README.md

---
language: ch
language_name: Chamorro
language_family: austronesian_oceanic_other
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_oceanic_other
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: 0.0563
  - name: vocabulary_size
    type: vocab
    value: 0
generated: 2026-01-03
---

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

This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Chamorro** 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** | 3.977x | 3.99 | 0.0998% | 38,069 |
| **16k** | 4.248x 🏆 | 4.26 | 0.1066% | 35,644 |

### Tokenization Examples

Below are sample sentences tokenized with each vocabulary size:

**Sample 1:** `+Afghanistan 125px Anthem: Millī سرود 300px Afghanistan capitat Kabul. Guåha na ...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁+ af ghanistan ▁ 1 2 5 px ▁anthem : ... (+21 more)` | 31 |
| 16k | `▁+ afghanistan ▁ 1 2 5 px ▁anthem : ▁millī ... (+20 more)` | 30 |

**Sample 2:** `Cartersville, nasong-song gi Estados Unidos. Guåha 19,731 na tataogues na popula...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁carters ville , ▁nasong - song ▁gi ▁estados ▁unidos . ... (+18 more)` | 28 |
| 16k | `▁cartersville , ▁nasong - song ▁gi ▁estados ▁unidos . ▁guåha ... (+17 more)` | 27 |

**Sample 3:** `Waleska, nasong-song gi Estados Unidos. Guåha 644 na tataogues na populasion i s...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁wa les ka , ▁nasong - song ▁gi ▁estados ▁unidos ... (+16 more)` | 26 |
| 16k | `▁waleska , ▁nasong - song ▁gi ▁estados ▁unidos . ▁guåha ... (+14 more)` | 24 |


### Key Findings

- **Best Compression:** 16k achieves 4.248x compression
- **Lowest UNK Rate:** 8k with 0.0998% 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 | 178 | 7.48 | 491 | 68.4% | 100.0% |
| **2-gram** | Subword | 227 | 7.83 | 866 | 71.1% | 100.0% |
| **3-gram** | Word | 133 | 7.06 | 577 | 70.8% | 100.0% |
| **3-gram** | Subword | 1,279 | 10.32 | 4,533 | 36.5% | 79.7% |
| **4-gram** | Word | 156 | 7.29 | 834 | 66.8% | 100.0% |
| **4-gram** | Subword | 3,664 | 11.84 | 12,412 | 26.2% | 57.0% |
| **5-gram** | Word | 102 🏆 | 6.67 | 583 | 72.6% | 100.0% |
| **5-gram** | Subword | 5,287 | 12.37 | 16,015 | 24.4% | 49.4% |

### Top 5 N-grams by Size

**2-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `i sengsong` | 364 |
| 2 | `nu i` | 329 |
| 3 | `i senso` | 310 |
| 4 | `na populasion` | 309 |
| 5 | `populasion i` | 308 |

**3-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `nu i senso` | 308 |
| 2 | `na populasion i` | 304 |
| 3 | `na tataogues na` | 304 |
| 4 | `tataogues na populasion` | 304 |
| 5 | `i sengsong nu` | 299 |

**4-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `na tataogues na populasion` | 304 |
| 2 | `tataogues na populasion i` | 303 |
| 3 | `sengsong nu i senso` | 299 |
| 4 | `i sengsong nu i` | 299 |
| 5 | `populasion i sengsong nu` | 299 |

**5-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `na tataogues na populasion i` | 303 |
| 2 | `populasion i sengsong nu i` | 299 |
| 3 | `i sengsong nu i senso` | 299 |
| 4 | `na populasion i sengsong nu` | 299 |
| 5 | `tataogues na populasion i sengsong` | 298 |

**2-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `a _` | 4,908 |
| 2 | `i _` | 4,194 |
| 3 | `n a` | 2,916 |
| 4 | `a n` | 2,801 |
| 5 | `_ i` | 2,765 |

**3-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ i _` | 2,248 |
| 2 | `_ n a` | 1,823 |
| 3 | `n a _` | 1,562 |
| 4 | `_ g i` | 1,298 |
| 5 | `_ m a` | 1,144 |

**4-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ n a _` | 1,357 |
| 2 | `_ g i _` | 959 |
| 3 | `s o n g` | 952 |
| 4 | `_ i _ s` | 793 |
| 5 | `o n g _` | 758 |

