Upload all models and assets for th (latest)
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- .gitattributes +7 -0
- README.md +773 -0
- models/embeddings/aligned/th_128d.bin +3 -0
- models/embeddings/aligned/th_128d.meta.json +1 -0
- models/embeddings/aligned/th_128d.projection.npy +3 -0
- models/embeddings/aligned/th_128d_metadata.json +8 -0
- models/embeddings/aligned/th_32d.bin +3 -0
- models/embeddings/aligned/th_32d.meta.json +1 -0
- models/embeddings/aligned/th_32d.projection.npy +3 -0
- models/embeddings/aligned/th_32d_metadata.json +8 -0
- models/embeddings/aligned/th_64d.bin +3 -0
- models/embeddings/aligned/th_64d.meta.json +1 -0
- models/embeddings/aligned/th_64d.projection.npy +3 -0
- models/embeddings/aligned/th_64d_metadata.json +8 -0
- models/embeddings/monolingual/th_128d.bin +3 -0
- models/embeddings/monolingual/th_128d.meta.json +1 -0
- models/embeddings/monolingual/th_128d_metadata.json +16 -0
- models/embeddings/monolingual/th_32d.bin +3 -0
- models/embeddings/monolingual/th_32d.meta.json +1 -0
- models/embeddings/monolingual/th_32d_metadata.json +16 -0
- models/embeddings/monolingual/th_64d.bin +3 -0
- models/embeddings/monolingual/th_64d.meta.json +1 -0
- models/embeddings/monolingual/th_64d_metadata.json +16 -0
- models/subword_markov/th_markov_ctx1_subword.parquet +3 -0
- models/subword_markov/th_markov_ctx1_subword_metadata.json +7 -0
- models/subword_markov/th_markov_ctx2_subword.parquet +3 -0
- models/subword_markov/th_markov_ctx2_subword_metadata.json +7 -0
- models/subword_markov/th_markov_ctx3_subword.parquet +3 -0
- models/subword_markov/th_markov_ctx3_subword_metadata.json +7 -0
- models/subword_markov/th_markov_ctx4_subword.parquet +3 -0
- models/subword_markov/th_markov_ctx4_subword_metadata.json +7 -0
- models/subword_ngram/th_2gram_subword.parquet +3 -0
- models/subword_ngram/th_2gram_subword_metadata.json +7 -0
- models/subword_ngram/th_3gram_subword.parquet +3 -0
- models/subword_ngram/th_3gram_subword_metadata.json +7 -0
- models/subword_ngram/th_4gram_subword.parquet +3 -0
- models/subword_ngram/th_4gram_subword_metadata.json +7 -0
- models/subword_ngram/th_5gram_subword.parquet +3 -0
- models/subword_ngram/th_5gram_subword_metadata.json +7 -0
- models/tokenizer/th_tokenizer_16k.model +3 -0
- models/tokenizer/th_tokenizer_16k.vocab +0 -0
- models/tokenizer/th_tokenizer_32k.model +3 -0
- models/tokenizer/th_tokenizer_32k.vocab +0 -0
- models/tokenizer/th_tokenizer_64k.model +3 -0
- models/tokenizer/th_tokenizer_64k.vocab +0 -0
- models/tokenizer/th_tokenizer_8k.model +3 -0
- models/tokenizer/th_tokenizer_8k.vocab +0 -0
- models/vocabulary/th_vocabulary.parquet +3 -0
- models/vocabulary/th_vocabulary_metadata.json +17 -0
- models/vocabulary/th_vocabulary_top.parquet +3 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,10 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
visualizations/embedding_similarity.png filter=lfs diff=lfs merge=lfs -text
|
| 37 |
+
visualizations/embedding_tsne_multilingual.png filter=lfs diff=lfs merge=lfs -text
|
| 38 |
+
visualizations/performance_dashboard.png filter=lfs diff=lfs merge=lfs -text
|
| 39 |
+
visualizations/position_encoding_comparison.png filter=lfs diff=lfs merge=lfs -text
|
| 40 |
+
visualizations/tsne_sentences.png filter=lfs diff=lfs merge=lfs -text
|
| 41 |
+
visualizations/tsne_words.png filter=lfs diff=lfs merge=lfs -text
|
| 42 |
+
visualizations/zipf_law.png filter=lfs diff=lfs merge=lfs -text
|
README.md
ADDED
|
@@ -0,0 +1,773 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language: th
|
| 3 |
+
language_name: Thai
|
| 4 |
+
language_family: taikadai_southwestern
|
| 5 |
+
tags:
|
| 6 |
+
- wikilangs
|
| 7 |
+
- nlp
|
| 8 |
+
- tokenizer
|
| 9 |
+
- embeddings
|
| 10 |
+
- n-gram
|
| 11 |
+
- markov
|
| 12 |
+
- wikipedia
|
| 13 |
+
- feature-extraction
|
| 14 |
+
- sentence-similarity
|
| 15 |
+
- tokenization
|
| 16 |
+
- n-grams
|
| 17 |
+
- markov-chain
|
| 18 |
+
- text-mining
|
| 19 |
+
- fasttext
|
| 20 |
+
- babelvec
|
| 21 |
+
- vocabulous
|
| 22 |
+
- vocabulary
|
| 23 |
+
- monolingual
|
| 24 |
+
- family-taikadai_southwestern
|
| 25 |
+
license: mit
|
| 26 |
+
library_name: wikilangs
|
| 27 |
+
pipeline_tag: text-generation
|
| 28 |
+
datasets:
|
| 29 |
+
- omarkamali/wikipedia-monthly
|
| 30 |
+
dataset_info:
|
| 31 |
+
name: wikipedia-monthly
|
| 32 |
+
description: Monthly snapshots of Wikipedia articles across 300+ languages
|
| 33 |
+
metrics:
|
| 34 |
+
- name: best_compression_ratio
|
| 35 |
+
type: compression
|
| 36 |
+
value: 4.749
|
| 37 |
+
- name: best_isotropy
|
| 38 |
+
type: isotropy
|
| 39 |
+
value: 0.8475
|
| 40 |
+
- name: vocabulary_size
|
| 41 |
+
type: vocab
|
| 42 |
+
value: 0
|
| 43 |
+
generated: 2026-01-17
|
| 44 |
+
---
|
| 45 |
+
|
| 46 |
+
# Thai - Wikilangs Models
|
| 47 |
+
## Comprehensive Research Report & Full Ablation Study
|
| 48 |
+
|
| 49 |
+
This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Thai** Wikipedia data.
|
| 50 |
+
We analyze tokenizers, n-gram models, Markov chains, vocabulary statistics, and word embeddings.
|
| 51 |
+
|
| 52 |
+
## 📋 Repository Contents
|
| 53 |
+
|
| 54 |
+
### Models & Assets
|
| 55 |
+
|
| 56 |
+
- Tokenizers (8k, 16k, 32k, 64k)
|
| 57 |
+
- N-gram models (2, 3, 4, 5-gram)
|
| 58 |
+
- Markov chains (context of 1, 2, 3, 4 and 5)
|
| 59 |
+
- Subword N-gram and Markov chains
|
| 60 |
+
- Embeddings in various sizes and dimensions (aligned and unaligned)
|
| 61 |
+
- Language Vocabulary
|
| 62 |
+
- Language Statistics
|
| 63 |
+
|
| 64 |
+

|
| 65 |
+
|
| 66 |
+
### Analysis and Evaluation
|
| 67 |
+
|
| 68 |
+
- [1. Tokenizer Evaluation](#1-tokenizer-evaluation)
|
| 69 |
+
- [2. N-gram Model Evaluation](#2-n-gram-model-evaluation)
|
| 70 |
+
- [3. Markov Chain Evaluation](#3-markov-chain-evaluation)
|
| 71 |
+
- [4. Vocabulary Analysis](#4-vocabulary-analysis)
|
| 72 |
+
- [5. Word Embeddings Evaluation](#5-word-embeddings-evaluation)
|
| 73 |
+
- [6. Morphological Analysis (Experimental)](#6--morphological-analysis-experimental)
|
| 74 |
+
- [7. Summary & Recommendations](#7-summary--recommendations)
|
| 75 |
+
- [Metrics Glossary](#appendix-metrics-glossary--interpretation-guide)
|
| 76 |
+
- [Visualizations Index](#visualizations-index)
|
| 77 |
+
|
| 78 |
+
---
|
| 79 |
+
## 1. Tokenizer Evaluation
|
| 80 |
+
|
| 81 |
+

