README.md

---
language: bug
language_name: Buginese
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: 4.927
  - name: best_isotropy
    type: isotropy
    value: 0.0849
  - name: vocabulary_size
    type: vocab
    value: 0
generated: 2026-01-03
---

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

This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Buginese** 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** | 4.286x | 4.31 | 0.4928% | 36,732 |
| **16k** | 4.517x | 4.55 | 0.5194% | 34,850 |
| **32k** | 4.927x 🏆 | 4.96 | 0.5665% | 31,952 |

### Tokenization Examples

Below are sample sentences tokenized with each vocabulary size:

**Sample 1:** `Dammartin-sur-Meuse iyanaritu séuwa komun ri déparetema Haute-Marne ri Perancis....`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁dam martin - sur - meuse ▁iyanaritu ▁séuwa ▁komun ▁ri ... (+22 more)` | 32 |
| 16k | `▁dammartin - sur - meuse ▁iyanaritu ▁séuwa ▁komun ▁ri ▁déparetema ... (+21 more)` | 31 |
| 32k | `▁dammartin - sur - meuse ▁iyanaritu ▁séuwa ▁komun ▁ri ▁déparetema ... (+21 more)` | 31 |

**Sample 2:** `Bussières iyanaritu séuwa komun ri déparetema Yonne ri Perancis. Ita to Komun ri...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁bussières ▁iyanaritu ▁séuwa ▁komun ▁ri ▁déparetema ▁yonne ▁ri ▁perancis . ... (+11 more)` | 21 |
| 16k | `▁bussières ▁iyanaritu ▁séuwa ▁komun ▁ri ▁déparetema ▁yonne ▁ri ▁perancis . ... (+11 more)` | 21 |
| 32k | `▁bussières ▁iyanaritu ▁séuwa ▁komun ▁ri ▁déparetema ▁yonne ▁ri ▁perancis . ... (+11 more)` | 21 |

**Sample 3:** `Pujols iyanaritu séuwa komun ri déparetema Gironde ri Perancis. Ita to Komun ri ...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁pujols ▁iyanaritu ▁séuwa ▁komun ▁ri ▁déparetema ▁gironde ▁ri ▁perancis . ... (+11 more)` | 21 |
| 16k | `▁pujols ▁iyanaritu ▁séuwa ▁komun ▁ri ▁déparetema ▁gironde ▁ri ▁perancis . ... (+11 more)` | 21 |
| 32k | `▁pujols ▁iyanaritu ▁séuwa ▁komun ▁ri ▁déparetema ▁gironde ▁ri ▁perancis . ... (+11 more)` | 21 |


### Key Findings

- **Best Compression:** 32k achieves 4.927x compression
- **Lowest UNK Rate:** 8k with 0.4928% 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 | 75 🏆 | 6.23 | 1,721 | 84.8% | 98.5% |
| **2-gram** | Subword | 167 | 7.39 | 2,161 | 81.3% | 99.5% |
| **3-gram** | Word | 118 | 6.89 | 2,060 | 74.9% | 98.6% |
| **3-gram** | Subword | 511 | 9.00 | 10,879 | 62.7% | 89.5% |
| **4-gram** | Word | 229 | 7.84 | 4,999 | 61.5% | 96.5% |
| **4-gram** | Subword | 938 | 9.87 | 41,989 | 58.6% | 80.3% |
| **5-gram** | Word | 304 | 8.25 | 4,200 | 51.5% | 97.0% |
| **5-gram** | Subword | 1,221 | 10.25 | 76,709 | 57.6% | 78.3% |

### Top 5 N-grams by Size

**2-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `komun ri` | 40,953 |
| 2 | `ri déparetema` | 25,713 |
| 3 | `kategori komun` | 15,118 |
| 4 | `ita to` | 13,903 |
| 5 | `to komun` | 13,889 |

**3-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `komun ri déparetema` | 25,709 |
| 2 | `kategori komun ri` | 15,117 |
| 3 | `to komun ri` | 13,889 |
| 4 | `ita to komun` | 13,889 |
| 5 | `iyanaritu séuwa komun` | 13,324 |

**4-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `to komun ri déparetema` | 13,889 |
| 2 | `ita to komun ri` | 13,889 |
| 3 | `perancis ita to komun` | 12,104 |
| 4 | `iyanaritu séuwa komun ri` | 11,780 |
| 5 | `séuwa komun ri déparetema` | 11,779 |

**5-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `ita to komun ri déparetema` | 13,889 |
| 2 | `perancis ita to komun ri` | 12,104 |
| 3 | `iyanaritu séuwa komun ri déparetema` | 11,779 |
| 4 | `ri perancis ita to komun` | 10,125 |
| 5 | `to komun ri déparetema haute` | 1,825 |

**2-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `r i` | 90,059 |
| 2 | `a _` | 63,515 |
| 3 | `i _` | 58,114 |
| 4 | `_ r` | 57,562 |
| 5 | `t e` | 57,375 |

**3-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ r i` | 56,241 |
| 2 | `r i _` | 55,684 |
| 3 | `m u n` | 43,031 |
| 4 | `u n _` | 42,981 |
| 5 | `k o m` | 42,817 |

