Upload all models and assets for ary (20251001)
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- .gitattributes +1 -0
- README.md +305 -138
- models/embeddings/monolingual/ary_128d.bin +2 -2
- models/embeddings/monolingual/ary_128d_metadata.json +5 -3
- models/embeddings/monolingual/ary_32d.bin +2 -2
- models/embeddings/monolingual/ary_32d_metadata.json +5 -3
- models/embeddings/monolingual/ary_64d.bin +2 -2
- models/embeddings/monolingual/ary_64d_metadata.json +5 -3
- models/subword_markov/ary_markov_ctx1_subword.parquet +2 -2
- models/subword_markov/ary_markov_ctx1_subword_metadata.json +2 -2
- models/subword_markov/ary_markov_ctx2_subword.parquet +2 -2
- models/subword_markov/ary_markov_ctx2_subword_metadata.json +2 -2
- models/subword_markov/ary_markov_ctx3_subword.parquet +2 -2
- models/subword_markov/ary_markov_ctx3_subword_metadata.json +2 -2
- models/subword_markov/ary_markov_ctx4_subword.parquet +2 -2
- models/subword_markov/ary_markov_ctx4_subword_metadata.json +2 -2
- models/subword_ngram/ary_2gram_subword.parquet +2 -2
- models/subword_ngram/ary_2gram_subword_metadata.json +2 -2
- models/subword_ngram/ary_3gram_subword.parquet +2 -2
- models/subword_ngram/ary_3gram_subword_metadata.json +2 -2
- models/subword_ngram/ary_4gram_subword.parquet +2 -2
- models/subword_ngram/ary_4gram_subword_metadata.json +2 -2
- models/tokenizer/ary_tokenizer_16k.model +2 -2
- models/tokenizer/ary_tokenizer_16k.vocab +0 -0
- models/tokenizer/ary_tokenizer_32k.model +2 -2
- models/tokenizer/ary_tokenizer_32k.vocab +0 -0
- models/tokenizer/ary_tokenizer_64k.model +2 -2
- models/tokenizer/ary_tokenizer_64k.vocab +0 -0
- models/tokenizer/ary_tokenizer_8k.model +2 -2
- models/tokenizer/ary_tokenizer_8k.vocab +0 -0
- models/vocabulary/ary_vocabulary.parquet +2 -2
- models/vocabulary/ary_vocabulary_metadata.json +10 -9
- models/word_markov/ary_markov_ctx1_word.parquet +2 -2
- models/word_markov/ary_markov_ctx1_word_metadata.json +2 -2
- models/word_markov/ary_markov_ctx2_word.parquet +2 -2
- models/word_markov/ary_markov_ctx2_word_metadata.json +2 -2
- models/word_markov/ary_markov_ctx3_word.parquet +2 -2
- models/word_markov/ary_markov_ctx3_word_metadata.json +2 -2
- models/word_markov/ary_markov_ctx4_word.parquet +2 -2
- models/word_markov/ary_markov_ctx4_word_metadata.json +2 -2
- models/word_ngram/ary_2gram_word.parquet +2 -2
- models/word_ngram/ary_2gram_word_metadata.json +2 -2
- models/word_ngram/ary_3gram_word.parquet +2 -2
- models/word_ngram/ary_3gram_word_metadata.json +2 -2
- models/word_ngram/ary_4gram_word.parquet +2 -2
- models/word_ngram/ary_4gram_word_metadata.json +2 -2
- visualizations/embedding_isotropy.png +0 -0
- visualizations/embedding_norms.png +0 -0
- visualizations/embedding_similarity.png +2 -2
- visualizations/markov_branching.png +0 -0
.gitattributes
CHANGED
|
@@ -38,3 +38,4 @@ visualizations/performance_dashboard.png filter=lfs diff=lfs merge=lfs -text
|
|
| 38 |
visualizations/tsne_sentences.png filter=lfs diff=lfs merge=lfs -text
|
| 39 |
visualizations/tsne_words.png filter=lfs diff=lfs merge=lfs -text
|
| 40 |
visualizations/zipf_law.png filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 38 |
visualizations/tsne_sentences.png filter=lfs diff=lfs merge=lfs -text
|
| 39 |
visualizations/tsne_words.png filter=lfs diff=lfs merge=lfs -text
|
| 40 |
visualizations/zipf_law.png filter=lfs diff=lfs merge=lfs -text
|
| 41 |
+
visualizations/position_encoding_comparison.png filter=lfs diff=lfs merge=lfs -text
|
README.md
CHANGED
|
@@ -23,14 +23,14 @@ dataset_info:
|
|
| 23 |
metrics:
|
| 24 |
- name: best_compression_ratio
|
| 25 |
type: compression
|
| 26 |
-
value:
|
| 27 |
- name: best_isotropy
|
| 28 |
type: isotropy
|
| 29 |
-
value: 0.
|
| 30 |
- name: vocabulary_size
|
| 31 |
type: vocab
|
| 32 |
-
value:
|
| 33 |
-
generated:
|
| 34 |
---
|
| 35 |
|
| 36 |
# Moroccan Arabic - Wikilangs Models
|
|
@@ -44,12 +44,13 @@ We analyze tokenizers, n-gram models, Markov chains, vocabulary statistics, and
|
|
| 44 |
### Models & Assets
|
| 45 |
|
| 46 |
- Tokenizers (8k, 16k, 32k, 64k)
|
| 47 |
-
- N-gram models (2, 3, 4-gram)
|
| 48 |
-
- Markov chains (context of 1, 2, 3 and
|
| 49 |
- Subword N-gram and Markov chains
|
| 50 |
-
- Embeddings in various sizes and dimensions
|
| 51 |
- Language Vocabulary
|
| 52 |
- Language Statistics
|
|
|
|
| 53 |

|
| 54 |
|
| 55 |
### Analysis and Evaluation
|
|
@@ -59,7 +60,8 @@ We analyze tokenizers, n-gram models, Markov chains, vocabulary statistics, and
|
|
| 59 |
- [3. Markov Chain Evaluation](#3-markov-chain-evaluation)
|
| 60 |
- [4. Vocabulary Analysis](#4-vocabulary-analysis)
|
| 61 |
- [5. Word Embeddings Evaluation](#5-word-embeddings-evaluation)
|
| 62 |
-
- [6.
|
|
|
|
| 63 |
- [Metrics Glossary](#appendix-metrics-glossary--interpretation-guide)
|
| 64 |
- [Visualizations Index](#visualizations-index)
|
| 65 |
|
|
@@ -68,56 +70,57 @@ We analyze tokenizers, n-gram models, Markov chains, vocabulary statistics, and
|
|
| 68 |
|
| 69 |

|
| 70 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 71 |
### Results
|
| 72 |
|
| 73 |
| Vocab Size | Compression | Avg Token Len | UNK Rate | Total Tokens |
|
| 74 |
|------------|-------------|---------------|----------|--------------|
|
| 75 |
-
| **8k** | 3.
|
| 76 |
-
| **16k** | 3.
|
| 77 |
-
| **32k** |
|
| 78 |
-
| **64k** |
|
| 79 |
|
| 80 |
### Tokenization Examples
|
| 81 |
|
| 82 |
Below are sample sentences tokenized with each vocabulary size:
|
| 83 |
|
| 84 |
-
**Sample 1:**
|
| 85 |
-
|
| 86 |
-
مصادر
|
| 87 |
-
|
| 88 |
-
تصنيف:زيادة 1954
|
| 89 |
-
تصنيف:ناس حيين...`
|
| 90 |
|
| 91 |
| Vocab | Tokens | Count |
|
| 92 |
|-------|--------|-------|
|
| 93 |
-
| 8k |
|
| 94 |
-
| 16k |
|
| 95 |
-
| 32k |
|
| 96 |
-
| 64k |
|
| 97 |
|
| 98 |
-
**Sample 2:**
|
| 99 |
|
| 100 |
| Vocab | Tokens | Count |
|
| 101 |
|-------|--------|-------|
|
| 102 |
-
| 8k |
|
| 103 |
-
| 16k |
|
| 104 |
-
| 32k |
|
| 105 |
-
| 64k |
|
| 106 |
|
| 107 |
-
**Sample 3:**
|
| 108 |
|
| 109 |
| Vocab | Tokens | Count |
|
| 110 |
|-------|--------|-------|
|
| 111 |
-
| 8k |
|
| 112 |
-
| 16k |
|
| 113 |
-
| 32k |
|
| 114 |
-
| 64k |
|
| 115 |
|
| 116 |
|
| 117 |
### Key Findings
|
| 118 |
|
| 119 |
-
- **Best Compression:** 64k achieves
|
| 120 |
-
- **Lowest UNK Rate:** 8k with 0.
|
| 121 |
- **Trade-off:** Larger vocabularies improve compression but increase model size
|
| 122 |
- **Recommendation:** 32k vocabulary provides optimal balance for production use
|
| 123 |
|
|
@@ -126,57 +129,89 @@ Below are sample sentences tokenized with each vocabulary size:
|
|
| 126 |
|
| 127 |