|
| 82 |
+
|
| 83 |
+

|
| 84 |
+
|
| 85 |
+

|
| 86 |
+
|
| 87 |
+

|
| 88 |
+
|
| 89 |
+
### Results
|
| 90 |
+
|
| 91 |
+
| Vocab Size | Compression | Avg Token Len | UNK Rate | Total Tokens |
|
| 92 |
+
|------------|-------------|---------------|----------|--------------|
|
| 93 |
+
| **8k** | 3.339x | 3.36 | 0.1132% | 2,229,178 |
|
| 94 |
+
| **16k** | 3.862x | 3.88 | 0.1309% | 1,927,473 |
|
| 95 |
+
| **32k** | 4.323x | 4.35 | 0.1466% | 1,722,046 |
|
| 96 |
+
| **64k** | 4.749x 🏆 | 4.78 | 0.1610% | 1,567,500 |
|
| 97 |
+
|
| 98 |
+
### Tokenization Examples
|
| 99 |
+
|
| 100 |
+
Below are sample sentences tokenized with each vocabulary size:
|
| 101 |
+
|
| 102 |
+
**Sample 1:** `ไลฟ์การ์ดแห่งหาดบอนได เป็นสารคดีจากออสเตรเลียนำเสนอการทำงานตลอด 24 ชั่วโมงของไลฟ...`
|
| 103 |
+
|
| 104 |
+
| Vocab | Tokens | Count |
|
| 105 |
+
|-------|--------|-------|
|
| 106 |
+
| 8k | `▁ไล ฟ์ การ์ ด แห่ง หาด บอน ได ▁เป็น สาร ... (+24 more)` | 34 |
|
| 107 |
+
| 16k | `▁ไล ฟ์ การ์ด แห่ง หาด บอน ได ▁เป็นสาร คดี จาก ... (+19 more)` | 29 |
|
| 108 |
+
| 32k | `▁ไลฟ์ การ์ด แห่ง หาด บอน ได ▁เป็นสาร คดี จาก ออสเตร ... (+18 more)` | 28 |
|
| 109 |
+
| 64k | `▁ไลฟ์ การ์ด แห่ง หาด บอน ได ▁เป็นสาร คดี จาก ออสเตรเลียน ... (+17 more)` | 27 |
|
| 110 |
+
|
| 111 |
+
**Sample 2:** `32 อาจหมายถึง: 32 (ตัวเลข) 32 ก่อนคริสตศักราช, 32, และอื่นๆ 32 (เพลง) ,เพลงในปี ...`
|
| 112 |
+
|
| 113 |
+
| Vocab | Tokens | Count |
|
| 114 |
+
|-------|--------|-------|
|
| 115 |
+
| 8k | `▁ 3 2 ▁อาจหมายถึง : ▁ 3 2 ▁( ตัว ... (+28 more)` | 38 |
|
| 116 |
+
| 16k | `▁ 3 2 ▁อาจหมายถึง : ▁ 3 2 ▁( ตัวเลข ... (+27 more)` | 37 |
|
| 117 |
+
| 32k | `▁ 3 2 ▁อาจหมายถึง : ▁ 3 2 ▁( ตัวเลข ... (+25 more)` | 35 |
|
| 118 |
+
| 64k | `▁ 3 2 ▁อาจหมายถึง : ▁ 3 2 ▁( ตัวเลข ... (+24 more)` | 34 |
|
| 119 |
+
|
| 120 |
+
**Sample 3:** `Molopanthera เป็นสกุลของพืชดอกที่อยู่ในวงศ์ Rubiaceae. ถิ่นกำเนิดของมันคือ บราซิ...`
|
| 121 |
+
|
| 122 |
+
| Vocab | Tokens | Count |
|
| 123 |
+
|-------|--------|-------|
|
| 124 |
+
| 8k | `▁m ol op anth era ▁เป็นสกุลของ พืชดอก ที่อยู่ในวงศ์ ▁r ub ... (+24 more)` | 34 |
|
| 125 |
+
| 16k | `▁mol op anthera ▁เป็นสกุลของ พืชดอก ที่อยู่ในวงศ์ ▁rub iaceae . ▁ถิ่นกําเนิด ... (+17 more)` | 27 |
|
| 126 |
+
| 32k | `▁mol op anthera ▁เป็นสกุลของ พืชดอก ที่อยู่ในวงศ์ ▁rubiaceae . ▁ถิ่นกําเนิด ของมันคือ ... (+14 more)` | 24 |
|
| 127 |
+
| 64k | `▁mol op anthera ▁เป็นสกุลของ พืชดอก ที่อยู่ในวงศ์ ▁rubiaceae . ▁ถิ่นกําเนิด ของมันคือ ... (+14 more)` | 24 |
|
| 128 |
+
|
| 129 |
+
|
| 130 |
+
### Key Findings
|
| 131 |
+
|
| 132 |
+
- **Best Compression:** 64k achieves 4.749x compression
|
| 133 |
+
- **Lowest UNK Rate:** 8k with 0.1132% unknown tokens
|
| 134 |
+
- **Trade-off:** Larger vocabularies improve compression but increase model size
|
| 135 |
+
- **Recommendation:** 32k vocabulary provides optimal balance for production use
|
| 136 |
+
|
| 137 |
+
---
|
| 138 |
+
## 2. N-gram Model Evaluation
|
| 139 |
+
|
| 140 |
+

|
| 141 |
+
|
| 142 |
+

|
| 143 |
+
|
| 144 |
+

|
| 145 |
+
|
| 146 |
+
### Results
|
| 147 |
+
|
| 148 |
+
| N-gram | Variant | Perplexity | Entropy | Unique N-grams | Top-100 Coverage | Top-1000 Coverage |
|
| 149 |
+
|--------|---------|------------|---------|----------------|------------------|-------------------|
|
| 150 |
+
| **2-gram** | Word | 56,310 | 15.78 | 475,306 | 16.2% | 28.1% |
|
| 151 |
+
| **2-gram** | Subword | 2,438 🏆 | 11.25 | 124,885 | 27.9% | 71.1% |
|
| 152 |
+
| **3-gram** | Word | 160,871 | 17.30 | 713,993 | 10.6% | 19.4% |
|
| 153 |
+
| **3-gram** | Subword | 27,338 | 14.74 | 1,000,290 | 10.1% | 31.1% |
|
| 154 |
+
| **4-gram** | Word | 529,813 | 19.02 | 1,376,813 | 3.4% | 10.2% |
|
| 155 |
+
| **4-gram** | Subword | 174,441 | 17.41 | 4,905,540 | 5.4% | 17.4% |
|
| 156 |
+
| **5-gram** | Word | 577,241 | 19.14 | 1,093,587 | 2.6% | 7.1% |
|
| 157 |
+
| **5-gram** | Subword | 676,357 | 19.37 | 11,885,834 | 3.2% | 11.3% |
|
| 158 |
+
|
| 159 |
+
### Top 5 N-grams by Size
|
| 160 |
+
|
| 161 |
+
**2-grams (Word):**
|
| 162 |
+
|
| 163 |
+
| Rank | N-gram | Count |
|
| 164 |
+
|------|--------|-------|
|
| 165 |
+
| 1 | `พ ศ` | 586,670 |
|
| 166 |
+
| 2 | `ค ศ` | 304,560 |
|
| 167 |
+
| 3 | `อ้างอิง แหล่งข้อมูลอื่น` | 46,447 |
|
| 168 |
+
| 4 | `of the` | 42,755 |
|
| 169 |
+
| 5 | `ศ พ` | 32,101 |
|
| 170 |
+
|
| 171 |
+
**3-grams (Word):**
|
| 172 |
+
|
| 173 |
+
| Rank | N-gram | Count |
|
| 174 |
+
|------|--------|-------|
|
| 175 |
+
| 1 | `ศ พ ศ` | 31,957 |
|
| 176 |
+
| 2 | `พ ศ พ` | 27,195 |
|
| 177 |
+
| 3 | `ศ ค ศ` | 25,879 |
|
| 178 |
+
| 4 | `ธันวาคม พ ศ` | 21,330 |
|
| 179 |
+
| 5 | `ตุลาคม พ ศ` | 21,250 |
|
| 180 |
+
|
| 181 |
+
**4-grams (Word):**
|
| 182 |
+
|
| 183 |
+
| Rank | N-gram | Count |
|
| 184 |
+
|------|--------|-------|
|
| 185 |
+
| 1 | `พ ศ พ ศ` | 27,071 |
|
| 186 |
+
| 2 | `พ ศ ค ศ` | 20,164 |
|
| 187 |
+
| 3 | `0 0 0 0` | 7,943 |
|
| 188 |
+
| 4 | `ค ศ ค ศ` | 4,813 |
|
| 189 |
+
| 5 | `อ้างอิง แหล่งข้อมูลอื่น พ ศ` | 4,336 |
|
| 190 |
+
|
| 191 |
+
**5-grams (Word):**
|
| 192 |
+
|
| 193 |
+
| Rank | N-gram | Count |
|
| 194 |
+
|------|--------|-------|
|
| 195 |
+
| 1 | `ศ พ ศ พ ศ` | 4,329 |
|
| 196 |
+
| 2 | `พ ศ พ ศ พ` | 4,251 |
|
| 197 |
+
| 3 | `ศ พ ศ ค ศ` | 3,779 |
|
| 198 |
+
| 4 | `พ ศ พ ศ ค` | 3,510 |
|
| 199 |
+
| 5 | `0 0 0 0 0` | 3,345 |
|
| 200 |
+
|
| 201 |
+
**2-grams (Subword):**
|
| 202 |
+
|
| 203 |
+
| Rank | N-gram | Count |
|
| 204 |
+
|------|--------|-------|
|
| 205 |
+
| 1 | `อ ง` | 3,386,500 |
|
| 206 |
+
| 2 | `า ร` | 3,061,397 |
|
| 207 |
+
| 3 | `ก า` | 2,892,062 |
|
| 208 |
+
| 4 | `ร ะ` | 2,734,121 |
|
| 209 |
+
| 5 | `น _` | 2,476,484 |
|
| 210 |
+
|
| 211 |
+
**3-grams (Subword):**
|
| 212 |
+
|
| 213 |
+
| Rank | N-gram | Count |
|
| 214 |
+
|------|--------|-------|
|
| 215 |
+
| 1 | `ก า ร` | 2,154,969 |
|
| 216 |
+
| 2 | `เ ป็ น` | 1,461,135 |
|
| 217 |
+
| 3 | `แ ล ะ` | 1,456,554 |
|
| 218 |
+
| 4 | `ข อ ง` | 1,220,921 |
|
| 219 |
+
| 5 | `ป ร ะ` | 1,178,596 |
|
| 220 |
+
|
| 221 |
+
**4-grams (Subword):**
|
| 222 |
+
|
| 223 |
+
| Rank | N-gram | Count |
|
| 224 |
+
|------|--------|-------|
|
| 225 |
+
| 1 | `. ศ . _` | 887,086 |
|
| 226 |
+
| 2 | `_ แ ล ะ` | 845,421 |
|
| 227 |
+
| 3 | `พ . ศ .` | 598,362 |
|
| 228 |
+
| 4 | `_ พ . ศ` | 554,703 |
|
| 229 |
+
| 5 | `ค ว า ม` | 480,722 |
|
| 230 |
+
|
| 231 |
+
**5-grams (Subword):**
|
| 232 |
+
|
| 233 |
+
| Rank | N-gram | Count |
|
| 234 |
+
|------|--------|-------|
|
| 235 |
+
| 1 | `พ . ศ . _` | 572,671 |
|
| 236 |
+
| 2 | `_ พ . ศ .` | 553,928 |
|
| 237 |
+
| 3 | `ค . ศ . _` | 311,390 |
|
| 238 |
+
| 4 | `_ ค . ศ .` | 268,747 |
|
| 239 |
+
| 5 | `ป ร ะ เ ท` | 260,009 |
|
| 240 |
+
|
| 241 |
+
|
| 242 |
+
### Key Findings
|
| 243 |
+
|
| 244 |
+
- **Best Perplexity:** 2-gram (subword) with 2,438
|
| 245 |
+
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
|
| 246 |
+
- **Coverage:** Top-1000 patterns cover ~11% of corpus
|
| 247 |
+
- **Recommendation:** 4-gram or 5-gram for best predictive performance
|
| 248 |
+
|
| 249 |
+
---
|
| 250 |
+
## 3. Markov Chain Evaluation
|
| 251 |
+
|
| 252 |
+

