Upload all models and assets for co (20251001)

README.md

@@ -23,14 +23,14 @@ dataset_info:
 metrics:
   - name: best_compression_ratio
     type: compression
-    value: 4.112
+    value: 4.197
   - name: best_isotropy
     type: isotropy
-    value: 0.8279
+    value: 0.8272
   - name: vocabulary_size
     type: vocab
-    value: 60410
-generated: 2025-12-28
+    value: 0
+generated: 2026-01-03
 ---
 
 # CO - Wikilangs Models
@@ -44,12 +44,13 @@ We analyze tokenizers, n-gram models, Markov chains, vocabulary statistics, and
 ### Models & Assets
 
 - Tokenizers (8k, 16k, 32k, 64k)
-- N-gram models (2, 3, 4-gram)
-- Markov chains (context of 1, 2, 3 and 4)
+- 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
+- Embeddings in various sizes and dimensions (aligned and unaligned)
 - Language Vocabulary
 - Language Statistics
+
 ![Performance Dashboard](visualizations/performance_dashboard.png)
 
 ### Analysis and Evaluation
@@ -59,7 +60,8 @@ We analyze tokenizers, n-gram models, Markov chains, vocabulary statistics, and
 - [3. Markov Chain Evaluation](#3-markov-chain-evaluation)
 - [4. Vocabulary Analysis](#4-vocabulary-analysis)
 - [5. Word Embeddings Evaluation](#5-word-embeddings-evaluation)
-- [6. Summary & Recommendations](#6-summary--recommendations)
+- [6. Morphological Analysis (Experimental)](#6-morphological-analysis)
+- [7. Summary & Recommendations](#7-summary--recommendations)
 - [Metrics Glossary](#appendix-metrics-glossary--interpretation-guide)
 - [Visualizations Index](#visualizations-index)
 
@@ -68,59 +70,57 @@ We analyze tokenizers, n-gram models, Markov chains, vocabulary statistics, and
 
 ![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** | 3.411x | 3.34 | 0.0362% | 381,463 |
-| **16k** | 3.668x | 3.59 | 0.0389% | 354,761 |
-| **32k** | 3.899x | 3.81 | 0.0414% | 333,726 |
-| **64k** | 4.112x 🏆 | 4.02 | 0.0436% | 316,458 |
+| **8k** | 3.418x | 3.42 | 0.0321% | 368,161 |
+| **16k** | 3.691x | 3.69 | 0.0346% | 340,883 |
+| **32k** | 3.970x | 3.97 | 0.0372% | 316,946 |
+| **64k** | 4.197x 🏆 | 4.20 | 0.0394% | 299,775 |
 
 ### Tokenization Examples
 
 Below are sample sentences tokenized with each vocabulary size:
 
-**Sample 1:** `hè una cumuna spagnola di a pruvincia di Soria, in a cumunità autonoma di Castig...`
+**Sample 1:** `Agliana hè una cumuna toscana di a pruvincia di Pistoia. Teni abitanti cumuna di...`
 
 | Vocab | Tokens | Count |
 |-------|--------|-------|
-| 8k | `▁hè ▁una ▁cumuna ▁spagnola ▁di ▁a ▁pruvincia ▁di ▁soria , ... (+14 more)` | 24 |
-| 16k | `▁hè ▁una ▁cumuna ▁spagnola ▁di ▁a ▁pruvincia ▁di ▁soria , ... (+14 more)` | 24 |
-| 32k | `▁hè ▁una ▁cumuna ▁spagnola ▁di ▁a ▁pruvincia ▁di ▁soria , ... (+14 more)` | 24 |
-| 64k | `▁hè ▁una ▁cumuna ▁spagnola ▁di ▁a ▁pruvincia ▁di ▁soria , ... (+14 more)` | 24 |
-
-**Sample 2:** `Aries Spears hè un attore americanu.
-
- Biugrafia 
+| 8k | `▁a gli ana ▁hè ▁una ▁cumuna ▁toscana ▁di ▁a ▁pruvincia ... (+10 more)` | 20 |
+| 16k | `▁a gli ana ▁hè ▁una ▁cumuna ▁toscana ▁di ▁a ▁pruvincia ... (+8 more)` | 18 |
+| 32k | `▁agli ana ▁hè ▁una ▁cumuna ▁toscana ▁di ▁a ▁pruvincia ▁di ... (+7 more)` | 17 |
+| 64k | `▁agliana ▁hè ▁una ▁cumuna ▁toscana ▁di ▁a ▁pruvincia ▁di ▁pistoia ... (+6 more)` | 16 |
 
- Da vede dinò 
- Listinu di l'...`
+**Sample 2:** `Monteriggioni hè una cumuna toscana di a pruvincia di Siena.Teni 7.877 abitanti....`
 
 | Vocab | Tokens | Count |
 |-------|--------|-------|
-| 8k | `▁ari es ▁spe ars ▁hè ▁un ▁attore ▁americanu . ▁biugrafia ... (+22 more)` | 32 |
-| 16k | `▁ari es ▁spe ars ▁hè ▁un ▁attore ▁americanu . ▁biugrafia ... (+22 more)` | 32 |
-| 32k | `▁ari es ▁spe ars ▁hè ▁un ▁attore ▁americanu . ▁biugrafia ... (+22 more)` | 32 |
-| 64k | `▁aries ▁spears ▁hè ▁un ▁attore ▁americanu . ▁biugrafia ▁da ▁vede ... (+16 more)` | 26 |
+| 8k | `▁monte ri ggi oni ▁hè ▁una ▁cumuna ▁toscana ▁di ▁a ... (+15 more)` | 25 |
+| 16k | `▁monte ri ggi oni ▁hè ▁una ▁cumuna ▁toscana ▁di ▁a ... (+15 more)` | 25 |
+| 32k | `▁monte ri ggi oni ▁hè ▁una ▁cumuna ▁toscana ▁di ▁a ... (+15 more)` | 25 |
+| 64k | `▁monteriggioni ▁hè ▁una ▁cumuna ▁toscana ▁di ▁a ▁pruvincia ▁di ▁siena ... (+12 more)` | 22 |
 