|
| 128 |
|
|
|
|
|
|
|
| 129 |

|
| 130 |
|
| 131 |
### Results
|
| 132 |
|
| 133 |
-
| N-gram | Perplexity | Entropy | Unique N-grams | Top-100 Coverage | Top-1000 Coverage |
|
| 134 |
-
|
| 135 |
-
| **2-gram** |
|
| 136 |
-
| **2-gram** |
|
| 137 |
-
| **3-gram** |
|
| 138 |
-
| **3-gram** |
|
| 139 |
-
| **4-gram** |
|
| 140 |
-
| **4-gram** |
|
| 141 |
|
| 142 |
### Top 5 N-grams by Size
|
| 143 |
|
| 144 |
-
**2-grams:**
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 145 |
|
| 146 |
| Rank | N-gram | Count |
|
| 147 |
|------|--------|-------|
|
| 148 |
-
| 1 |
|
| 149 |
-
| 2 |
|
| 150 |
-
| 3 |
|
| 151 |
-
| 4 |
|
| 152 |
-
| 5 |
|
| 153 |
|
| 154 |
-
**
|
| 155 |
|
| 156 |
| Rank | N-gram | Count |
|
| 157 |
|------|--------|-------|
|
| 158 |
-
| 1 |
|
| 159 |
-
| 2 |
|
| 160 |
-
| 3 |
|
| 161 |
-
| 4 |
|
| 162 |
-
| 5 |
|
| 163 |
|
| 164 |
-
**
|
| 165 |
|
| 166 |
| Rank | N-gram | Count |
|
| 167 |
|------|--------|-------|
|
| 168 |
-
| 1 |
|
| 169 |
-
| 2 |
|
| 170 |
-
| 3 |
|
| 171 |
-
| 4 |
|
| 172 |
-
| 5 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 173 |
|
| 174 |
|
| 175 |
### Key Findings
|
| 176 |
|
| 177 |
-
- **Best Perplexity:** 2-gram with
|
| 178 |
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
|
| 179 |
-
- **Coverage:** Top-1000 patterns cover ~
|
| 180 |
- **Recommendation:** 4-gram or 5-gram for best predictive performance
|
| 181 |
|
| 182 |
---
|
|
@@ -184,55 +219,86 @@ Below are sample sentences tokenized with each vocabulary size:
|
|
| 184 |
|
| 185 |

|
| 186 |
|
|
|
|
|
|
|
| 187 |

|
| 188 |
|
| 189 |
### Results
|
| 190 |
|
| 191 |
-
| Context | Avg Entropy | Perplexity | Branching Factor | Unique Contexts | Predictability |
|
| 192 |
-
|
| 193 |
-
| **1** | 0.
|
| 194 |
-
| **1** | 1.
|
| 195 |
-
| **2** | 0.
|
| 196 |
-
| **2** | 0.
|
| 197 |
-
| **3** | 0.
|
| 198 |
-
| **3** | 0.
|
| 199 |
-
| **4** | 0.
|
| 200 |
-
| **4** | 0.
|
|
|
|
|
|
|
| 201 |
|
| 202 |
-
|
| 203 |
|
| 204 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 205 |
|
| 206 |
**Context Size 1:**
|
| 207 |
|
| 208 |
-
1.
|
| 209 |
-
2.
|
| 210 |
-
3.
|
| 211 |
|
| 212 |
**Context Size 2:**
|
| 213 |
|
| 214 |
-
1.
|
| 215 |
-
2.
|
| 216 |
-
3.
|
| 217 |
|
| 218 |
**Context Size 3:**
|
| 219 |
|
| 220 |
-
1.
|
| 221 |
-
2.
|
| 222 |
-
3.
|
| 223 |
|
| 224 |
**Context Size 4:**
|
| 225 |
|
| 226 |
-
1.
|
| 227 |
-
2.
|
| 228 |
-
3.
|
| 229 |
|
| 230 |
|
| 231 |
### Key Findings
|
| 232 |
|
| 233 |
-
- **Best Predictability:** Context-4 with
|
| 234 |
- **Branching Factor:** Decreases with context size (more deterministic)
|
| 235 |
-
- **Memory Trade-off:** Larger contexts require more storage (
|
| 236 |
- **Recommendation:** Context-3 or Context-4 for text generation
|
| 237 |
|
| 238 |
---
|
|
@@ -248,64 +314,64 @@ Below are text samples generated from each Markov chain model:
|
|
| 248 |
|
| 249 |
| Metric | Value |
|
| 250 |
|--------|-------|
|
| 251 |
-
| Vocabulary Size |
|
| 252 |
-
| Total Tokens |
|
| 253 |
-
| Mean Frequency |
|
| 254 |
| Median Frequency | 4 |
|
| 255 |
-
| Frequency Std Dev |
|
| 256 |
|
| 257 |
### Most Common Words
|
| 258 |
|
| 259 |
| Rank | Word | Frequency |
|
| 260 |
|------|------|-----------|
|
| 261 |
-
| 1 | ف |
|
| 262 |
-
| 2 | د |
|
| 263 |
-
| 3 | و |
|
| 264 |
-
| 4 |
|
| 265 |
-
| 5 |
|
| 266 |
-
| 6 |
|
| 267 |
-
| 7 |
|
| 268 |
-
| 8 |
|
| 269 |
-
| 9 |
|
| 270 |
-
| 10 | ب |
|
| 271 |
|
| 272 |
### Least Common Words (from vocabulary)
|
| 273 |
|
| 274 |
| Rank | Word | Frequency |
|
| 275 |
|------|------|-----------|
|
| 276 |
-
| 1 |
|
| 277 |
-
| 2 |
|
| 278 |
-
| 3 |
|
| 279 |
-
| 4 |
|
| 280 |
-
| 5 |
|
| 281 |
-
| 6 |
|
| 282 |
-
| 7 |
|
| 283 |
-
| 8 |
|
| 284 |
-
| 9 |
|
| 285 |
-
| 10 |
|
| 286 |
|
| 287 |
### Zipf's Law Analysis
|
| 288 |
|
| 289 |
| Metric | Value |
|
| 290 |
|--------|-------|
|
| 291 |
-
| Zipf Coefficient | 1.
|
| 292 |
-
| R² (Goodness of Fit) | 0.
|
| 293 |
| Adherence Quality | **excellent** |
|
| 294 |
|
| 295 |
### Coverage Analysis
|
| 296 |
|
| 297 |
| Top N Words | Coverage |
|
| 298 |
|-------------|----------|
|
| 299 |
-
| Top 100 |
|
| 300 |
-
| Top 1,000 |
|
| 301 |
-
| Top 5,000 |
|
| 302 |
-
| Top 10,000 |
|
| 303 |
|
| 304 |
### Key Findings
|
| 305 |
|
| 306 |
-
- **Zipf Compliance:** R²=0.
|
| 307 |
-
- **High Frequency Dominance:** Top 100 words cover
|
| 308 |
-
- **Long Tail:**
|
| 309 |
|
| 310 |
---
|
| 311 |
## 5. Word Embeddings Evaluation
|
|
@@ -318,24 +384,122 @@ Below are text samples generated from each Markov chain model:
|
|
| 318 |
|
| 319 |