|
| 253 |
+
|
| 254 |
+

|
| 255 |
+
|
| 256 |
+

|
| 257 |
+
|
| 258 |
+
### Results
|
| 259 |
+
|
| 260 |
+
| Context | Variant | Avg Entropy | Perplexity | Branching Factor | Unique Contexts | Predictability |
|
| 261 |
+
|---------|---------|-------------|------------|------------------|-----------------|----------------|
|
| 262 |
+
| **1** | Word | 0.2321 | 1.175 | 2.38 | 8,268,387 | 76.8% |
|
| 263 |
+
| **1** | Subword | 0.8922 | 1.856 | 12.22 | 37,876 | 10.8% |
|
| 264 |
+
| **2** | Word | 0.1165 | 1.084 | 1.32 | 19,576,764 | 88.4% |
|
| 265 |
+
| **2** | Subword | 0.6125 | 1.529 | 5.30 | 462,626 | 38.7% |
|
| 266 |
+
| **3** | Word | 0.0518 | 1.037 | 1.11 | 25,779,145 | 94.8% |
|
| 267 |
+
| **3** | Subword | 0.5564 | 1.471 | 3.91 | 2,452,254 | 44.4% |
|
| 268 |
+
| **4** | Word | 0.0248 🏆 | 1.017 | 1.05 | 28,430,641 | 97.5% |
|
| 269 |
+
| **4** | Subword | 0.4718 | 1.387 | 2.77 | 9,576,634 | 52.8% |
|
| 270 |
+
|
| 271 |
+
### Generated Text Samples (Word-based)
|
| 272 |
+
|
| 273 |
+
Below are text samples generated from each word-based Markov chain model:
|
| 274 |
+
|
| 275 |
+
**Context Size 1:**
|
| 276 |
+
|
| 277 |
+
1. `ศ 829 575b 220 นั้นมีการใช้กระสุนที่จำกัด จึงผลิตปืนรุ่นนี้ออกมา และยังมีรุ่นย่อยคือ พี วีส์บิกแอดเว...`
|
| 278 |
+
2. `พ ศ 12 12 34 1 พฤศจิกายน เมื่อวันที่ 16 ทีมสุดท้าย 8 เป็นต้นไป ค ศ เขายังได้ปีนข่านเทนกรี ในปี`
|
| 279 |
+
3. `1 จัดเป็นสี่กลุ่ม กลุ่มละ 4 9 ศรีราชา อำเภอศรีราชา จังหวัดชลบุรี ในครอบครัวที่มีพี่น้อง 5 พฤษภาคม สิ...`
|
| 280 |
+
|
| 281 |
+
**Context Size 2:**
|
| 282 |
+
|
| 283 |
+
1. `พ ศ ได้รับอนุมัติจากมหาเถรสมาคมให้ปรับปรุงสภาพวัดให้ดีขึ้น ปี พ ศ ค ศ พระเจ้าอิชต์วานที่ 1 ประเทศฮัง...`
|
| 284 |
+
2. `ค ศ ปัจจุบัน ละครชุด ปีเรื่องบทร่วมกับออกอากาศอ้างอิงพ ศ ความทรงจำที่ไม่อาจลืม ตอน บันทึกท่องเที่ยวท...`
|
| 285 |
+
3. `of the usaf retrieved 20 october เจ้าชายโทโมฮิโตะแห่งมิกาซะสิ้นพระชนม์เมื่อวันที่ 6 มิถุนายน พ ศ อ้า...`
|
| 286 |
+
|
| 287 |
+
**Context Size 3:**
|
| 288 |
+
|
| 289 |
+
1. `พ ศ พ ศ และครั้งที่สอง ประมาณ พ ศ 31 สิงหาคม พ ศ วิทยาลัยโทรคมนาคมนนทบุรี รับนักศึกษาจาก การสอบคัดเล...`
|
| 290 |
+
2. `ศ พ ศ ค ศ พระเจ้าแฟร์ดีนันท์ที่ 4 แห่งชาวโรมัน 8 กันยายน ค ศ เป็นที่รู้จักในชื่อ ไป๋ ลู่ เป็นนักแสดง...`
|
| 291 |
+
3. `ศ ค ศ พระองค์เจ้าสุวพักตร์วิไลยพรรณ ประสูติ 2 พฤษภาคม พ ศ พ ศ ไอ จี คอมมิวนิเคชัน`
|
| 292 |
+
|
| 293 |
+
**Context Size 4:**
|
| 294 |
+
|
| 295 |
+
1. `พ ศ พ ศ เป็นปราชญ์และนักคิดอิสระที่ได้รับการกล่าวถึงอย่างกว้างขวาง เมื่อยังเป็นเด็กหนุ่ม ทางสมาคมเทว...`
|
| 296 |
+
2. `พ ศ ค ศ ดีแลน มินเน็ตต์ นักแสดงและนักดนตรีชาวอเมริกัน ปรินซ์ แอมพอนซา นักฟุตบอลชาวกานา ซานะ มินาโตซา...`
|
| 297 |
+
3. `0 0 0 0 ไมได้เข้าร่วมแข่งขัน 4 0 0 0 0 4 21 17 4 26 4 ระนอง 16 6`
|
| 298 |
+
|
| 299 |
+
|
| 300 |
+
### Generated Text Samples (Subword-based)
|
| 301 |
+
|
| 302 |
+
Below are text samples generated from each subword-based Markov chain model:
|
| 303 |
+
|
| 304 |
+
**Context Size 1:**
|
| 305 |
+
|
| 306 |
+
1. `_คยงข_ในที่เป็นกอว่า`
|
| 307 |
+
2. `ารตียนผิดละคาระบโด`
|
| 308 |
+
3. `นาณฑลาระปยีปดารรน`
|
| 309 |
+
|
| 310 |
+
**Context Size 2:**
|
| 311 |
+
|
| 312 |
+
1. `องครั้งที่รู้จักระชนิดนึงนิ`
|
| 313 |
+
2. `ารไปส.ธ.90.0_ของเ`
|
| 314 |
+
3. `การสช.อีกครั้งมีว่า_เป`
|
| 315 |
+
|
| 316 |
+
**Context Size 3:**
|
| 317 |
+
|
| 318 |
+
1. `การใด_ๆ_broad_mete`
|
| 319 |
+
2. `เป็นการใช้อย่างต่างจาก`
|
| 320 |
+
3. `และครั้งแรมชาติขึ้นบก_แ`
|
| 321 |
+
|
| 322 |
+
**Context Size 4:**
|
| 323 |
+
|
| 324 |
+
1. `.ศ._สหราชอาณาเขตจตุจั`
|
| 325 |
+
2. `_และไม่สามารถป้องกันแล`
|
| 326 |
+
3. `พ.ศ._76_<small>(ไทย`
|
| 327 |
+
|
| 328 |
+
|
| 329 |
+
### Key Findings
|
| 330 |
+
|
| 331 |
+
- **Best Predictability:** Context-4 (word) with 97.5% predictability
|
| 332 |
+
- **Branching Factor:** Decreases with context size (more deterministic)
|
| 333 |
+
- **Memory Trade-off:** Larger contexts require more storage (9,576,634 contexts)
|
| 334 |
+
- **Recommendation:** Context-3 or Context-4 for text generation
|
| 335 |
+
|
| 336 |
+
---
|
| 337 |
+
## 4. Vocabulary Analysis
|
| 338 |
+
|
| 339 |
+