-**Sample 3:** `Arliss Howard (18 ottobre 1954) hè un attore americanu.
-
-Da vede dinò
-Listinu di...`
+**Sample 3:** `Sean Justin Penn hè un attore americanu. Biugrafia Da vede dinò The Thin Red Lin...`
 
 | Vocab | Tokens | Count |
 |-------|--------|-------|
-| 8k | `▁ar li ss ▁ho ward ▁( 1 8 ▁ottobre ▁ ... (+23 more)` | 33 |
-| 16k | `▁ar li ss ▁ho ward ▁( 1 8 ▁ottobre ▁ ... (+23 more)` | 33 |
-| 32k | `▁ar li ss ▁howard ▁( 1 8 ▁ottobre ▁ 1 ... (+20 more)` | 30 |
-| 64k | `▁arli ss ▁howard ▁( 1 8 ▁ottobre ▁ 1 9 ... (+19 more)` | 29 |
+| 8k | `▁s ean ▁j us tin ▁pen n ▁hè ▁un ▁attore ... (+22 more)` | 32 |
+| 16k | `▁sean ▁jus tin ▁pen n ▁hè ▁un ▁attore ▁americanu . ... (+16 more)` | 26 |
+| 32k | `▁sean ▁jus tin ▁penn ▁hè ▁un ▁attore ▁americanu . ▁biugrafia ... (+13 more)` | 23 |
+| 64k | `▁sean ▁justin ▁penn ▁hè ▁un ▁attore ▁americanu . ▁biugrafia ▁da ... (+12 more)` | 22 |
 
 
 ### Key Findings
 
-- **Best Compression:** 64k achieves 4.112x compression
-- **Lowest UNK Rate:** 8k with 0.0362% unknown tokens
+- **Best Compression:** 64k achieves 4.197x compression
+- **Lowest UNK Rate:** 8k with 0.0321% unknown tokens
 - **Trade-off:** Larger vocabularies improve compression but increase model size
 - **Recommendation:** 32k vocabulary provides optimal balance for production use
 
@@ -129,57 +129,89 @@ Listinu di...`
 
 ![N-gram Perplexity](visualizations/ngram_perplexity.png)
 
+![N-gram Unique](visualizations/ngram_unique.png)
+
 ![N-gram Coverage](visualizations/ngram_coverage.png)
 
 ### Results
 
-| N-gram | Perplexity | Entropy | Unique N-grams | Top-100 Coverage | Top-1000 Coverage |
-|--------|------------|---------|----------------|------------------|-------------------|
-| **2-gram** | 8,586 🏆 | 13.07 | 59,847 | 24.1% | 47.7% |
-| **2-gram** | 252 🏆 | 7.98 | 3,720 | 68.2% | 99.4% |
-| **3-gram** | 25,561 | 14.64 | 108,167 | 12.3% | 31.9% |
-| **3-gram** | 1,976 | 10.95 | 27,040 | 26.7% | 74.8% |
-| **4-gram** | 50,965 | 15.64 | 187,731 | 9.1% | 24.8% |
-| **4-gram** | 10,378 | 13.34 | 132,259 | 13.2% | 40.9% |
+| N-gram | Variant | Perplexity | Entropy | Unique N-grams | Top-100 Coverage | Top-1000 Coverage |
+|--------|---------|------------|---------|----------------|------------------|-------------------|
+| **2-gram** | Word | 9,228 | 13.17 | 49,319 | 22.0% | 44.8% |
+| **2-gram** | Subword | 221 🏆 | 7.79 | 3,181 | 71.2% | 99.6% |
+| **3-gram** | Word | 24,246 | 14.57 | 83,012 | 11.1% | 30.7% |
+| **3-gram** | Subword | 1,706 | 10.74 | 22,404 | 28.4% | 77.6% |
+| **4-gram** | Word | 41,538 | 15.34 | 136,699 | 9.4% | 25.8% |
+| **4-gram** | Subword | 9,044 | 13.14 | 107,042 | 13.9% | 42.6% |
 
 ### Top 5 N-grams by Size
 
-**2-grams:**
+**2-grams (Word):**
 
 | Rank | N-gram | Count |
 |------|--------|-------|
-| 1 | `l '` | 45,334 |
-| 2 | `di u` | 18,690 |
-| 3 | `di a` | 18,510 |
-| 4 | `d '` | 14,383 |
-| 5 | `di l` | 13,245 |
+| 1 | `di u` | 18,783 |
+| 2 | `di a` | 18,523 |
+| 3 | `di l` | 13,279 |
+| 4 | `di i` | 10,605 |
+| 5 | `à u` | 9,199 |
 
-**3-grams:**
+**3-grams (Word):**
 
 | Rank | N-gram | Count |
 |------|--------|-------|
-| 1 | `di l '` | 12,757 |
-| 2 | `à l '` | 5,757 |
-| 3 | `a famiglia di` | 4,354 |
-| 4 | `hè una spezia` | 3,364 |
-| 5 | `com ' è` | 3,225 |
+| 1 | `a famiglia di` | 4,349 |
+| 2 | `hè una spezia` | 3,355 |
+| 3 | `di a famiglia` | 2,698 |
+| 4 | `hè una pianta` | 2,612 |
+| 5 | `una spezia di` | 2,287 |
 
