Upload all models and assets for iu (latest)

.gitattributes

@@ -33,3 +33,10 @@ 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
+visualizations/tsne_sentences.png filter=lfs diff=lfs merge=lfs -text
+visualizations/tsne_words.png filter=lfs diff=lfs merge=lfs -text
+visualizations/zipf_law.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,764 @@
+---
+language: iu
+language_name: Inuktitut
+language_family: eskimoaleut
+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-eskimoaleut
+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: 3.905
+  - name: best_isotropy
+    type: isotropy
+    value: 0.2183
+  - name: vocabulary_size
+    type: vocab
+    value: 0
+generated: 2026-01-10
+---
+
+# Inuktitut - Wikilangs Models
+## Comprehensive Research Report & Full Ablation Study
+
+This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Inuktitut** 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** | 3.015x | 3.02 | 0.1769% | 75,744 |
+| **16k** | 3.468x | 3.47 | 0.2035% | 65,854 |
+| **32k** | 3.905x 🏆 | 3.91 | 0.2292% | 58,476 |
+
+### Tokenization Examples
+
+Below are sample sentences tokenized with each vocabulary size:
+
+**Sample 1:** `ᕙᐃᔅᐳᒃ ᒥᐊᓕᒐᐃᑦ ᓄᓇᖓᓐᓂ ᖃᕆᑕᐅᔭᒃᑯᑦ ᑐᑭᓯᒋᐊᕐᕕᒃ ᓴᖅᑭᑕᐅᓚᐅᖅᓯᒪᔪᖅ ᒫᒃ ᓵᑯᐴᒡᒧᑦ. ᕙᐃᔅᐳᒃ ᑐᓴᐅᒪᔭᐅᓂᖅᐹᖑᕗᖅ ...`
+
+| Vocab | Tokens | Count |
+|-------|--------|-------|
+| 8k | `▁ᕙᐃᔅᐳᒃ ▁ᒥᐊᓕᒐᐃᑦ ▁ᓄᓇᖓᓐᓂ ▁ᖃᕆᑕᐅᔭᒃᑯᑦ ▁ᑐᑭᓯᒋᐊ ᕐᕕᒃ ▁ᓴᖅᑭᑕᐅᓚᐅᖅᓯᒪᔪᖅ ▁ᒫᒃ ▁ᓵᑯ ᐴᒡ ... (+16 more)` | 26 |
+| 16k | `▁ᕙᐃᔅᐳᒃ ▁ᒥᐊᓕᒐᐃᑦ ▁ᓄᓇᖓᓐᓂ ▁ᖃᕆᑕᐅᔭᒃᑯᑦ ▁ᑐᑭᓯᒋᐊ ᕐᕕᒃ ▁ᓴᖅᑭᑕᐅᓚᐅᖅᓯᒪᔪᖅ ▁ᒫᒃ ▁ᓵᑯᐴᒡᒧᑦ . ... (+10 more)` | 20 |
+| 32k | `▁ᕙᐃᔅᐳᒃ ▁ᒥᐊᓕᒐᐃᑦ ▁ᓄᓇᖓᓐᓂ ▁ᖃᕆᑕᐅᔭᒃᑯᑦ ▁ᑐᑭᓯᒋᐊᕐᕕᒃ ▁ᓴᖅᑭᑕᐅᓚᐅᖅᓯᒪᔪᖅ ▁ᒫᒃ ▁ᓵᑯᐴᒡᒧᑦ . ▁ᕙᐃᔅᐳᒃ ... (+7 more)` | 17 |
+
+**Sample 2:** `ᐅᓵᐃᐅ—[ᖃᓪᓗᓈᑎᑐᑦ—Ohio]— ) ᐃᑎᐊᔪᑦ ᐃᓗᐊᓂ. ᐅᓵᐃᐅ ᐃᓄᖁᑎ ᐊᒥᐊᓕᑲ. ᐊᐅᓚᑦᑎᔩᑦ ᓯᕗᓕᖅᑎᖓᑦ-ᓄᓇᓖᑦ ᑰᕉᒻᐴᔅ «...`
+
+| Vocab | Tokens | Count |
+|-------|--------|-------|
+| 8k | `▁ᐅᓵᐃᐅ —[ ᖃᓪᓗᓈᑎᑐᑦ — ohio ]— ▁) ▁ᐃᑎᐊᔪᑦ ▁ᐃᓗᐊᓂ . ... (+27 more)` | 37 |
+| 16k | `▁ᐅᓵᐃᐅ —[ ᖃᓪᓗᓈᑎᑐᑦ — ohio ]— ▁) ▁ᐃᑎᐊᔪᑦ ▁ᐃᓗᐊᓂ . ... (+22 more)` | 32 |
+| 32k | `▁ᐅᓵᐃᐅ —[ ᖃᓪᓗᓈᑎᑐᑦ — ohio ]— ▁) ▁ᐃᑎᐊᔪᑦ ▁ᐃᓗᐊᓂ . ... (+22 more)` | 32 |
+
+**Sample 3:** `ᐊᐅᑦᓯᓇᖅᑐᖅ ᓱᓕᐊᖅ ᐊᓂᖅᐸᓈᖅᑑᔭᖅᑐᖅ ᐅᓚᐱᑉᐹ ᓴᐳᒻᒥᕚ ᑎᒥ. ᐅᑉᔭᒃᐳᖅ ᐊᓐᓄᕌᓂᒃ`
+
