README.md

---
language: hak
language_name: Hakka Chinese
language_family: sinitic_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-sinitic_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: 2.827
  - name: best_isotropy
    type: isotropy
    value: 0.8359
  - name: vocabulary_size
    type: vocab
    value: 0
generated: 2026-01-10
---

# Hakka Chinese - Wikilangs Models
## Comprehensive Research Report & Full Ablation Study

This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Hakka Chinese** 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 |
|------------|-------------|---------------|----------|--------------|
| **32k** | 2.723x | 2.73 | 0.0000% | 159,401 |
| **64k** | 2.827x 🏆 | 2.83 | 0.0000% | 153,537 |

### Tokenization Examples

Below are sample sentences tokenized with each vocabulary size:

**Sample 1:** `Theodore Roosevelt () he Mî-koet ke thi 26-ngim chúng-thúng, chṳ chhai-ngim. chú...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 32k | `▁theod ore ▁ro ose vel t ▁() ▁he ▁mî - ... (+28 more)` | 38 |
| 64k | `▁theodore ▁roosevelt ▁() ▁he ▁mî - koet ▁ke ▁thi ▁ ... (+24 more)` | 34 |

**Sample 2:** `Kí-hoi he kôn-chṳ̂ ke thi 36 chak, chhai kâng-chṳ́ ke thèu-chhièn lâu vú-sut ke ...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 32k | `▁kí - hoi ▁he ▁kôn - chṳ̂ ▁ke ▁thi ▁ ... (+21 more)` | 31 |
| 64k | `▁kí - hoi ▁he ▁kôn - chṳ̂ ▁ke ▁thi ▁ ... (+21 more)` | 31 |

**Sample 3:** `Ngi-yông-fa-than (二氧化碳) he khûng-hi lî-tú ke yit chúng hi-thí, fa-ho̍k-sit he CO...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 32k | `▁ngi - yông - fa - than ▁( 二 氧 ... (+26 more)` | 36 |
| 64k | `▁ngi - yông - fa - than ▁( 二氧化碳 ) ... (+23 more)` | 33 |


### Key Findings

- **Best Compression:** 64k achieves 2.827x compression
- **Lowest UNK Rate:** 32k with 0.0000% 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 | 2,928 | 11.52 | 12,963 | 30.2% | 63.0% |
| **2-gram** | Subword | 299 🏆 | 8.22 | 5,288 | 67.0% | 97.8% |
| **3-gram** | Word | 3,725 | 11.86 | 19,089 | 27.9% | 60.7% |
| **3-gram** | Subword | 1,606 | 10.65 | 19,348 | 33.3% | 78.9% |
| **4-gram** | Word | 4,712 | 12.20 | 29,249 | 25.7% | 59.3% |
| **4-gram** | Subword | 5,871 | 12.52 | 69,776 | 20.0% | 56.6% |
| **5-gram** | Word | 3,701 | 11.85 | 22,039 | 25.8% | 63.5% |
| **5-gram** | Subword | 13,255 | 13.69 | 118,405 | 14.2% | 43.7% |

### Top 5 N-grams by Size

**2-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `ngìn khiéu` | 4,633 |
| 2 | `li̍t sṳ́` | 3,847 |
| 3 | `ke yit` | 3,582 |
| 4 | `thi lî` | 3,401 |
| 5 | `sṳ́ thi` | 2,991 |

**3-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `li̍t sṳ́ thi` | 2,989 |
| 2 | `sṳ́ thi lî` | 2,986 |
| 3 | `ngoi phu lièn` | 2,672 |
| 4 | `phu lièn kiet` | 2,229 |
| 5 | `hàng chṳn khî` | 2,051 |

**4-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `li̍t sṳ́ thi lî` | 2,985 |
| 2 | `ngoi phu lièn kiet` | 2,228 |
| 3 | `hàng chṳn khî va̍k` | 1,813 |
| 4 | `phìn fông kûng lî` | 1,797 |
| 5 | `kháu vùn hien ngoi` | 1,788 |

