Upload all models and assets for ng (latest)

.gitattributes

@@ -33,3 +33,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+visualizations/embedding_similarity.png filter=lfs diff=lfs merge=lfs -text
+visualizations/embedding_tsne_multilingual.png filter=lfs diff=lfs merge=lfs -text
+visualizations/performance_dashboard.png filter=lfs diff=lfs merge=lfs -text
+visualizations/position_encoding_comparison.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,698 @@
+---
+language: ng
+language_name: Ndonga
+language_family: bantu_central
+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-bantu_central
+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: 2.981
+  - name: best_isotropy
+    type: isotropy
+    value: 0.0034
+  - name: vocabulary_size
+    type: vocab
+    value: 0
+generated: 2026-01-10
+---
+
+# Ndonga - Wikilangs Models
+## Comprehensive Research Report & Full Ablation Study
+
+This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Ndonga** 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** | 2.981x 🏆 | 2.98 | 1.0627% | 13,080 |
+
+### Tokenization Examples
+
+Below are sample sentences tokenized with each vocabulary size:
+
+
+### Key Findings
+
+- **Best Compression:** 8k achieves 2.981x compression
+- **Lowest UNK Rate:** 8k with 1.0627% 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 | 17 | 4.12 | 22 | 100.0% | 100.0% |
+| **2-gram** | Subword | 286 | 8.16 | 589 | 60.1% | 100.0% |
+| **3-gram** | Word | 13 | 3.74 | 23 | 100.0% | 100.0% |
+| **3-gram** | Subword | 1,258 | 10.30 | 2,328 | 29.4% | 80.5% |
+| **4-gram** | Word | 16 | 4.02 | 29 | 100.0% | 100.0% |
+| **4-gram** | Subword | 2,459 | 11.26 | 4,677 | 22.5% | 61.8% |
+| **5-gram** | Word | 9 🏆 | 3.17 | 15 | 100.0% | 100.0% |
+| **5-gram** | Subword | 2,457 | 11.26 | 4,586 | 24.2% | 59.4% |
+
+### Top 5 N-grams by Size
+
+**2-grams (Word):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `nowy dwór` | 35 |
+| 2 | `dwór królewski` | 35 |
+| 3 | `na uuthemba` | 31 |
+| 4 | `omuntu kehe` | 29 |
+| 5 | `oku na` | 29 |
+
+**3-grams (Word):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `nowy dwór królewski` | 35 |
+| 2 | `omuntu kehe oku` | 27 |
+| 3 | `kehe oku na` | 27 |
+| 4 | `oku na uuthemba` | 26 |
+| 5 | `zh min nan` | 12 |
+
+**4-grams (Word):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `omuntu kehe oku na` | 27 |
+| 2 | `kehe oku na uuthemba` | 24 |
+| 3 | `nekofungama ar sefala angubo` | 3 |
+| 4 | `harranga nekofungama ar sefala` | 3 |
+| 5 | `ast harranga nekofungama ar` | 3 |
+
+**5-grams (Word):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `omuntu kehe oku na uuthemba` | 24 |
+| 2 | `harranga nekofungama ar sefala angubo` | 3 |
+| 3 | `ast harranga nekofungama ar sefala` | 3 |
+| 4 | `nekofungama ar sefala angubo andusat` | 3 |
+| 5 | `kape na nando omuntu e` | 3 |
+
+**2-grams (Subword):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `a _` | 1,406 |
+| 2 | `a n` | 583 |
+| 3 | `e _` | 427 |
+| 4 | `n g` | 419 |
+| 5 | `e n` | 411 |
+
+**3-grams (Subword):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `i a _` | 277 |
+| 2 | `n a _` | 275 |
+| 3 | `e r s` | 197 |
+| 4 | `e n _` | 193 |
+| 5 | `t e r` | 177 |
+
+**4-grams (Subword):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `e r s e` | 175 |
+| 2 | `t e r s` | 169 |
+| 3 | `r s e n` | 169 |
+| 4 | `e t e r` | 169 |
+| 5 | `u e t e` | 168 |
+
+**5-grams (Subword):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `e r s e n` | 169 |
+| 2 | `t e r s e` | 169 |
