁cumhuriyetiniñ ▁kuşnarenko ... (+13 more)` | 23 |
| 16k | `▁İs lanovo ▁() ▁- ▁rusiyede , ▁başqırtistan ▁cumhuriyetiniñ ▁kuşnarenko ▁rayonında ... (+12 more)` | 22 |
| 32k | `▁İs lanovo ▁() ▁- ▁rusiyede , ▁başqırtistan ▁cumhuriyetiniñ ▁kuşnarenko ▁rayonında ... (+12 more)` | 22 |
| 64k | `▁İslanovo ▁() ▁- ▁rusiyede , ▁başqırtistan ▁cumhuriyetiniñ ▁kuşnarenko ▁rayonında ▁bir ... (+11 more)` | 21 |

**Sample 2:** `Drujbivka () - Ukrainanıñ Jıtomır vilâyetinde Korosten rayonında bir köy. Ealisi...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁druj bivka ▁() ▁- ▁ukrainanıñ ▁jıtomır ▁vilâyetinde ▁korosten ▁rayonında ▁bir ... (+12 more)` | 22 |
| 16k | `▁druj bivka ▁() ▁- ▁ukrainanıñ ▁jıtomır ▁vilâyetinde ▁korosten ▁rayonında ▁bir ... (+12 more)` | 22 |
| 32k | `▁druj bivka ▁() ▁- ▁ukrainanıñ ▁jıtomır ▁vilâyetinde ▁korosten ▁rayonında ▁bir ... (+12 more)` | 22 |
| 64k | `▁drujbivka ▁() ▁- ▁ukrainanıñ ▁jıtomır ▁vilâyetinde ▁korosten ▁rayonında ▁bir ▁köy ... (+11 more)` | 21 |

**Sample 3:** `Koltunovka () - Rusiyeniñ Belgorod vilâyetinde, Alekseyevka rayonında bir köy. E...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁kol tun ovka ▁() ▁- ▁rusiyeniñ ▁belgorod ▁vilâyetinde , ▁alekseyevka ... (+15 more)` | 25 |
| 16k | `▁kol tun ovka ▁() ▁- ▁rusiyeniñ ▁belgorod ▁vilâyetinde , ▁alekseyevka ... (+15 more)` | 25 |
| 32k | `▁kol tun ovka ▁() ▁- ▁rusiyeniñ ▁belgorod ▁vilâyetinde , ▁alekseyevka ... (+15 more)` | 25 |
| 64k | `▁koltunovka ▁() ▁- ▁rusiyeniñ ▁belgorod ▁vilâyetinde , ▁alekseyevka ▁rayonında ▁bir ... (+13 more)` | 23 |


### Key Findings

- **Best Compression:** 64k achieves 4.779x compression
- **Lowest UNK Rate:** 8k with 0.2038% 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 | 849 | 9.73 | 10,213 | 56.1% | 74.4% |
| **2-gram** | Subword | 348 🏆 | 8.44 | 3,878 | 63.4% | 98.0% |
| **3-gram** | Word | 1,276 | 10.32 | 13,301 | 49.1% | 71.8% |
| **3-gram** | Subword | 2,220 | 11.12 | 29,221 | 33.1% | 71.8% |
| **4-gram** | Word | 4,190 | 12.03 | 31,513 | 31.9% | 54.7% |
| **4-gram** | Subword | 7,833 | 12.94 | 131,199 | 26.0% | 52.3% |
| **5-gram** | Word | 6,061 | 12.57 | 29,487 | 24.1% | 48.5% |
| **5-gram** | Subword | 16,690 | 14.03 | 285,107 | 23.4% | 46.1% |

### Top 5 N-grams by Size

**2-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `ealisiniñ sayısı` | 20,740 |
| 2 | `rayonında bir` | 17,352 |
| 3 | `meskün yerler` | 12,883 |
| 4 | `bir köy` | 10,061 |
| 5 | `köy ealisiniñ` | 9,139 |