**5-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ i _ s e` | 690 |
| 2 | `i _ s e n` | 687 |
| 3 | `s o n g _` | 653 |
| 4 | `_ u n i d` | 463 |
| 5 | `u n i d o` | 448 |


### Key Findings

- **Best Perplexity:** 5-gram (word) with 102
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
- **Coverage:** Top-1000 patterns cover ~49% 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 | 0.4903 | 1.405 | 2.61 | 5,477 | 51.0% |
| **1** | Subword | 1.0984 | 2.141 | 7.88 | 223 | 0.0% |
| **2** | Word | 0.1693 | 1.125 | 1.32 | 14,138 | 83.1% |
| **2** | Subword | 1.1342 | 2.195 | 5.32 | 1,755 | 0.0% |
| **3** | Word | 0.0592 | 1.042 | 1.09 | 18,443 | 94.1% |
| **3** | Subword | 0.7400 | 1.670 | 2.81 | 9,321 | 26.0% |
| **4** | Word | 0.0211 🏆 | 1.015 | 1.03 | 19,853 | 97.9% |
| **4** | Subword | 0.3920 | 1.312 | 1.72 | 26,122 | 60.8% |

### Generated Text Samples (Word-based)

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

**Context Size 1:**

1. `i saddok segua ya siha gi i mayot maelihi gobietna i mundo ma li e società`
2. `na populasion i senso unidos guåha 296 na agronomia i senso bibliografia riferensia horst lehne and`
3. `gi i sengsong nu i patgon siha ma usa ginen i dos gi islan sumatra pekanbaru`

**Context Size 2:**

1. `i sengsong nu i senso unidos`
2. `nu i senso para i fondo gaige hålom hånom hao kalan guihan gue gi iya estados unidos`
3. `na populasion i sengsong nu i senso unidos`

**Context Size 3:**

1. `na tataogues na populasion i sengsong nu i senso unidos`
2. `na populasion i sengsong nu i senso website sanhiyong siha rome`
3. `tataogues na populasion i sengsong nu i senso yeet website sanhiyong siha commons coronel fabriciano`

**Context Size 4:**

1. `na tataogues na populasion i sengsong nu i senso unidos`
2. `tataogues na populasion i sengsong nu i senso unidos`
3. `na populasion i sengsong nu i senso unidos`


### Generated Text Samples (Subword-based)

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

**Context Size 1:**

1. `_yia_a_mesotinio`
2. `a_dorn._ikug._s_`
3. `nusot_fai_i_i_gs`

**Context Size 2:**

1. `a_para_ediu_nasto`
2. `i_me":_ki,_vícite`
3. `na'i_achamane_pås`

**Context Size 3:**

1. `_i_semak_senggen_c`
2. `_na_pat_gi_wikike'`
3. `na_taogues_na_gi_k`

**Context Size 4:**

1. `_na_populasion_yan_`
2. `_gi_para_u_matungo'`
3. `song_nu_i_sengsong_`


### Key Findings

- **Best Predictability:** Context-4 (word) with 97.9% predictability
- **Branching Factor:** Decreases with context size (more deterministic)
- **Memory Trade-off:** Larger contexts require more storage (26,122 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 | 1,919 |
| Total Tokens | 22,562 |
| Mean Frequency | 11.76 |
| Median Frequency | 3 |
| Frequency Std Dev | 73.53 |

### Most Common Words

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | i | 2,319 |
| 2 | na | 1,511 |
| 3 | gi | 974 |
| 4 | unidos | 448 |
| 5 | yan | 436 |
| 6 | sengsong | 370 |
| 7 | guåha | 356 |
| 8 | nu | 335 |
| 9 | ni | 334 |
| 10 | populasion | 331 |

### Least Common Words (from vocabulary)

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | säger | 2 |
| 2 | ett | 2 |
| 3 | så | 2 |
| 4 | du | 2 |
| 5 | skate | 2 |
| 6 | med | 2 |
| 7 | smaskiga | 2 |
| 8 | löken | 2 |
| 9 | tychy | 2 |
| 10 | museon | 2 |

### Zipf's Law Analysis

| Metric | Value |
|--------|-------|
| Zipf Coefficient | 0.9547 |
| R² (Goodness of Fit) | 0.986088 |
| Adherence Quality | **excellent** |

