README.md

---
language: ty
language_name: Tahitian
language_family: austronesian_polynesian
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_polynesian
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: 3.561
  - name: best_isotropy
    type: isotropy
    value: 0.0301
  - name: vocabulary_size
    type: vocab
    value: 0
generated: 2026-01-11
---

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

This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Tahitian** 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** | 3.455x | 3.48 | 0.1990% | 40,695 |
| **16k** | 3.561x 🏆 | 3.59 | 0.2052% | 39,479 |

### Tokenization Examples

Below are sample sentences tokenized with each vocabulary size:

**Sample 1:** `’O Sant Miquel de Campmajor te hō’ē ’oire iti nō Tatarūnia. mau ’oire iti nō Tat...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁’ o ▁sant ▁miquel ▁de ▁camp major ▁te ▁hō ’ ... (+13 more)` | 23 |
| 16k | `▁’ o ▁sant ▁miquel ▁de ▁campmajor ▁te ▁hō ’ ē ... (+12 more)` | 22 |

**Sample 2:** `Ò Hakahau te òire rahi aè no Ua Pou i Pōrīnetia farāni. ènata`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁ò ▁hakahau ▁te ▁òire ▁rahi ▁aè ▁no ▁ua ▁pou ▁i ... (+4 more)` | 14 |
| 16k | `▁ò ▁hakahau ▁te ▁òire ▁rahi ▁aè ▁no ▁ua ▁pou ▁i ... (+4 more)` | 14 |

**Sample 3:** `’O te hō’ē ’oire iti nō Soria. mau ’oire iti nō Soria`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁’ o ▁te ▁hō ’ ē ▁’ oire ▁iti ▁nō ... (+8 more)` | 18 |
| 16k | `▁’ o ▁te ▁hō ’ ē ▁’ oire ▁iti ▁nō ... (+8 more)` | 18 |


### Key Findings

- **Best Compression:** 16k achieves 3.561x compression
- **Lowest UNK Rate:** 8k with 0.1990% 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 | 422 | 8.72 | 1,465 | 56.1% | 93.7% |
| **2-gram** | Subword | 157 🏆 | 7.29 | 1,040 | 80.2% | 99.9% |
| **3-gram** | Word | 804 | 9.65 | 2,559 | 47.1% | 82.0% |
| **3-gram** | Subword | 845 | 9.72 | 5,412 | 45.0% | 86.0% |
| **4-gram** | Word | 1,231 | 10.27 | 4,355 | 43.3% | 70.6% |
| **4-gram** | Subword | 2,588 | 11.34 | 15,928 | 29.6% | 67.7% |
| **5-gram** | Word | 874 | 9.77 | 3,200 | 48.9% | 75.5% |
| **5-gram** | Subword | 4,432 | 12.11 | 22,183 | 24.4% | 57.5% |

### Top 5 N-grams by Size

**2-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `i te` | 2,291 |
| 2 | `te mau` | 1,467 |
| 3 | `o te` | 1,091 |
| 4 | `oire iti` | 927 |
| 5 | `iti nō` | 927 |

**3-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `oire iti nō` | 927 |
| 2 | `te hō ē` | 509 |
| 3 | `hō ē oire` | 499 |
| 4 | `ē oire iti` | 472 |
| 5 | `mau oire iti` | 455 |

**4-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `te hō ē oire` | 499 |
| 2 | `ē oire iti nō` | 472 |
| 3 | `hō ē oire iti` | 472 |
| 4 | `mau oire iti nō` | 455 |
| 5 | `oire iti nō tatarūnia` | 451 |

**5-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `te hō ē oire iti` | 472 |
| 2 | `hō ē oire iti nō` | 472 |
| 3 | `o te hō ē oire` | 228 |
| 4 | `mau oire iti nō tatarūnia` | 226 |
| 5 | `tatarūnia mau oire iti nō` | 225 |

**2-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `e _` | 14,698 |
| 2 | `_ t` | 13,338 |
| 3 | `a _` | 13,137 |
| 4 | `t e` | 10,307 |
| 5 | `i _` | 9,732 |

**3-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ t e` | 8,324 |
| 2 | `t e _` | 8,198 |
| 3 | `_ m a` | 4,382 |
| 4 | `_ i _` | 3,923 |
| 5 | `i _ t` | 3,482 |

**4-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ t e _` | 7,569 |
| 2 | `i _ t e` | 2,795 |
| 3 | `_ i _ t` | 2,691 |
| 4 | `e _ m a` | 2,393 |
| 5 | `t e _ m` | 2,331 |

**5-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `i _ t e _` | 2,677 |
| 2 | `_ i _ t e` | 2,345 |
| 3 | `t e _ m a` | 2,135 |
| 4 | `_ t e _ m` | 2,091 |
| 5 | `_ m a u _` | 2,082 |