-**4-grams:**
+**4-grams (Word):**
 
 | Rank | N-gram | Count |
 |------|--------|-------|
-| 1 | `di a famiglia di` | 2,632 |
+| 1 | `di a famiglia di` | 2,629 |
 | 2 | `a famiglia di i` | 2,171 |
-| 3 | `hè una spezia di` | 2,067 |
-| 4 | `à wikimedia commons .` | 1,947 |
-| 5 | `annantu à wikimedia commons` | 1,947 |
+| 3 | `hè una spezia di` | 2,061 |
+| 4 | `annantu à wikimedia commons` | 1,945 |
+| 5 | `à wikimedia commons di` | 1,923 |
+
+**2-grams (Subword):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `i _` | 432,981 |
+| 2 | `a _` | 404,157 |
+| 3 | `u _` | 316,081 |
+| 4 | `_ d` | 246,351 |
+| 5 | `d i` | 217,005 |
+
+**3-grams (Subword):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `_ d i` | 173,124 |
+| 2 | `d i _` | 152,141 |
+| 3 | `_ i n` | 82,653 |
+| 4 | `_ u _` | 81,426 |
+| 5 | `_ a _` | 72,871 |
+
+**4-grams (Subword):**
+
+| Rank | N-gram | Count |
+|------|--------|-------|
+| 1 | `_ d i _` | 143,493 |
+| 2 | `_ i n _` | 57,416 |
+| 3 | `a _ d i` | 45,268 |
+| 4 | `_ h è _` | 44,732 |
+| 5 | `i _ d i` | 35,176 |
 
 
 ### Key Findings
 
-- **Best Perplexity:** 2-gram with 252
+- **Best Perplexity:** 2-gram (subword) with 221
 - **Entropy Trend:** Decreases with larger n-grams (more predictable)
-- **Coverage:** Top-1000 patterns cover ~41% of corpus
+- **Coverage:** Top-1000 patterns cover ~43% of corpus
 - **Recommendation:** 4-gram or 5-gram for best predictive performance
 
 ---
@@ -187,55 +219,86 @@ Listinu di...`
 
 ![Markov Entropy](visualizations/markov_entropy.png)
 
+![Markov Contexts](visualizations/markov_contexts.png)
+
 ![Markov Branching](visualizations/markov_branching.png)
 
 ### Results
 
-| Context | Avg Entropy | Perplexity | Branching Factor | Unique Contexts | Predictability |
-|---------|-------------|------------|------------------|-----------------|----------------|
-| **1** | 0.8152 | 1.760 | 5.44 | 128,497 | 18.5% |
-| **1** | 1.0047 | 2.007 | 7.61 | 1,338 | 0.0% |
-| **2** | 0.3514 | 1.276 | 1.95 | 698,773 | 64.9% |
-| **2** | 0.9540 | 1.937 | 5.63 | 10,183 | 4.6% |
-| **3** | 0.1551 | 1.114 | 1.31 | 1,359,938 | 84.5% |
-| **3** | 0.8234 | 1.770 | 4.05 | 57,282 | 17.7% |
-| **4** | 0.0762 🏆 | 1.054 | 1.13 | 1,780,901 | 92.4% |
-| **4** | 0.6487 🏆 | 1.568 | 2.82 | 232,155 | 35.1% |
+| Context | Variant | Avg Entropy | Perplexity | Branching Factor | Unique Contexts | Predictability |
+|---------|---------|-------------|------------|------------------|-----------------|----------------|
+| **1** | Word | 0.8949 | 1.860 | 5.60 | 123,267 | 10.5% |
+| **1** | Subword | 1.0516 | 2.073 | 8.41 | 976 | 0.0% |
+| **2** | Word | 0.3102 | 1.240 | 1.80 | 688,381 | 69.0% |
+| **2** | Subword | 0.9618 | 1.948 | 5.61 | 8,204 | 3.8% |
+| **3** | Word | 0.1337 | 1.097 | 1.25 | 1,235,287 | 86.6% |
+| **3** | Subword | 0.7919 | 1.731 | 3.99 | 46,007 | 20.8% |
+| **4** | Word | 0.0622 🏆 | 1.044 | 1.10 | 1,541,605 | 93.8% |
+| **4** | Subword | 0.6473 | 1.566 | 2.90 | 183,668 | 35.3% |
 
-### Generated Text Samples
+### Generated Text Samples (Word-based)
 
-Below are text samples generated from each Markov chain model:
+Below are text samples generated from each word-based Markov chain model:
 
 **Context Size 1:**
 
-1. `di l ' edda l ' uparatori devi u rè unificadori , u rodanu è à`
-2. `, i bordi di a so pusizioni giugrafica . mentri chì và , ti pristaraghju u`
-3. `. a so nomu di a femina face di a prima u primu piannu fundata u`
+1. `di l onore di culombu facciandine e rete di pianta di u guvernu à spessu in`
+2. `u so foglie basale lanceulate è difficili suprattuttu vocalici senza fà la tira glǝ munnǝ è`
+3. `a gioia quandu ridatti i casci sò forsi statu indipindente di l india di sporti ecunumia`
 
 **Context Size 2:**
 
-1. `l ' esemplare più grossu mai truvatu fubbe tumbatu in angola in u mediterraniu . faci parti`
-2. `di u cubu di u 2011 , airbnb hà raccoltu $ 119 , 8 % in isuzu`
-3. `di a nutazioni pusiziunali micca assà diffarenti rispettu à l ' animali chì si pò mintuvà :`
+1. `di u muvimentu di a pruvincia agira aidone assoro calascibetta caltanissetta cl gangi pa leonforte n...`
+2. `di a corsica nordu africa uccidintali in sudafrica è in europa meridiunale è cintrale burhinus oedic...`
+3. `di l annu avenimenti in corsica jeanmonod d gamisans j flora corsica 2 ed edisud noti altri`
 
