Upload folder using huggingface_hub

README.md

@@ -0,0 +1,21 @@
+# wilai-2.5 (draft)
+
+This folder contains prepared artifacts for the `wilai-2.5` release (development draft).
+
+Contents:
+- `config.json` (copied from v2.0, may be adjusted)
+- `tokenizer.json` (canonical tokenizer copied from v2.0)
+- `pytorch_model.bin` (converted or to-be-converted checkpoint for v2.5)
+
+Notes:
+- The v2 model uses a new modular layer `SecretLayer` and a SwiGLU MLP (see `wilai_transformers/modeling_wilai_v2.py`).
+- This implementation is experimental and intended for testing. Do not publish publicly without review if you intend the `SecretLayer` to remain private.
+
+How to test locally:
+
+```powershell
+$env:PYTHONPATH='D:\GitHub\Ai-Thai-wilai'
+python -c "from wilai_transformers.modeling_wilai_v2 import WilaiForCausalLMV2; print('import ok')"
+```
+
+When you're ready, I can upload this folder to Hugging Face as `JonusNattapong/wilai-2.5` and include the converted checkpoint. If you want me to upload now, confirm and I'll push the prepared files to the new repo.

config.json

@@ -0,0 +1,32 @@
+{
+  "_name_or_path": "JonusNattapong/wilai-2.0",
+  "model_type": "wilai",
+  "vocab_size": 1000,
+  "hidden_size": 512,
+  "num_hidden_layers": 6,
+  "num_attention_heads": 8,
+  "max_position_embeddings": 128,
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-5,
+  "embd_pdrop": 0.1,
+  "attn_pdrop": 0.1,
+  "resid_pdrop": 0.1,
+  "activation_function": "gelu",
+  "n_inner": null,
+  "use_cache": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "pad_token_id": 0,
+  "tie_word_embeddings": false,
+  "use_gradient_checkpointing": true,
+  "use_mixed_precision": false,
+  "gradient_clip_norm": 1.0,
+  "model_name": "wilai-2.0",
+  "model_version": "1.0.0",
+  "description": "Thai-specific GPT-like language model",
+  "created_at": "2025-10-27T11:44:35.229069",
+  "training_completed": true,
+  "total_steps": 16600,
+  "best_val_loss": 2.707622994680773,
+  "training_time": 58012.46856427193
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:854f68930029c898337c6dd9720c26cb17a4da95020db03f2819b4a11eb1e8bb
+size 80046702

tokenizer.json

@@ -0,0 +1,1022 @@
+{
+  "version": "1.0",
+  "truncation": null,
+  "padding": null,
+  "added_tokens": [],
+  "normalizer": null,
+  "pre_tokenizer": {
+    "type": "Whitespace"
+  },
+  "post_processor": null,
+  "decoder": {
+    "type": "WordPiece",
+    "prefix": "",
+    "cleanup": true
+  },
+  "model": {
+    "type": "WordLevel",
+    "vocab": {
+      "<unk>": 0,
+      "<s>": 1,
+      "</s>": 2,
+      "[UNK]": 3,
+      "[CLS]": 4,
+      "[SEP]": 5,
+      "[PAD]": 6,
+      "[MASK]": 7,
+      "▁เ": 8,
+      "▁ก": 9,
+      "▁แ": 10,
