---
language: gur
language_name: Frafra
language_family: atlantic_gur
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-atlantic_gur
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: 4.001
  - name: best_isotropy
    type: isotropy
    value: 0.7704
  - name: vocabulary_size
    type: vocab
    value: 0
generated: 2026-01-10
---

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

This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Frafra** 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** | 3.687x | 3.69 | 0.1485% | 403,994 |
| **16k** | 3.867x | 3.87 | 0.1558% | 385,154 |
| **32k** | 4.001x 🏆 | 4.00 | 0.1612% | 372,255 |

### Tokenization Examples

Below are sample sentences tokenized with each vocabulary size:

**Sample 1:** `Buɣum Chuɣu de la de'eŋo n boi northern Ghana so'olum. Yelesi'a n bo de'eŋo la p...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁bu ɣ um ▁ch u ɣ u ▁de ▁la ▁de ... (+22 more)` | 32 |
| 16k | `▁bu ɣ um ▁chu ɣ u ▁de ▁la ▁de ' ... (+21 more)` | 31 |
| 32k | `▁bu ɣ um ▁chu ɣ u ▁de ▁la ▁de ' ... (+21 more)` | 31 |

**Sample 2:** `David Acquah' de la Gaana boole ŋwɛ'ara Club Tuuma A Solemitiŋa Tuuma A Miŋa Vom`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁david ▁acquah ' ▁de ▁la ▁gaana ▁boole ▁ŋwɛ ' ara ... (+8 more)` | 18 |
| 16k | `▁david ▁acquah ' ▁de ▁la ▁gaana ▁boole ▁ŋwɛ ' ara ... (+8 more)` | 18 |
| 32k | `▁david ▁acquah ' ▁de ▁la ▁gaana ▁boole ▁ŋwɛ ' ara ... (+8 more)` | 18 |

**Sample 3:** `William Du Bois Yaw Salhi Kumi (May 5, yuure ken dɛla Koo Kumi.`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁william ▁du ▁boi s ▁yaw ▁sal hi ▁kumi ▁( may ... (+9 more)` | 19 |
| 16k | `▁william ▁du ▁boi s ▁yaw ▁sal hi ▁kumi ▁( may ... (+9 more)` | 19 |
| 32k | `▁william ▁du ▁bois ▁yaw ▁salhi ▁kumi ▁( may ▁ 5 ... (+7 more)` | 17 |


### Key Findings

- **Best Compression:** 32k achieves 4.001x compression
- **Lowest UNK Rate:** 8k with 0.1485% 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 | 2,984 | 11.54 | 12,149 | 29.4% | 60.4% |
| **2-gram** | Subword | 241 🏆 | 7.92 | 2,090 | 68.4% | 99.3% |
| **3-gram** | Word | 9,118 | 13.15 | 23,058 | 15.5% | 40.4% |
| **3-gram** | Subword | 1,660 | 10.70 | 15,739 | 33.3% | 76.7% |
| **4-gram** | Word | 22,484 | 14.46 | 43,960 | 9.9% | 26.4% |
| **4-gram** | Subword | 7,120 | 12.80 | 67,011 | 19.0% | 50.9% |
| **5-gram** | Word | 20,312 | 14.31 | 34,263 | 9.1% | 25.3% |
| **5-gram** | Subword | 18,752 | 14.19 | 135,527 | 13.4% | 36.8% |

### Top 5 N-grams by Size

**2-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `la puan` | 6,048 |
| 2 | `de la` | 5,275 |
| 3 | `ti ba` | 4,735 |
| 4 | `n de` | 3,480 |
| 5 | `yuunɛ la` | 3,371 |

**3-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `yuunɛ la puan` | 2,827 |
| 2 | `e zo e` | 1,083 |
| 3 | `zo e zo` | 1,080 |
| 4 | `la puan a` | 938 |
| 5 | `ba yi ira` | 814 |

**4-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `zo e zo e` | 1,079 |
| 2 | `ti ba yi ira` | 779 |
| 3 | `yuunɛ la puan a` | 641 |
| 4 | `of the 4th republic` | 580 |
| 5 | `parliament of the 4th` | 573 |

**5-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `parliament of the 4th republic` | 573 |
| 2 | `ti ba yi ira ti` | 369 |
| 3 | `nɛreba parliament of the 4th` | 297 |
| 4 | `nalɛgeriba nɛreba parliament of the` | 292 |
| 5 | `lɔgerɔ nalɛgeriba nɛreba parliament of` | 266 |

