README.md

---
language: bi
language_name: Bislama
language_family: germanic_west_anglofrisian
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-germanic_west_anglofrisian
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: 4.441
  - name: best_isotropy
    type: isotropy
    value: 0.0691
  - name: vocabulary_size
    type: vocab
    value: 0
generated: 2026-01-03
---

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

This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Bislama** 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** | 4.034x | 4.06 | 0.1436% | 45,948 |
| **16k** | 4.441x 🏆 | 4.46 | 0.1581% | 41,742 |

### Tokenization Examples

Below are sample sentences tokenized with each vocabulary size:

**Sample 1:** `Spiro Theodore "Ted" Agnew (9 Novemba – 17 Septemba em i politikis blong Yunaete...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁spi ro ▁theodore ▁" ted " ▁agnew ▁( 9 ▁novemba ... (+19 more)` | 29 |
| 16k | `▁spiro ▁theodore ▁" ted " ▁agnew ▁( 9 ▁novemba ▁– ... (+18 more)` | 28 |

**Sample 2:** `Xi Jinping (boen i hed blong stet blong Jaena. blong Stet blong Jaena`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁xi ▁jinping ▁( boen ▁i ▁hed ▁blong ▁stet ▁blong ▁jaena ... (+5 more)` | 15 |
| 16k | `▁xi ▁jinping ▁( boen ▁i ▁hed ▁blong ▁stet ▁blong ▁jaena ... (+5 more)` | 15 |

**Sample 3:** `Miori Ichikawa (boen 12 Februari em i bin woman blong singsing blong Japan. woma...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁mi ori ▁ich ika wa ▁( boen ▁ 1 2 ... (+16 more)` | 26 |
| 16k | `▁miori ▁ichikawa ▁( boen ▁ 1 2 ▁februari ▁em ▁i ... (+13 more)` | 23 |


### Key Findings

- **Best Compression:** 16k achieves 4.441x compression
- **Lowest UNK Rate:** 8k with 0.1436% 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 | 362 | 8.50 | 1,045 | 58.8% | 99.0% |
| **2-gram** | Subword | 208 🏆 | 7.70 | 976 | 73.9% | 100.0% |
| **3-gram** | Word | 494 | 8.95 | 1,403 | 53.1% | 92.1% |
| **3-gram** | Subword | 1,176 | 10.20 | 5,825 | 38.3% | 79.5% |
| **4-gram** | Word | 875 | 9.77 | 2,432 | 44.2% | 77.7% |
| **4-gram** | Subword | 3,512 | 11.78 | 19,179 | 28.6% | 58.3% |
| **5-gram** | Word | 727 | 9.51 | 1,831 | 46.0% | 82.2% |
| **5-gram** | Subword | 5,192 | 12.34 | 26,363 | 25.9% | 52.6% |

### Top 5 N-grams by Size

**2-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `hem i` | 741 |
| 2 | `stet blong` | 731 |
| 3 | `em i` | 611 |
| 4 | `blong amerika` | 599 |
| 5 | `blong yunaeted` | 537 |

**3-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `stet blong amerika` | 585 |
| 2 | `blong yunaeted stet` | 481 |
| 3 | `yunaeted stet blong` | 481 |
| 4 | `blong singsing blong` | 291 |
| 5 | `blong hem i` | 259 |

**4-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `yunaeted stet blong amerika` | 479 |
| 2 | `blong yunaeted stet blong` | 472 |
| 3 | `akta blong yunaeted stet` | 210 |
| 4 | `woman blong singsing blong` | 181 |
| 5 | `blong singsing blong japan` | 150 |

**5-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `blong yunaeted stet blong amerika` | 471 |
| 2 | `akta blong yunaeted stet blong` | 210 |
| 3 | `woman blong singsing blong japan` | 129 |
| 4 | `em i woman blong singsing` | 100 |
| 5 | `i woman blong singsing blong` | 96 |

**2-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `o n` | 9,097 |
| 2 | `n g` | 8,801 |
| 3 | `l o` | 8,033 |
| 4 | `g _` | 7,960 |
| 5 | `_ b` | 7,074 |