 **Context Size 3:**
 
-1. `di l ' arte è di a fine di listessu annu deposita à l ' iranu . porphyrio`
-2. `à l ' università di versailles , hè cunnisciutu sottu u nome di locu vicinu di u vechju`
-3. `a famiglia di e liliaceae . descrizzione lucalisazione referenze ligami categoria : cumuna di corsic...`
+1. `a famiglia di e papaveraceae si distingue da i so piccioli sò ritti ramificati è cuparti à pela`
+2. `hè una spezia di pianta chì face parte di a famiglia di i labrinae ss articulu pruveni in`
+3. `di a famiglia di i poaceae discrizzioni poa bulbosa hè prisenti in l alpi i pirenei è i`
 
 **Context Size 4:**
 
-1. `di a famiglia di i ranunculaceae . discrizzioni ranunculus kuepferi hè una spezia di acellu chì face...`
-2. `a famiglia di i laniidae , chì cumprende dinù altre spezie d ' aythya . si ciba per u`
-3. `hè una spezia di acellu chì face parte di a famiglia di i polypodiaceae . ' ssa pianta priferisce`
+1. `di a famiglia di e polygonaceae si distingue da e so fiurarelli rusulatu pallidu à purpureu ragruppa...`
+2. `a famiglia di i dryopteridaceae ss articulu pruveni in parti o in tutalità da l articulu currispunde...`
+3. `hè una spezia di pianta arbacea vivaci appartinendu à a famiglia di i fabaceae discrizzioni ornithop...`
+
+
+### Generated Text Samples (Subword-based)
+
+Below are text samples generated from each subword-based Markov chain model:
+
+**Context Size 1:**
+
+1. `_pinaga._i_cavac`
+2. `isa_se_ssa_villa`
+3. `ara_dinum'a_dici`
+
+**Context Size 2:**
+
+1. `i_fece_fiazionduc`
+2. `a_hè_a_chì_nalegu`
+3. `u_50px_le_d'isi_f`
+
+**Context Size 3:**
+
+1. `_di_"forte_pianu_p`
+2. `di_incamplica_è_fa`
+3. `_in_atlantunimentr`
+
+**Context Size 4:**
+
+1. `_di_frutti_in_corsa`
+2. `_in_u_harrisparterà`
+3. `a_di_lunghjadori_ri`
 
 
 ### Key Findings
 
-- **Best Predictability:** Context-4 with 92.4% predictability
+- **Best Predictability:** Context-4 (word) with 93.8% predictability
 - **Branching Factor:** Decreases with context size (more deterministic)
-- **Memory Trade-off:** Larger contexts require more storage (232,155 contexts)
+- **Memory Trade-off:** Larger contexts require more storage (183,668 contexts)
 - **Recommendation:** Context-3 or Context-4 for text generation
 
 ---
@@ -251,64 +314,64 @@ Below are text samples generated from each Markov chain model:
 
 | Metric | Value |
 |--------|-------|
-| Vocabulary Size | 60,410 |
-| Total Tokens | 2,262,071 |
-| Mean Frequency | 37.45 |
+| Vocabulary Size | 58,612 |
+| Total Tokens | 2,193,141 |
+| Mean Frequency | 37.42 |
 | Median Frequency | 4 |
-| Frequency Std Dev | 967.51 |
+| Frequency Std Dev | 979.13 |
 
 ### Most Common Words
 
 | Rank | Word | Frequency |
 |------|------|-----------|
-| 1 | di | 143,631 |
-| 2 | u | 84,326 |
-| 3 | a | 76,229 |
-| 4 | è | 67,228 |
-| 5 | in | 58,950 |
-| 6 | à | 58,500 |
-| 7 | l | 48,443 |
-| 8 | hè | 46,128 |
-| 9 | i | 45,172 |
-| 10 | da | 24,646 |
+| 1 | di | 143,885 |
+| 2 | u | 84,171 |
+| 3 | a | 75,994 |
+| 4 | è | 66,959 |
+| 5 | in | 58,823 |
+| 6 | à | 58,335 |
+| 7 | l | 48,252 |
+| 8 | hè | 45,746 |
+| 9 | i | 45,068 |
+| 10 | da | 24,631 |
 
 ### Least Common Words (from vocabulary)
 
 | Rank | Word | Frequency |
 |------|------|-----------|
-| 1 | corsehttps | 2 |
-| 2 | secchia | 2 |
-| 3 | bergeries | 2 |
-| 4 | douleur | 2 |
-| 5 | bouvier | 2 |
-| 6 | spezialità | 2 |
-| 7 | alerta | 2 |
-| 8 | francebleu | 2 |
-| 9 | emissions | 2 |
-| 10 | ꦈꦠꦩ | 2 |
+| 1 | zampigiallu | 2 |
+| 2 | lepeletier | 2 |
+| 3 | nigrithorax | 2 |
+| 4 | crabro | 2 |
+| 5 | entomologhi | 2 |
+| 6 | priculusità | 2 |
+| 7 | apiarie | 2 |
+| 8 | cottura | 2 |
+| 9 | risuttati | 2 |
+| 10 | tippicu | 2 |
 
 ### Zipf's Law Analysis
 
 | Metric | Value |
 |--------|-------|
-| Zipf Coefficient | 1.0521 |
-| R² (Goodness of Fit) | 0.997001 |
+| Zipf Coefficient | 1.0564 |
+| R² (Goodness of Fit) | 0.996983 |
 | Adherence Quality | **excellent** |
 
 ### Coverage Analysis
 
 | Top N Words | Coverage |
 |-------------|----------|
-| Top 100 | 48.2% |
-| Top 1,000 | 68.9% |
-| Top 5,000 | 83.6% |
-| Top 10,000 | 89.1% |
+| Top 100 | 48.8% |
+| Top 1,000 | 69.5% |
+| Top 5,000 | 84.0% |
+| Top 10,000 | 89.4% |
 
