language: ka
language_name: Georgian
language_family: kartvelian
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-kartvelian
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: 5.034
- name: best_isotropy
type: isotropy
value: 0.7869
- name: vocabulary_size
type: vocab
value: 0
generated: 2026-01-10T00:00:00.000Z
Georgian - Wikilangs Models
Comprehensive Research Report & Full Ablation Study
This repository contains NLP models trained and evaluated by Wikilangs, specifically on Georgian Wikipedia data. We analyze tokenizers, n-gram models, Markov chains, vocabulary statistics, and word embeddings.
๐ 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
Analysis and Evaluation
- 1. Tokenizer Evaluation
- 2. N-gram Model Evaluation
- 3. Markov Chain Evaluation
- 4. Vocabulary Analysis
- 5. Word Embeddings Evaluation
- 6. Morphological Analysis (Experimental)
- 7. Summary & Recommendations
- Metrics Glossary
- Visualizations Index
1. Tokenizer Evaluation
Results
| Vocab Size | Compression | Avg Token Len | UNK Rate | Total Tokens |
|---|---|---|---|---|
| 8k | 3.607x | 3.61 | 0.0743% | 1,129,507 |
| 16k | 4.126x | 4.13 | 0.0850% | 987,428 |
| 32k | 4.611x | 4.61 | 0.0950% | 883,432 |
| 64k | 5.034x ๐ | 5.04 | 0.1037% | 809,192 |
Tokenization Examples
Below are sample sentences tokenized with each vocabulary size:
Sample 1: แแ () โ แแแแ แ แแกแ แแ แแแฃแ แแแแฌแแ แแแแแจแ. แแ แแ แแก แแ แแแฃแแ แแแ แแแแขแ แแแ แแฃแแ แแแแแกแ. ...
| Vocab | Tokens | Count |
|---|---|---|
| 8k | โแแ โ() โโ โแแแแ แ โแแกแ โแแ แแ แฃแ โแแแแฌแแ แแแแแจแ . ... (+19 more) |
29 |
| 16k | โแแ โ() โโ โแแแแ แ โแแกแ โแแ แแแฃแ โแแแแฌแแ แแแแแจแ . โแแ ... (+16 more) |
26 |
| 32k | โแแ โ() โโ โแแแแ แ โแแกแ โแแ แแแฃแ โแแแแฌแแ แแแแแจแ . โแแ ... (+13 more) |
23 |
| 64k | โแแ โ() โโ โแแแแ แ โแแกแ โแแ แแแฃแ โแแแแฌแแ แแแแแจแ . โแแ โแแ แแก ... (+12 more) |
22 |
Sample 2: แฎแแแแแแฃแแแงแ () โ แกแแคแแแ แแแแ แแแแฏแแแจแ, แคแฃแแฃแแแก แ แแแแแจแ. แ แแกแฃแ แกแแแ แแแขแแ แแแขแจแ แ แแแ...
| Vocab | Tokens | Count |
|---|---|---|
| 8k | โแฎ แแแแ แ แฃแแ แง แ โ() โโ โแกแแคแแแ โแแแแ แแแแฏแแแจแ ... (+10 more) |
20 |
| 16k | โแฎ แแแแ แ แฃแแ แงแ โ() โโ โแกแแคแแแ โแแแแ แแแแฏแแแจแ , ... (+9 more) |
19 |
| 32k | โแฎ แแแแ แแฃแแ แงแ โ() โโ โแกแแคแแแ โแแแแ แแแแฏแแแจแ , โแคแฃแแฃแแแก ... (+6 more) |
16 |
| 64k | โแฎ แแแแ แแฃแแแงแ โ() โโ โแกแแคแแแ โแแแแ แแแแฏแแแจแ , โแคแฃแแฃแแแก โแ แแแแแจแ ... (+5 more) |
15 |
Sample 3: แแแ แแแแแ () โ แกแแคแแแ แแแแ แแแแฏแแแจแ, แคแฃแแฃแแแก แ แแแแแจแ. แ แแกแฃแ แกแแแ แแแขแแ แแแขแจแ แ แแแแแแก...