**4-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ r i _` | 55,382 |
| 2 | `o m u n` | 42,738 |
| 3 | `k o m u` | 42,737 |
| 4 | `m u n _` | 42,682 |
| 5 | `n _ r i` | 41,406 |

**5-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `k o m u n` | 42,737 |
| 2 | `o m u n _` | 42,672 |
| 3 | `n _ r i _` | 41,389 |
| 4 | `u n _ r i` | 40,955 |
| 5 | `m u n _ r` | 40,953 |


### Key Findings

- **Best Perplexity:** 2-gram (word) with 75
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
- **Coverage:** Top-1000 patterns cover ~78% 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 | 0.5091 | 1.423 | 2.20 | 33,150 | 49.1% |
| **1** | Subword | 0.6409 | 1.559 | 6.02 | 1,114 | 35.9% |
| **2** | Word | 0.1228 | 1.089 | 1.21 | 72,762 | 87.7% |
| **2** | Subword | 0.6769 | 1.599 | 3.79 | 6,702 | 32.3% |
| **3** | Word | 0.0488 | 1.034 | 1.07 | 87,846 | 95.1% |
| **3** | Subword | 0.6926 | 1.616 | 3.05 | 25,381 | 30.7% |
| **4** | Word | 0.0142 🏆 | 1.010 | 1.02 | 93,544 | 98.6% |
| **4** | Subword | 0.5499 | 1.464 | 2.16 | 77,409 | 45.0% |

### Generated Text Samples (Word-based)

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

**Context Size 1:**

1. `ri haute loire rocé roches avrillé caa guillaucourt guillemont guizancourt guyencourt saulcourt iyan...`
2. `komun ri déparetema dordogne ri déparetema somme ri lino kaminang maégai napunnai peddang malampe si...`
3. `déparetema aube ri déparetema vosges kategori komun ri manoraŋna perancis ita to komun ri perancis i...`

**Context Size 2:**

1. `komun ri ardennes`
2. `ri déparetema somme ri perancis ita to komun ri finistère`
3. `kategori komun ri déparetema somme kategori komun ri déparetema haute saône kategori komun ri gard`

**Context Size 3:**

1. `komun ri déparetema somme ri perancis ita to komun ri déparetema somme ri perancis ita to komun ri`
2. `kategori komun ri guadeloupe`
3. `ita to komun ri déparetema eure et loir kategori komun ri hautes pyrénées`

**Context Size 4:**

1. `to komun ri déparetema ain kategori komun ri ain`
2. `ita to komun ri déparetema vosges ri perancis ita to komun ri déparetema gard ri perancis ita to kom...`
3. `perancis ita to komun ri déparetema haute saône ri perancis ita to komun ri déparetema yvelines kate...`


### Generated Text Samples (Subword-based)

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

**Context Size 1:**

1. `_te_raweri:korom`
2. `apajesaniritori_`
3. `resèséun_i:ko_ay`

**Context Size 2:**

1. `ritu_séuwa_katema`
2. `a_agny-saônes_bin`
3. `i_dépari_lancis_s`

**Context Size 3:**

1. `_ri_aisnes_kategor`
2. `ri_déparetema_eurc`
3. `mun_ri_allers_kate`

**Context Size 4:**

1. `_ri_déparetema_côte`
2. `omun_ri_ain_vignoll`
3. `komun_ri_déparetema`


### Key Findings

- **Best Predictability:** Context-4 (word) with 98.6% predictability
- **Branching Factor:** Decreases with context size (more deterministic)
- **Memory Trade-off:** Larger contexts require more storage (77,409 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 | 13,449 |
| Total Tokens | 358,170 |
| Mean Frequency | 26.63 |
| Median Frequency | 2 |
| Frequency Std Dev | 718.89 |

### Most Common Words

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | ri | 55,392 |
| 2 | komun | 42,679 |
| 3 | déparetema | 27,244 |
| 4 | kategori | 15,395 |
| 5 | to | 14,029 |
| 6 | ita | 13,904 |
| 7 | iyanaritu | 13,505 |
| 8 | séuwa | 13,393 |
| 9 | perancis | 12,636 |
| 10 | haute | 6,206 |

### Least Common Words (from vocabulary)

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | museum | 2 |
| 2 | tychy | 2 |
| 3 | tangnga | 2 |
| 4 | miniaturowej | 2 |
| 5 | sztuki | 2 |
| 6 | profesjonalnej | 2 |
| 7 | wideo | 2 |
| 8 | nietypowe | 2 |
| 9 | sztalugi | 2 |
| 10 | zapałek | 2 |

### Zipf's Law Analysis

| Metric | Value |
|--------|-------|
| Zipf Coefficient | 0.9102 |
| R² (Goodness of Fit) | 0.956494 |
| Adherence Quality | **excellent** |

### Coverage Analysis

| Top N Words | Coverage |
|-------------|----------|
| Top 100 | 83.1% |
| Top 1,000 | 89.7% |
| Top 5,000 | 95.1% |
| Top 10,000 | 98.1% |