+      "▁ส": 11,
+      "▁ท": 12,
+      "าร": 13,
+      "▁อ": 14,
+      "อง": 15,
+      "▁ใ": 16,
+      "▁ม": 17,
+      "ี่": 18,
+      "ระ": 19,
+      "▁ค": 20,
+      "ํา": 21,
+      "▁ไ": 22,
+      "▁ข": 23,
+      "▁ป": 24,
+      "่า": 25,
+      "▁จ": 26,
+      "▁ห": 27,
+      "▁ต": 28,
+      "▁พ": 29,
+      "▁ร": 30,
+      "▁โ": 31,
+      "ัน": 32,
+      "▁ที่": 33,
+      "▁น": 34,
+      "▁ใน": 35,
+      "ละ": 36,
+      "▁การ": 37,
+      "ีย": 38,
+      "▁ว": 39,
+      "็น": 40,
+      "▁เป": 41,
+      "้า": 42,
+      "่ง": 43,
+      "▁และ": 44,
+      "าม": 45,
+      "าก": 46,
+      "▁เป็น": 47,
+      "ับ": 48,
+      "่อ": 49,
+      "▁ของ": 50,
+      "▁ด": 51,
+      "าย": 52,
+      "ือ": 53,
+      "้น": 54,
+      "ร์": 55,
+      "▁มี": 56,
+      "ัง": 57,
+      "▁ช": 58,
+      "▁บ": 59,
+      "าน": 60,
+      "ัก": 61,
+      "▁ประ": 62,
+      "าง": 63,
+      "▁ล": 64,
+      "ด้": 65,
+      "ภา": 66,
+      "ู่": 67,
+      "▁ย": 68,
+      "่น": 69,
+      "ิน": 70,
+      "รร": 71,
+      "▁ได้": 72,
+      "ื่อ": 73,
+      "าว": 74,
+      "ริ": 75,
+      "่ว": 76,
+      "▁ซ": 77,
+      "ทย": 78,
+      "หม": 79,
+      "้ง": 80,
+      "อก": 81,
+      "คร": 82,
+      "้อ": 83,
+      "การ": 84,
+      "ติ": 85,
+      "่าง": 86,
+      "ั้ง": 87,
+      "ัด": 88,
+      "าส": 89,
+      "▁ผ": 90,
+      "an": 91,
+      "▁หม": 92,
+      "วาม": 93,
+      "เท": 94,
+      "▁ภา": 95,
+      "▁โด": 96,
+      "ษา": 97,
+      "าช": 98,
+      "▁โดย": 99,
+      "er": 100,
+      "ึ่ง": 101,
+      "ตร": 102,
+      "on": 103,
+      "▁เก": 104,
+      "ัย": 105,
+      "▁เพ": 106,
+      "▁ความ": 107,
+      "in": 108,
+      "ศาส": 109,
+      "ิด": 110,
+      "อร์": 111,
+      "นา": 112,
+      "▁เม": 113,
+      "่ม": 114,
+      "าล": 115,
+      "▁ถ": 116,
+      "วด": 117,
+      "▁วิ": 118,
+      "▁ระ": 119,
+      "ที่": 120,
+      "ัว": 121,
+      "▁อย": 122,
+      "▁เข": 123,
+      "ตร์": 124,
+      "หล": 125,
+      "่อง": 126,
+      "▁ไม": 127,
+      "เทศ": 128,
+      "ี้": 129,
+      "▁กล": 130,
+      "ู้": 131,
+      "หน": 132,
+      "▁เร": 133,
+      "ึง": 134,
+      "▁พระ": 135,
+      "or": 136,
+      "ประ": 137,
+      "ห้": 138,
+      "▁หมวด": 139,
+      "บบ": 140,
+      "at": 141,
+      "en": 142,
+      "หมู่": 143,
+      "ศาสตร์": 144,
+      "▁จาก": 145,
+      "้ว": 146,
+      "▁หมวดหมู่": 147,
+      "ช้": 148,
+      "▁เช": 149,
+      "ั้น": 150,
+      "ุด": 151,
+      "ิต": 152,
+      "▁S": 153,
+      "อน": 154,
+      "al": 155,
+      "้าง": 156,
+      "es": 157,
+      "▁ว่า": 158,
+      "ิง": 159,
+      "▁ทาง": 160,
+      "รือ": 161,
+      "▁จะ": 162,
+      "ar": 163,