| Vocab | Tokens | Count |
|---|---|---|
| 8k | โแแแ แแแ แแ โ() โโ โแกแแคแแแ โแแแแ แแแแฏแแแจแ , โแค แฃแ ... (+7 more) |
17 |
| 16k | โแแแ แแแ แแ โ() โโ โแกแแคแแแ โแแแแ แแแแฏแแแจแ , โแค แฃแ ... (+7 more) |
17 |
| 32k | โแแแ แแแ แแ โ() โโ โแกแแคแแแ โแแแแ แแแแฏแแแจแ , โแคแฃแแฃแแแก โแ แแแแแจแ ... (+5 more) |
15 |
| 64k | โแแแ แแแ แแ โ() โโ โแกแแคแแแ โแแแแ แแแแฏแแแจแ , โแคแฃแแฃแแแก โแ แแแแแจแ ... (+5 more) |
15 |
Key Findings
- Best Compression: 64k achieves 5.034x compression
- Lowest UNK Rate: 8k with 0.0743% 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
Results
| N-gram | Variant | Perplexity | Entropy | Unique N-grams | Top-100 Coverage | Top-1000 Coverage |
|---|---|---|---|---|---|---|
| 2-gram | Word | 211,247 | 17.69 | 839,931 | 5.0% | 14.8% |
| 2-gram | Subword | 423 ๐ | 8.73 | 18,881 | 58.4% | 96.6% |
| 3-gram | Word | 297,944 | 18.18 | 972,504 | 4.8% | 13.7% |
| 3-gram | Subword | 3,918 | 11.94 | 159,884 | 21.8% | 60.3% |
| 4-gram | Word | 509,839 | 18.96 | 1,548,052 | 4.2% | 12.7% |
| 4-gram | Subword | 22,811 | 14.48 | 936,455 | 10.0% | 32.5% |
| 5-gram | Word | 359,649 | 18.46 | 1,104,097 | 4.9% | 14.4% |
| 5-gram | Subword | 88,814 | 16.44 | 2,995,844 | 5.2% | 19.7% |
Top 5 N-grams by Size
2-grams (Word):
| Rank | N-gram | Count |
|---|---|---|
| 1 | แ แแกแฃแ แกแแแ แแแขแแ แแแขแจแ |
100,310 |
| 2 | แแฎแแแแ แแแ แแแแ |
38,767 |
| 3 | แแ แ แแ แแ |
38,192 |
| 4 | แแ แกแฎแแ |
20,983 |
| 5 | of the |
20,829 |
3-grams (Word):
| Rank | N-gram | Count |
|---|---|---|
| 1 | แฌแแแก แแแแแชแแแแแแ แแแกแแฎแแแแแ |
9,844 |
| 2 | แ แแกแฃแ แกแแแ แแแขแแ แแแขแจแ แกแฅแแแแ |
8,451 |
| 3 | แแแฃแฎแแแแแแ แแแแกแ แ แแ |
8,239 |
| 4 | แแคแแชแแแแฃแ แ แกแแแขแ แกแฅแแแแ |
7,888 |
| 5 | แแแแแแ แแแแก แแฆแแแก แแแแแแแ |
7,694 |
4-grams (Word):
| Rank | N-gram | Count |
|---|---|---|
| 1 | แแแชแ แฌ แแฎแแแแ แแแ แแแแ |
6,676 |
| 2 | แแฃแแแชแแแแแแขแแขแแก แแคแแชแแแแฃแ แ แกแแแขแ แกแฅแแแแ |
5,942 |
| 3 | แแคแแชแแแแฃแ แ แกแแแขแ แกแฅแแแแ แแฃแแแชแแแแแแขแแขแแแ |
5,878 |
| 4 | แแ แแ แแฎแแแแ แแแ แแแแ |
5,763 |
| 5 | แญแแแแญแแแแแกแแแ แแ แแ แ แแ แแ แแแแ แกแแฎแแแแ |
5,614 |
5-grams (Word):
| Rank | N-gram | Count |
|---|---|---|
| 1 | แแฃแแแชแแแแแแขแแขแแก แแคแแชแแแแฃแ แ แกแแแขแ แกแฅแแแแ แแฃแแแชแแแแแแขแแขแแแ |
5,860 |
| 2 | แแแแกแแก แญแแแแญแแแแแกแแแ แแ แแ แ แแ แแ แแแแ แกแแฎแแแแ |
5,614 |
| 3 | แแฌแแ แแ แแแแกแแก แญแแแแญแแแแแกแแแ แแ แแ แ แแ แแ |