### Key Findings

- **Zipf Compliance:** R²=0.9565 indicates excellent adherence to Zipf's law
- **High Frequency Dominance:** Top 100 words cover 83.1% of corpus
- **Long Tail:** 3,449 words needed for remaining 1.9% 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 | 0.0849 🏆 | 0.7683 | N/A | N/A |
| **mono_64d** | 64 | 0.0269 | 0.6385 | N/A | N/A |
| **mono_128d** | 128 | 0.0039 | 0.6251 | N/A | N/A |
| **aligned_32d** | 32 | 0.0849 | 0.7636 | 0.0000 | 0.0300 |
| **aligned_64d** | 64 | 0.0269 | 0.6542 | 0.0120 | 0.1200 |
| **aligned_128d** | 128 | 0.0039 | 0.6125 | 0.0300 | 0.1620 |

### Key Findings

- **Best Isotropy:** mono_32d with 0.0849 (more uniform distribution)
- **Semantic Density:** Average pairwise similarity of 0.6770. Lower values indicate better semantic separation.
- **Alignment Quality:** Aligned models achieve up to 3.0% 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 | **0.239** | 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 |
|--------|----------|
| `-ma` | marson, massoins, maël |
| `-mo` | montégut, moncale, morton |
| `-ch` | chépy, cheylard, chatel |

#### Productive Suffixes
| Suffix | Examples |
|--------|----------|
| `-s` | siprus, massoins, hiis |
| `-e` | épagne, aizanville, vesle |
| `-es` | barges, vellèches, laspènes |
| `-le` | aizanville, vesle, gameville |
| `-lle` | aizanville, gameville, girondelle |
| `-rt` | begnécourt, hinacourt, bouzincourt |
| `-urt` | begnécourt, hinacourt, bouzincourt |
| `-ourt` | begnécourt, hinacourt, bouzincourt |

### 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 |
|------|----------|------------------|----------|
| `ngka` | 1.51x | 20 contexts | angka, engka, éngka |
| `appa` | 1.55x | 15 contexts | cappa, nappa, lappa |
| `engk` | 1.57x | 9 contexts | engka, engkaé, engkai |
| `seng` | 1.50x | 10 contexts | aseng, siseng, naseng |
| `asen` | 1.46x | 8 contexts | aseng, asenna, naseng |
| `unna` | 1.46x | 6 contexts | punna, punnai, umunna |
| `enna` | 1.46x | 5 contexts | asenna, sisenna, lalenna |
| `yana` | 1.38x | 5 contexts | iyana, iyanaé, iyanae |
| `iyan` | 1.37x | 5 contexts | iyana, iyanaé, iyanae |

### 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 |
|--------|--------|-----------|----------|
| `-ch` | `-s` | 56 words | chaulnes, champdeniers |
| `-ch` | `-e` | 46 words | châtaigneraie, chabre |
| `-ma` | `-e` | 44 words | maritime, maire |
| `-ma` | `-s` | 43 words | mainvilliers, mandres |
| `-mo` | `-s` | 41 words | molins, moulines |
| `-ch` | `-es` | 40 words | chaulnes, chamvres |
| `-mo` | `-e` | 19 words | motteville, moulière |
| `-ma` | `-es` | 18 words | mandres, maulichères |
| `-mo` | `-on` | 18 words | monthodon, montfaucon |
| `-mo` | `-rt` | 13 words | montlibert, montescourt |

### 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 |
|------|-----------------|------------|------|
| lagardelle | **`lagarde-lle`** | 4.5 | `lagarde` |
| motteville | **`mo-ttev-ille`** | 3.0 | `ttev` |
| chalencon | **`ch-alenc-on`** | 3.0 | `alenc` |
| champignelles | **`ch-ampignell-es`** | 3.0 | `ampignell` |
| chamarandes | **`ch-amarand-es`** | 3.0 | `amarand` |
| martinsart | **`ma-rtinsa-rt`** | 3.0 | `rtinsa` |
| manancourt | **`ma-nanc-ourt`** | 3.0 | `nanc` |
| charleville | **`ch-arlev-ille`** | 3.0 | `arlev` |
| montheries | **`mo-ntheri-es`** | 3.0 | `ntheri` |
| marseille | **`ma-rsei-lle`** | 3.0 | `rsei` |
| champvallon | **`ch-ampvall-on`** | 3.0 | `ampvall` |
| monthodon | **`mo-nthod-on`** | 3.0 | `nthod` |
| mazerolles | **`ma-zeroll-es`** | 3.0 | `zeroll` |
| chevrières | **`ch-evrièr-es`** | 3.0 | `evrièr` |
| montagnes | **`mo-ntagn-es`** | 3.0 | `ntagn` |

### 6.6 Linguistic Interpretation

> **Automated Insight:**
The language Buginese 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 | **32k BPE** | Best compression (4.93x) |
| N-gram | **2-gram** | Lowest perplexity (75) |
| Markov | **Context-4** | Highest predictability (98.6%) |
| 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 19:48:58*