+      "หว": 164,
+      "ียง": 165,
+      "▁ภาษา": 166,
+      "▁หน": 167,
+      "▁ใช้": 168,
+      "รรม": 169,
+      "▁ปี": 170,
+      "▁อา": 171,
+      "ื่อง": 172,
+      "▁มา": 173,
+      "มา": 174,
+      "▁A": 175,
+      "ียน": 176,
+      "▁เท": 177,
+      "ัฐ": 178,
+      "▁หล": 179,
+      "▁หรือ": 180,
+      "▁พศ": 181,
+      "อบ": 182,
+      "▁ซึ่ง": 183,
+      "▁กับ": 184,
+      "เล": 185,
+      "ัญ": 186,
+      "ําน": 187,
+      "้าน": 188,
+      "าะ": 189,
+      "▁เด": 190,
+      "▁C": 191,
+      "▁เส": 192,
+      "เม": 193,
+      "▁ไม่": 194,
+      "ย์": 195,
+      "ําเ": 196,
+      "th": 197,
+      "ี่ย": 198,
+      "วน": 199,
+      "ต่": 200,
+      "รับ": 201,
+      "้อง": 202,
+      "▁ศ": 203,
+      "▁คือ": 204,
+      "่วน": 205,
+      "คม": 206,
+      "▁T": 207,
+      "▁สม": 208,
+      "▁ผู้": 209,
+      "ึก": 210,
+      "▁เล": 211,
+      "รี": 212,
+      "ูล": 213,
+      "ือง": 214,
+      "▁ให้": 215,
+      "▁ฟ": 216,
+      "กร": 217,
+      "ic": 218,
+      "▁ภ": 219,
+      "ต์": 220,
+      "รง": 221,
+      "is": 222,
+      "งาน": 223,
+      "ธิ": 224,
+      "เด": 225,
+      "วง": 226,
+      "▁ประเทศ": 227,
+      "กล": 228,
+      "▁ทํา": 229,
+      "▁ๆ": 230,
+      "ดับ": 231,
+      "▁ไป": 232,
+      "จาก": 233,
+      "▁M": 234,
+      "it": 235,
+      "ิก": 236,
+      "ูก": 237,
+      "ion": 238,
+      "▁เอ": 239,
+      "ุ่ม": 240,
+      "มพ": 241,
+      "▁อยู่": 242,
+      "้อม": 243,
+      "▁P": 244,
+      "ชา": 245,
+      "ลา": 246,
+      "ึ้น": 247,
+      "ัต": 248,
+      "อย": 249,
+      "▁เว": 250,
+      "เร": 251,
+      "▁ตัว": 252,
+      "ณะ": 253,
+      "▁o": 254,
+      "▁มห": 255,
+      "ิทย": 256,
+      "▁สถ": 257,
+      "ณ์": 258,
+      "วก": 259,
+      "ื่น": 260,
+      "▁ส่วน": 261,
+      "▁แต่": 262,
+      "โล": 263,
+      "่าน": 264,
+      "▁สาม": 265,
+      "▁นิ": 266,
+      "ักษ": 267,
+      "้ํา": 268,
+      "ห่ง": 269,
+      "▁ออก": 270,
+      "▁วิทย": 271,
+      "▁เน": 272,
+      "▁of": 273,
+      "▁นัก": 274,
+      "▁วัน": 275,
+      "้วย": 276,
+      "เข": 277,
+      "าลัย": 278,
+      "▁เมื่อ": 279,
+      "หญ": 280,
+      "▁ธ": 281,
+      "▁ทั้ง": 282,
+      "▁มาก": 283,
+      "▁I": 284,
+      "รก": 285,
+      "▁อย่าง": 286,
+      "▁รัฐ": 287,
+      "เต": 288,
+      "▁คศ": 289,
+      "าศ": 290,
+      "▁ต่อ": 291,
+      "▁ยัง": 292,
+      "ัติ": 293,
+      "เก": 294,
+      "▁ไทย": 295,
+      "ูป": 296,
+      "ว่า": 297,
+      "▁ด้วย": 298,
+      "ุม": 299,
+      "▁B": 300,
+      "▁พื": 301,
+      "ันธ": 302,
+      "ุก": 303,
+      "ภาพ": 304,
+      "▁เพื่อ": 305,
+      "le": 306,
+      "มิ": 307,