5,614 |
| 4 | แคแแฎแกแแฎแกแ แแแแแ แขแแแแก แแฌแแ แแ แแแแกแแก แญแแแแญแแแแแกแแแ แแ |
5,614 |
| 5 | แขแแแแก แแฌแแ แแ แแแแกแแก แญแแแแญแแแแแกแแแ แแ แแ แ |
5,614 |
2-grams (Subword):
| Rank | N-gram | Count |
|---|---|---|
| 1 | แก _ |
7,040,665 |
| 2 | แ แก |
7,006,427 |
| 3 | แ _ |
6,580,536 |
| 4 | แ _ |
4,984,758 |
| 5 | แ แ |
4,979,269 |
3-grams (Subword):
| Rank | N-gram | Count |
|---|---|---|
| 1 | แ แก _ |
5,017,584 |
| 2 | แ แ แ |
2,375,826 |
| 3 | _ แ แ |
2,051,321 |
| 4 | _ แก แ |
1,739,078 |
| 5 | แ แ _ |
1,650,928 |
4-grams (Subword):
| Rank | N-gram | Count |
|---|---|---|
| 1 | _ แ แ _ |
1,089,218 |
| 2 | แ แ แก _ |
984,947 |
| 3 | แ แ แ แก |
883,610 |
| 4 | แ แ แ _ |
738,436 |
| 5 | แ แก _ แ |
733,039 |
5-grams (Subword):
| Rank | N-gram | Count |
|---|---|---|
| 1 | แ แ แ แก _ |
734,333 |
| 2 | แ _ แ แ _ |
430,206 |
| 3 | , _ แ แ แ |
406,399 |
| 4 | แ แก _ แก แ |
350,334 |
| 5 | แ แ แฃ แ แ |
307,583 |
Key Findings
- Best Perplexity: 2-gram (subword) with 423
- Entropy Trend: Decreases with larger n-grams (more predictable)
- Coverage: Top-1000 patterns cover ~20% of corpus
- Recommendation: 4-gram or 5-gram for best predictive performance
3. Markov Chain Evaluation
Results
| Context | Variant | Avg Entropy | Perplexity | Branching Factor | Unique Contexts | Predictability |
|---|---|---|---|---|---|---|
| 1 | Word | 0.9133 | 1.883 | 9.96 | 1,644,802 | 8.7% |
| 1 | Subword | 1.0685 | 2.097 | 7.59 | 8,326 | 0.0% |
| 2 | Word | 0.2778 | 1.212 | 1.75 | 16,356,120 | 72.2% |
| 2 | Subword | 0.7948 | 1.735 | 5.53 | 63,087 | 20.5% |
| 3 | Word | 0.0801 | 1.057 | 1.15 | 28,502,543 | 92.0% |
| 3 | Subword | 0.7992 | 1.740 | 4.61 | 348,410 | 20.1% |
| 4 | Word | 0.0282 ๐ | 1.020 | 1.04 | 32,622,645 | 97.2% |
| 4 | Subword | 0.7081 | 1.634 | 3.48 | 1,604,353 | 29.2% |
Generated Text Samples (Word-based)
Below are text samples generated from each word-based Markov chain model:
Context Size 1:
แแ แคแแกแขแแแแแแก roadburn festival at war chapter 8 6 7 489 แฎแแแแแแแแแจแ แแแฌแแคแฃแแ แแ แแแกแแแแฃแ แ แแแ แแฃแแแฌแแแก แแแ แแแจแ แแฃแ แแแ modern philology v แกแแฃแแฃแแแแแแ แซแแแแ แแแขแแ แขแแ แแแแกแแแ แแก แแจแแแแแแแก แแแแแขแแแชแแแแจแ ...แฌแแแก แขแแซแ แแก แแ แแแแขแ แชแแแแแขแฅแแแจแแก แฎแกแแแ แแก แกแแแแแ แแแ 0 1 แแแแแแแ แ แแแแแแแแ แ แแฃแแแชแแแแแแขแแขแจแ แญแแแแก แแฃแฎแ...