|
| 340 |
+
|
| 341 |
+

|
| 342 |
+
|
| 343 |
+

|
| 344 |
+
|
| 345 |
+
### Statistics
|
| 346 |
+
|
| 347 |
+
| Metric | Value |
|
| 348 |
+
|--------|-------|
|
| 349 |
+
| Vocabulary Size | 1,276,542 |
|
| 350 |
+
| Total Tokens | 26,332,909 |
|
| 351 |
+
| Mean Frequency | 20.63 |
|
| 352 |
+
| Median Frequency | 3 |
|
| 353 |
+
| Frequency Std Dev | 1261.31 |
|
| 354 |
+
|
| 355 |
+
### Most Common Words
|
| 356 |
+
|
| 357 |
+
| Rank | Word | Frequency |
|
| 358 |
+
|------|------|-----------|
|
| 359 |
+
| 1 | ศ | 920,143 |
|
| 360 |
+
| 2 | พ | 595,465 |
|
| 361 |
+
| 3 | 1 | 351,475 |
|
| 362 |
+
| 4 | ค | 314,624 |
|
| 363 |
+
| 5 | 2 | 306,676 |
|
| 364 |
+
| 6 | 3 | 247,910 |
|
| 365 |
+
| 7 | the | 217,279 |
|
| 366 |
+
| 8 | 4 | 172,069 |
|
| 367 |
+
| 9 | ๆ | 171,685 |
|
| 368 |
+
| 10 | of | 169,227 |
|
| 369 |
+
|
| 370 |
+
### Least Common Words (from vocabulary)
|
| 371 |
+
|
| 372 |
+
| Rank | Word | Frequency |
|
| 373 |
+
|------|------|-----------|
|
| 374 |
+
| 1 | เบอร์ลินฟาร์มาซูติคอลอินดัสตรี้ | 2 |
|
| 375 |
+
| 2 | มีเลขซีเทนขั้นต่ำที่ | 2 |
|
| 376 |
+
| 3 | น้ำมันดีเซลหมุนเวียน | 2 |
|
| 377 |
+
| 4 | neste | 2 |
|
| 378 |
+
| 5 | เฮกซาดีเคน | 2 |
|
| 379 |
+
| 6 | เฮปตาเมทิลโนเนน | 2 |
|
| 380 |
+
| 7 | เครื่องทดสอบคุณภาพการจุดระเบิด | 2 |
|
| 381 |
+
| 8 | เครื่องมือนี้ใช้วิธีเรียบง่ายกว่าและแข็งแกร่งกว่าในการวัดเลขซีเทนเมื่อเทียบกับ | 2 |
|
| 382 |
+
| 9 | บ้านเก้าเลี้ยว | 2 |
|
| 383 |
+
| 10 | ชุมชนบ้านเก้าเลี้ยว | 2 |
|
| 384 |
+
|
| 385 |
+
### Zipf's Law Analysis
|
| 386 |
+
|
| 387 |
+
| Metric | Value |
|
| 388 |
+
|--------|-------|
|
| 389 |
+
| Zipf Coefficient | 0.9360 |
|
| 390 |
+
| R² (Goodness of Fit) | 0.999043 |
|
| 391 |
+
| Adherence Quality | **excellent** |
|
| 392 |
+
|
| 393 |
+
### Coverage Analysis
|
| 394 |
+
|
| 395 |
+
| Top N Words | Coverage |
|
| 396 |
+
|-------------|----------|
|
| 397 |
+
| Top 100 | 29.7% |
|
| 398 |
+
| Top 1,000 | 45.1% |
|
| 399 |
+
| Top 5,000 | 57.6% |
|
| 400 |
+
| Top 10,000 | 63.4% |
|
| 401 |
+
|
| 402 |
+
### Key Findings
|
| 403 |
+
|
| 404 |
+
- **Zipf Compliance:** R²=0.9990 indicates excellent adherence to Zipf's law
|
| 405 |
+
- **High Frequency Dominance:** Top 100 words cover 29.7% of corpus
|
| 406 |
+
- **Long Tail:** 1,266,542 words needed for remaining 36.6% coverage
|
| 407 |
+
|
| 408 |
+
---
|
| 409 |
+
## 5. Word Embeddings Evaluation
|
| 410 |
+
|
| 411 |
+

|
| 412 |
+
|
| 413 |
+

|
| 414 |
+
|
| 415 |
+

|
| 416 |
+
|
| 417 |
+

|
| 418 |
+
|
| 419 |
+
|
| 420 |
+
### 5.1 Cross-Lingual Alignment
|
| 421 |
+
|
| 422 |
+