### Key Findings

- **Best Perplexity:** 2-gram (subword) with 157
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
- **Coverage:** Top-1000 patterns cover ~58% 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 | 0.5459 | 1.460 | 3.10 | 6,845 | 45.4% |
| **1** | Subword | 1.3451 | 2.541 | 10.23 | 217 | 0.0% |
| **2** | Word | 0.2579 | 1.196 | 1.61 | 21,075 | 74.2% |
| **2** | Subword | 1.0621 | 2.088 | 5.21 | 2,216 | 0.0% |
| **3** | Word | 0.1372 | 1.100 | 1.25 | 33,605 | 86.3% |
| **3** | Subword | 0.7095 | 1.635 | 2.86 | 11,525 | 29.1% |
| **4** | Word | 0.0708 🏆 | 1.050 | 1.11 | 41,713 | 92.9% |
| **4** | Subword | 0.3985 | 1.318 | 1.79 | 32,904 | 60.2% |

### Generated Text Samples (Word-based)

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

**Context Size 1:**

1. `te matahiti te reo wiwi me te apooraa ua riro oia ana i te fare haapiiraa`
2. `i ma il sung te repupirita no ghana e te haere oia ei tauturu a era`
3. `e rave a te matahiti ua huru o te purūmu hātua e rave e tae atu`

**Context Size 2:**

1. `i te 27 no mē tai ivuaro peretiteni o te taata nei e nina atoa hia o`
2. `te mau mea atoa ta na ïa i rave no te mau tupuna i afa i mai`
3. `o te repūpirita michael sata 23 no tiurai herēni peretiteni o te papori me te aro o`

**Context Size 3:**

1. `oire iti nō tatarūnia mau oire iti nō tatarūnia mau oire iti nō soria mau oire iti nō`
2. `te hō ē oire iti nō tatarūnia mau oire iti nō soria mau oire iti nō soria mau`
3. `hō ē oire iti nō tatarūnia mau oire iti nō fenua marite huira atira 681 090 ta ata`

**Context Size 4:**

1. `te hō ē oire iti nō soria mau oire iti nō soria mau oire iti nō tatarūnia mau oire`
2. `ē oire iti nō soria mau oire iti nō soria mau oire iti nō tatarūnia mau oire iti nō`
3. `hō ē oire iti nō tatarūnia mau oire iti nō tatarūnia mau oire iti nō soria mau oire iti`


### Generated Text Samples (Subword-based)

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

**Context Size 1:**

1. `_fētitoino_nurât`
2. `a,_tetu_hnafena_`
3. `i_i_nō_tē_no,_oa`

**Context Size 2:**

1. `e_te_14_nov._utom`
2. `_te_ia_te_mē_’oia`
3. `a_ra_faapera,_ó_t`

**Context Size 3:**

1. `_te_aorené_paraa_f`
2. `te_faata_no_tupu_p`
3. `_mau_fāna_nei_o_tu`

**Context Size 4:**

1. `_te_mau_poritita_na`
2. `i_te_repūpirita_mot`
3. `_i_te_di_rave_rapaa`


### Key Findings

- **Best Predictability:** Context-4 (word) with 92.9% predictability
- **Branching Factor:** Decreases with context size (more deterministic)
- **Memory Trade-off:** Larger contexts require more storage (32,904 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 | 2,668 |
| Total Tokens | 62,941 |
| Mean Frequency | 23.59 |
| Median Frequency | 3 |
| Frequency Std Dev | 206.25 |

### Most Common Words

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | te | 7,849 |
| 2 | i | 4,463 |
| 3 | e | 2,642 |
| 4 | o | 2,217 |
| 5 | no | 2,165 |
| 6 | mau | 2,091 |
| 7 | a | 1,691 |
| 8 | nō | 1,108 |
| 9 | oire | 1,030 |
| 10 | iti | 946 |

### Least Common Words (from vocabulary)

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | antitumor | 2 |
| 2 | mcgill | 2 |
| 3 | polanyi | 2 |
| 4 | stanford | 2 |
| 5 | lehn | 2 |
| 6 | uttar | 2 |
| 7 | pradesh | 2 |
| 8 | papu | 2 |
| 9 | tarutaru | 2 |
| 10 | anavai | 2 |

### Zipf's Law Analysis

| Metric | Value |
|--------|-------|
| Zipf Coefficient | 1.1618 |
| R² (Goodness of Fit) | 0.985563 |
| Adherence Quality | **excellent** |

### Coverage Analysis

| Top N Words | Coverage |
|-------------|----------|
| Top 100 | 72.0% |
| Top 1,000 | 93.5% |
| Top 5,000 | 0.0% |
| Top 10,000 | 0.0% |