Context Size 2:
แ แแกแฃแ แกแแแ แแแขแแ แแแขแจแ แแคแแชแแแแฃแ แ แกแแแขแ แกแฅแแแแ แแฃแแแชแแแแแแขแแขแแแ แฅแแแแฅแแแ แแ แฌแแแก แแก แแแแฌแแแแก แแแขแแ แแฃแ แ...แแฎแแแแ แแแ แแแแ แแแแแก แแแ แแแกแ แแ แแแแแแแก แจแขแแขแแแ แ แแกแฃแ แกแแแ แแแขแแ แแแขแจแ แแคแแชแแแแฃแ แ แกแแแขแ แฐแแ แแแ แ แแฃแกแแแแก...แแ แ แแ แแ แแแแชแแแขแแ แ แกแฃแแ แแ แแ แแ แแแแแ แชแแฎแแกแแแแแ แแ แจแแฌแงแแแขแ แแฎแแแแ แแแจแแ แแฅแแแแแแแแ แแแ แแฃ แแฏแแฎแก แกแแ...
Context Size 3:
แฌแแแก แแแแแชแแแแแแ แแแกแแฎแแแแแ 84 469 แแแแแแแแก แจแแแแแแแแ แคแแ แแแแ 358 แแ แแแกแแฎแแแแแ 61 418 แแแแแแแแ แฌแแแก...แแแฃแฎแแแแแแ แแแแกแ แ แแ แแแ แแแแแแแแแแขแฃแ แแ แแแ แแแฃแ แแแแ แแแแแแแแแแขแแแ แจแแแซแแแแ แแแแแแแงแแแแ แกแแแแขแแแแแแก ...แ แแกแฃแ แกแแแ แแแขแแ แแแขแจแ แกแฅแแแแ แแแแฆแแ แแแแ 25 แแแแแกแขแ records แแก แจแแแกแ แฃแแแแแแแ records แแก แจแแแกแ แฃแแแแแแแ ...
Context Size 4:
แแแชแ แฌ แแฎแแแแ แแแ แแแแ แกแแแแแแแก แแฃแแแชแแแแแแขแแขแแแ แ แแกแฃแ แกแแแ แแแขแแ แแแขแจแ แแแแแแกแแก แแฃแแแชแแแแแแขแแขแแก แแคแแชแแ...แแฃแแแชแแแแแแขแแขแแก แแคแแชแแแแฃแ แ แกแแแขแ แกแฅแแแแ แแฃแแแชแแแแแแขแแขแแแ แฅแแแแฅแแแ แแแแ แแแแ แแแแ แแก แจแแแชแแแแ แกแแแแแ แแ...แแ แแ แแฎแแแแ แแแ แแแแ แแแแฃแแแแแก แฅแแแแฅแแแแก แกแแ แ แแกแฃแ แกแแแ แแแขแแ แแแขแจแ แฅแแแแฅแแก แแแแแ แแแแก แกแแแขแ แฅแแแแฅแแก แกแ...