+| Vocab | Tokens | Count |
+|-------|--------|-------|
+| 8k | `▁ᐊᐅᑦᓯᓇᖅᑐᖅ ▁ᓱᓕᐊᖅ ▁ᐊᓂᖅᐸᓈᖅᑑᔭᖅᑐᖅ ▁ᐅᓚᐱ ᑉᐹ ▁ᓴᐳᒻᒥᕚ ▁ᑎᒥ . ▁ᐅᑉᔭᒃᐳᖅ ▁ᐊᓐᓄᕌᓂᒃ` | 10 |
+| 16k | `▁ᐊᐅᑦᓯᓇᖅᑐᖅ ▁ᓱᓕᐊᖅ ▁ᐊᓂᖅᐸᓈᖅᑑᔭᖅᑐᖅ ▁ᐅᓚᐱᑉᐹ ▁ᓴᐳᒻᒥᕚ ▁ᑎᒥ . ▁ᐅᑉᔭᒃᐳᖅ ▁ᐊᓐ��ᕌᓂᒃ` | 9 |
+| 32k | `▁ᐊᐅᑦᓯᓇᖅᑐᖅ ▁ᓱᓕᐊᖅ ▁ᐊᓂᖅᐸᓈᖅᑑᔭᖅᑐᖅ ▁ᐅᓚᐱᑉᐹ ▁ᓴᐳᒻᒥᕚ ▁ᑎᒥ . ▁ᐅᑉᔭᒃᐳᖅ ▁ᐊᓐᓄᕌᓂᒃ` | 9 |
+
+
+### Key Findings
+
+- **Best Compression:** 32k achieves 3.905x compression
+- **Lowest UNK Rate:** 8k with 0.1769% 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 | 93 🏆 | 6.54 | 126 | 90.8% | 100.0% |
+| **2-gram** | Subword | 962 | 9.91 | 3,039 | 37.0% | 87.0% |
+| **3-gram** | Word | 130 | 7.03 | 174 | 73.9% | 100.0% |
+| **3-gram** | Subword | 5,020 | 12.29 | 12,029 | 15.7% | 49.7% |
+| **4-gram** | Word | 694 | 9.44 | 794 | 25.0% | 100.0% |
+| **4-gram** | Subword | 14,093 | 13.78 | 28,526 | 8.8% | 30.5% |
+| **5-gram** | Word | 607 | 9.25 | 676 | 24.5% | 100.0% |
+| **5-gram** | Subword | 19,229 | 14.23 | 32,493 | 7.1% | 24.4% |
+
+### Top 5 N-grams by Size
+
+**2-grams (Word):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `san marino` | 73 |
+| 2 | `of the` | 55 |
+| 3 | `ᖄᖓᒍᑦ ᖄᖓᒍᑦ` | 55 |
+| 4 | `ᑭᒻᒧᑦ ᐅᖅᓯᖅ` | 47 |
+| 5 | `ᑕᕆᐅᑉ ᐊᑭᐊᓂ` | 44 |
+
+**3-grams (Word):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ` | 51 |
+| 2 | `ᑭᒻᒧᑦ ᐅᖅᓯᖅ www` | 30 |
+| 3 | `ᐃᓄᖁᑎ ᐊᒥᐊᓕᑲ ᐊᐅᓚᑦᑎᔩᑦ` | 22 |
+| 4 | `ᐊᐅᓚᑦᑎᔩᑦ ᓯᕗᓕᖅᑎᖓᑦ ᓄᓇᓖᑦ` | 22 |
+| 5 | `ᐊᒥᐊᓕᑲ ᐊᐅᓚᑦᑎᔩᑦ ᓯᕗᓕᖅᑎᖓᑦ` | 22 |
+
+**4-grams (Word):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ` | 48 |
+| 2 | `ᐃᓄᖁᑎ ᐊᒥᐊᓕᑲ ᐊᐅᓚᑦᑎᔩᑦ ᓯᕗᓕᖅᑎᖓᑦ` | 22 |
+| 3 | `ᐊᒥᐊᓕᑲ ᐊᐅᓚᑦᑎᔩᑦ ᓯᕗᓕᖅᑎᖓᑦ ᓄᓇᓖᑦ` | 22 |
+| 4 | `ᓄᓇᓖᑦ ᑭᒻᒧᑦ ᐅᖅᓯᖅ www` | 20 |
+| 5 | `the grand and general` | 10 |
+
+**5-grams (Word):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ` | 45 |
+| 2 | `ᐃᓄᖁᑎ ᐊᒥᐊᓕᑲ ᐊᐅᓚᑦᑎᔩᑦ ᓯᕗᓕᖅᑎᖓᑦ ᓄᓇᓖᑦ` | 22 |
+| 3 | `the grand and general council` | 10 |
+| 4 | `ᓄᓇ frameless upright 0 3` | 7 |
+| 5 | `o canada we stand on` | 5 |
+
+**2-grams (Subword):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `ᑦ _` | 4,757 |
+| 2 | `_ ᐊ` | 3,099 |
+| 3 | `ᖅ _` | 2,694 |
+| 4 | `_ ᐃ` | 2,386 |
+| 5 | `, _` | 2,385 |
+
+**3-grams (Subword):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `ᐊ ᒻ ᒪ` | 851 |