**2-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `a _` | 167,038 |
| 2 | `l a` | 58,490 |
| 3 | `_ l` | 56,125 |
| 4 | `e _` | 52,651 |
| 5 | `i _` | 52,108 |

**3-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `l a _` | 48,700 |
| 2 | `_ l a` | 47,930 |
| 3 | `_ t i` | 22,894 |
| 4 | `t i _` | 21,274 |
| 5 | `n a _` | 19,826 |

**4-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ l a _` | 42,166 |
| 2 | `_ y u u` | 16,124 |
| 3 | `_ t i _` | 15,515 |
| 4 | `a _ l a` | 12,811 |
| 5 | `_ p u a` | 11,224 |

**5-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ p u a n` | 11,191 |
| 2 | `a _ l a _` | 10,944 |
| 3 | `e _ l a _` | 8,770 |
| 4 | `a _ p u a` | 8,569 |
| 5 | `_ y u u m` | 8,354 |


### Key Findings

- **Best Perplexity:** 2-gram (subword) with 241
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
- **Coverage:** Top-1000 patterns cover ~37% 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 | 0.7873 | 1.726 | 5.18 | 34,791 | 21.3% |
| **1** | Subword | 0.8475 | 1.799 | 6.78 | 735 | 15.3% |
| **2** | Word | 0.2846 | 1.218 | 1.80 | 180,038 | 71.5% |
| **2** | Subword | 0.9784 | 1.970 | 5.94 | 4,984 | 2.2% |
| **3** | Word | 0.1408 | 1.102 | 1.29 | 323,151 | 85.9% |
| **3** | Subword | 0.8530 | 1.806 | 3.93 | 29,621 | 14.7% |
| **4** | Word | 0.0663 🏆 | 1.047 | 1.11 | 415,146 | 93.4% |
| **4** | Subword | 0.5923 | 1.508 | 2.47 | 116,449 | 40.8% |

### Generated Text Samples (Word-based)

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

**Context Size 1:**

1. `la kolesov gee malum dugelegɔ lɔgerɔ ba yi a gce o loe e la za a`
2. `a characteristically thick dough covered by yaba badoe about alex segbefia 16 years 2 form world`
3. `ti fu san bɔna tiŋsuka se sɛba iŋa n me bɔ ɔra roads and former swansea`

**Context Size 2:**

1. `la puan indihiang tiŋa tasikmalaya tiŋa la puan la a yuuma la wa tiŋa a kiŋɛ a`
2. `de la se em n yuum de la são francisco xavier ti ŋwana wa yuum pa ase`
3. `ti ba yi ira b a economic la pɔlitisi nanana wa a kiŋɛ a sukuu katɛ de`

**Context Size 3:**

1. `yuunɛ la puan bawumia yuum niɛ la dr matthew opoku prempeh ba yuun dugɛ e la yuunɛ la`
2. `zo e zo e n de sorts of amulets tigera wa n de mina a wan ta am`
3. `e zo e n nyaa boi ti nɛrawoo yuun mina ti a dena se em la dɔla de`

**Context Size 4:**

1. `zo e zo e daa ka tari tuuma nya daa eŋɛ ba puti ira ti koloni zuoduma la daa`
2. `ti ba yi ira ti tyre fitting la cold calling la tuuma bɔna ford dagenham a kelum yuum tum`
3. `yuunɛ la puan a le to e sɛtifiketi bɔna koosego la ligeri yɛla washington yunivɛsiti of world bank m`


### Generated Text Samples (Subword-based)

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

**Context Size 1:**

1. `_a_iela_b_talena`
2. `arseryɛra_laa_d_`
3. `era_n_hrɛ_ss"_n,`

**Context Size 2:**

1. `a_yuum_._ti_sɛ_we`
2. `la_zo'ela_buum_la`
3. `_lɔgembese’eloobi`

**Context Size 3:**

1. `la_a_yuum_toni_la,`
2. `_la_la_pa'am_tiŋa_`
3. `_til_of_ghama_at_t`

**Context Size 4:**

1. `_la_puan,_ba_kɔm_ba`
2. `_yuuni_yuum_ta_paat`
3. `_ti_ba_gee_"efua_tu`


### Key Findings

- **Best Predictability:** Context-4 (word) with 93.4% predictability
- **Branching Factor:** Decreases with context size (more deterministic)
- **Memory Trade-off:** Larger contexts require more storage (116,449 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 | 15,750 |
| Total Tokens | 531,469 |
| Mean Frequency | 33.74 |
| Median Frequency | 4 |
| Frequency Std Dev | 489.14 |