|
| 320 |
|
| 321 |
-
### Model Comparison
|
| 322 |
|
| 323 |
-
|
| 324 |
-
|
| 325 |
-
|
| 326 |
-
|
| 327 |
-
|
| 328 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 329 |
|
| 330 |
### Key Findings
|
| 331 |
|
| 332 |
-
- **Best Isotropy:** mono_32d with 0.
|
| 333 |
-
- **
|
| 334 |
-
- **
|
| 335 |
-
- **Recommendation:**
|
| 336 |
|
| 337 |
---
|
| 338 |
-
## 6.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 339 |
|
| 340 |

|
| 341 |
|
|
@@ -343,11 +507,12 @@ Below are text samples generated from each Markov chain model:
|
|
| 343 |
|
| 344 |
| Component | Recommended | Rationale |
|
| 345 |
|-----------|-------------|-----------|
|
| 346 |
-
| Tokenizer | **
|
| 347 |
-
| N-gram | **
|
| 348 |
-
| Markov | **Context-4** | Highest predictability (
|
| 349 |
| Embeddings | **100d** | Balanced semantic capture and isotropy |
|
| 350 |
|
|
|
|
| 351 |
---
|
| 352 |
## Appendix: Metrics Glossary & Interpretation Guide
|
| 353 |
|
|
@@ -537,7 +702,8 @@ If you use these models in your research, please cite:
|
|
| 537 |
author = {Kamali, Omar},
|
| 538 |
title = {Wikilangs: Open NLP Models for Wikipedia Languages},
|
| 539 |
year = {2025},
|
| 540 |
-
|
|
|
|
| 541 |
url = {https://huggingface.co/wikilangs}
|
| 542 |
institution = {Omneity Labs}
|
| 543 |
}
|
|
@@ -553,7 +719,8 @@ MIT License - Free for academic and commercial use.
|
|
| 553 |
- 🤗 Models: [huggingface.co/wikilangs](https://huggingface.co/wikilangs)
|
| 554 |
- 📊 Data: [wikipedia-monthly](https://huggingface.co/datasets/omarkamali/wikipedia-monthly)
|
| 555 |
- 👤 Author: [Omar Kamali](https://huggingface.co/omarkamali)
|
|
|
|
| 556 |
---
|
| 557 |
*Generated by Wikilangs Models Pipeline*
|
| 558 |
|
| 559 |
-
*Report Date:
|
|
|
|
| 23 |
metrics:
|
| 24 |
- name: best_compression_ratio
|
| 25 |
type: compression
|
| 26 |
+
value: 4.180
|
| 27 |
- name: best_isotropy
|
| 28 |
type: isotropy
|
| 29 |
+
value: 0.8384
|
| 30 |
- name: vocabulary_size
|
| 31 |
type: vocab
|
| 32 |
+
value: 0
|
| 33 |
+
generated: 2026-01-03
|
| 34 |
---
|
| 35 |
|
| 36 |
# Moroccan Arabic - Wikilangs Models
|
|
|
|
| 44 |
### Models & Assets
|
| 45 |
|
| 46 |
- Tokenizers (8k, 16k, 32k, 64k)
|
| 47 |
+
- N-gram models (2, 3, 4, 5-gram)
|
| 48 |
+
- Markov chains (context of 1, 2, 3, 4 and 5)
|
| 49 |
- Subword N-gram and Markov chains
|
| 50 |
+
- Embeddings in various sizes and dimensions (aligned and unaligned)
|
| 51 |
- Language Vocabulary
|
| 52 |
- Language Statistics
|
| 53 |
+
|
| 54 |

|
| 55 |
|
| 56 |
### Analysis and Evaluation
|
|
|
|
| 60 |
- [3. Markov Chain Evaluation](#3-markov-chain-evaluation)
|
| 61 |
- [4. Vocabulary Analysis](#4-vocabulary-analysis)
|
| 62 |
- [5. Word Embeddings Evaluation](#5-word-embeddings-evaluation)
|
| 63 |
+
- [6. Morphological Analysis (Experimental)](#6-morphological-analysis)
|
| 64 |
+
- [7. Summary & Recommendations](#7-summary--recommendations)
|
| 65 |
- [Metrics Glossary](#appendix-metrics-glossary--interpretation-guide)
|
| 66 |
- [Visualizations Index](#visualizations-index)
|
| 67 |
|
|
|
|
| 70 |
|
| 71 |

|
| 72 |
|
| 73 |
+

|
| 74 |
+
|
| 75 |
+

|
| 76 |
+
|
| 77 |
+

|
| 78 |
+
|
| 79 |
### Results
|
| 80 |
|
| 81 |
| Vocab Size | Compression | Avg Token Len | UNK Rate | Total Tokens |
|
| 82 |
|------------|-------------|---------------|----------|--------------|
|
| 83 |
+
| **8k** | 3.512x | 3.52 | 0.0922% | 278,716 |
|
| 84 |
+
| **16k** | 3.778x | 3.78 | 0.0992% | 259,059 |
|
| 85 |
+
| **32k** | 4.002x | 4.01 | 0.1051% | 244,561 |
|
| 86 |
+
| **64k** | 4.180x 🏆 | 4.18 | 0.1098% | 234,163 |
|
| 87 |
|
| 88 |
### Tokenization Examples
|
| 89 |
|
| 90 |
Below are sample sentences tokenized with each vocabulary size:
|
| 91 |
|
| 92 |
+
**Sample 1:** `مصادر شوف تا داريجة تاريخ لكتابة ب داريجة ليستة د لمكتوبات ب داريجة ليستة د لكتو...`
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 93 |
|
| 94 |
| Vocab | Tokens | Count |
|
| 95 |
|-------|--------|-------|
|
| 96 |
+
| 8k | `▁مصادر ▁شوف ▁تا ▁داريجة ▁تاريخ ▁لكتابة ▁ب ▁داريجة ▁ليستة ▁د ... (+22 more)` | 32 |
|
| 97 |
+
| 16k | `▁مصادر ▁شوف ▁تا ▁داريجة ▁تاريخ ▁لكتابة ▁ب ▁داريجة ▁ليستة ▁د ... (+20 more)` | 30 |
|
| 98 |
+
| 32k | `▁مصادر ▁شوف ▁تا ▁داريجة ▁تاريخ ▁لكتابة ▁ب ▁داريجة ▁ليستة ▁د ... (+20 more)` | 30 |
|
| 99 |
+
| 64k | `▁مصادر ▁شوف ▁تا ▁داريجة ▁تاريخ ▁لكتابة ▁ب ▁داريجة ▁ليستة ▁د ... (+20 more)` | 30 |
|
| 100 |
|
| 101 |
+
**Sample 2:** `أمين رباطي (مزيود ف يوليوز هو كوايري مغريبي. مصادر مغريبي د رجال حيين`
|
| 102 |
|
| 103 |
| Vocab | Tokens | Count |
|
| 104 |
|-------|--------|-------|
|
| 105 |
+
| 8k | `▁أمين ▁رباط ي ▁( مزيود ▁ف ▁يوليوز ▁هو ▁كوايري ▁مغريبي ... (+6 more)` | 16 |
|
| 106 |
+
| 16k | `▁أمين ▁رباط ي ▁( مزيود ▁ف ▁يوليوز ▁هو ▁كوايري ▁مغريبي ... (+6 more)` | 16 |
|
| 107 |
+
| 32k | `▁أ��ين ▁رباطي ▁( مزيود ▁ف ▁يوليوز ▁هو ▁كوايري ▁مغريبي . ... (+5 more)` | 15 |
|
| 108 |
+
| 64k | `▁أمين ▁رباطي ▁( مزيود ▁ف ▁يوليوز ▁هو ▁كوايري ▁مغريبي . ... (+5 more)` | 15 |
|
| 109 |
|
| 110 |
+
**Sample 3:** `هادي صفحة د التوضيح، كلمة دوري يمكن يكونو عندها هاد لمعاني: طابلو دوري دوري أبطا...`
|
| 111 |
|
| 112 |
| Vocab | Tokens | Count |
|
| 113 |
|-------|--------|-------|
|
| 114 |
+
| 8k | `▁هادي ▁صفحة ▁د ▁التوضيح ، ▁كلمة ▁دوري ▁يمكن ▁يكونو ▁عندها ... (+10 more)` | 20 |
|
| 115 |
+
| 16k | `▁هادي ▁صفحة ▁د ▁التوضيح ، ▁كلمة ▁دوري ▁يمكن ▁يكونو ▁عندها ... (+9 more)` | 19 |
|
| 116 |
+
| 32k | `▁هادي ▁صفحة ▁د ▁التوضيح ، ▁كلمة ▁دوري ▁يمكن ▁يكونو ▁عندها ... (+9 more)` | 19 |
|
| 117 |
+
| 64k | `▁هادي ▁صفحة ▁د ▁التوضيح ، ▁كلمة ▁دوري ▁يمكن ▁يكونو ▁عندها ... (+9 more)` | 19 |
|
| 118 |
|
| 119 |
|
| 120 |
### Key Findings
|
| 121 |
|
| 122 |
+
- **Best Compression:** 64k achieves 4.180x compression
|
| 123 |
+
- **Lowest UNK Rate:** 8k with 0.0922% unknown tokens
|
| 124 |
- **Trade-off:** Larger vocabularies improve compression but increase model size
|
| 125 |
- **Recommendation:** 32k vocabulary provides optimal balance for production use
|
| 126 |
|
|
|
|
| 129 |
|
| 130 |