**5-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `vùn hien ngoi phu lièn` | 1,787 |
| 2 | `kháu vùn hien ngoi phu` | 1,787 |
| 3 | `hien ngoi phu lièn kiet` | 1,716 |
| 4 | `li̍t sṳ́ thi lî hì` | 1,440 |
| 5 | `sṳ́ thi lî hì hèu` | 1,440 |

**2-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `n g` | 101,597 |
| 2 | `c h` | 73,339 |
| 3 | `_ k` | 56,623 |
| 4 | `n -` | 53,747 |
| 5 | `_ c` | 43,912 |

**3-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ c h` | 41,840 |
| 2 | `n g -` | 33,183 |
| 3 | `- c h` | 29,450 |
| 4 | `c h h` | 27,982 |
| 5 | `n g _` | 25,297 |

**4-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ c h h` | 17,463 |
| 2 | `_ k e _` | 17,195 |
| 3 | `- n g i` | 11,569 |
| 4 | `û n g -` | 10,810 |
| 5 | `_ h e _` | 10,165 |

**5-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ n g ì n` | 7,584 |
| 2 | `n g i è n` | 6,756 |
| 3 | `- k o e t` | 6,022 |
| 4 | `n g ì n -` | 5,605 |
| 5 | `_ t h i -` | 5,586 |


### Key Findings

- **Best Perplexity:** 2-gram (subword) with 299
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
- **Coverage:** Top-1000 patterns cover ~44% 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 | 0.5070 | 1.421 | 4.84 | 32,806 | 49.3% |
| **1** | Subword | 0.3333 | 1.260 | 2.63 | 26,193 | 66.7% |
| **2** | Word | 0.3144 | 1.243 | 1.80 | 157,697 | 68.6% |
| **2** | Subword | 0.2363 | 1.178 | 1.75 | 68,456 | 76.4% |
| **3** | Word | 0.1163 | 1.084 | 1.22 | 281,462 | 88.4% |
| **3** | Subword | 0.2644 | 1.201 | 1.82 | 119,003 | 73.6% |
| **4** | Word | 0.0498 🏆 | 1.035 | 1.08 | 339,015 | 95.0% |
| **4** | Subword | 0.3010 | 1.232 | 1.70 | 216,161 | 69.9% |

### Generated Text Samples (Word-based)

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

**Context Size 1:**

1. `ke yit têu ke ngìn ya he thòi vân thòi vân ngiùn hòng khûng kûng lî`
2. `he chûng koet si chhôn thai khûng thiet lu khiéu yok 4 ngie̍t 6 170 phìn`
3. `sṳ he hk diamond hill lâu au chû piên sṳ kón lî tú phài miàng thòng`

**Context Size 2:**

1. `ngìn khiéu yok thúng kie mien chit he chṳ́ chit chhûi thung vu̍t 動物 45 fish ǹg`
2. `li̍t sṳ́ thi lî hàng kín tiám chiá moi sàng sṳ yû assisi città di castello foligno`
3. `ke yit chak khiun chúng mien chit 89 44 phìn fông kûng lî ngìn khiéu he 644`

**Context Size 3:**

1. `li̍t sṳ́ thi lî ngìn khiéu 15 van ngìn khiéu me̍t thu mî chak phìn fông kûng lî`
2. `sṳ́ thi lî vùn fa kau yuk tshâm kháu vùn hien ngoi phu lièn kiet khi̍p thi khî`
3. `ngoi phu lièn kiet kâm suk tsṳn fú mióng lu`

**Context Size 4:**

1. `li̍t sṳ́ thi lî vùn fa kau yuk tshâm kháu vùn hien ngoi phu lièn kiet tho`
2. `phìn fông kûng lî ngìn khiéu li̍t sṳ́ thi lî kîn chi ngìn khiéu hàng chṳn khî va̍k ngìn`
3. `kháu vùn hien ngoi phu lièn kiet ngî chhu̍k khiung fò koet`