### Most Common Words

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | la | 45,893 |
| 2 | a | 16,970 |
| 3 | ti | 15,755 |
| 4 | n | 14,415 |
| 5 | ba | 12,540 |
| 6 | de | 11,579 |
| 7 | puan | 11,117 |
| 8 | yuum | 7,135 |
| 9 | e | 6,343 |
| 10 | wa | 5,603 |

### Least Common Words (from vocabulary)

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | jurgen | 2 |
| 2 | martini | 2 |
| 3 | mcmullan | 2 |
| 4 | penina | 2 |
| 5 | mlama | 2 |
| 6 | richards | 2 |
| 7 | amowi | 2 |
| 8 | rotimi | 2 |
| 9 | watts | 2 |
| 10 | windley | 2 |

### Zipf's Law Analysis

| Metric | Value |
|--------|-------|
| Zipf Coefficient | 1.2037 |
| R² (Goodness of Fit) | 0.996962 |
| Adherence Quality | **excellent** |

### Coverage Analysis

| Top N Words | Coverage |
|-------------|----------|
| Top 100 | 57.7% |
| Top 1,000 | 82.5% |
| Top 5,000 | 93.9% |
| Top 10,000 | 97.7% |

### Key Findings

- **Zipf Compliance:** R²=0.9970 indicates excellent adherence to Zipf's law
- **High Frequency Dominance:** Top 100 words cover 57.7% of corpus
- **Long Tail:** 5,750 words needed for remaining 2.3% 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 | 0.7704 🏆 | 0.3622 | N/A | N/A |
| **mono_64d** | 64 | 0.5062 | 0.3302 | N/A | N/A |
| **mono_128d** | 128 | 0.1445 | 0.3114 | N/A | N/A |
| **aligned_32d** | 32 | 0.7704 | 0.3520 | 0.0340 | 0.1900 |
| **aligned_64d** | 64 | 0.5062 | 0.3219 | 0.0640 | 0.3020 |
| **aligned_128d** | 128 | 0.1445 | 0.3190 | 0.1120 | 0.3520 |

### Key Findings

- **Best Isotropy:** mono_32d with 0.7704 (more uniform distribution)
- **Semantic Density:** Average pairwise similarity of 0.3328. Lower values indicate better semantic separation.
- **Alignment Quality:** Aligned models achieve up to 11.2% 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 | **-0.314** | Low formulaic content | - |

### 6.2 Affix Inventory (Productive Units)

These are the most productive prefixes and suffixes identified by sampling the vocabulary for global substitutability patterns. A unit is considered an affix if stripping it leaves a valid stem that appears in other contexts.

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

#### Productive Suffixes
| Suffix | Examples |
|--------|----------|
| `-a` | solemitiŋa, nangooma, bawadua |

### 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 |
|------|----------|------------------|----------|
| `gera` | 1.96x | 37 contexts | ɛgera, ãgera, ugera |
| `ɔger` | 1.60x | 30 contexts | bɔgerɛ, tɔgera, yɔgera |
| `iger` | 1.64x | 25 contexts | niger, digeri, tigera |
| `atio` | 1.94x | 14 contexts | nation, nations, station |
| `rega` | 1.64x | 22 contexts | ɛrega, ãarega, tɛrega |
| `elum` | 1.81x | 15 contexts | belum, celum, kelum |
| `tion` | 1.85x | 13 contexts | action, option, nation |
| `segɔ` | 1.67x | 16 contexts | osegɔ, isegɔ, ɔsegɔ |
| `reba` | 1.62x | 17 contexts | ireba, ɛreba, areba |
| `gerɔ` | 2.03x | 9 contexts | sɔgerɔ, logerɔ, pɔgerɔ |
| `ɛger` | 1.54x | 17 contexts | ɛgera, pɛgerɛ, sɛgerɛ |
| `aana` | 1.73x | 12 contexts | gaana, paana, baana |

### 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`).

*Insufficient data for recursive segmentation.*


### 6.6 Linguistic Interpretation

> **Automated Insight:**
The language Frafra 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 (4.00x) |
| N-gram | **2-gram** | Lowest perplexity (241) |
| Markov | **Context-4** | Highest predictability (93.4%) |
| 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: 2026-01-10 00:37:19*