|
| 131 |
|
| 132 |
+

|
| 133 |
+
|
| 134 |

|
| 135 |
|
| 136 |
### Results
|
| 137 |
|
| 138 |
+
| N-gram | Variant | Perplexity | Entropy | Unique N-grams | Top-100 Coverage | Top-1000 Coverage |
|
| 139 |
+
|--------|---------|------------|---------|----------------|------------------|-------------------|
|
| 140 |
+
| **2-gram** | Word | 6,129 | 12.58 | 35,218 | 24.5% | 53.4% |
|
| 141 |
+
| **2-gram** | Subword | 415 🏆 | 8.70 | 5,585 | 58.6% | 96.6% |
|
| 142 |
+
| **3-gram** | Word | 4,994 | 12.29 | 39,702 | 28.5% | 58.9% |
|
| 143 |
+
| **3-gram** | Subword | 3,624 | 11.82 | 41,944 | 23.5% | 61.8% |
|
| 144 |
+
| **4-gram** | Word | 6,987 | 12.77 | 63,706 | 28.4% | 55.4% |
|
| 145 |
+
| **4-gram** | Subword | 18,675 | 14.19 | 204,568 | 12.3% | 37.2% |
|
| 146 |
|
| 147 |
### Top 5 N-grams by Size
|
| 148 |
|
| 149 |
+
**2-grams (Word):**
|
| 150 |
+
|
| 151 |
+
| Rank | N-gram | Count |
|
| 152 |
+
|------|--------|-------|
|
| 153 |
+
| 1 | `واصلة ل` | 8,540 |
|
| 154 |
+
| 2 | `نسبة د` | 7,170 |
|
| 155 |
+
| 3 | `ف لمغريب` | 6,247 |
|
| 156 |
+
| 4 | `ف إقليم` | 6,016 |
|
| 157 |
+
| 5 | `ف نسبة` | 4,265 |
|
| 158 |
+
|
| 159 |
+
**3-grams (Word):**
|
| 160 |
|
| 161 |
| Rank | N-gram | Count |
|
| 162 |
|------|--------|-------|
|
| 163 |
+
| 1 | `ف نسبة د` | 4,264 |
|
| 164 |
+
| 2 | `فيها مصدر و` | 3,236 |
|
| 165 |
+
| 3 | `و نسبة د` | 2,894 |
|
| 166 |
+
| 4 | `مصدر و بايت` | 2,856 |
|
| 167 |
+
| 5 | `اللي خدامين ف` | 2,759 |
|
| 168 |
|
| 169 |
+
**4-grams (Word):**
|
| 170 |
|
| 171 |
| Rank | N-gram | Count |
|
| 172 |
|------|--------|-------|
|
| 173 |
+
| 1 | `فيها مصدر و بايت` | 2,856 |
|
| 174 |
+
| 2 | `نسبة نّاس اللي خدامين` | 2,705 |
|
| 175 |
+
| 3 | `نّاس اللي خدامين ف` | 2,593 |
|
| 176 |
+
| 4 | `على حساب لإحصاء الرسمي` | 2,501 |
|
| 177 |
+
| 5 | `لعاداد د سّكان ديالو` | 2,500 |
|
| 178 |
|
| 179 |
+
**2-grams (Subword):**
|
| 180 |
|
| 181 |
| Rank | N-gram | Count |
|
| 182 |
|------|--------|-------|
|
| 183 |
+
| 1 | `ا ل` | 293,281 |
|
| 184 |
+
| 2 | `_ ل` | 265,615 |
|
| 185 |
+
| 3 | `ة _` | 209,034 |
|
| 186 |
+
| 4 | `_ ا` | 180,710 |
|
| 187 |
+
| 5 | `_ م` | 141,509 |
|
| 188 |
+
|
| 189 |
+
**3-grams (Subword):**
|
| 190 |
+
|
| 191 |
+
| Rank | N-gram | Count |
|
| 192 |
+
|------|--------|-------|
|
| 193 |
+
| 1 | `_ ا ل` | 176,897 |
|
| 194 |
+
| 2 | `_ ف _` | 80,240 |
|
| 195 |
+
| 3 | `_ د _` | 57,749 |
|
| 196 |
+
| 4 | `_ و _` | 57,033 |
|
| 197 |
+
| 5 | `ا ت _` | 56,985 |
|
| 198 |
+
|
| 199 |
+
**4-grams (Subword):**
|
| 200 |
+
|
| 201 |
+
| Rank | N-gram | Count |
|
| 202 |
+
|------|--------|-------|
|
| 203 |
+
| 1 | `_ د ي ا` | 43,807 |
|
| 204 |
+
| 2 | `د ي ا ل` | 43,597 |
|
| 205 |
+
| 3 | `ي ا ل _` | 30,362 |
|
| 206 |
+
| 4 | `د _ ا ل` | 29,177 |
|
| 207 |
+
| 5 | `_ م ن _` | 25,265 |
|
| 208 |
|
| 209 |
|
| 210 |
### Key Findings
|
| 211 |
|
| 212 |
+
- **Best Perplexity:** 2-gram (subword) with 415
|
| 213 |
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
|
| 214 |
+
- **Coverage:** Top-1000 patterns cover ~37% of corpus
|
| 215 |
- **Recommendation:** 4-gram or 5-gram for best predictive performance
|
| 216 |
|
| 217 |
---
|
|
|
|
| 219 |
|
| 220 |