+| 2 | `_ ᐊ ᒻ` | 837 |
+| 3 | `_ ᓄ ᓇ` | 816 |
+| 4 | `ᓂ ᒃ _` | 784 |
+| 5 | `ᑦ _ ᐊ` | 710 |
+
+**4-grams (Subword):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `_ ᐊ ᒻ ᒪ` | 833 |
+| 2 | `ᐊ ᒻ ᒪ _` | 420 |
+| 3 | `ᐊ ᒻ ᒪ ᓗ` | 407 |
+| 4 | `ᖅ ᑐ ᖅ _` | 405 |
+| 5 | `ᒻ ᒪ ᓗ _` | 385 |
+
+**5-grams (Subword):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `_ ᐊ ᒻ ᒪ _` | 418 |
+| 2 | `_ ᐊ ᒻ ᒪ ᓗ` | 400 |
+| 3 | `ᐊ ᒻ ᒪ ᓗ _` | 385 |
+| 4 | `_ t h e _` | 346 |
+| 5 | `ᑦ _ ᐊ ᒻ ᒪ` | 218 |
+
+
+### Key Findings
+
+- **Best Perplexity:** 2-gram (word) with 93
+- **Entropy Trend:** Decreases with larger n-grams (more predictable)
+- **Coverage:** Top-1000 patterns cover ~24% 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 | 0.3388 | 1.265 | 1.76 | 15,002 | 66.1% |
+| **1** | Subword | 1.4995 | 2.827 | 13.51 | 541 | 0.0% |
+| **2** | Word | 0.0479 | 1.034 | 1.07 | 26,047 | 95.2% |
+| **2** | Subword | 0.9813 | 1.974 | 4.39 | 7,301 | 1.9% |
+| **3** | Word | 0.0129 | 1.009 | 1.02 | 27,517 | 98.7% |
+| **3** | Subword | 0.5441 | 1.458 | 2.22 | 31,981 | 45.6% |
+| **4** | Word | 0.0049 🏆 | 1.003 | 1.01 | 27,602 | 99.5% |
+| **4** | Subword | 0.3121 | 1.242 | 1.55 | 70,999 | 68.8% |
+
+### Generated Text Samples (Word-based)
+
+Below are text samples generated from each word-based Markov chain model:
+
+**Context Size 1:**
+
+1. `ᐊᒻᒪ ᐱᕈᖅᓯᐊᖅ ᑭᒃᑯᑦ ᐅᐊᑎᒌᓯᕆᒥᒻᒧᑦ ᓴᐃᓇᒃᑭᐅᔪᖅ ᐊᑎᖃᕐᒥᑕᐅᓂᖏᓐᓂᒃ ᐊᓂᔨᖃᕆᔪᑦ ᐅᓂᖅᑕᖃᕐᑕᐅᔪᑦ ᐅᑎᓇᐅᖃᑎᒌᑦ ᐱᒻᒥᕐᒥᐅᑕᓗᑉ ᑭᒻᒧᑦ ᐅᖅᓯᖅ www...`
+2. `ᐊᒻᒪᓗ ᐊᐅᓚᓃᑦ ᐱᓕᕆᖃᑎᒌᖃᑦᑕᖅᑐᑦ ᐋᖅᑭᐅᒪᑎᑦᑎᓂᐊᕐᓗᓂ ᐊᖏᕐᕋᒥᒃ ᐅᓗᕆᐊᓇᙱᑦᑐᒃᑯᑦ ᓲᕐᓗ ᕕᑐᕆᑯ ᐃᓇᓗᒃᑲ ᐃᒡᓗᓐᓂ ᓄᓇᖃᖅᐳᑦ ᐸᑏᑎ ᐃᓕᓚᐅᖅᑕᕋ ᐊᐅᓚ...`
+3. `the roman republic the sammarinese fascist government declared war on their passports citation neede...`
+
+**Context Size 2:**
+
+1. `san marino appealed to pope boniface viii against the contribution demands by the legate papal gover...`
+2. `ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᒥᑭᔫᖕᓂᒃ ᑐᐊᑎᐊᓂᒃ ᐊᒻᒪ ᐃᓛᓐᓂᒃᑯᑦ ᓴᓇᔭᐅᕙᒃᖢᑎᒃ ᒑᑲᒧᓕᒧᑦ ᓵᓪᓴᒧᑦ ᓂᐅᓐᔅᒧᑦ ᐊᒻᒪ ᓯᓚᓐᑐᒧᑦ ᑯᕆᓐᑐ ᒪᑉᐱ...`
+3. `of the european union it is the fifth smallest country in europe after vatican city and state`
+
+**Context Size 3:**
+