+      "ทร": 308,
+      "ค์": 309,
+      "▁คํา": 310,
+      "้อมูล": 311,
+      "ํานวน": 312,
+      "ิจ": 313,
+      "ึกษา": 314,
+      "เป": 315,
+      "▁คน": 316,
+      "▁กลุ่ม": 317,
+      "▁แห่ง": 318,
+      "▁ดาว": 319,
+      "าบ": 320,
+      "เว": 321,
+      "▁จํานวน": 322,
+      "ารถ": 323,
+      "ro": 324,
+      "หวัด": 325,
+      "หญ่": 326,
+      "▁จัด": 327,
+      "▁เช่น": 328,
+      "▁p": 329,
+      "ต้": 330,
+      "คล": 331,
+      "▁นี้": 332,
+      "▁N": 333,
+      "ังหวัด": 334,
+      "ราช": 335,
+      "ัส": 336,
+      "▁อื่น": 337,
+      "and": 338,
+      "าร์": 339,
+      "▁ตาม": 340,
+      "▁กร": 341,
+      "วัน": 342,
+      "ol": 343,
+      "▁th": 344,
+      "ไป": 345,
+      "▁คร": 346,
+      "▁ราช": 347,
+      "▁รูป": 348,
+      "▁H": 349,
+      "หาร": 350,
+      "เซ": 351,
+      "▁บริ": 352,
+      "▁D": 353,
+      "วม": 354,
+      "ดี": 355,
+      "กัน": 356,
+      "▁โล": 357,
+      "re": 358,
+      "▁F": 359,
+      "▁ก็": 360,
+      "▁กระ": 361,
+      "ฤษ": 362,
+      "▁R": 363,
+      "▁ฝ": 364,
+      "ถึง": 365,
+      "สุด": 366,
+      "ั่ง": 367,
+      "▁L": 368,
+      "▁ฮ": 369,
+      "▁แบบ": 370,
+      "▁ยุ": 371,
+      "▁ผล": 372,
+      "ส์": 373,
+      "▁สุ": 374,
+      "▁แล": 375,
+      "▁ขึ้น": 376,
+      "▁f": 377,
+      "จ้า": 378,
+      "ิล": 379,
+      "ิว": 380,
+      "น์": 381,
+      "ูมิ": 382,
+      "▁สามารถ": 383,
+      "▁G": 384,
+      "▁สํา": 385,
+      "▁จังหวัด": 386,
+      "▁s": 387,
+      "้อย": 388,
+      "▁แม": 389,
+      "ีน": 390,
+      "▁ต่าง": 391,
+      "เส": 392,
+      "▁อี": 393,
+      "ภอ": 394,
+      "▁ถูก": 395,
+      "ําเภอ": 396,
+      "อล": 397,
+      "▁อุ": 398,
+      "ะวัน": 399,
+      "ฐาน": 400,
+      "ีว": 401,
+      "as": 402,
+      "▁อําเภอ": 403,
+      "▁ชาว": 404,
+      "▁E": 405,
+      "ลง": 406,
+      "อด": 407,
+      "▁สิ": 408,
+      "▁เคร": 409,
+      "่าย": 410,
+      "▁ตั้ง": 411,
+      "▁c": 412,
+      "ญา": 413,
+      "รม": 414,
+      "▁เข้า": 415,
+      "ิส": 416,
+      "▁เซ": 417,
+      "us": 418,
+      "he": 419,
+      "ังก": 420,
+      "▁ระบบ": 421,
+      "็ก": 422,
+      "ให้": 423,
+      "ิทยาลัย": 424,
+      "ประเทศ": 425,
+      "นุ": 426,
+      "▁มหาว": 427,
+      "▁มหาวิทยาลัย": 428,
+      "สม": 429,
+      "ation": 430,
+      "เตอร์": 431,
+      "ปร": 432,
+      "นี้": 433,
+      "จจ": 434,
+      "▁นั้น": 435,
+      "ศึกษา": 436,
+      "ร้าง": 437,
+      "▁ถึง": 438,
+      "▁พื้น": 439,
+      "ัฒ": 440,
+      "el": 441,
+      "วย": 442,
+      "ูง": 443,
+      "▁เรีย": 444,
+      "ตั้ง": 445,
+      "ur": 446,
+      "ตก": 447,
+      "็จ": 448,
+      "เช": 449,