|
| 423 |
+
|
| 424 |
+

|
| 425 |
+
|
| 426 |
+
|
| 427 |
+
### 5.2 Model Comparison
|
| 428 |
+
|
| 429 |
+
| Model | Dimension | Isotropy | Semantic Density | Alignment R@1 | Alignment R@10 |
|
| 430 |
+
|-------|-----------|----------|------------------|---------------|----------------|
|
| 431 |
+
| **mono_32d** | 32 | 0.8475 | 0.3288 | N/A | N/A |
|
| 432 |
+
| **mono_64d** | 64 | 0.8400 | 0.2631 | N/A | N/A |
|
| 433 |
+
| **mono_128d** | 128 | 0.8225 | 0.1868 | N/A | N/A |
|
| 434 |
+
| **aligned_32d** | 32 | 0.8475 🏆 | 0.3296 | 0.2180 | 0.6440 |
|
| 435 |
+
| **aligned_64d** | 64 | 0.8400 | 0.2600 | 0.4200 | 0.7840 |
|
| 436 |
+
| **aligned_128d** | 128 | 0.8225 | 0.1907 | 0.4680 | 0.8680 |
|
| 437 |
+
|
| 438 |
+
### Key Findings
|
| 439 |
+
|
| 440 |
+
- **Best Isotropy:** aligned_32d with 0.8475 (more uniform distribution)
|
| 441 |
+
- **Semantic Density:** Average pairwise similarity of 0.2598. Lower values indicate better semantic separation.
|
| 442 |
+
- **Alignment Quality:** Aligned models achieve up to 46.8% R@1 in cross-lingual retrieval.
|
| 443 |
+
- **Recommendation:** 128d aligned for best cross-lingual performance
|
| 444 |
+
|
| 445 |
+
---
|
| 446 |
+
## 6. Morphological Analysis (Experimental)
|
| 447 |
+
|
| 448 |
+
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.
|
| 449 |
+
|
| 450 |
+
### 6.1 Productivity & Complexity
|
| 451 |
+
|
| 452 |
+
| Metric | Value | Interpretation | Recommendation |
|
| 453 |
+
|--------|-------|----------------|----------------|
|
| 454 |
+
| Productivity Index | **5.000** | High morphological productivity | Reliable analysis |
|
| 455 |
+
| Idiomaticity Gap | **-0.367** | Low formulaic content | - |
|
| 456 |
+
|
| 457 |
+
### 6.2 Affix Inventory (Productive Units)
|
| 458 |
+
|
| 459 |
+
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.
|
| 460 |
+
|
| 461 |
+
#### Productive Prefixes
|
| 462 |
+
| Prefix | Examples |
|
| 463 |
+
|--------|----------|
|
| 464 |
+
| `-และ` | และได้รับแต่งตั้งเป็นรัฐมนตรีที่ไม่ประจำกระทรวง, และพาทีมเข้าชิงชนะเลิศ, และคริสต์ศาสนา |
|
| 465 |
+
| `-เ` | เจ้าเมืองนครเขื่อนขันธ์คนที่, เฉิงเตี๋ยอี, เวลวิชเซีย |
|
| 466 |
+
| `-โ` | โรงเรียนพระปฐมวิทยาลัย, โรงเรียนเวียงกาหลงวิทยา, โอมมอก |
|
| 467 |
+
| `-แ` | แม่เม็ง, และได้รับแต่งตั้งเป็นรัฐมนตรีที่ไม่ประจำกระทรวง, และพาทีมเข้าชิงชนะเลิศ |
|
| 468 |
+
| `-อ` | อควาครอส, อำเภอทุ่งย���งแดง, อุดมศิลป์ |
|
| 469 |
+
| `-ส` | สมเด็จพระสันตะปาปาธีโอดอร์ที่, สถิรวํโส, สาขาวิชาศิลปกรรม |
|
| 470 |
+
| `-ก` | การลอบฆ่า, กาญจนพัฒน์, การประมูลคลื่น |
|
| 471 |
+
| `-ค` | คิคาวาดะ, คิดประเสริฐ, ครอบคลุมพื้นที่ตำบลสงยางทั้งตำบล |
|
| 472 |
+
|
| 473 |
+
#### Productive Suffixes
|
| 474 |
+
| Suffix | Examples |
|
| 475 |
+
|--------|----------|
|
| 476 |
+
| `-น` | ที่โด่งดังเช่น, ตั้งอยู่บนเชิงเทิน, นักเรียนห้องกิฟต์รุ่น |
|
| 477 |
+
| `-ง` | แม่เม็ง, และได้รับแต่งตั้งเป็นรัฐมนตรีที่ไม่ประจำกระทรวง, อำเภอทุ่งยางแดง |
|
| 478 |
+
| `-า` | การลอบฆ่า, โรงเรียนเวียงกาหลงวิทยา, อดีตนายกรัฐมนตรีแคนาดา |
|
| 479 |
+
| `-ย` | นำชัย, โรงเรียนพระปฐมวิทยาลัย, ชีวิตช่วงปลาย |
|
| 480 |
+
| `-ก` | ล้อมรอบด้วยกนกเปลวเพลิงด้านบนมีรัศมีประกอบแปดแฉก, โอมมอก, เนื่องในโอกาสพระราชพิธีกาญจนาภิเษก |
|
| 481 |
+
| `-ม` | หิมาลัยยิม, มอบโดยกระทรวงวัฒนธรรม, สาขาวิชาศิลปกรรม |
|
| 482 |
+
| `-อง` | และในคืนนั้นเอง, กาเอต็อง, นักแสดงจากเรื่อง |
|
| 483 |
+
| `-ร` | หงษ์ขจร, จะพูดถึงความรู้สึกเชื่อใจได้อย่างไร, และนายทหาร |
|
| 484 |
+
|
| 485 |
+
### 6.3 Bound Stems (Lexical Roots)
|
| 486 |
+
|
| 487 |
+
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.
|
| 488 |
+
|
| 489 |
+
| Stem | Cohesion | Substitutability | Examples |
|
| 490 |
+
|------|----------|------------------|----------|
|
| 491 |
+
| `การแ` | 2.17x | 65 contexts | การแลก, การแปล, การแตก |
|
| 492 |
+
| `ของเ` | 1.49x | 196 contexts | ของเล, ของเจ, ของเอ |
|
| 493 |
+
| `พระร` | 2.07x | 33 contexts | พระรถ, พระราม, พระราช |
|
| 494 |
+
| `การเ` | 1.55x | 93 contexts | การเย, การเดท, การเอา |
|
| 495 |
+
| `ศาสต` | 1.82x | 46 contexts | ศาสตา, ศาสตรา, ศาสตร์ |
|
| 496 |
+
| `าการ` | 1.45x | 100 contexts | อาการ, บาการี, คาการิ |
|
| 497 |
+
| `นการ` | 1.48x | 86 contexts | ธนการ, ในการ, แผนการ |
|
| 498 |
+
| `ประก` | 1.46x | 84 contexts | ประกบ, ประการ, ประกิจ |
|
| 499 |
+
| `โรงเ` | 2.92x | 8 contexts | โรงเจ, โรงเข้, โรงเรีย |
|
| 500 |
+
| `ประเ` | 1.42x | 83 contexts | ประเถท, ประเภท, ประเทศ |
|
| 501 |
+
| `งจาก` | 1.44x | 72 contexts | บางจาก, อิงจาก, ทางจาก |
|
| 502 |
+
| `ระเท` | 1.66x | 38 contexts | กระเทย, กระเทศ, พระเทพ |
|
| 503 |
+
|
| 504 |
+
### 6.4 Affix Compatibility (Co-occurrence)
|
| 505 |
+
|
| 506 |
+
This table shows which prefixes and suffixes most frequently co-occur on the same stems, revealing the 'stacking' rules of the language's morphology.
|
| 507 |
+
|
| 508 |
+
| Prefix | Suffix | Frequency | Examples |
|
| 509 |
+
|--------|--------|-----------|----------|
|
| 510 |
+
| `-เ` | `-น` | 101 words | เพื่อเป็นการรักษากำลังและไพร่พลทหารของตนเองไว้สำหรับการศึกอื่น, เหยาจื่อจิน |
|
| 511 |
+
| `-เ` | `-ง` | 84 words | เธอได้ออกเพลง, เจิ้นเสียง |
|
| 512 |
+
| `-เ` | `-า` | 80 words | เข็มข้าหลวงเดิมราชกิจจานุเบกษา, เจ้าหญิงโรมานอฟสกายา |
|
| 513 |
+
| `-เ` | `-ย` | 53 words | เป็นภาษาไทยอีกด้วย, เพลงดาบแม่น้ำร้อยสาย |
|
| 514 |
+
| `-และ` | `-น` | 52 words | และตำบลบ้านแหวน, และเปลี่ยนชื่อไปเป็น |
|
| 515 |
+
| `-โ` | `-ง` | 50 words | โดยใช้เครื่องบินโบอิง, โรงเรียนตะโกดอนหญ้านาง |
|
| 516 |
+
| `-โ` | `-น` | 45 words | โคเฮ็น, โฟกส์วาเกน |
|
| 517 |
+
| `-ก` | `-น` | 44 words | กุลธน, การแบ่งชนชั้น |
|
| 518 |
+
| `-โ` | `-า` | 42 words | โรงเรียนศรีสมบูรณ์วิทยา, โดยมีวัตถุประสงค์เพื่อเป็นสถาบันการศึกษา |
|
| 519 |
+
| `-แ` | `-น` | 41 words | และตำบลบ้านแหวน, แม่ฮ่องสอน |
|
| 520 |
+
|
| 521 |
+
### 6.5 Recursive Morpheme Segmentation
|
| 522 |
+
|
| 523 |
+
Using **Recursive Hierarchical Substitutability**, we decompose complex words into their constituent morphemes. This approach handles nested affixes (e.g., `prefix-prefix-root-suffix`).
|
| 524 |
+
|
| 525 |
+
| Word | Suggested Split | Confidence | Stem |
|
| 526 |
+
|------|-----------------|------------|------|
|
| 527 |
+
| การทำคะแนน | **`การทำคะแ-น-น`** | 7.5 | `น` |
|
| 528 |
+
| เจ้าจอมสุวัทนา | **`เจ้าจอมสุวัท-น-า`** | 7.5 | `น` |
|
| 529 |
+
| หัวอกชาวนา | **`หัวอกชาว-น-า`** | 7.5 | `น` |
|
| 530 |
+
| อิริยาปถบรรพ | **`อิริยาปถบร-ร-พ`** | 7.5 | `ร` |
|
| 531 |
+
| ตำบลพลวงสองนาง | **`ตำบลพลวงสอง-น-าง`** | 7.5 | `น` |
|
| 532 |
+
| เส้นทางทรนง | **`เส้นทางทร-น-ง`** | 7.5 | `น` |
|
| 533 |
+
| อเล็กซานดรอฟนา | **`อเล็กซานดรอฟ-น-า`** | 7.5 | `น` |
|
| 534 |
+
| คาจิโดกิอาร์มส | **`คาจิโดกิอาร์-ม-ส`** | 7.5 | `ม` |
|
| 535 |
+
| และเซเรนา | **`และเซเร-น-า`** | 7.5 | `น` |
|
| 536 |
+
| และคูลลิแนน | **`และคูลลิแ-น-น`** | 7.5 | `น` |
|
| 537 |
+
| ฟุชิกิดาเนะ | **`ฟุชิกิดาเ-น-ะ`** | 7.5 | `น` |
|
| 538 |
+
| ตำบลม่วงงาม | **`ตำบลม่วง-ง-าม`** | 6.0 | `ตำบลม่วง` |
|
| 539 |
+
| และตำบลหมื่นไวย | **`และ-ตำบลหมื่นไวย`** | 4.5 | `ตำบลหมื่นไวย` |
|
| 540 |
+
| และได้รับสมญาว่า | **`และ-ได้รับสมญาว่า`** | 4.5 | `ได้รับสมญาว่า` |
|
| 541 |
+
| และประทับอยู่ | **`และ-ประทับอยู่`** | 4.5 | `ประทับอยู่` |
|
| 542 |
+
|
| 543 |
+
### 6.6 Linguistic Interpretation
|
| 544 |
+
|
| 545 |
+
> **Automated Insight:**
|
| 546 |
+
The language Thai shows high morphological productivity. The subword models are significantly more efficient than word models, suggesting a rich system of affixation or compounding.
|
| 547 |
+
|
| 548 |
+
---
|
| 549 |
+
## 7. Summary & Recommendations
|
| 550 |
+
|
| 551 |
+