**3-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `n g _` | 7,816 |
| 2 | `o n g` | 7,315 |
| 3 | `l o n` | 7,271 |
| 4 | `_ b l` | 5,295 |
| 5 | `b l o` | 5,265 |

**4-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `o n g _` | 7,216 |
| 2 | `l o n g` | 7,207 |
| 3 | `_ b l o` | 5,255 |
| 4 | `b l o n` | 5,031 |
| 5 | `_ l o n` | 2,154 |

**5-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `l o n g _` | 7,179 |
| 2 | `b l o n g` | 5,030 |
| 3 | `_ b l o n` | 5,028 |
| 4 | `_ l o n g` | 2,151 |
| 5 | `e m _ i _` | 1,374 |


### Key Findings

- **Best Perplexity:** 2-gram (subword) with 208
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
- **Coverage:** Top-1000 patterns cover ~53% 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 | 0.5784 | 1.493 | 3.02 | 8,408 | 42.2% |
| **1** | Subword | 0.9577 | 1.942 | 6.51 | 362 | 4.2% |
| **2** | Word | 0.1997 | 1.148 | 1.41 | 25,020 | 80.0% |
| **2** | Subword | 0.9916 | 1.988 | 5.13 | 2,350 | 0.8% |
| **3** | Word | 0.0750 | 1.053 | 1.13 | 34,806 | 92.5% |
| **3** | Subword | 0.7944 | 1.734 | 3.18 | 12,029 | 20.6% |
| **4** | Word | 0.0323 🏆 | 1.023 | 1.05 | 38,812 | 96.8% |
| **4** | Subword | 0.4624 | 1.378 | 1.90 | 38,112 | 53.8% |

### Generated Text Samples (Word-based)

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

**Context Size 1:**

1. `blong miusik grup i praem minista blong pasifik tu kristianiti islam jeinisim i praem minista blong`
2. `i stap wetem graon kavremap 29 septemba hem hemi sapraesm ol pipol likem kakae we i`
3. `long septemba i stap mekem afta blong et et i wan fruit kakae we ol komposisen`

**Context Size 2:**

1. `hem i wan miusik grup stet blong philippines blong stet blong amerika man blong singsing blong japan`
2. `stet blong peru bik kaontri long saot blong yurop we i stap araon 860 090 external links`
3. `em i bin transletem niu testeman i kam mo watchem kustom danis wetem good fren pipol`

**Context Size 3:**

1. `yunaeted stet blong amerika akta blong yunaeted stet blong amerika risos long internet www vilnius l...`
2. `blong yunaeted stet blong amerika blong yunaeted stet blong amerika akta blong yunaeted stet blong a...`
3. `blong singsing blong taelan woman blong singsing blong japan woman blong singsing blong japan man bl...`

**Context Size 4:**

1. `blong yunaeted stet blong amerika akta blong yunaeted stet blong amerika blong stet blong yunaeted s...`
2. `yunaeted stet blong amerika bara lyle crist images of america alliance arcadia publishing s 41 isbn ...`
3. `akta blong yunaeted stet blong amerika akta blong yunaeted stet blong amerika akta blong yunaeted st...`


### Generated Text Samples (Subword-based)

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

**Context Size 1:**

1. `_stakthae_m_blon`
2. `ak_25paryulgraju`
3. `ng_lons_i_we_d_p`

**Context Size 2:**

1. `ong_yun_wosing_i_`
2. `ng_noasol_ww.cita`
3. `long_en_lon_i_sol`

**Context Size 3:**

1. `ng_nara_(cano_red_`
2. `ong_wan_blong_mius`
3. `long_(long_blong_y`

**Context Size 4:**

1. `ong_nolej,_televis_`
2. `long_gud_fasin_muha`
3. `_blong_stet_blong_s`


### Key Findings

- **Best Predictability:** Context-4 (word) with 96.8% predictability
- **Branching Factor:** Decreases with context size (more deterministic)
- **Memory Trade-off:** Larger contexts require more storage (38,112 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 | 3,106 |
| Total Tokens | 48,839 |
| Mean Frequency | 15.72 |
| Median Frequency | 3 |
| Frequency Std Dev | 125.16 |