+| 3 | `u e t e r` | 168 |
+| 4 | `e t e r s` | 168 |
+| 5 | `r s e n _` | 167 |
+
+
+### Key Findings
+
+- **Best Perplexity:** 5-gram (word) with 9
+- **Entropy Trend:** Decreases with larger n-grams (more predictable)
+- **Coverage:** Top-1000 patterns cover ~59% 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 | 0.4936 | 1.408 | 2.30 | 2,515 | 50.6% |
+| **1** | Subword | 0.5935 | 1.509 | 3.07 | 1,104 | 40.6% |
+| **2** | Word | 0.0333 | 1.023 | 1.06 | 5,756 | 96.7% |
+| **2** | Subword | 0.4561 | 1.372 | 2.43 | 3,389 | 54.4% |
+| **3** | Word | 0.0092 | 1.006 | 1.02 | 6,060 | 99.1% |
+| **3** | Subword | 0.4100 | 1.329 | 1.84 | 8,218 | 59.0% |
+| **4** | Word | 0.0036 🏆 | 1.002 | 1.01 | 6,160 | 99.6% |
+| **4** | Subword | 0.2372 | 1.179 | 1.40 | 15,074 | 76.3% |
+
+### Generated Text Samples (Word-based)
+
+Below are text samples generated from each word-based Markov chain model:
+
+**Context Size 1:**
+
+1. `uetersen nds asien li azië nn geografi sw jamhuri ya uvuneka kutya otashi gandja uuthemba wokugamenw...`
+2. `wikipedia id turki sq uetersen tl turkiya crh asiya hak asasi manusia io kulturo es uetersen`
+3. `na nando omuntu kehe ngoka ha baibûl hak ngùi kî pak khô haw ākia he אנגלית`
+
+**Context Size 2:**
+
+1. `nowy dwór królewski tr nowy dwór królewski en nowy dwór królewski nn nowy dwór królewski pt nowy`
+2. `dwór królewski en nowy dwór królewski nn nowy dwór królewski en nowy dwór królewski et nowy dwór`
+3. `na uuthemba womuthika omwaanawa gwonkalamwenyo memanguluko iya andjagana uuna iilyo yiilongo ya uvun...`
+
+**Context Size 3:**
+
+1. `nowy dwór królewski ff nowy dwór królewski tum nowy dwór królewski pl nowy dwór królewski de nowy dw...`
+2. `kehe oku na uuthemba womuthika omwaanawa gwonkalamwenyo gwa yeleka uukolele nonkalo ombwanawa ye mwe...`
+3. `omuntu kehe oku na uuthemba welandulathano iyopankalathano nolyomuuyuni moka uuthemba nemanguluko nd...`
+
+**Context Size 4:**
+
+1. `omuntu kehe oku na uuthemba wokutota nokuninga oshilyo shehangano iyaaniilonga opo a gamene uuwanawa...`
+2. `kehe oku na uuthemba womafutilo ngele okwa kulupa nenge a mona oshiponga moshilongo she nenge paigwa...`
+3. `ar sefala angubo andusat ace bahsa inggréh af engels ak english als englische sprache am እንግሊዝኛ an i...`
+
+
+### Generated Text Samples (Subword-based)
+
+Below are text samples generated from each subword-based Markov chain model:
+
+**Context Size 1:**
+
+1. `_inoghe_अधिकारों_sk:`
+2. `a:tesen_ndulidur`
+3. `entueneburs'at_b`
+
+**Context Size 2:**
+
+1. `a_a_vica_op_an_uu`
+2. `an_she:ויקיפדיה_l`
+3. `e_papublisencia_k`
+
+**Context Size 3:**
+
+1. `ia_bm:hadan_mwl:bi`
+2. `na_nga_nomakwa_uvu`
+3. `ersen_wu_li:una_oy`
+
+**Context Size 4:**
+
+1. `ersen_wuukwa,_a_kut`
+2. `etersen_su:wikipiki`
+3. `tersele_nokoompumbi`
+
+
+### Key Findings
+
+- **Best Predictability:** Context-4 (word) with 99.6% predictability
+- **Branching Factor:** Decreases with context size (more deterministic)
+- **Memory Trade-off:** Larger contexts require more storage (15,074 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 | 648 |
+| Total Tokens | 4,436 |
+| Mean Frequency | 6.85 |
+| Median Frequency | 4 |
+| Frequency Std Dev | 10.46 |
+
+### Most Common Words
+
+| Rank | Word | Frequency |
+|------|------|-----------|
+| 1 | uetersen | 168 |
+| 2 | wikipedia | 87 |
+| 3 | na | 78 |
+| 4 | ghana | 71 |
+| 5 | uuthemba | 50 |
+| 6 | asia | 49 |
+| 7 | pigazzano | 47 |
+| 8 | zh | 44 |
+| 9 | de | 42 |
+| 10 | kehe | 37 |
+