+1. `ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᒃᑲᓐᓂᖅ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ`
+2. `ᑭᒻᒧᑦ ᐅᖅᓯᖅ www sd gov`
+3. `ᐃᓄᖁᑎ ᐊᒥᐊᓕᑲ ᐊᐅᓚᑦᑎᔩᑦ ᓯᕗᓕᖅᑎᖓᑦ ᓄᓇᓖᑦ ᐲᕐ ᖃᓪᓗᓈᑎᑐᑦ pierre ᓄᓇᓖᑦ ᑭᒻᒧᑦ ᐅᖅᓯᖅ www ok gov`
+
+**Context Size 4:**
+
+1. `ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ ᖄᖓᒍᑦ`
+2. `ᐊᒥᐊᓕᑲ ᐊᐅᓚᑦᑎᔩᑦ ᓯᕗᓕᖅᑎᖓᑦ ᓄᓇᓖᑦ ᐴᕐᑦᓛᓐᑦ ᖃᓪᓗᓈᑎᑐᑦ portland ᓄᓇᓖᑦ ᑭᒻᒧᑦ ᐅᖅᓯᖅ www nv gov`
+3. `ᐃᓄᖁᑎ ᐊᒥᐊᓕᑲ ᐊᐅᓚᑦᑎᔩᑦ ᓯᕗᓕᖅᑎᖓᑦ ᓄᓇᓖᑦ ᓂᐅ ᐆᕐᓖᓐᔅ ᖃᓪᓗᓈᑎᑐᑦ new orleans ᓄᓇᓖᑦ ᑭᒻᒧᑦ ᐅᖅᓯᖅ www idaho gov`
+
+
+### Generated Text Samples (Subword-based)
+
+Below are text samples generated from each subword-based Markov chain model:
+
+**Context Size 1:**
+
+1. `_ᕕᒃ_ᒐᔪᑎᓪᓗᑕᑭᓯᐊᒻᒪ_`
+2. `ᖅᑐᒃ_ontunixiteco`
+3. `ᑦᑕ)_ᑲ,_ᖄᕐᒥᓱᐊᕈᑎᐊᒻ`
+
+**Context Size 2:**
+
+1. `ᑦ_(ᐱᓚᕈ,_ᑎᑎᓪᓗᓕᖃᖅᐳᑦ`
+2. `_ᐊᐅᐸᐃᒡ_ᐊᖕᓇᖅ_ᑭᓕᐊᑉ_`
+3. `ᖅ_ᑕᐃᑲᓂᐸ,_ᓄᓇᓖᑦ_ᐃᒡᓗ`
+
+**Context Size 3:**
+
+1. `ᐊᒻᒪᓗ_ᕿᓚᒃ._ᓴᓂᑭᓗᐊᕐᒥ.`
+2. `_ᐊᒻᒪ_ᐃᓗᐊᓃᑐᓂ._ᐃᓚᖃᖅᑐ`
+3. `_ᓄᓇᖃᐃᓐᓇᕆᐊᓚᐅᖅᐳᖅ_ᐊᕋᕕ`
+
+**Context Size 4:**
+
+1. `_ᐊᒻᒪ_ᑎᓴᒪᓂᒃ_ᓄᓇᒥᐅᑕᐅᕗᑦ`
+2. `ᐊᒻᒪ_ᑕᑯᑦᑎᐊᔪᐃᓐᓇᕐᒥᒃ_ᐱᖃ`
+3. `ᐊᒻᒪᓗ_ᖁᕕᐊᓱᖕᓂᖅ")ᐃᙱᐅᓯᖓ`
+
+
+### Key Findings
+
+- **Best Predictability:** Context-4 (word) with 99.5% predictability
+- **Branching Factor:** Decreases with context size (more deterministic)
+- **Memory Trade-off:** Larger contexts require more storage (70,999 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 | 3,802 |
+| Total Tokens | 18,925 |
+| Mean Frequency | 4.98 |
+| Median Frequency | 2 |
+| Frequency Std Dev | 13.99 |
+
+### Most Common Words
+
+| Rank | Word | Frequency |
+|------|------|-----------|
+| 1 | ᐊᒻᒪ | 424 |
+| 2 | ᐊᒻᒪᓗ | 392 |
+| 3 | the | 353 |
+| 4 | of | 210 |
+| 5 | ᐃᓄᐃᑦ | 139 |
+| 6 | and | 131 |
+| 7 | ᐅᕝᕙᓘᓐᓃᑦ | 114 |
+| 8 | in | 106 |
+| 9 | ᖃᓪᓗᓈᑎᑐᑦ | 104 |
+| 10 | to | 98 |
+
+### Least Common Words (from vocabulary)
+
+| Rank | Word | Frequency |
+|------|------|-----------|
+| 1 | ᑕᑯᔭᒐᖃᕐᕕᐅᔪᑦ | 2 |
+| 2 | ᒥᐅᓯᐅ | 2 |
+| 3 | ᓴᒃᑯᑐᖃᕐᓄᑦ | 2 |
+| 4 | ᓴᕕᕋᔭᓄᑦ | 2 |
+| 5 | ᐊᒥᐊᖅᑕᐅᓯᒪᔪᑦ | 2 |
+| 6 | ᑭᐊᕋᒥ | 2 |
+| 7 | ᔨᐊᓇᕆ | 2 |
+| 8 | ᓴᓇᓐᖑᐊᒐᐃᑦ | 2 |
+| 9 | ᓅᑉᐸᓪᓕᐊᔪᓄᑦ | 2 |
+| 10 | ᓄᓇᒥᐅᑕᓕᕆᓂᕐᒧᑦ | 2 |
+
+### Zipf's Law Analysis
+
+| Metric | Value |
+|--------|-------|
+| Zipf Coefficient | 0.6869 |