|
| 221 |
|
| 222 |
+

|
| 223 |
+
|
| 224 |

|
| 225 |
|
| 226 |
### Results
|
| 227 |
|
| 228 |
+
| Context | Variant | Avg Entropy | Perplexity | Branching Factor | Unique Contexts | Predictability |
|
| 229 |
+
|---------|---------|-------------|------------|------------------|-----------------|----------------|
|
| 230 |
+
| **1** | Word | 0.8416 | 1.792 | 5.23 | 162,378 | 15.8% |
|
| 231 |
+
| **1** | Subword | 1.1133 | 2.163 | 8.05 | 2,149 | 0.0% |
|
| 232 |
+
| **2** | Word | 0.2252 | 1.169 | 1.49 | 849,251 | 77.5% |
|
| 233 |
+
| **2** | Subword | 0.8048 | 1.747 | 4.99 | 17,291 | 19.5% |
|
| 234 |
+
| **3** | Word | 0.0625 | 1.044 | 1.10 | 1,262,316 | 93.8% |
|
| 235 |
+
| **3** | Subword | 0.8001 | 1.741 | 4.09 | 86,361 | 20.0% |
|
| 236 |
+
| **4** | Word | 0.0215 🏆 | 1.015 | 1.04 | 1,391,141 | 97.9% |
|
| 237 |
+
| **4** | Subword | 0.6559 | 1.576 | 2.83 | 352,807 | 34.4% |
|
| 238 |
+
|
| 239 |
+
### Generated Text Samples (Word-based)
|
| 240 |
|
| 241 |
+
Below are text samples generated from each word-based Markov chain model:
|
| 242 |
|
| 243 |
+
**Context Size 1:**
|
| 244 |
+
|
| 245 |
+
1. `ف لجولة اللولة ديالو ماسك ب الريحة فاميلة ديال لوغات الأمازيغية هويتنا الوطنية بحال بنادم بشكل`
|
| 246 |
+
2. `د الشوماج واصلة ل كانت وحدة من جيهت بّاه إيرول ماسك أسس جمعية الشرف هو اللعاب`
|
| 247 |
+
3. `و بايت زادهوم داريجابوت 19 فاش كانو كايطراو ف نسبة لبطالة نّاس نّشيطين لّي يقدرو يخدمو`
|
| 248 |
+
|
| 249 |
+
**Context Size 2:**
|
| 250 |
+
|
| 251 |
+
1. `واصلة ل 5 و عدد لفاميلات تزاد ب 12 2 لمشاركات ف كأس افريقيا في البطولة ديال`
|
| 252 |
+
2. `نسبة د الناس النشيطين ف دوار أمرس واصلة ل 96 3 و نسبة د الجواج ف امزرو`
|
| 253 |
+
3. `ف لمغريب ف إقليم تارودانت جهة سوس ماسة ف لمغريب ف إقليم وارزازات جهة درعا تافيلالت ساكنين`
|
| 254 |
+
|
| 255 |
+
**Context Size 3:**
|
| 256 |
+
|
| 257 |
+
1. `ف نسبة د الناس النشيطين ف دوار تامكونسي واصلة ل 49 7 و لموعدّال د لعمر عند الجواج`
|
| 258 |
+
2. `فيها مصدر و علاين بايت د الصويرة`
|
| 259 |
+
3. `و نسبة د الشوماج واصلة ل 14 7 نوطات مصادر ف لمغريب ف إقليم لحوز زادهوم داريجابوت`
|
| 260 |
+
|
| 261 |
+
**Context Size 4:**
|
| 262 |
+
|
| 263 |
+
1. `نسبة نّاس اللي خدامين ف دّولة ولا لبيطاليين اللي سبق ليهوم مصادر طنجة تطوان الحسيمة قروية ف إقليم لح...`
|
| 264 |
+
2. `نّاس اللي خدامين ف دّولة ولا لبيطاليين اللي سبق ليهوم خدمو 6 7 نسبة نّاس اللي خدامين ف لپريڤي`
|
| 265 |
+
3. `على حساب لإحصاء الرسمي د عام إحصائيات إحصائيات عامة عدد السكان ديال تمزاوروت تزاد ب 18 6 و عدد`
|
| 266 |
+
|
| 267 |
+
|
| 268 |
+
### Generated Text Samples (Subword-based)
|
| 269 |
+
|
| 270 |
+
Below are text samples generated from each subword-based Markov chain model:
|
| 271 |
|
| 272 |
**Context Size 1:**
|
| 273 |
|
| 274 |
+
1. `_"أكابي_مناتحسن_`
|
| 275 |
+
2. `ايلممرسية_اهة،_ل`
|
| 276 |
+
3. `لم"ليعن_لنف_لميم`
|
| 277 |
|
| 278 |
**Context Size 2:**
|
| 279 |
|
| 280 |
+
1. `ال_لليزنيز،_إسلة_`
|
| 281 |
+
2. `_لعام_نخب_ور_تقرو`
|
| 282 |
+
3. `ة_سويسها_كولا_بحو`
|
| 283 |
|
| 284 |
**Context Size 3:**
|
| 285 |
|
| 286 |
+
1. `_اللات،_سورين._لڭر`
|
| 287 |
+
2. `_ف_نسبة_شبه_ولكرور`
|
| 288 |
+
3. `_د_لعالمغريب._هوّ_و`
|
| 289 |
|
| 290 |
**Context Size 4:**
|
| 291 |
|
| 292 |
+
1. `_ديال_على_حساب_لإحص`
|
| 293 |
+
2. `ديالو،_(a)_–_bringe`
|
| 294 |
+
3. `يال_التاني_توفى_عوا`
|
| 295 |
|
| 296 |
|
| 297 |
### Key Findings
|
| 298 |
|
| 299 |
+
- **Best Predictability:** Context-4 (word) with 97.9% predictability
|
| 300 |
- **Branching Factor:** Decreases with context size (more deterministic)
|
| 301 |
+
- **Memory Trade-off:** Larger contexts require more storage (352,807 contexts)
|
| 302 |
- **Recommendation:** Context-3 or Context-4 for text generation
|
| 303 |
|
| 304 |
---
|
|
|
|
| 314 |
|
| 315 |
| Metric | Value |
|
| 316 |
|--------|-------|
|
| 317 |
+
| Vocabulary Size | 70,940 |
|
| 318 |
+
| Total Tokens | 1,845,717 |
|
| 319 |
+
| Mean Frequency | 26.02 |
|
| 320 |
| Median Frequency | 4 |
|
| 321 |
+
| Frequency Std Dev | 518.94 |
|
| 322 |
|
| 323 |
### Most Common Words
|
| 324 |
|
| 325 |
| Rank | Word | Frequency |
|
| 326 |
|------|------|-----------|
|
| 327 |
+
| 1 | ف | 80,525 |
|
| 328 |
+
| 2 | د | 57,913 |
|
| 329 |
+
| 3 | و | 57,274 |
|
| 330 |
+
| 4 | ديال | 29,978 |
|
| 331 |
+
| 5 | من | 25,568 |
|
| 332 |
+
| 6 | ل | 23,006 |
|
| 333 |
+
| 7 | على | 17,625 |
|
| 334 |
+
| 8 | لي | 17,540 |
|
| 335 |
+
| 9 | نسبة | 16,376 |
|
| 336 |
+
| 10 | ب | 16,161 |
|
| 337 |
|
| 338 |
### Least Common Words (from vocabulary)
|
| 339 |
|
| 340 |
| Rank | Word | Frequency |
|
| 341 |
|------|------|-----------|
|
| 342 |
+
| 1 | تعاونيات | 2 |
|
| 343 |
+
| 2 | خواني | 2 |
|
| 344 |
+
| 3 | والمصطلحات | 2 |
|
| 345 |
+
| 4 | والنقدية | 2 |
|
| 346 |
+
| 5 | شرقًا | 2 |
|
| 347 |
+
| 6 | غربًا | 2 |
|
| 348 |
+
| 7 | المتري | 2 |
|
| 349 |
+
| 8 | بالمدّ | 2 |
|
| 350 |
+
| 9 | والعبارات | 2 |
|
| 351 |
+
| 10 | الكرم | 2 |
|
| 352 |
|
| 353 |
### Zipf's Law Analysis
|
| 354 |
|
| 355 |
| Metric | Value |
|
| 356 |
|--------|-------|
|
| 357 |
+
| Zipf Coefficient | 1.0352 |
|
| 358 |
+
| R² (Goodness of Fit) | 0.998696 |
|
| 359 |
| Adherence Quality | **excellent** |
|
| 360 |
|
| 361 |
### Coverage Analysis
|
| 362 |
|
| 363 |
| Top N Words | Coverage |
|
| 364 |
|-------------|----------|
|
| 365 |
+
| Top 100 | 40.4% |
|
| 366 |
+
| Top 1,000 | 64.9% |
|
| 367 |
+
| Top 5,000 | 79.3% |
|
| 368 |
+
| Top 10,000 | 85.4% |
|
| 369 |
|
| 370 |
### Key Findings
|
| 371 |
|
| 372 |
+
- **Zipf Compliance:** R²=0.9987 indicates excellent adherence to Zipf's law
|
| 373 |
+
- **High Frequency Dominance:** Top 100 words cover 40.4% of corpus
|
| 374 |
+
- **Long Tail:** 60,940 words needed for remaining 14.6% coverage
|
| 375 |
|
| 376 |
---
|
| 377 |
## 5. Word Embeddings Evaluation
|
|
|
|
| 384 |
|
| 385 |