+### Least Common Words (from vocabulary)
+
+| Rank | Word | Frequency |
+|------|------|-----------|
+| 1 | turecko | 2 |
+| 2 | τουρκία | 2 |
+| 3 | tuirc | 2 |
+| 4 | तुर्किये | 2 |
+| 5 | germany | 2 |
+| 6 | ঘানা | 2 |
+| 7 | thumb | 2 |
+| 8 | italy | 2 |
+| 9 | piasensa | 2 |
+| 10 | двор | 2 |
+
+### Zipf's Law Analysis
+
+| Metric | Value |
+|--------|-------|
+| Zipf Coefficient | 0.8074 |
+| R² (Goodness of Fit) | 0.939699 |
+| Adherence Quality | **excellent** |
+
+### Coverage Analysis
+
+| Top N Words | Coverage |
+|-------------|----------|
+| Top 100 | 49.3% |
+| Top 1,000 | 0.0% |
+| Top 5,000 | 0.0% |
+| Top 10,000 | 0.0% |
+
+### Key Findings
+
+- **Zipf Compliance:** R²=0.9397 indicates excellent adherence to Zipf's law
+- **High Frequency Dominance:** Top 100 words cover 49.3% of corpus
+- **Long Tail:** -9,352 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
+
+![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 | 0.0034 🏆 | 0.0000 | N/A | N/A |
+| **mono_64d** | 64 | 0.0001 | 0.0000 | N/A | N/A |
+| **mono_128d** | 128 | 0.0000 | 0.0000 | N/A | N/A |
+| **aligned_32d** | 32 | 0.0034 | 0.0000 | 0.0000 | 0.0000 |
+| **aligned_64d** | 64 | 0.0001 | 0.0000 | 0.0000 | 0.0000 |
+| **aligned_128d** | 128 | 0.0000 | 0.0000 | 0.0000 | 0.0000 |
+
+### Key Findings
+
+- **Best Isotropy:** mono_32d with 0.0034 (more uniform distribution)
+- **Semantic Density:** Average pairwise similarity of 0.0000. Lower values indicate better semantic separation.
+- **Alignment Quality:** Aligned models evaluated but achieved 0% recall.
+- **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 | **1.124** | High morphological productivity | Reliable analysis |
+| Idiomaticity Gap | **0.683** | High formulaic/idiomatic content | - |
+
+### 6.2 Affix Inventory (Productive Units)
+
+These are the most productive prefixes and suffixes identified by sampling the vocabulary for global substitutability patterns. A unit is considered an affix if stripping it leaves a valid stem that appears in other contexts.
+
+#### Productive Prefixes
+| Prefix | Examples |
+|--------|----------|
+
+#### Productive Suffixes
+| Suffix | Examples |
+|--------|----------|
+| `-a` | mpoka, ehyia, kaa |
+
+### 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.
+
+*No significant affix co-occurrences detected.*
+
+
+### 6.5 Recursive Morpheme Segmentation
+
+Using **Recursive Hierarchical Substitutability**, we decompose complex words into their constituent morphemes. This approach handles nested affixes (e.g., `prefix-prefix-root-suffix`).
+
+| Word | Suggested Split | Confidence | Stem |
+|------|-----------------|------------|------|
+| universala | **`universal-a`** | 4.5 | `universal` |
+| geografia | **`geografi-a`** | 4.5 | `geografi` |
+| republika | **`republik-a`** | 4.5 | `republik` |
+| kwatelela | **`kwatelel-a`** | 1.5 | `kwatelel` |
+| manguluka | **`manguluk-a`** | 1.5 | `manguluk` |
+| wikipedya | **`wikipedy-a`** | 1.5 | `wikipedy` |
+| geographia | **`geographi-a`** | 1.5 | `geographi` |
+
+### 6.6 Linguistic Interpretation
+
+> **Automated Insight:**
+The language Ndonga shows moderate morphological complexity. There is a balanced trade-off between whole-word memorization and subword composition.
+
+> **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 | **8k BPE** | Best compression (2.98x) |
+| N-gram | **5-gram** | Lowest perplexity (9) |
+| Markov | **Context-4** | Highest predictability (99.6%) |
+| 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 14:50:35*