+      "▁ช่ว": 450,
+      "าท": 451,
+      "ing": 452,
+      "▁the": 453,
+      "ดง": 454,
+      "กรรม": 455,
+      "ชาติ": 456,
+      "ac": 457,
+      "▁เมือง": 458,
+      "ก่": 459,
+      "▁ด้าน": 460,
+      "om": 461,
+      "ักร": 462,
+      "▁แบ": 463,
+      "▁สห": 464,
+      "ทาง": 465,
+      "ื้อ": 466,
+      "ed": 467,
+      "ดน": 468,
+      "▁รวม": 469,
+      "ุ่น": 470,
+      "▁วิทยา": 471,
+      "▁ได้รับ": 472,
+      "▁แล้ว": 473,
+      "บุ": 474,
+      "วรร": 475,
+      "et": 476,
+      "ัล": 477,
+      "ูน": 478,
+      "▁ปั": 479,
+      "▁กัน": 480,
+      "ิม": 481,
+      "▁ชื่อ": 482,
+      "ent": 483,
+      "เน": 484,
+      "หนือ": 485,
+      "่าว": 486,
+      "▁ประก": 487,
+      "มาณ": 488,
+      "▁โลก": 489,
+      "▁ปร": 490,
+      "ือน": 491,
+      "่อน": 492,
+      "ธรรม": 493,
+      "▁ปัจจ": 494,
+      "▁t": 495,
+      "▁แส": 496,
+      "am": 497,
+      "หา": 498,
+      "ิท": 499,
+      "เศ": 500,
+      "▁ราย": 501,
+      "ัพ": 502,
+      "▁หนึ่ง": 503,
+      "ออก": 504,
+      "▁อัน": 505,
+      "▁m": 506,
+      "▁อีก": 507,
+      "โน": 508,
+      "▁เครื่อง": 509,
+      "ตุ": 510,
+      "▁O": 511,
+      "▁ดัง": 512,
+      "▁ข้อมูล": 513,
+      "ุบ": 514,
+      "ใหญ่": 515,
+      "เมือง": 516,
+      "ั่ว": 517,
+      "▁and": 518,
+      "il": 519,
+      "กับ": 520,
+      "▁W": 521,
+      "▁กํา": 522,
+      "ig": 523,
+      "▁บาง": 524,
+      "ุน": 525,
+      "ยาย": 526,
+      "▁อ้าง": 527,
+      "▁เรื่อง": 528,
+      "หรับ": 529,
+      "รัฐ": 530,
+      "▁สา": 531,
+      "หว่าง": 532,
+      "▁ระหว่าง": 533,
+      "ณา": 534,
+      "▁ภาค": 535,
+      "▁แหล": 536,
+      "▁d": 537,
+      "ิ่ม": 538,
+      "ความ": 539,
+      "▁ตะวัน": 540,
+      "จํา": 541,
+      "ากาศ": 542,
+      "▁K": 543,
+      "อม": 544,
+      "▁เรียก": 545,
+      "un": 546,
+      "ลี่ย": 547,
+      "ผล": 548,
+      "พิ": 549,
+      "▁วง": 550,
+      "ia": 551,
+      "ากร": 552,
+      "เรียน": 553,
+      "หม่": 554,
+      "ุบัน": 555,
+      "ชน": 556,
+      "um": 557,
+      "ยา": 558,
+      "▁ฉ": 559,
+      "▁The": 560,
+      "▁จึง": 561,
+      "▁ประช": 562,
+      "อิง": 563,
+      "่งข": 564,
+      "▁หลาย": 565,
+      "▁อง": 566,
+      "ันธ์": 567,
+      "▁อ้างอิง": 568,
+      "▁ปัจจุบัน": 569,
+      "▁เพิ่ม": 570,
+      "▁เห": 571,
+      "รา": 572,
+      "ห์": 573,
+      "▁อิน": 574,
+      "▁คณะ": 575,
+      "ec": 576,
+      "▁ที่สุด": 577,
+      "▁มิ": 578,
+      "▁U": 579,
+      "▁b": 580,
+      "ัม": 581,
+      "▁โคร": 582,
+      "ครอง": 583,
+      "ตรี": 584,
+      "▁ทําให้": 585,
+      "ุง": 586,
+      "▁ชน": 587,
+      "▁สี": 588,
+      "▁หลัก": 589,