|
| 386 |
|
|
|
|
| 387 |
|
| 388 |
+
### 5.1 Cross-Lingual Alignment
|
| 389 |
+
|
| 390 |
+
> *Note: Multilingual alignment visualization not available for this language.*
|
| 391 |
+
|
| 392 |
+
|
| 393 |
+
### 5.2 Model Comparison
|
| 394 |
+
|
| 395 |
+
| Model | Dimension | Isotropy | Semantic Density | Alignment R@1 | Alignment R@10 |
|
| 396 |
+
|-------|-----------|----------|------------------|---------------|----------------|
|
| 397 |
+
| **mono_32d** | 32 | 0.8384 🏆 | 0.3320 | N/A | N/A |
|
| 398 |
+
| **mono_64d** | 64 | 0.8149 | 0.2519 | N/A | N/A |
|
| 399 |
+
| **mono_128d** | 128 | 0.6695 | 0.2114 | N/A | N/A |
|
| 400 |
|
| 401 |
### Key Findings
|
| 402 |
|
| 403 |
+
- **Best Isotropy:** mono_32d with 0.8384 (more uniform distribution)
|
| 404 |
+
- **Semantic Density:** Average pairwise similarity of 0.2651. Lower values indicate better semantic separation.
|
| 405 |
+
- **Alignment Quality:** No aligned models evaluated in this run.
|
| 406 |
+
- **Recommendation:** 128d aligned for best cross-lingual performance
|
| 407 |
|
| 408 |
---
|
| 409 |
+
## 6. Morphological Analysis (Experimental)
|
| 410 |
+
|
| 411 |
+
> ⚠️ **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.
|
| 412 |
+
|
| 413 |
+
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.
|
| 414 |
+
|
| 415 |
+
### 6.1 Productivity & Complexity
|
| 416 |
+
|
| 417 |
+
| Metric | Value | Interpretation | Recommendation |
|
| 418 |
+
|--------|-------|----------------|----------------|
|
| 419 |
+
| Productivity Index | **0.000** | Low morphological productivity | ⚠️ Likely unreliable |
|
| 420 |
+
| Idiomaticity Gap | **-1.000** | Low formulaic content | - |
|
| 421 |
+
|
| 422 |
+
### 6.2 Affix Inventory (Productive Units)
|
| 423 |
+
|
| 424 |
+
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.
|
| 425 |
+
|
| 426 |
+
#### Productive Prefixes
|
| 427 |
+
| Prefix | Examples |
|
| 428 |
+
|--------|----------|
|
| 429 |
+
| `-ال` | التار, العادات, الواري |
|
| 430 |
+
| `-لم` | لموتقافين, لمحمية, لموتيفات |
|
| 431 |
+
| `-كا` | كايعطيهوم, كايتبناو, كايلمح |
|
| 432 |
+
|
| 433 |
+
#### Productive Suffixes
|
| 434 |
+
| Suffix | Examples |
|
| 435 |
+
|--------|----------|
|
| 436 |
+
| `-ات` | العادات, باللوغات, وزّعات |
|
| 437 |
+
| `-ية` | حيمائية, لافريقية, ليدارية |
|
| 438 |
+
| `-ين` | نّازيين, فالميادين, لموتقافين |
|
| 439 |
+
|
| 440 |
+
### 6.3 Bound Stems (Lexical Roots)
|
| 441 |
+
|
| 442 |
+
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.
|
| 443 |
+
|
| 444 |
+
| Stem | Cohesion | Substitutability | Examples |
|
| 445 |
+
|------|----------|------------------|----------|
|
| 446 |
+
| `انية` | 1.82x | 63 contexts | تانية, كانية, دانية |
|
| 447 |
+
| `الات` | 1.79x | 57 contexts | تالات, صالات, سالات |
|
| 448 |
+
| `جماع` | 1.93x | 37 contexts | تجماع, إجماع, جماعة |
|
| 449 |
+
| `لمغر` | 2.01x | 28 contexts | لمغرب, لمغربي, دلمغرب |
|
| 450 |
+
| `اللو` | 1.65x | 57 contexts | اللوت, اللوز, اللوح |
|
| 451 |
+
| `النا` | 1.64x | 55 contexts | النار, الناس, الناتو |
|
| 452 |
+
| `دهوم` | 2.21x | 16 contexts | ضدهوم, جهدهوم, بعدهوم |
|
| 453 |
+
| `مغري` | 2.02x | 18 contexts | مغرية, مغريب, مغريبي |
|
| 454 |
+
| `قليم` | 2.06x | 15 contexts | اقليم, فقليم, إقليم |
|
| 455 |
+
| `لجوا` | 1.76x | 24 contexts | لجواب, الجوا, لجوائر |
|
| 456 |
+
| `اميل` | 1.78x | 23 contexts | كاميل, عاميل, ݣاميلة |
|
| 457 |
+
| `إحصا` | 2.08x | 14 contexts | لإحصا, إحصاء, إحصائي |
|
| 458 |
+
|
| 459 |
+
### 6.4 Affix Compatibility (Co-occurrence)
|
| 460 |
+
|
| 461 |
+
This table shows which prefixes and suffixes most frequently co-occur on the same stems, revealing the 'stacking' rules of the language's morphology.
|
| 462 |
+
|
| 463 |
+
| Prefix | Suffix | Frequency | Examples |
|
| 464 |
+
|--------|--------|-----------|----------|
|
| 465 |
+
| `-ال` | `-ية` | 126 words | الكوانتية, الشهية |
|
| 466 |
+
| `-ال` | `-ات` | 123 words | العقوبات, الدبانيات |
|
| 467 |
+
| `-ال` | `-ين` | 70 words | الرينين, الثلاثين |
|
| 468 |
+
| `-لم` | `-ات` | 41 words | لمسراحيات, لمانيفولضات |
|
| 469 |
+
| `-لم` | `-ين` | 37 words | لمعروفين, لموليكيين |
|
| 470 |
+
| `-لم` | `-ية` | 18 words | لماركسية, لمرساوية |
|
| 471 |
+
| `-كا` | `-ين` | 2 words | كاتبيين, كالكيريين |
|
| 472 |
+
| `-كا` | `-ات` | 2 words | كارنيڤورات, كاريكاتورات |
|
| 473 |
+
|
| 474 |
+
### 6.5 Recursive Morpheme Segmentation
|
| 475 |
+
|
| 476 |
+
Using **Recursive Hierarchical Substitutability**, we decompose complex words into their constituent morphemes. This approach handles nested affixes (e.g., `prefix-prefix-root-suffix`).
|
| 477 |
+
|
| 478 |
+
| Word | Suggested Split | Confidence | Stem |
|
| 479 |
+
|------|-----------------|------------|------|
|
| 480 |
+
| لمعلوماتية | **`لم-علوم-ات-ية`** | 7.5 | `علوم` |
|
| 481 |
+
| الثلاثينات | **`ال-ثلاث-ين-ات`** | 7.5 | `ثلاث` |
|
| 482 |
+
| التأريخية | **`ال-تأريخ-ية`** | 6.0 | `تأريخ` |
|
| 483 |
+
| المهندسين | **`ال-مهندس-ين`** | 6.0 | `مهندس` |
|
| 484 |
+
| التيليفونات | **`ال-تيليفون-ات`** | 6.0 | `تيليفون` |
|
| 485 |
+
| السيشيلية | **`ال-سيشيل-ية`** | 6.0 | `سيشيل` |
|
| 486 |
+
| المجتمعين | **`ال-مجتمع-ين`** | 6.0 | `مجتمع` |
|
| 487 |
+
| التجهيزات | **`ال-تجهيز-ات`** | 6.0 | `تجهيز` |
|
| 488 |
+
| العثمانية | **`ال-عثمان-ية`** | 6.0 | `عثمان` |
|
| 489 |
+
| المعتقدات | **`ال-معتقد-ات`** | 6.0 | `معتقد` |
|
| 490 |
+
| البوليسية | **`ال-بوليس-ية`** | 6.0 | `بوليس` |
|
| 491 |
+
| التشكالات | **`ال-تشكال-ات`** | 6.0 | `تشكال` |
|
| 492 |
+
| المستشارين | **`ال-مستشار-ين`** | 6.0 | `مستشار` |
|
| 493 |
+
| السيركويات | **`ال-سيركوي-ات`** | 6.0 | `سيركوي` |
|
| 494 |
+
| التحضيرية | **`ال-تحضير-ية`** | 6.0 | `تحضير` |
|
| 495 |
+
|
| 496 |
+
### 6.6 Linguistic Interpretation
|
| 497 |
+
|
| 498 |
+
> **Automated Insight:**
|
| 499 |
+
The language Moroccan Arabic 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.
|
| 500 |
+
|
| 501 |
+
---
|
| 502 |
+
## 7. Summary & Recommendations
|
| 503 |
|
| 504 |