### Generated Text Samples (Subword-based)

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

**Context Size 1:**

1. `_sén_sṳ._sàngt-k`
2. `-n-pángîmìng-vùn`
3. `ngì-lî_ye_ongî_t`

**Context Size 2:**

1. `ngièn-sṳ̂nh_ngà_ke`
2. `chiên-khi_ng_mìn-`
3. `_ko_piân-khî_ho̍k_`

**Context Size 3:**

1. `_chak_vu̍t_sông_lâu`
2. `ng-thai-lièn-kiên-`
3. `-chhiung-lî_hì-tho`

**Context Size 4:**

1. `_chhai_sàng-chû_kaz`
2. `_ke_pu-nùng-sṳ_kâng`
3. `-ngim._chhṳ_yîn-kon`


### Key Findings

- **Best Predictability:** Context-4 (word) with 95.0% predictability
- **Branching Factor:** Decreases with context size (more deterministic)
- **Memory Trade-off:** Larger contexts require more storage (216,161 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 | 9,572 |
| Total Tokens | 587,243 |
| Mean Frequency | 61.35 |
| Median Frequency | 3 |
| Frequency Std Dev | 439.55 |

### Most Common Words

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | ke | 19,829 |
| 2 | he | 11,897 |
| 3 | sṳ | 11,428 |
| 4 | ngìn | 9,264 |
| 5 | lî | 8,292 |
| 6 | koet | 7,837 |
| 7 | yit | 7,446 |
| 8 | thi | 7,274 |
| 9 | khî | 6,944 |
| 10 | ngièn | 6,742 |

### Least Common Words (from vocabulary)

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | chài | 2 |
| 2 | then_séu | 2 |
| 3 | cṳ̀n | 2 |
| 4 | fta | 2 |
| 5 | gaya | 2 |
| 6 | 한국 | 2 |
| 7 | 신화 | 2 |
| 8 | kbo | 2 |
| 9 | 누리호 | 2 |
| 10 | rocket | 2 |

### Zipf's Law Analysis

| Metric | Value |
|--------|-------|
| Zipf Coefficient | 1.4428 |
| R² (Goodness of Fit) | 0.978164 |
| Adherence Quality | **excellent** |

### Coverage Analysis

| Top N Words | Coverage |
|-------------|----------|
| Top 100 | 55.1% |
| Top 1,000 | 92.0% |
| Top 5,000 | 98.3% |
| Top 10,000 | 0.0% |

### Key Findings

- **Zipf Compliance:** R²=0.9782 indicates excellent adherence to Zipf's law
- **High Frequency Dominance:** Top 100 words cover 55.1% of corpus
- **Long Tail:** -428 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 | 0.8359 🏆 | 0.3600 | N/A | N/A |
| **mono_64d** | 64 | 0.3973 | 0.3173 | N/A | N/A |
| **mono_128d** | 128 | 0.0725 | 0.3112 | N/A | N/A |
| **aligned_32d** | 32 | 0.8359 | 0.3657 | 0.0200 | 0.1480 |
| **aligned_64d** | 64 | 0.3973 | 0.3155 | 0.0440 | 0.2960 |
| **aligned_128d** | 128 | 0.0725 | 0.3193 | 0.0980 | 0.3700 |

### Key Findings

- **Best Isotropy:** mono_32d with 0.8359 (more uniform distribution)
- **Semantic Density:** Average pairwise similarity of 0.3315. Lower values indicate better semantic separation.
- **Alignment Quality:** Aligned models achieve up to 9.8% 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 | **0.453** | 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.

*No productive affixes detected.*


### 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 |
|------|----------|------------------|----------|
| `ióng` | 2.10x | 9 contexts | lióng, hióng, sióng |

### 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 Hakka Chinese 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 | **64k BPE** | Best compression (2.83x) |
| N-gram | **2-gram** | Lowest perplexity (299) |
| Markov | **Context-4** | Highest predictability (95.0%) |
| 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 02:10:12*