|
| 552 |
+
|
| 553 |
+
### Production Recommendations
|
| 554 |
+
|
| 555 |
+
| Component | Recommended | Rationale |
|
| 556 |
+
|-----------|-------------|-----------|
|
| 557 |
+
| Tokenizer | **64k BPE** | Best compression (4.75x) |
|
| 558 |
+
| N-gram | **2-gram** | Lowest perplexity (2,438) |
|
| 559 |
+
| Markov | **Context-4** | Highest predictability (97.5%) |
|
| 560 |
+
| Embeddings | **100d** | Balanced semantic capture and isotropy |
|
| 561 |
+
|
| 562 |
+
|
| 563 |
+
---
|
| 564 |
+
## Appendix: Metrics Glossary & Interpretation Guide
|
| 565 |
+
|
| 566 |
+
This section provides definitions, intuitions, and guidance for interpreting the metrics used throughout this report.
|
| 567 |
+
|
| 568 |
+
### Tokenizer Metrics
|
| 569 |
+
|
| 570 |
+
**Compression Ratio**
|
| 571 |
+
> *Definition:* The ratio of characters to tokens (chars/token). Measures how efficiently the tokenizer represents text.
|
| 572 |
+
>
|
| 573 |
+
> *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.
|
| 574 |
+
>
|
| 575 |
+
> *What to seek:* Higher is generally better for efficiency, but extremely high compression may indicate overly aggressive merging that loses morphological information.
|
| 576 |
+
|
| 577 |
+
**Average Token Length (Fertility)**
|
| 578 |
+
> *Definition:* Mean number of characters per token produced by the tokenizer.
|
| 579 |
+
>
|
| 580 |
+
> *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.
|
| 581 |
+
>
|
| 582 |
+
> *What to seek:* Balance between 2-5 characters for most languages. Arabic/morphologically-rich languages may benefit from slightly longer tokens.
|
| 583 |
+
|
| 584 |
+
**Unknown Token Rate (OOV Rate)**
|
| 585 |
+
> *Definition:* Percentage of tokens that map to the unknown/UNK token, indicating words the tokenizer cannot represent.
|
| 586 |
+
>
|
| 587 |
+
> *Intuition:* Lower OOV means better vocabulary coverage. High OOV indicates the tokenizer encounters many unseen character sequences.
|
| 588 |
+
>
|
| 589 |
+
> *What to seek:* Below 1% is excellent; below 5% is acceptable. BPE tokenizers typically achieve very low OOV due to subword fallback.
|
| 590 |
+
|
| 591 |
+
### N-gram Model Metrics
|
| 592 |
+
|
| 593 |
+
**Perplexity**
|
| 594 |
+
> *Definition:* Measures how "surprised" the model is by test data. Mathematically: 2^(cross-entropy). Lower values indicate better prediction.
|
| 595 |
+
>
|
| 596 |
+
> *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.
|
| 597 |
+
>
|
| 598 |
+
> *What to seek:* Lower is better. Perplexity decreases with larger n-grams (more context). Values vary widely by language and corpus size.
|
| 599 |
+
|
| 600 |
+
**Entropy**
|
| 601 |
+
> *Definition:* Average information content (in bits) needed to encode the next token given the context. Related to perplexity: perplexity = 2^entropy.
|
| 602 |
+
>
|
| 603 |
+
> *Intuition:* High entropy means high uncertainty/randomness; low entropy means predictable patterns. Natural language typically has entropy between 1-4 bits per character.
|
| 604 |
+
>
|
| 605 |
+
> *What to seek:* Lower entropy indicates more predictable text patterns. Entropy should decrease as n-gram size increases.
|
| 606 |
+
|
| 607 |
+
**Coverage (Top-K)**
|
| 608 |
+
> *Definition:* Percentage of corpus occurrences explained by the top K most frequent n-grams.
|
| 609 |
+
>
|
| 610 |
+
> *Intuition:* High coverage with few patterns indicates repetitive/formulaic text; low coverage suggests diverse vocabulary usage.
|
| 611 |
+
>
|
| 612 |
+
> *What to seek:* Depends on use case. For language modeling, moderate coverage (40-60% with top-1000) is typical for natural text.
|
| 613 |
+
|
| 614 |
+
### Markov Chain Metrics
|
| 615 |
+
|
| 616 |
+
**Average Entropy**
|
| 617 |
+
> *Definition:* Mean entropy across all contexts, measuring average uncertainty in next-word prediction.
|
| 618 |
+
>
|
| 619 |
+
> *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).
|
| 620 |
+
>
|
| 621 |
+
> *What to seek:* Decreasing entropy with larger context sizes. Very low entropy (<0.1) indicates highly deterministic transitions.
|
| 622 |
+
|
| 623 |
+
**Branching Factor**
|
| 624 |
+
> *Definition:* Average number of unique next tokens observed for each context.
|
| 625 |
+
>
|
| 626 |
+
> *Intuition:* High branching = many possible continuations (flexible but uncertain); low branching = few options (predictable but potentially repetitive).
|
| 627 |
+
>
|
| 628 |
+
> *What to seek:* Branching factor should decrease with context size. Values near 1.0 indicate nearly deterministic chains.
|
| 629 |
+
|
| 630 |
+
**Predictability**
|
| 631 |
+
> *Definition:* Derived metric: (1 - normalized_entropy) × 100%. Indicates how deterministic the model's predictions are.
|
| 632 |
+
>
|
| 633 |
+
> *Intuition:* 100% predictability means the next word is always certain; 0% means completely random. Real text falls between these extremes.
|
| 634 |
+
>
|
| 635 |
+
> *What to seek:* Higher predictability for text generation quality, but too high (>98%) may produce repetitive output.
|
| 636 |
+
|
| 637 |
+
### Vocabulary & Zipf's Law Metrics
|
| 638 |
+
|
| 639 |
+
**Zipf's Coefficient**
|
| 640 |
+
> *Definition:* The slope of the log-log plot of word frequency vs. rank. Zipf's law predicts this should be approximately -1.
|
| 641 |
+
>
|
| 642 |
+
> *Intuition:* A coefficient near -1 indicates the corpus follows natural language patterns where a few words are very common and most words are rare.
|
| 643 |
+
>
|
| 644 |
+
> *What to seek:* Values between -0.8 and -1.2 indicate healthy natural language distribution. Deviations may suggest domain-specific or artificial text.
|
| 645 |
+
|
| 646 |
+
**R² (Coefficient of Determination)**
|
| 647 |
+
> *Definition:* Measures how well the linear fit explains the frequency-rank relationship. Ranges from 0 to 1.
|
| 648 |
+
>
|
| 649 |
+
> *Intuition:* R² near 1.0 means the data closely follows Zipf's law; lower values indicate deviation from expected word frequency patterns.
|
| 650 |
+
>
|
| 651 |
+
> *What to seek:* R² > 0.95 is excellent; > 0.99 indicates near-perfect Zipf adherence typical of large natural corpora.
|
| 652 |
+
|
| 653 |
+
**Vocabulary Coverage**
|
| 654 |
+
> *Definition:* Cumulative percentage of corpus tokens accounted for by the top N words.
|
| 655 |
+
>
|
| 656 |
+
> *Intuition:* Shows how concentrated word usage is. If top-100 words cover 50% of text, the corpus relies heavily on common words.
|
| 657 |
+
>
|
| 658 |
+
> *What to seek:* Top-100 covering 30-50% is typical. Higher coverage indicates more repetitive text; lower suggests richer vocabulary.
|
| 659 |
+
|
| 660 |
+
### Word Embedding Metrics
|
| 661 |
+
|
| 662 |
+
**Isotropy**
|
| 663 |
+
> *Definition:* Measures how uniformly distributed vectors are in the embedding space. Computed as the ratio of minimum to maximum singular values.
|
| 664 |
+
>
|
| 665 |
+
> *Intuition:* High isotropy (near 1.0) means vectors spread evenly in all directions; low isotropy means vectors cluster in certain directions, reducing expressiveness.
|
| 666 |
+
>
|
| 667 |
+
> *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.
|
| 668 |
+
|
| 669 |
+
**Average Norm**
|
| 670 |
+
> *Definition:* Mean magnitude (L2 norm) of word vectors in the embedding space.
|
| 671 |
+
>
|
| 672 |
+
> *Intuition:* Indicates the typical "length" of vectors. Consistent norms suggest stable training; high variance may indicate some words are undertrained.
|
| 673 |
+
>
|
| 674 |
+
> *What to seek:* Relatively consistent norms across models. The absolute value matters less than consistency (low std deviation).
|
| 675 |
+
|
| 676 |
+
**Cosine Similarity**
|
| 677 |
+
> *Definition:* Measures angular similarity between vectors, ranging from -1 (opposite) to 1 (identical direction).
|
| 678 |
+
>
|
| 679 |
+