|
| 505 |
|
|
|
|
| 507 |
|
| 508 |
| Component | Recommended | Rationale |
|
| 509 |
|-----------|-------------|-----------|
|
| 510 |
+
| Tokenizer | **64k BPE** | Best compression (4.18x) |
|
| 511 |
+
| N-gram | **2-gram** | Lowest perplexity (415) |
|
| 512 |
+
| Markov | **Context-4** | Highest predictability (97.9%) |
|
| 513 |
| Embeddings | **100d** | Balanced semantic capture and isotropy |
|
| 514 |
|
| 515 |
+
|
| 516 |
---
|
| 517 |
## Appendix: Metrics Glossary & Interpretation Guide
|
| 518 |
|
|
|
|
| 702 |
author = {Kamali, Omar},
|
| 703 |
title = {Wikilangs: Open NLP Models for Wikipedia Languages},
|
| 704 |
year = {2025},
|
| 705 |
+
doi = {10.5281/zenodo.18073153},
|
| 706 |
+
publisher = {Zenodo},
|
| 707 |
url = {https://huggingface.co/wikilangs}
|
| 708 |
institution = {Omneity Labs}
|
| 709 |
}
|
|
|
|
| 719 |
- 🤗 Models: [huggingface.co/wikilangs](https://huggingface.co/wikilangs)
|
| 720 |
- 📊 Data: [wikipedia-monthly](https://huggingface.co/datasets/omarkamali/wikipedia-monthly)
|
| 721 |
- 👤 Author: [Omar Kamali](https://huggingface.co/omarkamali)
|
| 722 |
+
- 🤝 Sponsor: [Featherless AI](https://featherless.ai)
|
| 723 |
---
|
| 724 |
*Generated by Wikilangs Models Pipeline*
|
| 725 |
|
| 726 |
+
*Report Date: 2026-01-03 05:20:40*
|
models/embeddings/monolingual/ary_128d.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ec0c8ea941d4e589712d1341927459bcf79927b09c35a78a960c9f6d7e10e2d9
|
| 3 |
+
size 1056923528
|
models/embeddings/monolingual/ary_128d_metadata.json
CHANGED
|
@@ -3,11 +3,13 @@
|
|
| 3 |
"dimension": 128,
|
| 4 |
"version": "monolingual",
|
| 5 |
"training_params": {
|
| 6 |
-
"
|
| 7 |
"min_count": 5,
|
| 8 |
"window": 5,
|
| 9 |
"negative": 5,
|
| 10 |
-
"epochs": 5
|
|
|
|
|
|
|
| 11 |
},
|
| 12 |
-
"vocab_size":
|
| 13 |
}
|
|
|
|
| 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 |
},
|
| 14 |
+
"vocab_size": 31513
|
| 15 |
}
|
models/embeddings/monolingual/ary_32d.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eeac1e5f452245ffe93038f0b7258b2f1f16a3bbf65b19fe05b1a6db4fae8474
|
| 3 |
+
size 264721544
|
models/embeddings/monolingual/ary_32d_metadata.json
CHANGED
|
@@ -3,11 +3,13 @@
|
|
| 3 |
"dimension": 32,
|
| 4 |
"version": "monolingual",
|
| 5 |
"training_params": {
|
| 6 |
-
"
|
| 7 |
"min_count": 5,
|
| 8 |
"window": 5,
|
| 9 |
"negative": 5,
|
| 10 |
-
"epochs": 5
|
|
|
|
|
|
|
| 11 |
},
|
| 12 |
-
"vocab_size":
|
| 13 |
}
|
|
|
|
| 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 |
},
|
| 14 |
+
"vocab_size": 31513
|
| 15 |
}
|
models/embeddings/monolingual/ary_64d.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b01a57a7d1b52d182e7b0bb4f504588432f74b447abf3627c5c7314bbffab5f2
|
| 3 |
+
size 528788872
|
models/embeddings/monolingual/ary_64d_metadata.json
CHANGED
|
@@ -3,11 +3,13 @@
|
|
| 3 |
"dimension": 64,
|
| 4 |
"version": "monolingual",
|
| 5 |
"training_params": {
|
| 6 |
-
"
|
| 7 |
"min_count": 5,
|
| 8 |
"window": 5,
|
| 9 |
"negative": 5,
|
| 10 |
-
"epochs": 5
|
|
|
|
|
|
|
| 11 |
},
|
| 12 |
-
"vocab_size":
|
| 13 |
}
|
|
|
|
| 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 |
},
|
| 14 |
+
"vocab_size": 31513
|
| 15 |
}
|
models/subword_markov/ary_markov_ctx1_subword.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:459fa6b743dbcf395519af9bf18ba67b777ddd0e0cc864ac3bb6a7bf3189d793
|
| 3 |
+
size 133878
|
models/subword_markov/ary_markov_ctx1_subword_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"context_size": 1,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_contexts":
|
| 6 |
-
"total_transitions":
|
| 7 |
}
|
|
|
|
| 2 |
"context_size": 1,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_contexts": 2149,
|
| 6 |
+
"total_transitions": 10496838
|
| 7 |
}
|
models/subword_markov/ary_markov_ctx2_subword.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:552c831bedb22d7f7b1c00d88c34681f2f3325dd23a3f89d85c02eb0c4ac3281
|
| 3 |
+
size 749221
|
models/subword_markov/ary_markov_ctx2_subword_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"context_size": 2,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_contexts":
|
| 6 |
-
"total_transitions":
|
| 7 |
}
|
|
|
|
| 2 |
"context_size": 2,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_contexts": 17291,
|
| 6 |
+
"total_transitions": 10486078
|
| 7 |
}
|
models/subword_markov/ary_markov_ctx3_subword.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a0e7a2836b00fdfdf3ecab6fd3ff4f39a94e43eeb24ea7dfa6bbdea5ac77430
|
| 3 |
+
size 2667571
|
models/subword_markov/ary_markov_ctx3_subword_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"context_size": 3,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_contexts":
|
| 6 |
-
"total_transitions":
|
| 7 |
}
|
|
|
|
| 2 |
"context_size": 3,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_contexts": 86361,
|
| 6 |
+
"total_transitions": 10475318
|
| 7 |
}
|
models/subword_markov/ary_markov_ctx4_subword.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:85296dab212dcfcc7a9ab757176cd9fbd76a93f9401830d56004280e4a7377f8
|
| 3 |
+
size 8465099
|
models/subword_markov/ary_markov_ctx4_subword_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"context_size": 4,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_contexts":
|
| 6 |
-
"total_transitions":
|
| 7 |
}
|
|
|
|
| 2 |
"context_size": 4,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_contexts": 352807,
|
| 6 |
+
"total_transitions": 10464558
|
| 7 |
}
|
models/subword_ngram/ary_2gram_subword.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:78ea13aba22c9b182dee2ab8bd64573851cedf52939c98249335191a3f23f274
|
| 3 |
+
size 80021
|
models/subword_ngram/ary_2gram_subword_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"n": 2,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_ngrams":
|
| 6 |
-
"total_ngrams":
|
| 7 |
}
|
|
|
|
| 2 |
"n": 2,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_ngrams": 5585,
|
| 6 |
+
"total_ngrams": 10496838
|
| 7 |
}
|
models/subword_ngram/ary_3gram_subword.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:24b08ece7a38c3125e70f5a75f6ead1fd500f694cb05ad5c6248ba7b85fd5634
|
| 3 |
+
size 573468
|
models/subword_ngram/ary_3gram_subword_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"n": 3,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_ngrams":
|
| 6 |
-
"total_ngrams":
|
| 7 |
}
|
|
|
|
| 2 |
"n": 3,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_ngrams": 41944,
|
| 6 |
+
"total_ngrams": 10486078
|
| 7 |
}
|
models/subword_ngram/ary_4gram_subword.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c1dd4f7cdfe642d380522ea08b1e34ceadbe57558bcccafcd2c10d5783f15f75
|
| 3 |
+
size 2617929