 ### Key Findings
 
 - **Zipf Compliance:** R²=0.9970 indicates excellent adherence to Zipf's law
-- **High Frequency Dominance:** Top 100 words cover 48.2% of corpus
-- **Long Tail:** 50,410 words needed for remaining 10.9% coverage
+- **High Frequency Dominance:** Top 100 words cover 48.8% of corpus
+- **Long Tail:** 48,612 words needed for remaining 10.6% coverage
 
 ---
 ## 5. Word Embeddings Evaluation
@@ -321,24 +384,134 @@ Below are text samples generated from each Markov chain model:
 
 ![t-SNE Sentences](visualizations/tsne_sentences.png)
 
-### Model Comparison
 
-| Model | Vocab Size | Dimension | Avg Norm | Std Norm | Isotropy |
-|-------|------------|-----------|----------|----------|----------|
-| **mono_32d** | 32,860 | 32 | 4.128 | 1.155 | 0.8279 🏆 |
-| **mono_64d** | 32,860 | 64 | 4.723 | 1.023 | 0.8160 |
-| **mono_128d** | 32,860 | 128 | 5.374 | 0.857 | 0.7544 |
-| **embeddings_enhanced** | 0 | 0 | 0.000 | 0.000 | 0.0000 |
+### 5.1 Cross-Lingual Alignment
+
+> *Note: Multilingual alignment visualization not available for this language.*
+
+
+### 5.2 Model Comparison
+
+| Model | Dimension | Isotropy | Semantic Density | Alignment R@1 | Alignment R@10 |
+|-------|-----------|----------|------------------|---------------|----------------|
+| **mono_32d** | 32 | 0.8272 🏆 | 0.3408 | N/A | N/A |
+| **mono_64d** | 64 | 0.8188 | 0.2697 | N/A | N/A |
+| **mono_128d** | 128 | 0.7473 | 0.2166 | N/A | N/A |
 
 ### Key Findings
 
-- **Best Isotropy:** mono_32d with 0.8279 (more uniform distribution)
-- **Dimension Trade-off:** Higher dimensions capture more semantics but reduce isotropy
-- **Vocabulary Coverage:** All models cover 32,860 words
-- **Recommendation:** 100d for balanced semantic capture and efficiency
+- **Best Isotropy:** mono_32d with 0.8272 (more uniform distribution)
+- **Semantic Density:** Average pairwise similarity of 0.2757. Lower values indicate better semantic separation.
+- **Alignment Quality:** No aligned models evaluated in this run.
+- **Recommendation:** 128d aligned for best cross-lingual performance
 
 ---
-## 6. Summary & Recommendations
+## 6.  Morphological Analysis (Experimental)
+
+> ⚠️ **Warning:** This language shows low morphological productivity. The statistical signals used for this analysis may be noisy or less reliable than for morphologically rich languages.
+
+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 | **0.000** | Low morphological productivity | ⚠️ Likely unreliable |
+| Idiomaticity Gap | **-1.000** | Low formulaic content | - |
+
+### 6.2 Affix Inventory (Productive Units)
+
+These are the most productive prefixes and suffixes identified by sampling the vocabulary for global substitutability patterns. A unit is considered an affix if stripping it leaves a valid stem that appears in other contexts.
+
+#### Productive Prefixes
+| Prefix | Examples |
+|--------|----------|
+| `-cu` | cullisori, cunisciutu, cumbinatoria |
+| `-ca` | cartagene, caerulescens, casagliò |
+| `-ri` | rizomatosu, ritiranu, ricustruita |
+| `-in` | innù, ingannatu, ingaghjatu |
+| `-pr` | produtta, predita, produzzione |
+| `-ma` | macidonia, matrimonii, maestro |
+| `-di` | difendidori, differenziale, dicriscenti |
+| `-pa` | pavillon, paola, parentella |
+
+#### Productive Suffixes
+| Suffix | Examples |
+|--------|----------|
+| `-a` | occhjatana, mdina, illeghjittima |
+| `-i` | petrignani, quindici, cullisori |
+| `-u` | locudoresu, glaucu, fuculaghju |
+| `-e` | phryganae, christine, volume |
+| `-tu` | cunisciutu, bassistu, ingannatu |
+| `-ni` | petrignani, parsicuzioni, bizantini |
+| `-ti` | viditi, dalmati, stupefacenti |
+| `-ta` | avvilanata, szocialista, produtta |
+
+### 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 |
+|------|----------|------------------|----------|
+| `endu` | 2.27x | 73 contexts | fendu, vendu, dendu |
+| `enti` | 1.76x | 119 contexts | lenti, penti, menti |
+| `azio` | 1.86x | 55 contexts | tazio, lazio, orazio |
+| `aghj` | 1.50x | 141 contexts | aghje, aghju, aghja |
+| `ment` | 1.57x | 87 contexts | mentr, menti, menta |
+| `glia` | 1.64x | 69 contexts | aglia, figlia, voglia |
+| `zion` | 1.67x | 63 contexts | lezion, azione, nuzione |
+| `igli` | 1.44x | 112 contexts | figli, migli, cigli |
+| `tura` | 1.59x | 62 contexts | altura, matura, turaci |
+| `cors` | 1.85x | 33 contexts | corsa, corso, corsi |
+| `sica` | 1.55x | 37 contexts | fisica, sicani, musica |
+| `ific` | 1.48x | 43 contexts | edificà, pacific, unificà |
+
+### 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 |
+|--------|--------|-----------|----------|
+| `-cu` | `-u` | 75 words | cumandamentu, cutratu |
+| `-cu` | `-i` | 74 words | cuttoli, custituenti |
+| `-ri` | `-i` | 69 words | riunghji, riferimenti |
+| `-pr` | `-i` | 66 words | prufundamenti, primuri |
+| `-in` | `-i` | 66 words | infruttuosi, indoauropei |
+| `-cu` | `-a` | 64 words | cumpattezza, cunghjunghja |
+| `-ca` | `-u` | 63 words | cancelieru, cattru |
+| `-cu` | `-e` | 58 words | cuuperazione, cuscione |
+| `-ca` | `-a` | 54 words | capua, cathartica |
+| `-ri` | `-a` | 51 words | rivolta, ridotta |
+
+### 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 |
+|------|-----------------|------------|------|
+| incasciata | **`in-ca-scia-ta`** | 7.5 | `scia` |
+| mariteddu | **`ma-ri-teddu`** | 6.0 | `teddu` |
+| olivetani | **`olive-ta-ni`** | 6.0 | `olive` |
+| infattonu | **`in-fatto-nu`** | 6.0 | `fatto` |
+| indebulitu | **`in-debuli-tu`** | 6.0 | `debuli` |
+| cunvertuti | **`cu-nver-tu-ti`** | 4.5 | `nver` |
+| sustenenu | **`sustene-nu`** | 4.5 | `sustene` |
+| cunsultatu | **`cu-nsul-ta-tu`** | 4.5 | `nsul` |
+| dilimitatu | **`di-limi-ta-tu`** | 4.5 | `limi` |
+| reichardia | **`reichard-ia`** | 4.5 | `reichard` |
+| affissati | **`affissa-ti`** | 4.5 | `affissa` |
+| riabilità | **`ri-abilità`** | 4.5 | `abilità` |
+| siracusani | **`siracusa-ni`** | 4.5 | `siracusa` |
+| ripresenta | **`ri-pr-esen-ta`** | 4.5 | `esen` |
+| chjappani | **`chjappa-ni`** | 4.5 | `chjappa` |
+
+### 6.6 Linguistic Interpretation
+
+> **Automated Insight:**
+The language CO appears to be more isolating or has a highly fixed vocabulary. Word-level models perform nearly as well as subword models, indicating fewer productive morphological processes.
+
+---
+## 7. Summary & Recommendations
 