Generated Text Samples (Subword-based)
Below are text samples generated from each subword-based Markov chain model:
Context Size 1:
_แแแแแ;_แฌแแแแแก_แแ_แแแก_แแแแกแแฆแแแฃแแแกแ_แช_แแแฐแแก_แกแขแฉแ
Context Size 2:
แก_แแแแ แแแแแ แแแแ แแกแแก_แแฅแแแแแแก_แแแแ_แ_แแ แแแแแแ,_nalos
Context Size 3:
แแก_แฏแแ แแ_แแแจแ,_แ แแแแแแก_แฌแแ แแแแแก_แแ แแ_แแแแฅแชแ._แแแแแ แแแแ
Context Size 4:
_แแ_แแ_แแ แแแ._แแแแแจแแแก_แกแแญแแ แ._แฌแแแก._แฎแแแแกแแแ_แกแแ แแแก_แฌแงแแ
Key Findings
- Best Predictability: Context-4 (word) with 97.2% predictability
- Branching Factor: Decreases with context size (more deterministic)
- Memory Trade-off: Larger contexts require more storage (1,604,353 contexts)
- Recommendation: Context-3 or Context-4 for text generation
4. Vocabulary Analysis
Statistics
| Metric | Value |
|---|---|
| Vocabulary Size | 715,127 |
| Total Tokens | 37,347,310 |
| Mean Frequency | 52.22 |
| Median Frequency | 4 |
| Frequency Std Dev | 1624.79 |
Most Common Words
| Rank | Word | Frequency |
|---|---|---|
| 1 | แแ | 1,097,500 |
| 2 | แฌแแแก | 288,743 |
| 3 | แฌแแแก | 254,559 |
| 4 | แแงแ | 188,838 |
| 5 | แแก | 162,453 |
| 6 | แ แแแแแแช | 141,611 |
| 7 | the | 128,418 |
| 8 | แ แแ | 124,976 |
| 9 | 1 | 122,280 |
| 10 | แแแกแ | 118,080 |
Least Common Words (from vocabulary)
| Rank | Word | Frequency |
|---|---|---|
| 1 | pbsuccess | 2 |
| 2 | แแแแแฃแกแ | 2 |
| 3 | แแแ แแฐแแแ | 2 |
| 4 | แ แแฉแแแแก | 2 |
| 5 | peig | 2 |
| 6 | แแแแกแแก | 2 |
| 7 | smap | 2 |
| 8 | แแแแแขแแก | 2 |
| 9 | แแแแแแแแแแก | 2 |
| 10 | แแแแแ แแก | 2 |
Zipf's Law Analysis
| Metric | Value |
|---|---|
| Zipf Coefficient | 0.9374 |
| Rยฒ (Goodness of Fit) | 0.993145 |
| Adherence Quality | excellent |
Coverage Analysis
| Top N Words | Coverage |
|---|---|
| Top 100 | 19.3% |
| Top 1,000 | 42.2% |
| Top 5,000 | 62.0% |
| Top 10,000 | 70.3% |
Key Findings
- Zipf Compliance: Rยฒ=0.9931 indicates excellent adherence to Zipf's law
- High Frequency Dominance: Top 100 words cover 19.3% of corpus
- Long Tail: 705,127 words needed for remaining 29.7% coverage
5. Word Embeddings Evaluation
5.1 Cross-Lingual Alignment
5.2 Model Comparison
| Model | Dimension | Isotropy | Semantic Density | Alignment R@1 | Alignment R@10 |
|---|---|---|---|---|---|
| mono_32d | 32 | 0.7869 | 0.3479 | N/A | N/A |
| mono_64d | 64 | 0.7389 | 0.2989 | N/A | N/A |
| mono_128d | 128 | 0.6243 | 0.2604 | N/A | N/A |
| aligned_32d | 32 | 0.7869 ๐ | 0.3607 | 0.1100 | 0.4300 |
| aligned_64d | 64 | 0.7389 | 0.3031 | 0.2300 | 0.6300 |
| aligned_128d | 128 | 0.6243 | 0.2636 | 0.2980 | 0.7120 |
Key Findings
- Best Isotropy: aligned_32d with 0.7869 (more uniform distribution)
- Semantic Density: Average pairwise similarity of 0.3057. Lower values indicate better semantic separation.
- Alignment Quality: Aligned models achieve up to 29.8% R@1 in cross-lingual retrieval.
- Recommendation: 128d aligned for best cross-lingual performance
6. Morphological Analysis (Experimental)
This section presents an automated morphological analysis derived from the statistical divergence between word-level and subword-level models. By analyzing where subword predictability spikes and where word-level coverage fails, we can infer linguistic structures without supervised data.