|
models/subword_ngram/ary_4gram_subword_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"n": 4,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_ngrams":
|
| 6 |
-
"total_ngrams":
|
| 7 |
}
|
|
|
|
| 2 |
"n": 4,
|
| 3 |
"variant": "subword",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_ngrams": 204568,
|
| 6 |
+
"total_ngrams": 10475318
|
| 7 |
}
|
models/tokenizer/ary_tokenizer_16k.model
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2b2380e695c2c3ff7f9eec16ecc69d1a82452d80f0c0884635d764801141b6e5
|
| 3 |
+
size 559353
|
models/tokenizer/ary_tokenizer_16k.vocab
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
models/tokenizer/ary_tokenizer_32k.model
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:334ec9fb02ebfd6451b7961710b7f94c763275ce5022015434501d439ac06ad4
|
| 3 |
+
size 894340
|
models/tokenizer/ary_tokenizer_32k.vocab
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
models/tokenizer/ary_tokenizer_64k.model
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:228ac76778adf794a1d8de7ed6648a2bbfd80b7498be5530f090627f915c0436
|
| 3 |
+
size 1593462
|
models/tokenizer/ary_tokenizer_64k.vocab
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
models/tokenizer/ary_tokenizer_8k.model
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fa0d6cbad79ebc5208a11f4bc61982c6584a2f87a08a41d8c2c90d8c0941ea81
|
| 3 |
+
size 396437
|
models/tokenizer/ary_tokenizer_8k.vocab
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
models/vocabulary/ary_vocabulary.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e6a100896fb6a926f24685f0b974ea8144580d8025c46f0451fa0eb2a1cccb88
|
| 3 |
+
size 1278866
|
models/vocabulary/ary_vocabulary_metadata.json
CHANGED
|
@@ -1,16 +1,17 @@
|
|
| 1 |
{
|
| 2 |
"language": "ary",
|
| 3 |
-
"vocabulary_size":
|
|
|
|
| 4 |
"statistics": {
|
| 5 |
-
"type_token_ratio": 0.
|
| 6 |
"coverage": {
|
| 7 |
-
"top_100": 0.
|
| 8 |
-
"top_1000": 0.
|
| 9 |
-
"top_5000": 0.
|
| 10 |
-
"top_10000": 0.
|
| 11 |
},
|
| 12 |
-
"hapax_count":
|
| 13 |
-
"hapax_ratio": 0.
|
| 14 |
-
"total_documents":
|
| 15 |
}
|
| 16 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"language": "ary",
|
| 3 |
+
"vocabulary_size": 70940,
|
| 4 |
+
"variant": "full",
|
| 5 |
"statistics": {
|
| 6 |
+
"type_token_ratio": 0.08383333066622203,
|
| 7 |
"coverage": {
|
| 8 |
+
"top_100": 0.38467935557772986,
|
| 9 |
+
"top_1000": 0.6186641695622032,
|
| 10 |
+
"top_5000": 0.7558627838344147,
|
| 11 |
+
"top_10000": 0.8141512107566836
|
| 12 |
},
|
| 13 |
+
"hapax_count": 91460,
|
| 14 |
+
"hapax_ratio": 0.5631773399014778,
|
| 15 |
+
"total_documents": 10760
|
| 16 |
}
|
| 17 |
}
|
models/word_markov/ary_markov_ctx1_word.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7c9d4b1eb5fe667ef9075fa89ad5a4ea4ea33df8dba16dd4243bd25ca92d4926
|
| 3 |
+
size 9279588
|
models/word_markov/ary_markov_ctx1_word_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"context_size": 1,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_contexts":
|
| 6 |
-
"total_transitions":
|
| 7 |
}
|
|
|
|
| 2 |
"context_size": 1,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_contexts": 162378,
|
| 6 |
+
"total_transitions": 1926417
|
| 7 |
}
|
models/word_markov/ary_markov_ctx2_word.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29a939f97ef93392311582b9ad3f3b729f547581513acb58fd36113bf5b295e2
|
| 3 |
+
size 23360695
|
models/word_markov/ary_markov_ctx2_word_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"context_size": 2,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_contexts":
|
| 6 |
-
"total_transitions":
|
| 7 |
}
|
|
|
|
| 2 |
"context_size": 2,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_contexts": 849251,
|
| 6 |
+
"total_transitions": 1915657
|
| 7 |
}
|
models/word_markov/ary_markov_ctx3_word.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4fc3870d90e96a6d0f094fa6751c6f78ed9b0e520e72f7882513945cab8f1373
|
| 3 |
+
size 32010448
|
models/word_markov/ary_markov_ctx3_word_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"context_size": 3,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_contexts":
|
| 6 |
-
"total_transitions":
|
| 7 |
}
|
|
|
|
| 2 |
"context_size": 3,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_contexts": 1262316,
|
| 6 |
+
"total_transitions": 1904897
|
| 7 |
}
|
models/word_markov/ary_markov_ctx4_word.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:526783a2fed42680ed61c32c03a44df01d06d2396568770540907bffacc2d4b5
|
| 3 |
+
size 38343134
|
models/word_markov/ary_markov_ctx4_word_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"context_size": 4,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_contexts":
|
| 6 |
-
"total_transitions":
|
| 7 |
}
|
|
|
|
| 2 |
"context_size": 4,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_contexts": 1391141,
|
| 6 |
+
"total_transitions": 1894137
|
| 7 |
}
|
models/word_ngram/ary_2gram_word.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bfeda04510e2c553b8ed89b81167eb362bcaad06b1cd9106f259c8829124e7bd
|
| 3 |
+
size 673580
|
models/word_ngram/ary_2gram_word_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"n": 2,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_ngrams":
|
| 6 |
-
"total_ngrams":
|
| 7 |
}
|
|
|
|
| 2 |
"n": 2,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_ngrams": 35218,
|
| 6 |
+
"total_ngrams": 1926417
|
| 7 |
}
|
models/word_ngram/ary_3gram_word.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:42e065ff96cac0f99796887363b07f7ed4c89751e6ffab760b5684e38bd7c11c
|
| 3 |
+
size 851553
|
models/word_ngram/ary_3gram_word_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"n": 3,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_ngrams":
|
| 6 |
-
"total_ngrams":
|
| 7 |
}
|
|
|
|
| 2 |
"n": 3,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_ngrams": 39702,
|
| 6 |
+
"total_ngrams": 1915657
|
| 7 |
}
|
models/word_ngram/ary_4gram_word.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4bad83de9507de5b6f2183bc341880b4b18a95728eb2e74422e18e4dc7410f5f
|
| 3 |
+
size 1506246
|
models/word_ngram/ary_4gram_word_metadata.json
CHANGED
|
@@ -2,6 +2,6 @@
|
|
| 2 |
"n": 4,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
-
"unique_ngrams":
|
| 6 |
-
"total_ngrams":
|
| 7 |
}
|
|
|
|
| 2 |
"n": 4,
|
| 3 |
"variant": "word",
|
| 4 |
"language": "ary",
|
| 5 |
+
"unique_ngrams": 63706,
|
| 6 |
+
"total_ngrams": 1904897
|
| 7 |
}
|
visualizations/embedding_isotropy.png
CHANGED
|
|
visualizations/embedding_norms.png
CHANGED
|
|
visualizations/embedding_similarity.png
CHANGED
|
Git LFS Details
|
|
Git LFS Details
|
visualizations/markov_branching.png
CHANGED
|
|