**3-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `rayonında bir köy` | 9,314 |
| 2 | `bir köy ealisiniñ` | 9,139 |
| 3 | `köy ealisiniñ sayısı` | 9,139 |
| 4 | `rayonındaki meskün yerler` | 5,591 |
| 5 | `kişi meskün yerler` | 4,604 |

**4-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `bir köy ealisiniñ sayısı` | 9,139 |
| 2 | `rayonında bir köy ealisiniñ` | 8,985 |
| 3 | `bir köydir ealisiniñ sayısı` | 4,601 |
| 4 | `rayonında bir köydir ealisiniñ` | 4,565 |
| 5 | `i̇htar rayonındaki meskün yerler` | 3,615 |

**5-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `rayonında bir köy ealisiniñ sayısı` | 8,985 |
| 2 | `rayonında bir köydir ealisiniñ sayısı` | 4,565 |
| 3 | `kişi i̇htar rayonındaki meskün yerler` | 2,558 |
| 4 | `asırnıñ bir senesi vaqialar doğumlar` | 1,996 |
| 5 | `bir senesi vaqialar doğumlar ölümler` | 1,917 |

**2-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `i n` | 101,089 |
| 2 | `e r` | 95,398 |
| 3 | `a _` | 88,613 |
| 4 | `r _` | 84,598 |
| 5 | `. _` | 80,856 |

**3-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `i ñ _` | 43,406 |
| 2 | `n i ñ` | 42,914 |
| 3 | `l e r` | 42,891 |
| 4 | `n d e` | 35,848 |
| 5 | `e t i` | 35,643 |

**4-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `n i ñ _` | 42,657 |
| 2 | `i n d e` | 34,217 |
| 3 | `y e t i` | 30,830 |
| 4 | `ı n d a` | 30,087 |
| 5 | `_ b i r` | 29,643 |

**5-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `i n i ñ _` | 28,194 |
| 2 | `y e t i n` | 28,057 |
| 3 | `_ b i r _` | 27,628 |
| 4 | `r a y o n` | 26,921 |
| 5 | `_ r a y o` | 26,900 |


### Key Findings

- **Best Perplexity:** 2-gram (subword) with 348
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
- **Coverage:** Top-1000 patterns cover ~46% 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 | 0.6244 | 1.542 | 2.99 | 128,666 | 37.6% |
| **1** | Subword | 0.8852 | 1.847 | 6.85 | 1,505 | 11.5% |
| **2** | Word | 0.1302 | 1.094 | 1.24 | 383,467 | 87.0% |
| **2** | Subword | 0.9025 | 1.869 | 5.57 | 10,300 | 9.7% |
| **3** | Word | 0.0387 | 1.027 | 1.07 | 474,016 | 96.1% |
| **3** | Subword | 0.8153 | 1.760 | 3.87 | 57,358 | 18.5% |
| **4** | Word | 0.0242 🏆 | 1.017 | 1.05 | 502,796 | 97.6% |
| **4** | Subword | 0.6069 | 1.523 | 2.54 | 221,948 | 39.3% |

### Generated Text Samples (Word-based)

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

**Context Size 1:**

1. `bir cemaatı ukrainanıñ jıtomır vilâyetinde olevsk rayonında bir şeer şeklinde qasabalar vahruşev nog...`
2. `kişi rayonındaki meskün yerler köyler abatskoye rusiyeniñ hantı mansi muhtar cumhuriyetinıñ devlet g...`
3. `sayısı 0 kişi meskün yerler veloturizm iklim deñişmelerine çoq yüklü yükni yükniñ yüksek mölekulâr o...`

**Context Size 2:**

1. `ealisiniñ sayısı kişi senesi vilâyetindeki qasabalar`
2. `rayonında bir aul adıge habl calancük kiçik i̇ncik kavkazskiy pregradna üçköken habez erkin şeer rus...`
3. `bir köy oktâbr rayonınıñ merkezi ealisiniñ sayısı 202 939 kişi senesi atıflar rayonındaki meskün yer...`