### Coverage Analysis

| Top N Words | Coverage |
|-------------|----------|
| Top 100 | 63.2% |
| Top 1,000 | 91.3% |
| Top 5,000 | 0.0% |
| Top 10,000 | 0.0% |

### Key Findings

- **Zipf Compliance:** R²=0.9861 indicates excellent adherence to Zipf's law
- **High Frequency Dominance:** Top 100 words cover 63.2% of corpus
- **Long Tail:** -8,081 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 | 0.0563 🏆 | 0.6662 | N/A | N/A |
| **mono_64d** | 64 | 0.0067 | 0.8730 | N/A | N/A |
| **mono_128d** | 128 | 0.0017 | 0.8734 | N/A | N/A |
| **aligned_32d** | 32 | 0.0563 | 0.6862 | 0.0332 | 0.1848 |
| **aligned_64d** | 64 | 0.0067 | 0.8793 | 0.0095 | 0.1090 |
| **aligned_128d** | 128 | 0.0017 | 0.8561 | 0.0047 | 0.0853 |

### Key Findings

- **Best Isotropy:** mono_32d with 0.0563 (more uniform distribution)
- **Semantic Density:** Average pairwise similarity of 0.8057. Lower values indicate better semantic separation.
- **Alignment Quality:** Aligned models achieve up to 3.3% 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 | **3.506** | High morphological productivity | Reliable analysis |
| Idiomaticity Gap | **1.025** | 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` | manamerikanu, maisang, manmafa |

#### Productive Suffixes
| Suffix | Examples |
|--------|----------|
| `-a` | sina, finta, nangga |
| `-n` | ayman, guguan, direchon |
| `-on` | direchon, mision, museon |
| `-an` | ayman, guguan, geran |
| `-ia` | iglesia, cecilia, diktionaria |
| `-ion` | mision, administration, nasion |

### 6.3 Bound Stems (Lexical Roots)

Bound stems are high-frequency subword units that are semantically cohesive but rarely appear as standalone words. These often correspond to the 'core' of a word that requires inflection or derivation to be valid.

*No significant bound stems detected.*


### 6.4 Affix Compatibility (Co-occurrence)

This table shows which prefixes and suffixes most frequently co-occur on the same stems, revealing the 'stacking' rules of the language's morphology.

| Prefix | Suffix | Frequency | Examples |
|--------|--------|-----------|----------|
| `-ma` | `-a` | 17 words | manmafa, mafana |
| `-ma` | `-n` | 13 words | mangginen, manmatutuhon |
| `-ma` | `-an` | 6 words | masasangan, maneran |
| `-ma` | `-on` | 4 words | manmatutuhon, matutuhon |
| `-ma` | `-ia` | 1 words | malaysia, maria |

### 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 |
|------|-----------------|------------|------|
| makonsidera | **`ma-konsidera`** | 4.5 | `konsidera` |
| manmatutuhon | **`ma-nmatutuh-on`** | 3.0 | `nmatutuh` |
| matutuhon | **`ma-tutuh-on`** | 3.0 | `tutuh` |
| masasangan | **`ma-sasang-an`** | 3.0 | `sasang` |
| pennsylvania | **`pennsylv-an-ia`** | 3.0 | `pennsylv` |
| manofisinan | **`ma-nofisin-an`** | 3.0 | `nofisin` |
| manguayan | **`ma-nguay-an`** | 3.0 | `nguay` |
| machulijan | **`ma-chulij-an`** | 3.0 | `chulij` |
| manamerikanu | **`ma-namerikanu`** | 1.5 | `namerikanu` |
| diktionaria | **`diktionar-ia`** | 1.5 | `diktionar` |
| administration | **`administrat-ion`** | 1.5 | `administrat` |
| misionarion | **`misionar-ion`** | 1.5 | `misionar` |
| mangginen | **`ma-ngginen`** | 1.5 | `ngginen` |
| toneladan | **`tonelad-an`** | 1.5 | `tonelad` |
| wikimedia | **`wikimed-ia`** | 1.5 | `wikimed` |

### 6.6 Linguistic Interpretation

> **Automated Insight:**
The language Chamorro 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 (4.25x) |
| N-gram | **5-gram** | Lowest perplexity (102) |
| Markov | **Context-4** | Highest predictability (97.9%) |
| 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 20:18:48*