+      "▁ไม้": 590,
+      "▁แน": 591,
+      "ut": 592,
+      "▁ดู": 593,
+      "▁ช่วง": 594,
+      "ด์": 595,
+      "ตัว": 596,
+      "ใต้": 597,
+      "่วย": 598,
+      "▁กิ": 599,
+      "ดิ": 600,
+      "▁ภาพ": 601,
+      "นาด": 602,
+      "▁เบ": 603,
+      "้าย": 604,
+      "▁สถาน": 605,
+      "ag": 606,
+      "าด": 607,
+      "ือก": 608,
+      "▁นํา": 609,
+      "▁โรง": 610,
+      "็บ": 611,
+      "▁In": 612,
+      "▁ใบ": 613,
+      "นิ": 614,
+      "▁เฉ": 615,
+      "▁in": 616,
+      "▁n": 617,
+      "▁เกิด": 618,
+      "▁เกี่ย": 619,
+      "พาะ": 620,
+      "สาร": 621,
+      "▁แผ": 622,
+      "▁เขต": 623,
+      "แก่": 624,
+      "▁วันที่": 625,
+      "▁ง": 626,
+      "เบ": 627,
+      "▁ภูมิ": 628,
+      "เซีย": 629,
+      "ุล": 630,
+      "ต้น": 631,
+      "จะ": 632,
+      "หมาย": 633,
+      "ศ์": 634,
+      "▁แบ่ง": 635,
+      "iv": 636,
+      "เฉ": 637,
+      "▁ไว": 638,
+      "▁ภาย": 639,
+      "▁คล": 640,
+      "▁น้ํา": 641,
+      "▁สําหรับ": 642,
+      "▁แก": 643,
+      "▁เขา": 644,
+      "im": 645,
+      "ทน": 646,
+      "▁a": 647,
+      "▁สร้าง": 648,
+      "ัท": 649,
+      "อมพ": 650,
+      "▁สูง": 651,
+      "ใจ": 652,
+      "▁บุ": 653,
+      "คัญ": 654,
+      "บาล": 655,
+      "▁โครง": 656,
+      "ษฐ": 657,
+      "▁เทค": 658,
+      "ลาย": 659,
+      "▁ตอน": 660,
+      "▁หมาย": 661,
+      "▁หน่วย": 662,
+      "โร": 663,
+      "แบบ": 664,
+      "▁เค": 665,
+      "▁ได้แก่": 666,
+      "ังกฤษ": 667,
+      "เมตร": 668,
+      "▁สัง": 669,
+      "รรค": 670,
+      "ด้วย": 671,
+      "ot": 672,
+      "้อน": 673,
+      "่วม": 674,
+      "▁ลํา": 675,
+      "ศาสนา": 676,
+      "ชื่อ": 677,
+      "▁ประมาณ": 678,
+      "▁ในประเทศ": 679,
+      "em": 680,
+      "oc": 681,
+      "ist": 682,
+      "▁โอ": 683,
+      "อยู่": 684,
+      "▁ประกอบ": 685,
+      "ันท": 686,
+      "อมพิว": 687,
+      "อมพิวเตอร์": 688,
+      "st": 689,
+      "ีวิต": 690,
+      "▁ระดับ": 691,
+      "แต่": 692,
+      "▁ลง": 693,
+      "▁เจ้า": 694,
+      "▁พื้นที่": 695,
+      "▁ต้อง": 696,
+      "id": 697,
+      "▁จน": 698,
+      "ิทยา": 699,
+      "กครอง": 700,
+      "ลัง": 701,
+      "ริ่ม": 702,
+      "▁ขนาด": 703,
+      "ra": 704,
+      "ล์": 705,
+      "ลี่ยน": 706,
+      "ธี": 707,
+      "▁นา": 708,
+      "▁นอก": 709,
+      "▁บน": 710,
+      "▁J": 711,
+      "าติ": 712,
+      "▁เย": 713,
+      "▁จีน": 714,
+      "▁ญ": 715,
+      "ียว": 716,
+      "▁ทุก": 717,
+      "▁วิทยาศาสตร์": 718,
+      "จักร": 719,
+      "บุรี": 720,
+      "▁ทรง": 721,
+      "ัช": 722,
+      "▁ดี": 723,
+      "▁สิ่ง": 724,
+      "กา": 725,
+      "เจ้า": 726,
+      "ต่อ": 727,
+      "อากาศ": 728,