+| R² (Goodness of Fit) | 0.969855 |
+| Adherence Quality | **excellent** |
+
+### Coverage Analysis
+
+| Top N Words | Coverage |
+|-------------|----------|
+| Top 100 | 30.3% |
+| Top 1,000 | 65.3% |
+| Top 5,000 | 0.0% |
+| Top 10,000 | 0.0% |
+
+### Key Findings
+
+- **Zipf Compliance:** R²=0.9699 indicates excellent adherence to Zipf's law
+- **High Frequency Dominance:** Top 100 words cover 30.3% of corpus
+- **Long Tail:** -6,198 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 | 0.2183 | 0.4714 | N/A | N/A |
+| **mono_64d** | 64 | 0.0445 | 0.4570 | N/A | N/A |
+| **mono_128d** | 128 | 0.0046 | 0.4821 | N/A | N/A |
+| **aligned_32d** | 32 | 0.2183 🏆 | 0.4659 | 0.0189 | 0.1384 |
+| **aligned_64d** | 64 | 0.0445 | 0.4550 | 0.0314 | 0.1384 |
+| **aligned_128d** | 128 | 0.0046 | 0.4794 | 0.0503 | 0.1509 |
+
+### Key Findings
+
+- **Best Isotropy:** aligned_32d with 0.2183 (more uniform distribution)
+- **Semantic Density:** Average pairwise similarity of 0.4685. Lower values indicate better semantic separation.
+- **Alignment Quality:** Aligned models achieve up to 5.0% 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 | **3.097** | 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 |
+|--------|----------|
+| `-ᐊ` | ᐊᑐᖅᑕᐅᓯᒪᔪᖅ, ᐊᔅᑦᕌᓕᐊ, ᐊᓂᒍᖅᑎᓪᓗᒋᑦ |
+| `-ᐃ` | ᐃᓅᖃᑎᒌᑦ, ᐃᖃᑦᑐᖅ, ᐃᓱ |
+| `-ᐅ` | ᐅᓪᓗᓂᒃ, ᐅᓛᓴᒥ, ᐅᑭᐅᖃᓕᖅᑎᓪᓗᒋᑦ |
+| `-ᐅᖃ` | ᐅᖃᓕᒫᒐᓄᑦ, ᐅᖃᐅᓯᒃᓴᓂᖏᑦ, ᐅᖃᐅᓯᕐᖓᐅᑎᖃᕐᒪᑎᑕ |
+| `-ᓄᓇ` | ᓄᓇᕕᐅᑉ, ᓄᓇᖁᑎᖓᓂᒃ, ᓄᓇᙳᐊᖓ |
+| `-ᑕᐃ` | ᑕᐃᒫᑦᓴᐃᓐᓇᖅ, ᑕᐃᒃᓱᒪᓂ, ᑕᐃᒃᑯᓇᓂ |
+| `-ᐃᓄ` | ᐃᓄᒃ, ᐃᓄᒋᐊᓛᖑᓪᓗᓂ, ᐃᓄᖕᓂᒃ |
+| `-co` | coca, corporate, country |
+
+#### Productive Suffixes
+| Suffix | Examples |
+|--------|----------|
+| `-ᑦ` | ᖃᓚᒪᓐᖏᑑᓗᑎᓘᓐᓃᑦ, ᐃᓅᖃᑎᒌᑦ, ᐱᓕᕆᑦᑎᐊᕐᓂᖏᓐᓄᑦ |
+| `-ᖅ` | ᐃᖃᑦᑐᖅ, ᐊᑐᖅᑕᐅᓯᒪᔪᖅ, ᓯᐅᕋᖅ |
+| `-ᒃ` | ᓯᕗᓪᓕᖅᐹᒃ, ᐊᑕᐅᓯᕐᒥᒃ, ᐅᓪᓗᓂᒃ |
+| `-ᓂᒃ` | ᐅᓪᓗᓂᒃ, ᒥᓕᐊᓐᓂᒃ, ᓂᐊᖁᕐᓂᒃ |
+| `-ᑐᖅ` | ᐃᖃᑦᑐᖅ, ᓯᐅᕋᐅᔮᖅᑐᖅ, ᐃᓅᓕᖅᑐᖅ |
+| `-ᓄᑦ` | ᐱᓕᕆᑦᑎᐊᕐᓂᖏᓐᓄᑦ, ᑭᖑᓪᓕᖅᐹᖅᓯᐅᑎᓄᑦ, ᐊᑕᐅᓯᐅᖃᑎᒌᓄᑦ |
+| `-ᓂ` | ᓯᓚᑖᓂ, ᐃᓚᐅᙱᖦᖢᓂ, ᖃᓂᒋᔭᖓᓂ |
+| `-t` | aallatqiit, pitquhiinit, anngutikhaqanngittagaangat |
+
+### 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 |
+|------|----------|------------------|----------|
+| `ᕗᓪᓕᖅ` | 1.82x | 6 contexts | ᓯᕗᓪᓕᖅ, ᓯᕗᓪᓕᖅᐹᒃ, ᓯᕗᓪᓕᖅᐹᖅ |