models/embeddings/aligned/ng_128d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34bf2a399fd128262456f17fe7df20a1e629cc9c5143947550b6d3d9cd66c209
+size 1024056189

models/embeddings/aligned/ng_128d.meta.json

@@ -0,0 +1 @@
+{"lang": "ng", "dim": 128, "max_seq_len": 512, "is_aligned": true}

models/embeddings/aligned/ng_128d.projection.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e051c021f38d27c84ab4d6df44c89e9fbf3706c78ffd6fdf9b8d3763c34e7c5
+size 65664

models/embeddings/aligned/ng_128d_metadata.json

@@ -0,0 +1,8 @@
+{
+  "language": "ng",
+  "dimension": 128,
+  "version": "aligned",
+  "hub_language": "en",
+  "seed_vocab_size": 28,
+  "vocab_size": 54
+}

models/embeddings/aligned/ng_32d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa58d081dda754238780d73dc92ccd9120de5c25f3e7700818762f81f1912421
+size 256014717

models/embeddings/aligned/ng_32d.meta.json

@@ -0,0 +1 @@
+{"lang": "ng", "dim": 32, "max_seq_len": 512, "is_aligned": true}

models/embeddings/aligned/ng_32d.projection.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9cf7f47d8f65cbe8641d49e2139bec4461ffe42210102fe78b31e529c0f04a39
+size 4224

models/embeddings/aligned/ng_32d_metadata.json

@@ -0,0 +1,8 @@
+{
+  "language": "ng",
+  "dimension": 32,
+  "version": "aligned",
+  "hub_language": "en",
+  "seed_vocab_size": 28,
+  "vocab_size": 54
+}

models/embeddings/aligned/ng_64d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9408eae8053e83895873acb3dc744c69420075e2bdbd1f96b4425f7b1ad9e5f1
+size 512028541

models/embeddings/aligned/ng_64d.meta.json

@@ -0,0 +1 @@
+{"lang": "ng", "dim": 64, "max_seq_len": 512, "is_aligned": true}

models/embeddings/aligned/ng_64d.projection.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a53262ac6a66e2ba0fb6f696fa834f7c1926c9d514fbceeb93869590b4c9efa4
+size 16512