+      "▁ยุค": 729,
+      "เป็น": 730,
+      "ําเน": 731,
+      "▁ลักษ": 732,
+      "▁มาจาก": 733,
+      "ยน": 734,
+      "ไม": 735,
+      "ers": 736,
+      "▁ริ": 737,
+      "ยม": 738,
+      "ยี": 739,
+      "ังส": 740,
+      "▁โปร": 741,
+      "โนโล": 742,
+      "โนโลยี": 743,
+      "ณฑ": 744,
+      "ch": 745,
+      "วกับ": 746,
+      "▁เจ": 747,
+      "ขึ้น": 748,
+      "รู้": 749,
+      "▁สง": 750,
+      "▁ครั้ง": 751,
+      "ul": 752,
+      "รรณ": 753,
+      "▁ดอก": 754,
+      "วัติ": 755,
+      "มี": 756,
+      "เด็จ": 757,
+      "ิศ": 758,
+      "ะเล": 759,
+      "วิทยา": 760,
+      "ใน": 761,
+      "▁เส้น": 762,
+      "▁การป": 763,
+      "ธาน": 764,
+      "บัติ": 765,
+      "▁ความส": 766,
+      "▁ต้น": 767,
+      "▁พรรค": 768,
+      "▁ศาสนา": 769,
+      "og": 770,
+      "นตรี": 771,
+      "▁ดวง": 772,
+      "▁เหล": 773,
+      "▁St": 774,
+      "▁พิ": 775,
+      "▁แนว": 776,
+      "กรม": 777,
+      "▁ซี": 778,
+      "▁แรก": 779,
+      "เคร": 780,
+      "ได้": 781,
+      "▁ชั้น": 782,
+      "▁คอมพิวเตอร์": 783,
+      "▁ลักษณะ": 784,
+      "ัฒนา": 785,
+      "าการ": 786,
+      "เหนือ": 787,
+      "▁ใหม่": 788,
+      "▁ยาว": 789,
+      "รั่ง": 790,
+      "▁ทั่ว": 791,
+      "▁l": 792,
+      "▁สน": 793,
+      "▁การศึกษา": 794,
+      "วี": 795,
+      "▁ไว้": 796,
+      "เอ": 797,
+      "โด": 798,
+      "มพ์": 799,
+      "นั้น": 800,
+      "▁องค์": 801,
+      "▁เริ่ม": 802,
+      "ter": 803,
+      "นุ��": 804,
+      "▁รา": 805,
+      "▁ออ": 806,
+      "▁กรุง": 807,
+      "op": 808,
+      "▁อาจ": 809,
+      "อย่าง": 810,
+      "▁เทคโนโลยี": 811,
+      "ซ์": 812,
+      "ี่ป": 813,
+      "▁ก่อ": 814,
+      "▁ดิน": 815,
+      "ี่ปุ่น": 816,
+      "▁เสียง": 817,
+      "▁รับ": 818,
+      "▁ญี่ปุ่น": 819,
+      "าบัน": 820,
+      "▁หลัง": 821,
+      "▁ใหญ่": 822,
+      "▁ฝรั่ง": 823,
+      "เศส": 824,
+      "รษฐ": 825,
+      "▁สถาบัน": 826,
+      "ิทธิ": 827,
+      "็ม": 828,
+      "วรรษ": 829,
+      "▁ฝรั่งเศส": 830,
+      "ังสือ": 831,
+      "เดีย": 832,
+      "วบ": 833,
+      "ภูมิ": 834,
+      "ัญญา": 835,
+      "ทธ": 836,
+      "แล": 837,
+      "ายน": 838,
+      "ักษร": 839,
+      "ริส": 840,
+      "▁เต": 841,
+      "▁ติด": 842,
+      "▁เชิง": 843,
+      "▁พบ": 844,
+      "▁โค": 845,
+      "หนึ่ง": 846,
+      "▁อิ": 847,
+      "กิจ": 848,
+      "▁จุด": 849,
+      "กลาง": 850,
+      "ชีวิต": 851,
+      "▁สมัย": 852,
+      "าคม": 853,
+      "▁คุ": 854,
+      "เชีย": 855,
+      "▁เกี่ยวกับ": 856,
+      "ess": 857,
+      "ําร": 858,
+      "▁โร": 859,
+      "▁งาน": 860,
+      "▁เดือน": 861,
+      "แดน": 862,
+      "▁แหล่งข": 863,
+      "ียบ": 864,