 ![Performance Dashboard](visualizations/performance_dashboard.png)
 
@@ -346,11 +519,12 @@ Below are text samples generated from each Markov chain model:
 
 | Component | Recommended | Rationale |
 |-----------|-------------|-----------|
-| Tokenizer | **32k BPE** | Best compression (4.11x) with low UNK rate |
-| N-gram | **5-gram** | Lowest perplexity (252) |
-| Markov | **Context-4** | Highest predictability (92.4%) |
+| Tokenizer | **64k BPE** | Best compression (4.20x) |
+| N-gram | **2-gram** | Lowest perplexity (221) |
+| Markov | **Context-4** | Highest predictability (93.8%) |
 | Embeddings | **100d** | Balanced semantic capture and isotropy |
 
+
 ---
 ## Appendix: Metrics Glossary & Interpretation Guide
 
@@ -540,7 +714,8 @@ If you use these models in your research, please cite:
   author = {Kamali, Omar},
   title = {Wikilangs: Open NLP Models for Wikipedia Languages},
   year = {2025},
-  publisher = {HuggingFace},
+  doi = {10.5281/zenodo.18073153},
+  publisher = {Zenodo},
   url = {https://huggingface.co/wikilangs}
   institution = {Omneity Labs}
 }
@@ -556,7 +731,8 @@ MIT License - Free for academic and commercial use.
 - 🤗 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: 2025-12-28 23:12:11*
+*Report Date: 2026-01-03 10:28:53*

models/embeddings/monolingual/co_128d.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2031bd1cfa760da5eac047841642409977af029608423421c2a3ddfc0e72065e
-size 1058229953
+oid sha256:96cf2c345e68270f8dcdbe99946da8e1091f69fe819c6b9a2158c5e48ed9e52c
+size 1056877904

models/embeddings/monolingual/co_128d_metadata.json

@@ -3,11 +3,13 @@
   "dimension": 128,
   "version": "monolingual",
   "training_params": {
-    "dim": 128,
+    "algorithm": "skipgram",
     "min_count": 5,
     "window": 5,
     "negative": 5,
-    "epochs": 5
+    "epochs": 5,
+    "encoding_method": "rope",
+    "dim": 128
   },
-  "vocab_size": 32860
+  "vocab_size": 31563
 }

models/embeddings/monolingual/co_32d.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0499edff9dff70386e0e580846c3317c14a5ccf69c933f22b5f575af2dc898e8
-size 264993473
+oid sha256:1067ad298ad58d7164db3f4db9081298ba2b534ee2da4cd72c1b7dd84f187b81
+size 264637520

models/embeddings/monolingual/co_32d_metadata.json

@@ -3,11 +3,13 @@
   "dimension": 32,
   "version": "monolingual",
   "training_params": {
-    "dim": 32,
+    "algorithm": "skipgram",
     "min_count": 5,
     "window": 5,
     "negative": 5,
-    "epochs": 5
+    "epochs": 5,
+    "encoding_method": "rope",
+    "dim": 32
   },
-  "vocab_size": 32860
+  "vocab_size": 31563
 }

models/embeddings/monolingual/co_64d.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5e093d5c0a9be69a973aad2e726911eaa455515d867b6d6363c1d6b142999ae1
-size 529405633
+oid sha256:c664f3164ab6a4eac104c4ea7ff6212ebaa4e5e194ba8f7a21e25ea1c0d4871a
+size 528717648

models/embeddings/monolingual/co_64d_metadata.json

@@ -3,11 +3,13 @@
   "dimension": 64,
   "version": "monolingual",
   "training_params": {
-    "dim": 64,
+    "algorithm": "skipgram",
     "min_count": 5,
     "window": 5,
     "negative": 5,
-    "epochs": 5
+    "epochs": 5,
+    "encoding_method": "rope",
+    "dim": 64
   },
-  "vocab_size": 32860
+  "vocab_size": 31563
 }