models/embeddings/aligned/ng_64d_metadata.json

@@ -0,0 +1,8 @@
+{
+  "language": "ng",
+  "dimension": 64,
+  "version": "aligned",
+  "hub_language": "en",
+  "seed_vocab_size": 28,
+  "vocab_size": 54
+}

models/embeddings/monolingual/ng_128d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34bf2a399fd128262456f17fe7df20a1e629cc9c5143947550b6d3d9cd66c209
+size 1024056189

models/embeddings/monolingual/ng_128d.meta.json

@@ -0,0 +1 @@
+{"lang": "ng", "dim": 128, "max_seq_len": 512, "is_aligned": false}

models/embeddings/monolingual/ng_128d_metadata.json

@@ -0,0 +1,16 @@
+{
+  "language": "ng",
+  "dimension": 128,
+  "version": "monolingual",
+  "training_params": {
+    "algorithm": "skipgram",
+    "min_count": 5,
+    "window": 5,
+    "negative": 5,
+    "epochs": 5,
+    "encoding_method": "rope",
+    "dim": 128,
+    "threads": 5
+  },
+  "vocab_size": 54
+}

models/embeddings/monolingual/ng_32d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa58d081dda754238780d73dc92ccd9120de5c25f3e7700818762f81f1912421
+size 256014717

models/embeddings/monolingual/ng_32d.meta.json

@@ -0,0 +1 @@
+{"lang": "ng", "dim": 32, "max_seq_len": 512, "is_aligned": false}

models/embeddings/monolingual/ng_32d_metadata.json

@@ -0,0 +1,16 @@
+{
+  "language": "ng",
+  "dimension": 32,
+  "version": "monolingual",
+  "training_params": {
+    "algorithm": "skipgram",
+    "min_count": 5,
+    "window": 5,
+    "negative": 5,
+    "epochs": 5,
+    "encoding_method": "rope",
+    "dim": 32,
+    "threads": 5
+  },
+  "vocab_size": 54
+}

models/embeddings/monolingual/ng_64d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9408eae8053e83895873acb3dc744c69420075e2bdbd1f96b4425f7b1ad9e5f1
+size 512028541

models/embeddings/monolingual/ng_64d.meta.json

@@ -0,0 +1 @@
+{"lang": "ng", "dim": 64, "max_seq_len": 512, "is_aligned": false}

models/embeddings/monolingual/ng_64d_metadata.json

@@ -0,0 +1,16 @@
+{
+  "language": "ng",
+  "dimension": 64,
+  "version": "monolingual",
+  "training_params": {
+    "algorithm": "skipgram",
+    "min_count": 5,
+    "window": 5,
+    "negative": 5,
+    "epochs": 5,
+    "encoding_method": "rope",
+    "dim": 64,
+    "threads": 5
+  },
+  "vocab_size": 54
+}

models/subword_markov/ng_markov_ctx1_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:297baf033d4f418727d23b17b2108aba4409f4ea17bae25bd29f1580df216092
+size 33198

models/subword_markov/ng_markov_ctx1_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 1,
+  "variant": "subword",
+  "language": "ng",
+  "unique_contexts": 1104,
+  "total_transitions": 38341
+}

models/subword_markov/ng_markov_ctx2_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f30b0fa204f3f0a19f033013fa4c7884427471a22cbab856e5e408f33f630da6
+size 75678

models/subword_markov/ng_markov_ctx2_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 2,
+  "variant": "subword",
+  "language": "ng",
+  "unique_contexts": 3389,
+  "total_transitions": 38324
+}

models/subword_markov/ng_markov_ctx3_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47216f438bc6170fc0f32910e081ffaf456b5de4c17ffca2533aa3520fd1facb
+size 144981

models/subword_markov/ng_markov_ctx3_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 3,
+  "variant": "subword",
+  "language": "ng",
+  "unique_contexts": 8218,
+  "total_transitions": 38307
+}