+| `ᓯᕗᓪᓕ` | 1.82x | 5 contexts | ᓯᕗᓪᓕᖅ, ᓯᕗᓪᓕᕐᒥ, ᓯᕗᓪᓕᖅᐹᒃ |
+| `ᖅᓯᒪᔪ` | 1.50x | 6 contexts | ᓇᐃᓈᖅᓯᒪᔪᖅ, ᑎᑎᕋᖅᓯᒪᔪᖅ, ᑎᑎᕋᖅᓯᒪᔪᒥ |
+| `ᓯᒪᔪᖅ` | 1.72x | 4 contexts | ᐃᓚᓯᒪᔪᖅ, ᓴᓇᓯᒪᔪᖅ, ᓴᖅᑭᓯᒪᔪᖅ |
+| `ᖑᓪᓗᓂ` | 1.89x | 3 contexts | ᒥᑭᓛᖑᓪᓗᓂ, ᐊᖏᓛᖑᓪᓗᓂ, ᐊᖏᓛᖑᓪᓗᓂᓗ |
+
+### 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 |
+|--------|--------|-----------|----------|
+| `-ᐊ` | `-ᑦ` | 61 words | ᐊᓂᒍᖅᑎᓪᓗᒋᑦ, ᐊᐅᓚᑦᑎᐊᕈᓐᓃᖅᑐᑦ |
+| `-ᐃ` | `-ᖅ` | 47 words | ᐃᖃᑦᑐᖅ, ᐃᓅᓕᖅᑐᖅ |
+| `-ᐃ` | `-ᑦ` | 46 words | ᐃᓅᖃᑎᒌᑦ, ᐃᓯᒐᐃᑦ |
+| `-ᐅ` | `-ᑦ` | 41 words | ᐅᑭᐅᖃᓕᖅᑎᓪᓗᒋᑦ, ᐅᖃᓕᒫᒐᓄᑦ |
+| `-ᐊ` | `-ᖅ` | 37 words | ᐊᑐᖅᑕᐅᓯᒪᔪᖅ, ᐊᖏᓛᖑᔪᖅ |
+| `-ᐃ` | `-ᒃ` | 33 words | ᐃᓄᒃ, ᐃᓕᓐᓂᐊᕈᑎᒥᒃ |
+| `-ᐊ` | `-ᒃ` | 24 words | ᐊᑕᐅᓯᕐᒥᒃ, ᐊᑯᓕᕕᒃ |
+| `-ᐃ` | `-ᓂᒃ` | 19 words | ᐃᓄᖕᓂᒃ, ᐃᕐᕋᕕᖏᓐᓂᒃ |
+| `-ᐅ` | `-ᖅ` | 19 words | ᐅᐱᕐᖓᖅ, ᐅᖃᐅᓯᖅ |
+| `-ᐊ` | `-ᓂ` | 17 words | ᐊᑐᖅᑕᐅᓪᓗᓂ, ᐊᖏᔪᒻᒪᕆᐊᓘᓪᓗᓂ |
+
+### 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 |
+|------|-----------------|------------|------|
+| ᐋᖅᑭᒃᓯᒪᓂᖓᓄᑦ | **`ᐋᖅᑭᒃᓯᒪᓂᖓ-ᓄᑦ`** | 4.5 | `ᐋᖅᑭᒃᓯᒪᓂᖓ` |
+| presented | **`present-ed`** | 4.5 | `present` |
+| uniformed | **`uniform-ed`** | 4.5 | `uniform` |
+| ᓄᓇᓕᐸᐅᔭᖓᓄᑦ | **`ᓄᓇᓕᐸᐅᔭᖓ-ᓄᑦ`** | 4.5 | `ᓄᓇᓕᐸᐅᔭᖓ` |
+| ᑖᒃᓰᔭᐃᔭᕈᑎᑦ | **`ᑖᒃᓰᔭᐃᔭᕈᑎ-ᑦ`** | 4.5 | `ᑖᒃᓰᔭᐃᔭᕈᑎ` |
+| ᑖᒃᓰᔭᐃᔭᕈᑎᓄᑦ | **`ᑖᒃᓰᔭᐃᔭᕈᑎ-ᓄᑦ`** | 4.5 | `ᑖᒃᓰᔭᐃᔭᕈᑎ` |
+| ᑖᒃᓰᔭᐃᔭᕈᑎᓂᒃ | **`ᑖᒃᓰᔭᐃᔭᕈᑎ-ᓂᒃ`** | 4.5 | `ᑖᒃᓰᔭᐃᔭᕈᑎ` |
+| ᒫᓐᑎᕕᐅᓪᑐᒧᑦ | **`ᒫᓐᑎᕕᐅᓪᑐ-ᒧᑦ`** | 4.5 | `ᒫᓐᑎᕕᐅᓪᑐ` |
+| ᐊᕕᑦᑐᖅᓯᒪᔪᓂᑦ | **`ᐊᕕᑦᑐᖅᓯᒪᔪᓂ-ᑦ`** | 4.5 | `ᐊᕕᑦᑐᖅᓯᒪᔪᓂ` |
+| ᐃᓕᓐᓂᐊᕈᑎᒥᒃ | **`ᐃᓕᓐᓂᐊᕈᑎ-ᒥᒃ`** | 4.5 | `ᐃᓕᓐᓂᐊᕈᑎ` |
+| ᐃᓕᓐᓂᐊᖅᑎᓂᒃ | **`ᐃᓕᓐᓂᐊᖅᑎ-ᓂᒃ`** | 4.5 | `ᐃᓕᓐᓂᐊᖅᑎ` |
+| ᐋᖅᑭᒃᓯᒪᓂᖓᓂᒃ | **`ᐋᖅᑭᒃᓯᒪᓂᖓ-ᓂᒃ`** | 4.5 | `ᐋᖅᑭᒃᓯᒪᓂᖓ` |
+| ᐃᓚᒋᔭᐅᓕᖅᑐᖅ | **`ᐃᓚᒋᔭᐅᓕ-ᖅ-ᑐᖅ`** | 3.0 | `ᐃᓚᒋᔭᐅᓕ` |
+| ᐊᒥᐊᖅᑕᐅᓯᒪᔪᑦ | **`ᐊ-ᒥᐊᖅᑕᐅᓯᒪᔪ-ᑦ`** | 3.0 | `ᒥᐊᖅᑕᐅᓯᒪᔪ` |
+| ᐃᓄᑐᐃᓐᓇᕐᓂᒃ | **`ᐃᓄ-ᑐᐃᓐᓇᕐ-ᓂᒃ`** | 3.0 | `ᑐᐃᓐᓇᕐ` |
+
+### 6.6 Linguistic Interpretation
+
+> **Automated Insight:**
+The language Inuktitut 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 | **32k BPE** | Best compression (3.91x) |
+| N-gram | **2-gram** | Lowest perplexity (93) |
+| Markov | **Context-4** | Highest predictability (99.5%) |
+| 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 04:55:45*