models/subword_markov/co_markov_ctx1_subword.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:da6ee4e7b7a739f36ca48346a973fe93fbea5bd312654e88ed7d342a2bab158b
-size 79505
+oid sha256:e8b5e0c398d4f5b2969e1db1c671105a3ac5242a1d374d84d5b0069106642b3d
+size 65281

models/subword_markov/co_markov_ctx1_subword_metadata.json

@@ -2,6 +2,6 @@
   "context_size": 1,
   "variant": "subword",
   "language": "co",
-  "unique_contexts": 1338,
-  "total_transitions": 13921716
+  "unique_contexts": 976,
+  "total_transitions": 13219556
 }

models/subword_markov/co_markov_ctx2_subword.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0fc23d0cb39d2816f1c55c6b721f82a00000b47e3669e0f7ba67c349d8b6c7ad
-size 475061
+oid sha256:10d07b4e8c3757cc6b86c68087fcfea9af405d88517a6a0445f37de943be6b99
+size 363878

models/subword_markov/co_markov_ctx2_subword_metadata.json

@@ -2,6 +2,6 @@
   "context_size": 2,
   "variant": "subword",
   "language": "co",
-  "unique_contexts": 10183,
-  "total_transitions": 13913087
+  "unique_contexts": 8204,
+  "total_transitions": 13211059
 }

models/subword_markov/co_markov_ctx3_subword.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:01dda0faf7a4fd2527274349d4d0f2871938d54cd949ca563ff0d600aa16063a
-size 1796414
+oid sha256:ab0ac62add6f840e66c319753e1b42d3784d61773053b38c73b6b277df4cca3a
+size 1441137

models/subword_markov/co_markov_ctx3_subword_metadata.json

@@ -2,6 +2,6 @@
   "context_size": 3,
   "variant": "subword",
   "language": "co",
-  "unique_contexts": 57282,
-  "total_transitions": 13904458
+  "unique_contexts": 46007,
+  "total_transitions": 13202562
 }

models/subword_markov/co_markov_ctx4_subword.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d597e8410cb0bb1cdc86c793cb94c003b472854ba814995519e18d3efc5619ff
-size 5163799
+oid sha256:07c83362fcdbe265c5308c85c508b78deedca4f3c3c1ec1a96a44764934510ff
+size 4134190

models/subword_markov/co_markov_ctx4_subword_metadata.json

@@ -2,6 +2,6 @@
   "context_size": 4,
   "variant": "subword",
   "language": "co",
-  "unique_contexts": 232155,
-  "total_transitions": 13895829
+  "unique_contexts": 183668,
+  "total_transitions": 13194065
 }

models/subword_ngram/co_2gram_subword.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7c590fbecb8a6478958f0834c16ef9fa20994cd0d160f3be79aaf7b0a95db119
-size 49835
+oid sha256:899bc5e6584570580099e942da78aa713e4d3807f6ebea5a4ffd604efeea23f2
+size 42648

models/subword_ngram/co_2gram_subword_metadata.json

@@ -2,6 +2,6 @@
   "n": 2,
   "variant": "subword",
   "language": "co",
-  "unique_ngrams": 3720,
-  "total_ngrams": 13921716
+  "unique_ngrams": 3181,
+  "total_ngrams": 13219556
 }

models/subword_ngram/co_3gram_subword.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:58d1ad13e8ff8840d69a15f6a5270873cd07563a631a1a6116168bea8d4953c7
-size 345886
+oid sha256:7ed57d99baac180f6c586ad2f06d308b877fede514712ded41ed5e8989084810
+size 285900

models/subword_ngram/co_3gram_subword_metadata.json

@@ -2,6 +2,6 @@
   "n": 3,
   "variant": "subword",
   "language": "co",
-  "unique_ngrams": 27040,
-  "total_ngrams": 13913087
+  "unique_ngrams": 22404,
+  "total_ngrams": 13211059
 }

models/subword_ngram/co_4gram_subword.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ca7fb34ddff62fdaa0ea34b3527d9bc3d9b9e1b086724ccb1d057018c30e16bb
-size 1514811
+oid sha256:094c924c5f844afd2646ce2d24dd1cfc9060e5723e944365817cf9c12106c248
+size 1233900

models/subword_ngram/co_4gram_subword_metadata.json

@@ -2,6 +2,6 @@
   "n": 4,
   "variant": "subword",
   "language": "co",
-  "unique_ngrams": 132259,
-  "total_ngrams": 13904458
+  "unique_ngrams": 107042,
+  "total_ngrams": 13202562
 }

models/tokenizer/co_tokenizer_16k.model

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:35b62b7e9ff3a1d060192c00db2891e65de8eef6a5d8e81d9b96f9731f075306
-size 514115
+oid sha256:b34f2ba9c5b244ddd93aba34ada7fef4711b8397b21d33fd688d1c317cc7cb41
+size 516047

models/tokenizer/co_tokenizer_32k.model

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:09c59715681a237c025f2a6b31f3e716dd585e0e9754482be80ccac672433557
-size 802041
+oid sha256:f6ae7a171bd1aba13305f06c6f0f2b827daf529b4e36eed333e8c0c56302860c
+size 803640

models/tokenizer/co_tokenizer_64k.model

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a730b87621f6700258f75df8ae98e69b637aa8c1505c33f5fbdf6521adf511c5
-size 1375688
+oid sha256:17214768ca1dd2e6a0c00363e0983130aeccf918e9ce567d882f1537de050491
+size 1383670

models/tokenizer/co_tokenizer_8k.model

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7b49d54c08ab0b110c863d83bafc526ea2f48c68226f21a65baeac02c840810d
-size 375284
+oid sha256:73dc30b9584b2f1bd0dec231fe033922d1286e98bf5022421f69b9b07cc653e2
+size 376505