**Context Size 3:**

1. `rayonında bir köy ealisiniñ sayısı 394 kişi senesi atıflar rayonındaki meskün yerler köyler atıflar ...`
2. `bir köy ealisiniñ sayısı 593 kişi i̇htar rayonındaki meskün yerler köyler atıflar rayonındaki meskün...`
3. `köy ealisiniñ sayısı 828 kişi vilâyetindeki meskün yerler`

**Context Size 4:**

1. `bir köy ealisiniñ sayısı kişi vilâyetindeki meskün yerler`
2. `rayonında bir köy ealisiniñ sayısı 134 kişi vilâyetindeki meskün yerler`
3. `bir köydir ealisiniñ sayısı 25 kişi i̇htar rayonındaki meskün yerler vilâyetindeki şeer şeklinde qas...`


### Generated Text Samples (Subword-based)

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

**Context Size 1:**

1. `_()_-_qmı_mi._be`
2. `ariraye_altviyür`
3. `i._bişekayay._()`

**Context Size 2:**

1. `iniv-ufterlar,_ad`
2. `a_balisiyentılari`
3. `r_rusiyetingrayıs`

**Context Size 3:**

1. `iñ_sayısı_591_belg`
2. `niñ_sayısı_kir._ea`
3. `nde_dinde_ögrendi_`

**Context Size 4:**

1. `niñ_noviçi_bar._cev`
2. `inde_kontsev_artemi`
3. `yetinde_bir_qast_ma`


### Key Findings

- **Best Predictability:** Context-4 (word) with 97.6% predictability
- **Branching Factor:** Decreases with context size (more deterministic)
- **Memory Trade-off:** Larger contexts require more storage (221,948 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 | 51,458 |
| Total Tokens | 776,471 |
| Mean Frequency | 15.09 |
| Median Frequency | 3 |
| Frequency Std Dev | 272.01 |

### Most Common Words

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | bir | 27,753 |
| 2 | kişi | 20,857 |
| 3 | sayısı | 20,821 |
| 4 | ealisiniñ | 20,770 |
| 5 | rayonında | 17,392 |
| 6 | meskün | 13,506 |
| 7 | yerler | 12,926 |
| 8 | vilâyetinde | 12,440 |
| 9 | köy | 10,901 |
| 10 | rusiyeniñ | 9,597 |

### Least Common Words (from vocabulary)

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | зияде | 2 |
| 2 | atalarnıñ | 2 |
| 3 | kotsubınskıylar | 2 |
| 4 | yüneskonıñ | 2 |
| 5 | دیللر | 2 |
| 6 | ازبری | 2 |
| 7 | اولان | 2 |
| 8 | سامانچی | 2 |
| 9 | قیزی | 2 |
| 10 | samançı | 2 |

### Zipf's Law Analysis

| Metric | Value |
|--------|-------|
| Zipf Coefficient | 0.9856 |
| R² (Goodness of Fit) | 0.998043 |
| Adherence Quality | **excellent** |

### Coverage Analysis

| Top N Words | Coverage |
|-------------|----------|
| Top 100 | 45.6% |
| Top 1,000 | 63.8% |
| Top 5,000 | 78.2% |
| Top 10,000 | 84.4% |

### Key Findings

- **Zipf Compliance:** R²=0.9980 indicates excellent adherence to Zipf's law
- **High Frequency Dominance:** Top 100 words cover 45.6% of corpus
- **Long Tail:** 41,458 words needed for remaining 15.6% 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 | 0.7031 🏆 | 0.3722 | N/A | N/A |
| **mono_64d** | 64 | 0.4233 | 0.3424 | N/A | N/A |
| **mono_128d** | 128 | 0.1068 | 0.3377 | N/A | N/A |
| **aligned_32d** | 32 | 0.7031 | 0.3786 | 0.0140 | 0.1600 |
| **aligned_64d** | 64 | 0.4233 | 0.3386 | 0.0380 | 0.2140 |
| **aligned_128d** | 128 | 0.1068 | 0.3419 | 0.0560 | 0.2680 |

### Key Findings

- **Best Isotropy:** mono_32d with 0.7031 (more uniform distribution)
- **Semantic Density:** Average pairwise similarity of 0.3519. Lower values indicate better semantic separation.
- **Alignment Quality:** Aligned models achieve up to 5.6% 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 | **0.052** | 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.