models/embeddings/aligned/iu_128d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bbcdc098ee07b6adaf6cedc14fb2e0902baf8eda2884bbf1bf6a5449932d1025
+size 1024629510

models/embeddings/aligned/iu_128d.meta.json

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

models/embeddings/aligned/iu_128d.projection.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3dd4e8f7e8af48c7c9e90b3a5d87856458d1a3eb1e2d8a250707cf131a05f2be
+size 65664

models/embeddings/aligned/iu_128d_metadata.json

@@ -0,0 +1,8 @@
+{
+  "language": "iu",
+  "dimension": 128,
+  "version": "aligned",
+  "hub_language": "en",
+  "seed_vocab_size": 159,
+  "vocab_size": 601
+}

models/embeddings/aligned/iu_32d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eec8c678b658c84c6120533ab190e00ae645894c3490124fd77661dc3cbcbfdb
+size 256167942

models/embeddings/aligned/iu_32d.meta.json

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

models/embeddings/aligned/iu_32d.projection.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:082cd73db57664ac7a0b32d4764a137e39d020ccb5d498cb709746cdbd9cba55
+size 4224

models/embeddings/aligned/iu_32d_metadata.json

@@ -0,0 +1,8 @@
+{
+  "language": "iu",
+  "dimension": 32,
+  "version": "aligned",
+  "hub_language": "en",
+  "seed_vocab_size": 159,
+  "vocab_size": 601
+}

models/embeddings/aligned/iu_64d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b303bdc53b5b1eeae96f7f1e1e05c6a5da606814bb3c80e4a268037a174bdf0
+size 512321798

models/embeddings/aligned/iu_64d.meta.json

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

models/embeddings/aligned/iu_64d.projection.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e3db540e7e019f438d251b692085f4714039a5b3bb2dcaf8ffcd5dc2e2a96b0
+size 16512

models/embeddings/aligned/iu_64d_metadata.json

@@ -0,0 +1,8 @@
+{
+  "language": "iu",
+  "dimension": 64,
+  "version": "aligned",
+  "hub_language": "en",
+  "seed_vocab_size": 159,
+  "vocab_size": 601
+}

models/embeddings/monolingual/iu_128d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bbcdc098ee07b6adaf6cedc14fb2e0902baf8eda2884bbf1bf6a5449932d1025
+size 1024629510

models/embeddings/monolingual/iu_128d.meta.json

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

models/embeddings/monolingual/iu_128d_metadata.json

@@ -0,0 +1,16 @@
+{
+  "language": "iu",
+  "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": 601
+}

models/embeddings/monolingual/iu_32d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eec8c678b658c84c6120533ab190e00ae645894c3490124fd77661dc3cbcbfdb
+size 256167942

models/embeddings/monolingual/iu_32d.meta.json

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

models/embeddings/monolingual/iu_32d_metadata.json

@@ -0,0 +1,16 @@
+{
+  "language": "iu",
+  "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": 601
+}