models/subword_markov/ng_markov_ctx4_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb3e5cb24aabaaacc0f8e8680e997948f9312aba47d7375aee3edf8ca4e07335
+size 229202

models/subword_markov/ng_markov_ctx4_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 4,
+  "variant": "subword",
+  "language": "ng",
+  "unique_contexts": 15074,
+  "total_transitions": 38290
+}

models/subword_ngram/ng_2gram_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3799c029a3bf65b745ec32c27679ee1a5d3843f53b46739f065ec13d802b1729
+size 7920

models/subword_ngram/ng_2gram_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "n": 2,
+  "variant": "subword",
+  "language": "ng",
+  "unique_ngrams": 589,
+  "total_ngrams": 38341
+}

models/subword_ngram/ng_3gram_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f090be70859a8466b7bd471c964709db1599df83218c6edddd011139c6ef36f2
+size 24223

models/subword_ngram/ng_3gram_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "n": 3,
+  "variant": "subword",
+  "language": "ng",
+  "unique_ngrams": 2328,
+  "total_ngrams": 38324
+}

models/subword_ngram/ng_4gram_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c1a682779981406d6788f56d88cae10731da840fbfb06ca0848e91d204759e1
+size 52059

models/subword_ngram/ng_4gram_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "n": 4,
+  "variant": "subword",
+  "language": "ng",
+  "unique_ngrams": 4677,
+  "total_ngrams": 38307
+}

models/subword_ngram/ng_5gram_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:470c9a91a994623fba445802bfc4106d6dcefe3031163d2744104d5977372b72
+size 54535

models/subword_ngram/ng_5gram_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "n": 5,
+  "variant": "subword",
+  "language": "ng",
+  "unique_ngrams": 4586,
+  "total_ngrams": 38290
+}

models/tokenizer/ng_tokenizer_8k.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:004c7de7524ff61123c7c67c10ee7f6aa5781b98f2e0b1a5096bd56e82ddbb50
+size 367292

models/vocabulary/ng_vocabulary.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:583c51c99cbd4e09740abcdcc5df7576ce2cc8f7d16aea843b7775033cfdee22
+size 12196

models/vocabulary/ng_vocabulary_metadata.json

@@ -0,0 +1,15 @@
+{
+  "language": "ng",
+  "vocabulary_size": 648,
+  "variant": "full",
+  "statistics": {
+    "type_token_ratio": 0.39977816510854064,
+    "coverage": {
+      "top_100": 0.3463793376643955,
+      "top_1000": 0.7586753287909999
+    },
+    "hapax_count": 1875,
+    "hapax_ratio": 0.7431629013079667,
+    "total_documents": 17
+  }
+}

models/word_markov/ng_markov_ctx1_word.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4976d02f4a3b7f9cc67d5838410cd03eccc8de754cbeed9f8e625b7ca718171
+size 72378

models/word_markov/ng_markov_ctx1_word_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 1,
+  "variant": "word",
+  "language": "ng",
+  "unique_contexts": 2515,
+  "total_transitions": 6294
+}

models/word_markov/ng_markov_ctx2_word.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e32567790fc33a99257fd8adb8e7ad0ce3344ca7d3a3eb520e45837f02b8aa0e
+size 109970

models/word_markov/ng_markov_ctx2_word_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 2,
+  "variant": "word",
+  "language": "ng",
+  "unique_contexts": 5756,
+  "total_transitions": 6277
+}

models/word_markov/ng_markov_ctx3_word.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:baacbbfa77ed4700cf8770e496208dc10568de3ddfb7d70f04d448e9c80f629a
+size 129320

models/word_markov/ng_markov_ctx3_word_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 3,
+  "variant": "word",
+  "language": "ng",
+  "unique_contexts": 6060,
+  "total_transitions": 6260
+}

models/word_markov/ng_markov_ctx4_word.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:895d3a6d2039026439c7f454c74fe99ddf6e0ef21d4295442199e0aefa21c515
+size 141586