6.1 Productivity & Complexity
| Metric | Value | Interpretation | Recommendation |
|---|---|---|---|
| Productivity Index | 5.000 | High morphological productivity | Reliable analysis |
| Idiomaticity Gap | -0.271 | 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 |
|---|---|
-แกแ |
แกแแขแแแแแจแแ แแกแ, แกแแแแแซแฃแ แแกแ, แกแแแแ แแแแแแ |
-แ |
แแแขแแแแขแแแแ แแแฃแ, แแแ แแขแแฅแแแแแก, แแชแแฃแ แแก |
-แแ |
แแแแแแแแแแแแแแแแ, แแแแแชแแแแแแแแแแ, แแแแฎแ แแแแกแแแแก |
-แแ |
แแแแแแแแแแ, แแแแ แซแแแฃแ, แแแแงแ แแแแ |
-แ |
แแแแแแจแแ แแแแแแ, แแแแ แแแแแกแก, แแแแแจแแแแแ |
-แแ |
แแแแ แแแแแกแก, แแแแแจแแแแแ, แแแ แฏแแแแจแแแแ |
-แแ |
แแแแแแจแแ แแแแแแ, แแแแฅแแแแ, แแแแแแชแแแแแแแแ |
-แก |
แกแแขแแแแแจแแ แแกแ, แกแแแแแซแฃแ แแกแ, แกแแแแ แแแแแแ |
Productive Suffixes
| Suffix | Examples |
|---|---|
-แก |
แแ แแแแก, แแแแ แแแแแกแก, แแ แแฉแแก |
-แ |
แแแแแจแแแกแแ, แแแแแแแแแ แจแ, แกแแแแ แแแแแแ |
-แแก |
แแแ แแแแแก, แแแกแแ แแก, แงแฃแกแแ แแก |
-แ |
แแแ แฅแแแกแแ, แแแแแแแแชแแแแกแ, แกแแแแแซแฃแ แแกแ |
-แแ |
แแแแแแแแแ, แแ แแแแ, แแ แแขแแแแแแแแกแแแแ |
-แ |
แแแแแแแแแ, แแ แแแแ, แแ แแขแแแแแแแแกแแแแ |
-แแ |
แแแ แฏแแแแจแแแแ, แฉแแแแจแแ แแแฃแแ, แแแแงแ แแแแ |
-แกแ |
แแแแแแแแชแแแแกแ, แกแแแแแซแฃแ แแกแ, แขแงแแแแแแกแ |
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 |
|---|---|---|---|
แแแแก |
1.85x | 194 contexts | แฉแแแแก, แจแแแแก, แแแแแก |
แแแแ |
1.48x | 550 contexts | แแแแแ, แแแแแ, แฏแแแแ |
แแแแ |
2.33x | 52 contexts | แแแแแ, แจแแแแแ, แแแแแแ |
แแแแ |
1.41x | 544 contexts | แแแแแ, แแแแแ, แแแแแก |
แแแฃแ |
1.59x | 250 contexts | แฅแแแฃแ, แแ แแแฃแ, แฅแแแฃแแ |
แแแแ |
1.75x | 134 contexts | แแแแแแก, แแแแแแ, แฃแแแแแก |
แแ แแ |
1.68x | 147 contexts | แฅแแ แแ, แแแ แแแ, แฉแแ แแแ |
แ แแแ |
1.81x | 96 contexts | แแแ แแแ, แ แแแแแก, แฅแแ แแแ |
แแแแ |
1.63x | 118 contexts | แขแแแแ, แ แแแแ, แแแแแแ |
แแขแแ |
1.51x | 148 contexts | แฃแแขแแ , แแแขแแ , แแแขแแ |
แแแแ |
1.43x | 180 contexts | แกแแแแ, แแแแแ, แแแแแ |
แแแแฅ |
1.58x | 106 contexts | แฅแแแแฅ, แแแแแฅ, แแแแแฅแก |
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 |
|---|---|---|---|
-แ |
-แก |
186 words | แแแแแขแแแฃแ แแแแก, แแแกแฃแแแจแแแแก |
-แ |
-แ |
167 words | แแฎแแขแแ แแแจแ, แแแแแแฅแแแแแฃแ แ |
-แ |
-แ |
129 words | แแฆแแแแแแ, แแแแแแแแกแแแแแ |
-แ |
-แ |
124 words | แแแขแแแแฃแ แแจแ, แแขแแแแ |
-แ |
-แก |
120 words | แแแแแก, แแแแแ แกแแแแก |
-แแ |
-แ |
111 words | แแแแแแแแแ แ, แแแแ แแขแแแแแ |
-แ |
-แแก |
104 words | แแแแแขแแแฃแ แแแแก, แแฃแกแแแแแแแก |
-แ |
-แ |
93 words | แแแ แแแแแชแแ, แแแฃแแแแ |
-แก |
-แก |
88 words | แกแแกแแแแก, แกแแแแ แแก |
-แก |
-แ |
82 words | แกแแคแ แแฎแแจแ, แกแขแแ แแแกแแแแฃแแ |
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 |
|---|---|---|---|
| แแแแแ แแแแกแ | แแแแแ แแแ-แก-แ |
7.5 | แก |
| แแแแแแขแ แแแกแ | แแแแแแขแ แแ-แก-แ |
7.5 | แก |