models/embeddings/monolingual/iu_64d.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b303bdc53b5b1eeae96f7f1e1e05c6a5da606814bb3c80e4a268037a174bdf0
+size 512321798

models/embeddings/monolingual/iu_64d.meta.json

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

models/embeddings/monolingual/iu_64d_metadata.json

@@ -0,0 +1,16 @@
+{
+  "language": "iu",
+  "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": 601
+}

models/subword_markov/iu_markov_ctx1_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3f8e0392bceb9d1c0af0337b021bc6dc41a40d31732b5c1abb8af053c9c190b
+size 52428

models/subword_markov/iu_markov_ctx1_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 1,
+  "variant": "subword",
+  "language": "iu",
+  "unique_contexts": 541,
+  "total_transitions": 227882
+}

models/subword_markov/iu_markov_ctx2_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ab2a45e6de750ba3afec4fbad09e61a3aef6fadfa2fa66832bb44fb3b537d3f
+size 220128

models/subword_markov/iu_markov_ctx2_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 2,
+  "variant": "subword",
+  "language": "iu",
+  "unique_contexts": 7301,
+  "total_transitions": 227468
+}

models/subword_markov/iu_markov_ctx3_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3cf704e895b673e27b044d3dfbd47c63deee22788061a6c577ef60a0bcb8d962
+size 596983

models/subword_markov/iu_markov_ctx3_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 3,
+  "variant": "subword",
+  "language": "iu",
+  "unique_contexts": 31981,
+  "total_transitions": 227054
+}

models/subword_markov/iu_markov_ctx4_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:429fa4ae6ed8ac0c38b6d225c2beba5b2152fd0b3195920934a64b01a03bede5
+size 1123996

models/subword_markov/iu_markov_ctx4_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 4,
+  "variant": "subword",
+  "language": "iu",
+  "unique_contexts": 70999,
+  "total_transitions": 226640
+}

models/subword_ngram/iu_2gram_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f46928e81f9dfbfb5c65fcf673d61df90a96792eba3ce666709160eb859bbb6
+size 35017

models/subword_ngram/iu_2gram_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "n": 2,
+  "variant": "subword",
+  "language": "iu",
+  "unique_ngrams": 3039,
+  "total_ngrams": 227882
+}

models/subword_ngram/iu_3gram_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:efdcb343a1e04c0a70691799affa80b77945caab6050cd5c758ad2e93371ae0f
+size 147171

models/subword_ngram/iu_3gram_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "n": 3,
+  "variant": "subword",
+  "language": "iu",
+  "unique_ngrams": 12029,
+  "total_ngrams": 227468
+}

models/subword_ngram/iu_4gram_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd7fbd0b9e26b9ecaab4561ac1ad916de73de6ed3a941bace6a3fd653f047a95
+size 378811

models/subword_ngram/iu_4gram_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "n": 4,
+  "variant": "subword",
+  "language": "iu",
+  "unique_ngrams": 28526,
+  "total_ngrams": 227054
+}

models/subword_ngram/iu_5gram_subword.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcc6ebd08d689bed4a1d89027d38aa2455518c9675efb8ed87a903c735f32840
+size 451112

models/subword_ngram/iu_5gram_subword_metadata.json

@@ -0,0 +1,7 @@
+{
+  "n": 5,
+  "variant": "subword",
+  "language": "iu",
+  "unique_ngrams": 32493,
+  "total_ngrams": 226640
+}

models/tokenizer/iu_tokenizer_16k.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c68ed5b03abbd7f0aa4423d77f8b697240fc464909b85ae32c6656dd99e20775
+size 602777

models/tokenizer/iu_tokenizer_32k.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7dead12def3bf4bec79859375a38ba6cbbad05e4e45b7170ec5c2083221b8e9b
+size 941618

models/tokenizer/iu_tokenizer_8k.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d2a744d72034939e556b850a145584632e4b23e92e83cbf02ac3a78d158057a
+size 407778

models/vocabulary/iu_vocabulary.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:46c646245ad3a73bae0834f99ae0f0760b884594f52b5de0d946cff9e3164576
+size 71746

models/vocabulary/iu_vocabulary_metadata.json

@@ -0,0 +1,17 @@
+{
+  "language": "iu",
+  "vocabulary_size": 3802,
+  "variant": "full",
+  "statistics": {
+    "type_token_ratio": 0.5007263123142951,
+    "coverage": {
+      "top_100": 0.18903928689336413,
+      "top_1000": 0.4081875206338726,
+      "top_5000": 0.6643446682073292,
+      "top_10000": 0.8294156487289535
+    },
+    "hapax_count": 11365,
+    "hapax_ratio": 0.749324190677128,
+    "total_documents": 414
+  }
+}

models/word_markov/iu_markov_ctx1_word.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1d444fa2525ea7745343d67204c5983f24c9dc660690798d352cb0f2b0c2dcb
+size 501533

models/word_markov/iu_markov_ctx1_word_metadata.json

@@ -0,0 +1,7 @@
+{
+  "context_size": 1,
+  "variant": "word",
+  "language": "iu",
+  "unique_contexts": 15002,
+  "total_transitions": 29876
+}