### Most Common Words

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | blong | 5,030 |
| 2 | i | 3,201 |
| 3 | long | 2,145 |
| 4 | mo | 1,056 |
| 5 | hem | 1,010 |
| 6 | ol | 899 |
| 7 | wan | 870 |
| 8 | stet | 842 |
| 9 | amerika | 672 |
| 10 | em | 654 |

### Least Common Words (from vocabulary)

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | ftps | 2 |
| 2 | sftp | 2 |
| 3 | operating | 2 |
| 4 | guide | 2 |
| 5 | spesifikesen | 2 |
| 6 | firewall | 2 |
| 7 | sapot | 2 |
| 8 | lesin | 2 |
| 9 | sanem | 2 |
| 10 | extended | 2 |

### Zipf's Law Analysis

| Metric | Value |
|--------|-------|
| Zipf Coefficient | 1.0402 |
| R² (Goodness of Fit) | 0.989274 |
| Adherence Quality | **excellent** |

### Coverage Analysis

| Top N Words | Coverage |
|-------------|----------|
| Top 100 | 62.1% |
| Top 1,000 | 88.5% |
| Top 5,000 | 0.0% |
| 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 62.1% of corpus
- **Long Tail:** -6,894 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 | 0.0691 🏆 | 0.6642 | N/A | N/A |
| **mono_64d** | 64 | 0.0097 | 0.6595 | N/A | N/A |
| **mono_128d** | 128 | 0.0022 | 0.6755 | N/A | N/A |
| **aligned_32d** | 32 | 0.0691 | 0.6741 | 0.0060 | 0.0420 |
| **aligned_64d** | 64 | 0.0097 | 0.6519 | 0.0080 | 0.0860 |
| **aligned_128d** | 128 | 0.0022 | 0.6801 | 0.0200 | 0.0920 |

### Key Findings

- **Best Isotropy:** mono_32d with 0.0691 (more uniform distribution)
- **Semantic Density:** Average pairwise similarity of 0.6675. Lower values indicate better semantic separation.
- **Alignment Quality:** Aligned models achieve up to 2.0% 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 | **0.564** | 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 |
|--------|----------|

#### Productive Suffixes
| Suffix | Examples |
|--------|----------|
| `-en` | warren, truiden, paten |
| `-em` | katem, raonem, sanem |
| `-an` | ejukesan, busan, giaman |

### 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 |
|------|----------|------------------|----------|
| `amba` | 1.40x | 8 contexts | ambae, namba, stamba |

### 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 |
|------|-----------------|------------|------|
| republican | **`republic-an`** | 4.5 | `republic` |
| andastanem | **`andast-an-em`** | 3.0 | `andast` |
| niutesteman | **`niutest-em-an`** | 3.0 | `niutest` |
| komunikesen | **`komunikes-en`** | 1.5 | `komunikes` |
| oganaesesen | **`oganaeses-en`** | 1.5 | `oganaeses` |
| sustreksen | **`sustreks-en`** | 1.5 | `sustreks` |
| vaespresiden | **`vaespresid-en`** | 1.5 | `vaespresid` |
| populesen | **`popules-en`** | 1.5 | `popules` |
| ekshumesen | **`ekshumes-en`** | 1.5 | `ekshumes` |
| komposisen | **`komposis-en`** | 1.5 | `komposis` |
| konstitusen | **`konstitus-en`** | 1.5 | `konstitus` |
| sébastien | **`sébasti-en`** | 1.5 | `sébasti` |
| austronesian | **`austronesi-an`** | 1.5 | `austronesi` |
| divelopem | **`divelop-em`** | 1.5 | `divelop` |
| christian | **`christi-an`** | 1.5 | `christi` |

### 6.6 Linguistic Interpretation

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

> **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 | **16k BPE** | Best compression (4.44x) |
| N-gram | **2-gram** | Lowest perplexity (208) |
| Markov | **Context-4** | Highest predictability (96.8%) |
| 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 18:57:38*