| แแแแแแแขแแ แแแแ | แแแแแแแขแแ แแ-แ-แ |
7.5 | แ |
| แแแ แแแฎแแแแแ | แแแ แแแฎแแแ-แ-แ |
7.5 | แ |
| แแแชแแแแแแแแแกแ | แแแชแแแแแแแแ-แก-แ |
7.5 | แก |
| แแแแแฎแแแแแแ | แแแแแฎแแแ-แ-แแ |
7.5 | แ |
| แแแขแงแแแแแแแจแ | แแแขแงแแแแแ-แ-แจแ |
7.5 | แ |
| แแ แแกแขแแแ แแขแแ | แแ แแกแขแแแ แแข-แ-แ |
7.5 | แ |
| แแแแแฃแฎแแแแแแแก | แแแแแฃแฎแแแแแ-แ-แก |
7.5 | แ |
| ะฟัะพะฒะธะฝัะธะธแแกแ | ะฟัะพะฒะธะฝัะธะธแ-แก-แ |
7.5 | แก |
| แซแแแแแแกแแช | แซแแแแแ-แก-แแช |
7.5 | แก |
| แแแแคแแกแแแชแแแกแ | แแแแคแแกแแแชแแ-แก-แ |
7.5 | แก |
| แแ แแแแขแแแแจแ | แแ แแแแขแแ-แ-แจแ |
7.5 | แ |
| แแแแแแ แแกแขแแแ | แแแแแแ แแกแข-แ-แแ |
7.5 | แ |
| แฐแแแแแแแแก | แฐแแแแแแ-แ-แก |
7.5 | แ |
6.6 Linguistic Interpretation
Automated Insight: The language Georgian shows high morphological productivity. The subword models are significantly more efficient than word models, suggesting a rich system of affixation or compounding.
7. Summary & Recommendations
Production Recommendations
| Component | Recommended | Rationale |
|---|---|---|
| Tokenizer | 64k BPE | Best compression (5.03x) |
| N-gram | 2-gram | Lowest perplexity (423) |
| Markov | Context-4 | Highest predictability (97.2%) |
| 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) ร 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.
Rยฒ (Coefficient of Determination)
Definition: Measures how well the linear fit explains the frequency-rank relationship. Ranges from 0 to 1.
Intuition: Rยฒ near 1.0 means the data closely follows Zipf's law; lower values indicate deviation from expected word frequency patterns.
What to seek: Rยฒ > 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
- Compare within model families: Metrics are most meaningful when comparing models of the same type (e.g., 8k vs 64k tokenizer).
- Consider trade-offs: Better performance on one metric often comes at the cost of another (e.g., compression vs. OOV rate).
- Context matters: Optimal values depend on downstream tasks. Text generation may prioritize different metrics than classification.
- Corpus influence: All metrics are influenced by corpus characteristics. Wikipedia text differs from social media or literature.
- 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 - a monthly snapshot of Wikipedia articles across 300+ languages.
Project
A project by Wikilangs - Open-source NLP models for every Wikipedia language.
Maintainer
Citation
If you use these models in your research, please cite:
@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
- ๐ Website: wikilangs.org
- ๐ค Models: huggingface.co/wikilangs
- ๐ Data: wikipedia-monthly
- ๐ค Author: Omar Kamali
- ๐ค Sponsor: Featherless AI
Generated by Wikilangs Models Pipeline
Report Date: 2026-01-10 11:10:47



