> *Intuition:* Words with similar meanings should have high cosine similarity. This is the standard metric for semantic relatedness in embeddings.
|
| 680 |
+
>
|
| 681 |
+
> *What to seek:* Semantically related words should score > 0.5; unrelated words should be near 0. Synonyms often score > 0.7.
|
| 682 |
+
|
| 683 |
+
**t-SNE Visualization**
|
| 684 |
+
> *Definition:* t-Distributed Stochastic Neighbor Embedding - a dimensionality reduction technique that preserves local structure for visualization.
|
| 685 |
+
>
|
| 686 |
+
> *Intuition:* Clusters in t-SNE plots indicate groups of semantically related words. Spread indicates vocabulary diversity; tight clusters suggest semantic coherence.
|
| 687 |
+
>
|
| 688 |
+
> *What to seek:* Meaningful clusters (e.g., numbers together, verbs together). Avoid over-interpreting distances - t-SNE preserves local, not global, structure.
|
| 689 |
+
|
| 690 |
+
### General Interpretation Guidelines
|
| 691 |
+
|
| 692 |
+
1. **Compare within model families:** Metrics are most meaningful when comparing models of the same type (e.g., 8k vs 64k tokenizer).
|
| 693 |
+
2. **Consider trade-offs:** Better performance on one metric often comes at the cost of another (e.g., compression vs. OOV rate).
|
| 694 |
+
3. **Context matters:** Optimal values depend on downstream tasks. Text generation may prioritize different metrics than classification.
|
| 695 |
+
4. **Corpus influence:** All metrics are influenced by corpus characteristics. Wikipedia text differs from social media or literature.
|
| 696 |
+
5. **Language-specific patterns:** Morphologically rich languages (like Arabic) may show different optimal ranges than analytic languages.
|
| 697 |
+
|
| 698 |
+
|
| 699 |
+
### Visualizations Index
|
| 700 |
+
|
| 701 |
+
| Visualization | Description |
|
| 702 |
+
|---------------|-------------|
|
| 703 |
+
| Tokenizer Compression | Compression ratios by vocabulary size |
|
| 704 |
+
| Tokenizer Fertility | Average token length by vocabulary |
|
| 705 |
+
| Tokenizer OOV | Unknown token rates |
|
| 706 |
+
| Tokenizer Total Tokens | Total tokens by vocabulary |
|
| 707 |
+
| N-gram Perplexity | Perplexity by n-gram size |
|
| 708 |
+
| N-gram Entropy | Entropy by n-gram size |
|
| 709 |
+
| N-gram Coverage | Top pattern coverage |
|
| 710 |
+
| N-gram Unique | Unique n-gram counts |
|
| 711 |
+
| Markov Entropy | Entropy by context size |
|
| 712 |
+
| Markov Branching | Branching factor by context |
|
| 713 |
+
| Markov Contexts | Unique context counts |
|
| 714 |
+
| Zipf's Law | Frequency-rank distribution with fit |
|
| 715 |
+
| Vocab Frequency | Word frequency distribution |
|
| 716 |
+
| Top 20 Words | Most frequent words |
|
| 717 |
+
| Vocab Coverage | Cumulative coverage curve |
|
| 718 |
+
| Embedding Isotropy | Vector space uniformity |
|
| 719 |
+
| Embedding Norms | Vector magnitude distribution |
|
| 720 |
+
| Embedding Similarity | Word similarity heatmap |
|
| 721 |
+
| Nearest Neighbors | Similar words for key terms |
|
| 722 |
+
| t-SNE Words | 2D word embedding visualization |
|
| 723 |
+
| t-SNE Sentences | 2D sentence embedding visualization |
|
| 724 |
+
| Position Encoding | Encoding method comparison |
|
| 725 |
+
| Model Sizes | Storage requirements |
|
| 726 |
+
| Performance Dashboard | Comprehensive performance overview |
|
| 727 |
+
|
| 728 |
+
---
|
| 729 |
+
## About This Project
|
| 730 |
+
|
| 731 |
+
### Data Source
|
| 732 |
+
|
| 733 |
+
Models trained on [wikipedia-monthly](https://huggingface.co/datasets/omarkamali/wikipedia-monthly) - a monthly snapshot of Wikipedia articles across 300+ languages.
|
| 734 |
+
|
| 735 |
+
### Project
|
| 736 |
+
|
| 737 |
+
A project by **[Wikilangs](https://wikilangs.org)** - Open-source NLP models for every Wikipedia language.
|
| 738 |
+
|
| 739 |
+
### Maintainer
|
| 740 |
+
|
| 741 |
+
[Omar Kamali](https://omarkamali.com) - [Omneity Labs](https://omneitylabs.com)
|
| 742 |
+
|
| 743 |
+
### Citation
|
| 744 |
+
|
| 745 |
+
If you use these models in your research, please cite:
|
| 746 |
+
|
| 747 |
+
```bibtex
|
| 748 |
+
@misc{wikilangs2025,
|
| 749 |
+
author = {Kamali, Omar},
|
| 750 |
+
title = {Wikilangs: Open NLP Models for Wikipedia Languages},
|
| 751 |
+
year = {2025},
|
| 752 |
+
doi = {10.5281/zenodo.18073153},
|
| 753 |
+
publisher = {Zenodo},
|
| 754 |
+
url = {https://huggingface.co/wikilangs}
|
| 755 |
+
institution = {Omneity Labs}
|
| 756 |
+
}
|
| 757 |
+
```
|
| 758 |
+
|
| 759 |
+
### License
|
| 760 |
+
|
| 761 |
+
MIT License - Free for academic and commercial use.
|
| 762 |
+
|
| 763 |
+
### Links
|
| 764 |
+
|
| 765 |
+
- 🌐 Website: [wikilangs.org](https://wikilangs.org)
|
| 766 |
+
- 🤗 Models: [huggingface.co/wikilangs](https://huggingface.co/wikilangs)
|
| 767 |
+
- 📊 Data: [wikipedia-monthly](https://huggingface.co/datasets/omarkamali/wikipedia-monthly)
|
| 768 |
+
- 👤 Author: [Omar Kamali](https://huggingface.co/omarkamali)
|
| 769 |
+
- 🤝 Sponsor: [Featherless AI](https://featherless.ai)
|
| 770 |
+
---
|
| 771 |
+
*Generated by Wikilangs Models Pipeline*
|
| 772 |
+
|
| 773 |
+
*Report Date: 2026-01-17 15:56:15*
|
models/embeddings/aligned/th_128d.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:218532ffd1ab8a530b7f6a92aad9884b8243b5d87185ede81b63f299cddbde50
|
| 3 |
+
size 1501738674
|
models/embeddings/aligned/th_128d.meta.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"lang": "th", "dim": 128, "max_seq_len": 512, "is_aligned": true}
|
models/embeddings/aligned/th_128d.projection.npy
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:113bd1fa2e56d2694ab2de234ab52b529f7ae2d210182aa2e0c00869c76835e4
|
| 3 |
+
size 65664
|
models/embeddings/aligned/th_128d_metadata.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"language": "th",
|
| 3 |
+
"dimension": 128,
|
| 4 |
+
"version": "aligned",
|
| 5 |
+
"hub_language": "en",
|
| 6 |
+
"seed_vocab_size": 110471,
|
| 7 |
+
"vocab_size": 450025
|
| 8 |
+
}
|
models/embeddings/aligned/th_32d.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d07e515211b39a047f35fe0cbc2638179bc51c080901fb9fff2d1d0a4644119
|
| 3 |
+
size 388119474
|
models/embeddings/aligned/th_32d.meta.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"lang": "th", "dim": 32, "max_seq_len": 512, "is_aligned": true}
|
models/embeddings/aligned/th_32d.projection.npy
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5bf4ae9d3acbb3509391b52dbca4080b525745fca52cba52a8255d479a207036
|
| 3 |
+
size 4224
|
models/embeddings/aligned/th_32d_metadata.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"language": "th",
|
| 3 |
+
"dimension": 32,
|
| 4 |
+
"version": "aligned",
|
| 5 |
+
"hub_language": "en",
|
| 6 |
+
"seed_vocab_size": 110471,
|
| 7 |
+
"vocab_size": 450025
|
| 8 |
+
}
|
models/embeddings/aligned/th_64d.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:428f8a66daf50dd434a1e9ab558a26f5a244d30dd9820833a63c5fc65c2693f9
|
| 3 |
+
size 759325874
|
models/embeddings/aligned/th_64d.meta.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"lang": "th", "dim": 64, "max_seq_len": 512, "is_aligned": true}
|
models/embeddings/aligned/th_64d.projection.npy
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:de0386f466c0045e3352f51e12350b39f7866b809a1b24a83b8c5e44c7ccb1fa
|
| 3 |
+
size 16512
|
models/embeddings/aligned/th_64d_metadata.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"language": "th",
|
| 3 |
+
"dimension": 64,
|
| 4 |
+
"version": "aligned",
|
| 5 |
+
"hub_language": "en",
|
| 6 |
+
"seed_vocab_size": 110471,
|
| 7 |
+
"vocab_size": 450025
|
| 8 |
+
}
|
models/embeddings/monolingual/th_128d.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:218532ffd1ab8a530b7f6a92aad9884b8243b5d87185ede81b63f299cddbde50
|
| 3 |
+
size 1501738674
|
models/embeddings/monolingual/th_128d.meta.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"lang": "th", "dim": 128, "max_seq_len": 512, "is_aligned": false}