+      "เขียน": 865,
+      "▁กลาง": 866,
+      "▁ตั้งแต่": 867,
+      "▁แหล่งข้อมูล": 868,
+      "▁ปฏ": 869,
+      "▁แสดง": 870,
+      "▁ตะวันออก": 871,
+      "ce": 872,
+      "ณิต": 873,
+      "▁มนุษ": 874,
+      "เห": 875,
+      "ัมพ": 876,
+      "▁สาย": 877,
+      "▁สําคัญ": 878,
+      "ูนย์": 879,
+      "▁สอง": 880,
+      "ัฒน": 881,
+      "▁เนื้อ": 882,
+      "ขา": 883,
+      "▁ส่ง": 884,
+      "▁หมู่": 885,
+      "ดา": 886,
+      "พล": 887,
+      "ยะ": 888,
+      "ว์": 889,
+      "▁เง": 890,
+      "▁แป": 891,
+      "ํานา": 892,
+      "ไทย": 893,
+      "▁แค": 894,
+      "ํานัก": 895,
+      "เจ": 896,
+      "▁ตํา": 897,
+      "od": 898,
+      "▁เกาะ": 899,
+      "ี่ยว": 900,
+      "ฎี": 901,
+      "▁ร้อย": 902,
+      "ir": 903,
+      "ad": 904,
+      "▁แอ": 905,
+      "▁สาร": 906,
+      "▁ล้าน": 907,
+      "ดิน": 908,
+      "เขา": 909,
+      "หนด": 910,
+      "่อย": 911,
+      "▁ละ": 912,
+      "▁ลิ": 913,
+      "▁ปราก": 914,
+      "▁สมเด็จ": 915,
+      "พร": 916,
+      "หน้า": 917,
+      "พิมพ์": 918,
+      "ส่วน": 919,
+      "▁หัว": 920,
+      "ท์": 921,
+      "▁ไฟ": 922,
+      "▁V": 923,
+      "ิ่น": 924,
+      "โรป": 925,
+      "▁ประเ": 926,
+      "▁ประวัติ": 927,
+      "ันธุ": 928,
+      "▁จํา": 929,
+      "▁เอก": 930,
+      "▁กา": 931,
+      "กว่า": 932,
+      "คราม": 933,
+      "าชิก": 934,
+      "เสียง": 935,
+      "▁การส": 936,
+      "็ต": 937,
+      "บริ": 938,
+      "ษัท": 939,
+      "ํานาจ": 940,
+      "▁ปรากฏ": 941,
+      "▁อันดับ": 942,
+      "วล": 943,
+      "▁w": 944,
+      "รณ์": 945,
+      "หมด": 946,
+      "ณาจักร": 947,
+      "▁เขียน": 948,
+      "▁รูปแบบ": 949,
+      "ab": 950,
+      "▁คา": 951,
+      "▁เศ": 952,
+      "ฤษฎี": 953,
+      "▁นับ": 954,
+      "▁เฉพาะ": 955,
+      "▁ประชากร": 956,
+      "้ม": 957,
+      "โลเมตร": 958,
+      "▁หนังสือ": 959,
+      "▁โบ": 960,
+      "▁ร่วม": 961,
+      "▁เวลา": 962,
+      "เหล": 963,
+      "ริกา": 964,
+      "▁บริษัท": 965,
+      "▁สํานัก": 966,
+      "าป": 967,
+      "โย": 968,
+      "▁โม": 969,
+      "▁ท่า": 970,
+      "▁การค": 971,
+      "ew": 972,
+      "▁ที": 973,
+      "▁รี": 974,
+      "▁ทั้งหมด": 975,
+      "ay": 976,
+      "ow": 977,
+      "สน": 978,
+      "▁รถ": 979,
+      "▁เชื่อ": 980,
+      "ฟ้า": 981,
+      "ภท": 982,
+      "ุต": 983,
+      "▁ค่า": 984,
+      "▁นอกจาก": 985,
+      "ian": 986,
+      "▁อังกฤษ": 987,
+      "าะห์": 988,
+      "การณ์": 989,
+      "ity": 990,
+      "ทํา": 991,
+      "▁มนุษย์": 992,
+      "ิลป": 993,
+      "เวณ": 994,
+      "โลก": 995,
+      "▁ยู": 996,
+      "▁เหม": 997,
+      "▁เว็บ": 998,
+      "os": 999
+    },
+    "unk_token": "<unk>"
+  }
+}