### Key Findings

- **Zipf Compliance:** R²=0.9856 indicates excellent adherence to Zipf's law
- **High Frequency Dominance:** Top 100 words cover 72.0% of corpus
- **Long Tail:** -7,332 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 | 0.0301 | 0.6381 | N/A | N/A |
| **mono_64d** | 64 | 0.0049 | 0.6224 | N/A | N/A |
| **mono_128d** | 128 | 0.0009 | 0.6655 | N/A | N/A |
| **aligned_32d** | 32 | 0.0301 🏆 | 0.6684 | 0.0028 | 0.0499 |
| **aligned_64d** | 64 | 0.0049 | 0.6410 | 0.0028 | 0.0748 |
| **aligned_128d** | 128 | 0.0009 | 0.6513 | 0.0055 | 0.0914 |

### Key Findings

- **Best Isotropy:** aligned_32d with 0.0301 (more uniform distribution)
- **Semantic Density:** Average pairwise similarity of 0.6478. Lower values indicate better semantic separation.
- **Alignment Quality:** Aligned models achieve up to 0.6% 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 | **0.313** | 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 |
|--------|----------|
| `-t` | tau, tauaparauraa, tapearaa |
| `-a` | anuanua, apooraa, agnes |
| `-m` | mori, mesia, mǎta |
| `-p` | pou, piahi, ph |
| `-ta` | tau, tauaparauraa, tapearaa |
| `-ma` | maha, maurice, maoro |
| `-fa` | farii, fakarava, faaoreraa |
| `-pa` | paari, paradisiaca, paturaa |

#### Productive Suffixes
| Suffix | Examples |
|--------|----------|
| `-a` | fakarava, anuanua, apooraa |
| `-e` | òe, grace, ne |
| `-ia` | mesia, citrifolia, māìtihia |
| `-i` | fifi, farii, mori |
| `-aa` | apooraa, oraraa, itiraa |
| `-ra` | atira, mētera, tera |
| `-na` | ghana, taina, raihana |
| `-ta` | mǎta, poritita, rekoata |

### 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 |
|------|----------|------------------|----------|
| `anga` | 1.54x | 13 contexts | hangai, whanga, umanga |
| `ahit` | 1.37x | 7 contexts | tahiti, mahiti, tahito |
| `faah` | 1.39x | 6 contexts | faahi, faaho, faahou |
| `tira` | 1.37x | 6 contexts | atira, itiraa, raatira |
| `aama` | 1.36x | 5 contexts | raama, haamau, haamata |
| `haam` | 1.36x | 4 contexts | haamo, haamau, haamou |

### 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 |
|--------|--------|-----------|----------|
| `-t` | `-a` | 68 words | tauaparauraa, tapearaa |
| `-m` | `-a` | 49 words | mesia, mǎta |
| `-a` | `-a` | 44 words | anuanua, apooraa |
| `-fa` | `-a` | 41 words | fakarava, faaoreraa |
| `-p` | `-a` | 41 words | poritita, paradisiaca |
| `-fa` | `-aa` | 21 words | faaoreraa, faaotiraa |
| `-t` | `-aa` | 19 words | tauaparauraa, tapearaa |
| `-t` | `-ia` | 18 words | torovenia, tureia |
| `-t` | `-i` | 16 words | tauatini, tieti |
| `-p` | `-ia` | 15 words | pipiria, punaauia |

### 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 |
|------|-----------------|------------|------|
| tauaparauraa | **`tauaparaur-a-a`** | 7.5 | `a` |
| faaoreraa | **`faaorer-a-a`** | 7.5 | `a` |
| faaotiraa | **`faaotir-a-a`** | 7.5 | `a` |
| faaohiparaa | **`faaohipar-a-a`** | 7.5 | `a` |
| feruriraa | **`ferurir-a-a`** | 7.5 | `a` |
| faanavairaa | **`faanavair-a-a`** | 7.5 | `a` |
| boraginaceae | **`boraginace-a-e`** | 7.5 | `a` |
| faaûruraa | **`faaûrur-a-a`** | 7.5 | `a` |
| haaparuparu | **`haaparup-a-ru`** | 7.5 | `a` |
| haapiiraa | **`haapiir-a-a`** | 7.5 | `a` |
| faahororaa | **`faahoror-a-a`** | 7.5 | `a` |
| misionare | **`mision-a-re`** | 7.5 | `a` |
| faaineineraa | **`faaineiner-a-a`** | 7.5 | `a` |
| rapaauraa | **`rapaaur-a-a`** | 7.5 | `a` |
| faataaraa | **`faataa-ra-a`** | 7.5 | `ra` |

### 6.6 Linguistic Interpretation

> **Automated Insight:**
The language Tahitian 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 (3.56x) |
| N-gram | **2-gram** | Lowest perplexity (157) |
| Markov | **Context-4** | Highest predictability (92.9%) |
| 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: 2026-01-11 02:05:21*