|
models/embeddings/monolingual/th_128d_metadata.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"language": "th",
|
| 3 |
+
"dimension": 128,
|
| 4 |
+
"version": "monolingual",
|
| 5 |
+
"training_params": {
|
| 6 |
+
"algorithm": "skipgram",
|
| 7 |
+
"min_count": 5,
|
| 8 |
+
"window": 5,
|
| 9 |
+
"negative": 5,
|
| 10 |
+
"epochs": 5,
|
| 11 |
+
"encoding_method": "rope",
|
| 12 |
+
"dim": 128,
|
| 13 |
+
"threads": 32
|
| 14 |
+
},
|
| 15 |
+
"vocab_size": 450025
|
| 16 |
+
}
|
models/embeddings/monolingual/th_32d.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d07e515211b39a047f35fe0cbc2638179bc51c080901fb9fff2d1d0a4644119
|
| 3 |
+
size 388119474
|
models/embeddings/monolingual/th_32d.meta.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"lang": "th", "dim": 32, "max_seq_len": 512, "is_aligned": false}
|
models/embeddings/monolingual/th_32d_metadata.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"language": "th",
|
| 3 |
+
"dimension": 32,
|
| 4 |
+
"version": "monolingual",
|
| 5 |
+
"training_params": {
|
| 6 |
+
"algorithm": "skipgram",
|
| 7 |
+
"min_count": 5,
|
| 8 |
+
"window": 5,
|
| 9 |
+
"negative": 5,
|
| 10 |
+
"epochs": 5,
|
| 11 |
+
"encoding_method": "rope",
|
| 12 |
+
"dim": 32,
|
| 13 |
+
"threads": 32
|
| 14 |
+
},
|
| 15 |
+
"vocab_size": 450025
|
| 16 |
+
}
|
models/embeddings/monolingual/th_64d.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:428f8a66daf50dd434a1e9ab558a26f5a244d30dd9820833a63c5fc65c2693f9
|
| 3 |
+
size 759325874
|
models/embeddings/monolingual/th_64d.meta.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"lang": "th", "dim": 64, "max_seq_len": 512, "is_aligned": false}
|
models/embeddings/monolingual/th_64d_metadata.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"language": "th",
|
| 3 |
+
"dimension": 64,
|
| 4 |
+
"version": "monolingual",
|
| 5 |
+
"training_params": {
|
| 6 |
+
"algorithm": "skipgram",
|
| 7 |
+
"min_count": 5,
|
| 8 |
+
"window": 5,
|
| 9 |
+
"negative": 5,
|
| 10 |
+
"epochs": 5,
|
| 11 |
+
"encoding_method": "rope",
|
| 12 |
+
"dim": 64,
|
| 13 |
+
"threads": 32
|
| 14 |
+
},
|
| 15 |
+
"vocab_size": 450025
|
| 16 |
+
}
|
models/subword_markov/th_markov_ctx1_subword.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e8c3c5453d56bd117013be521d5bd13c0f76c36bf62924b3f4a51650ddf36db0
|
| 3 |
+
size 2622929
|
models/subword_markov/th_markov_ctx1_subword_metadata.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"context_size": 1,
|
| 3 |
+
"variant": "subword",
|
| 4 |
+
"language": "th",
|
| 5 |
+
"unique_contexts": 37876,
|
| 6 |
+
"total_transitions": 362921572
|
| 7 |
+
}
|
models/subword_markov/th_markov_ctx2_subword.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:05a9869caa69adab3f395475b46538abb776112280f5fc27f58133f67800d092
|
| 3 |
+
size 17295138
|
models/subword_markov/th_markov_ctx2_subword_metadata.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"context_size": 2,
|
| 3 |
+
"variant": "subword",
|
| 4 |
+
"language": "th",
|
| 5 |
+
"unique_contexts": 462626,
|
| 6 |
+
"total_transitions": 362743892
|
| 7 |
+
}
|
models/subword_markov/th_markov_ctx3_subword.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:345e53928053840a0c3b726f1f82f78d34a11dd93e17e02111a867a4b6c04662
|
| 3 |
+
size 79210403
|
models/subword_markov/th_markov_ctx3_subword_metadata.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"context_size": 3,
|
| 3 |
+
"variant": "subword",
|
| 4 |
+
"language": "th",
|
| 5 |
+
"unique_contexts": 2452254,
|
| 6 |
+
"total_transitions": 362566212
|
| 7 |
+
}
|
models/subword_markov/th_markov_ctx4_subword.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:747a271090dce700234a4dba1a5c09b7a185958ac62c1ef233f9aa4f051a9488
|
| 3 |
+
size 258530071
|
models/subword_markov/th_markov_ctx4_subword_metadata.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"context_size": 4,
|
| 3 |
+
"variant": "subword",
|
| 4 |
+
"language": "th",
|
| 5 |
+
"unique_contexts": 9576634,
|
| 6 |
+
"total_transitions": 362388532
|
| 7 |
+
}
|
models/subword_ngram/th_2gram_subword.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0aebbfe9888b4c8bab458ed075a9cdfd31f77d45eed5bdc69bb930cb3a731172
|
| 3 |
+
size 1736471
|
models/subword_ngram/th_2gram_subword_metadata.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"n": 2,
|
| 3 |
+
"variant": "subword",
|
| 4 |
+
"language": "th",
|
| 5 |
+
"unique_ngrams": 124885,
|
| 6 |
+
"total_ngrams": 362921572
|
| 7 |
+
}
|
models/subword_ngram/th_3gram_subword.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1c9eb816e19268fe6597eab793ca3a8d7328dd8da5e65854b2044eaf7b65fe32
|
| 3 |
+
size 14441806
|
models/subword_ngram/th_3gram_subword_metadata.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"n": 3,
|
| 3 |
+
"variant": "subword",
|
| 4 |
+
"language": "th",
|
| 5 |
+
"unique_ngrams": 1000290,
|
| 6 |
+
"total_ngrams": 362743892
|
| 7 |
+
}
|
models/subword_ngram/th_4gram_subword.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28c56d804950cf11d46c246b7993385a053bd380098680d51c77ec6b51878ee2
|
| 3 |
+
size 73797867
|
models/subword_ngram/th_4gram_subword_metadata.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"n": 4,
|
| 3 |
+
"variant": "subword",
|
| 4 |
+
"language": "th",
|
| 5 |
+
"unique_ngrams": 4905540,
|
| 6 |
+
"total_ngrams": 362566212
|
| 7 |
+
}
|
models/subword_ngram/th_5gram_subword.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:33364b72d5499fdee7bd3f680816358955dc6378191fddcc83325291ce024e24
|
| 3 |
+
size 188702220
|
models/subword_ngram/th_5gram_subword_metadata.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"n": 5,
|
| 3 |
+
"variant": "subword",
|
| 4 |
+
"language": "th",
|
| 5 |
+
"unique_ngrams": 11885834,
|
| 6 |
+
"total_ngrams": 362388532
|
| 7 |
+
}
|
models/tokenizer/th_tokenizer_16k.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:49b51e2966fdddc4c863f8cfd159805d5d4379b51de3a4cc8bc7d258051ef1ae
|
| 3 |
+
size 615860
|
models/tokenizer/th_tokenizer_16k.vocab
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
models/tokenizer/th_tokenizer_32k.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3a0a9933b171da690a414690b6a33024b84a22fd4a515ee531f8927166804b0
|
| 3 |
+
size 1052286
|
models/tokenizer/th_tokenizer_32k.vocab
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
models/tokenizer/th_tokenizer_64k.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:43b740523f25f695c94ba5d474ac068f42a4dd6adf397bfc22bf250b9c442e95
|
| 3 |
+
size 1974648
|
models/tokenizer/th_tokenizer_64k.vocab
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
models/tokenizer/th_tokenizer_8k.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f503f1d1cc5b183693467a5604153e463f100f9e179cd905c1777a1794af70c1
|
| 3 |
+
size 411357
|
models/tokenizer/th_tokenizer_8k.vocab
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
models/vocabulary/th_vocabulary.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:edf046cbdf559a751d6381793569a7d0cdc3b19f746370da0a7e5de3f36ddf3a
|
| 3 |
+
size 33232671
|
models/vocabulary/th_vocabulary_metadata.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"language": "th",
|
| 3 |
+
"vocabulary_size": 1276542,
|
| 4 |
+
"variant": "full",
|
| 5 |
+
"statistics": {
|
| 6 |
+
"type_token_ratio": 0.2484188561316107,
|
| 7 |
+
"coverage": {
|
| 8 |
+
"top_100": 0.23482958261161382,
|
| 9 |
+
"top_1000": 0.3563300316759696,
|
| 10 |
+
"top_5000": 0.45467962493242736,
|
| 11 |
+
"top_10000": 0.500678290847444
|
| 12 |
+
},
|
| 13 |
+
"hapax_count": 7005297,
|
| 14 |
+
"hapax_ratio": 0.8458624950328061,
|
| 15 |
+
"total_documents": 177680
|
| 16 |
+
}
|
| 17 |
+
}
|
models/vocabulary/th_vocabulary_top.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:60ecb1d1a6b66f6e5944dede98ceced8bd8af2d6cf5527a9eeb3a536ffafce17
|
| 3 |
+
size 25231396
|