#### Productive Prefixes
| Prefix | Examples |
|--------|----------|

#### Productive Suffixes
| Suffix | Examples |
|--------|----------|
| `-a` | terehova, biçura, observatoriya |
| `-ka` | novosölka, alekseyevka, kapustânka |
| `-vo` | korolövo, semenovo, hetovo |
| `-vka` | alekseyevka, dolgalovka, svetlovka |
| `-an` | turan, birobican, adlandırğan |
| `-ovo` | semenovo, hetovo, panfilovo |
| `-ye` | zapolnoye, smelıye, voznesenskoye |
| `-en` | keçirmegen, nevbetten, neogen |

### 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 |
|------|----------|------------------|----------|
| `leri` | 1.60x | 110 contexts | ileri, lerik, galeri |
| `rler` | 1.60x | 57 contexts | erler, yerler, derler |
| `siye` | 2.05x | 21 contexts | asiye, rusiye, tevsiye |
| `isin` | 1.57x | 31 contexts | episine, kerisin, reisini |
| `iniñ` | 1.64x | 26 contexts | eviniñ, iliniñ, eliniñ |
| `nesi` | 1.64x | 22 contexts | nesib, nesil, nesir |
| `eniñ` | 1.75x | 16 contexts | seniñ, heniñ, ekeniñ |
| `usiy` | 2.11x | 9 contexts | lusiya, rusiye, hususiy |
| `lâye` | 1.87x | 11 contexts | belâyev, gulâyev, vilâyet |
| `âyet` | 1.87x | 11 contexts | menâyet, vilâyet, şikâyet |
| `sini` | 1.70x | 14 contexts | siniy, sinip, aksini |
| `yeti` | 1.59x | 17 contexts | yetip, yetim, yetişe |

### 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`).

| Word | Suggested Split | Confidence | Stem |
|------|-----------------|------------|------|
| gazetanen | **`gazet-an-en`** | 6.0 | `gazet` |
| ananyevka | **`anan-ye-vka`** | 6.0 | `anan` |
| petrusnıñ | **`petrus-nıñ`** | 4.5 | `petrus` |
| vesiqalarınıñ | **`vesiqaları-nıñ`** | 4.5 | `vesiqaları` |
| nikiforovo | **`nikifor-ovo`** | 4.5 | `nikifor` |
| sistemasınıñ | **`sisteması-nıñ`** | 4.5 | `sisteması` |
| qısımlarınıñ | **`qısımları-nıñ`** | 4.5 | `qısımları` |
| borispolye | **`borispol-ye`** | 4.5 | `borispol` |
| programmanıñ | **`programma-nıñ`** | 4.5 | `programma` |
| gotlarnıñ | **`gotlar-nıñ`** | 4.5 | `gotlar` |
| qadılıqnıñ | **`qadılıq-nıñ`** | 4.5 | `qadılıq` |
| kopelânka | **`kopelân-ka`** | 4.5 | `kopelân` |
| mahsulatlarnıñ | **`mahsulatlar-nıñ`** | 4.5 | `mahsulatlar` |
| nigeriyanıñ | **`nigeriya-nıñ`** | 4.5 | `nigeriya` |
| qasabanıñ | **`qasaba-nıñ`** | 4.5 | `qasaba` |

### 6.6 Linguistic Interpretation

> **Automated Insight:**
The language Crimean Tatar 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 (4.78x) |
| N-gram | **2-gram** | Lowest perplexity (348) |
| Markov | **Context-4** | Highest predictability (97.6%) |
| 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: 2026-01-03 20:48:59*