models/vocabulary/co_vocabulary.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f4c6a7878646d33830ca0828c2e640504c8e9f8067222c4ff73a98a9b546a9f8
-size 1046993
+oid sha256:bf8b269f53cdfcf4bafb12ed650edcd581ad571e89880ac7222ad9b23086308c
+size 1013491

models/vocabulary/co_vocabulary_metadata.json

@@ -1,16 +1,17 @@
 {
   "language": "co",
-  "vocabulary_size": 60410,
+  "vocabulary_size": 58612,
+  "variant": "full",
   "statistics": {
-    "type_token_ratio": 0.05505110462735727,
+    "type_token_ratio": 0.05464431503278916,
     "coverage": {
-      "top_100": 0.4675895285944704,
-      "top_1000": 0.6685259531847793,
-      "top_5000": 0.8115519548689529,
-      "top_10000": 0.8655231968740638
+      "top_100": 0.4740403851170396,
+      "top_1000": 0.6748029483890198,
+      "top_5000": 0.8154347093397393,
+      "top_10000": 0.8683801088705445
     },
-    "hapax_count": 67855,
-    "hapax_ratio": 0.5290219467508673,
-    "total_documents": 8629
+    "hapax_count": 64770,
+    "hapax_ratio": 0.5249550177497528,
+    "total_documents": 8497
   }
 }

models/word_markov/co_markov_ctx1_word.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6949f9b425e3be961b19dd9a0c1f58c5696e86840cf4d9308fe0868394eb5496
-size 6003594
+oid sha256:eb1cae5d79a7a73adf2ff0919afeebd9f38f7943e4b16f5e68a2a24aa799f516
+size 5678726

models/word_markov/co_markov_ctx1_word_metadata.json

@@ -2,6 +2,6 @@
   "context_size": 1,
   "variant": "word",
   "language": "co",
-  "unique_contexts": 128497,
-  "total_transitions": 2816255
+  "unique_contexts": 123267,
+  "total_transitions": 2249414
 }

models/word_markov/co_markov_ctx2_word.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e93a9113eeda44bf0b2bb164f25bd6eacb2b758db0e5e0548020ff9deb32da36
-size 14617480
+oid sha256:793e28fa70d170a16ec64307b53718d88a3f178234de84df35691f3c269d0ab2
+size 14184219

models/word_markov/co_markov_ctx2_word_metadata.json

@@ -2,6 +2,6 @@
   "context_size": 2,
   "variant": "word",
   "language": "co",
-  "unique_contexts": 698773,
-  "total_transitions": 2807626
+  "unique_contexts": 688381,
+  "total_transitions": 2240917
 }

models/word_markov/co_markov_ctx3_word.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d7eec15299369b55c2bafb391ac7ba2d709806255c93867717fe9f5a27490303
-size 23131102
+oid sha256:3be679dc26e47d1c83affcb1f0bfa604d17eedd9e61c519e7edf6a807644fa78
+size 21498375

models/word_markov/co_markov_ctx3_word_metadata.json

@@ -2,6 +2,6 @@
   "context_size": 3,
   "variant": "word",
   "language": "co",
-  "unique_contexts": 1359938,
-  "total_transitions": 2798997
+  "unique_contexts": 1235287,
+  "total_transitions": 2232420
 }

models/word_markov/co_markov_ctx4_word.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2b81f07d48a5d5c30ced1ab8c79658be61832d09a3da8d653212846695af65ec
-size 29571363
+oid sha256:45fc9f862083178d82015d22958b5220424828a7df3f1ed4ce32c8a05e0a9a55
+size 26430133

models/word_markov/co_markov_ctx4_word_metadata.json

@@ -2,6 +2,6 @@
   "context_size": 4,
   "variant": "word",
   "language": "co",
-  "unique_contexts": 1780901,
-  "total_transitions": 2790371
+  "unique_contexts": 1541605,
+  "total_transitions": 2223923
 }

models/word_ngram/co_2gram_word.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d3890edf82d50cc4d4ba390f75ca19e8e7859cc5d9c555d49657f0223306edc8
-size 848542
+oid sha256:7aae2d0f40ab8a8875fecff95ea79de2f0b021415c2cd81ab367e3462b913e5f
+size 733819

models/word_ngram/co_2gram_word_metadata.json

@@ -2,6 +2,6 @@
   "n": 2,
   "variant": "word",
   "language": "co",
-  "unique_ngrams": 59847,
-  "total_ngrams": 2816255
+  "unique_ngrams": 49319,
+  "total_ngrams": 2249414
 }

models/word_ngram/co_3gram_word.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a3c7db65bf6666a3d88d972de2f776c3796d775b1ce984349b2a21497f9efa3c
-size 1557250
+oid sha256:678ddf4c171105837be92682f1c563d14c857c49ad8a221c632b94a4201f2155
+size 1263431

models/word_ngram/co_3gram_word_metadata.json

@@ -2,6 +2,6 @@
   "n": 3,
   "variant": "word",
   "language": "co",
-  "unique_ngrams": 108167,
-  "total_ngrams": 2807626
+  "unique_ngrams": 83012,
+  "total_ngrams": 2240917
 }

models/word_ngram/co_4gram_word.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c864a9125949dc68af0721ac2baa0a2c0f334474d9de22a25157676bfd24dc7f
-size 2832643
+oid sha256:d5ad2774dd632bee639c94f1e7595fb0df6d64aed7e2f4f57fc254ad95ba2d01
+size 2191498

models/word_ngram/co_4gram_word_metadata.json

@@ -2,6 +2,6 @@
   "n": 4,
   "variant": "word",
   "language": "co",
-  "unique_ngrams": 187731,
-  "total_ngrams": 2798997
+  "unique_ngrams": 136699,
+  "total_ngrams": 2232420
 }