Upload folder using huggingface_hub

README.md

@@ -110,11 +110,11 @@ You can finetune this model on your own dataset.
 
 ### Framework Versions
 - Python: 3.11.12
-- Sentence Transformers: 3.4.1
+- Sentence Transformers: 4.1.0
 - Transformers: 4.51.3
 - PyTorch: 2.6.0+cu124
 - Accelerate: 1.6.0
-- Datasets: 
+- Datasets: 2.14.4
 - Tokenizers: 0.21.1
 
 ## Citation

added_tokens.json

@@ -1,21 +1,32 @@
 {
+  "ASFR": 30654,
+  "BOP": 30648,
   "BPP": 30551,
   "BPS": 30521,
   "BRS": 30566,
   "Badan Pusat Statistik": 30522,
   "CIF": 30556,
+  "CMR": 30649,
+  "CPI": 30639,
   "EKSPOR": 30573,
   "Ekspor": 30552,
   "FOB": 30555,
+  "GDP": 30637,
+  "GK": 30626,
+  "GRDP": 30638,
+  "HDI": 30641,
   "HLS": 30543,
   "HS": 30554,
+  "IBS": 30624,
   "IHK": 30525,
   "IHP": 30527,
   "IHPB": 30528,
   "IMK": 30532,
   "IMPOR": 30574,
+  "IMR": 30650,
   "IPH": 30526,
   "IPM": 30537,
+  "IPP": 30640,
   "ITB": 30531,
   "ITK": 30530,
   "Impor": 30553,
@@ -27,37 +38,57 @@
   "KKL": 30560,
   "KRT": 30563,
   "Katalog": 30571,
+  "LFPR": 30644,
+  "LFS": 30647,
   "LPE": 30533,
   "LTN": 30549,
   "LTT": 30548,
+  "MDGs": 30645,
+  "METADATA": 30627,
+  "MMR": 30652,
+  "MYS": 30643,
   "Metadata": 30570,
   "NTP": 30529,
   "NTUP": 30550,
   "PDB": 30524,
   "PDRB": 30523,
   "PKL": 30559,
+  "PODES": 30628,
   "PUBLIKASI": 30576,
+  "Podes": 30623,
   "Publikasi": 30567,
   "RLS": 30544,
   "RT": 30564,
   "RW": 30565,
   "SAKERNAS": 30575,
   "SDGI": 30541,
+  "SDGs": 30646,
   "SDKI": 30540,
+  "SE": 30621,
   "SEKTORAL": 30579,
+  "SHK": 30619,
+  "SKHI": 30625,
+  "SP": 30620,
+  "SP2010": 30616,
   "SP2020": 30538,
+  "ST": 30622,
   "ST2013": 30547,
   "ST2023": 30546,
   "STATISTIK": 30578,
   "SUPAS": 30539,
   "SURVEI": 30581,
+  "SUSENAS": 30618,
   "SUTAS": 30545,
   "Sakernas": 30557,
   "Sektoral": 30569,
+  "Statistik": 30617,
   "Survei": 30572,
+  "TFR": 30653,
   "TPAK": 30558,
   "TPK": 30534,
   "TPT": 30535,
+  "U5MR": 30651,
+  "UHC": 30642,
   "UMP": 30536,
   "_Statistik": 30568,
   "bpp": 30604,
@@ -65,7 +96,10 @@
   "cif": 30606,
   "fob": 30605,
   "hls": 30596,
+  "ibs": 30635,
   "ihk": 30583,
+  "ihp": 30629,
+  "ihpb": 30630,
   "imk": 30587,
   "iph": 30584,
   "ipm": 30591,
@@ -81,15 +115,20 @@
   "ntp": 30585,
   "ntup": 30603,
   "pdrb": 30582,
+  "podes": 30634,
   "rls": 30597,
   "sakernas": 30607,
   "sdgi": 30595,
   "sdki": 30594,
   "sektoral": 30614,
+  "shk": 30633,
+  "skhi": 30636,
+  "sp2010": 30631,
   "sp2020": 30592,
   "st2013": 30600,
   "st2023": 30599,
   "supas": 30593,
+  "susenas": 30632,
   "sutas": 30598,
   "tpak": 30608,
   "tpk": 30589,

config.json

@@ -42,5 +42,5 @@
   "transformers_version": "4.51.3",
   "type_vocab_size": 2,
   "use_cache": true,
-  "vocab_size": 30616
+  "vocab_size": 30655
 }

config_sentence_transformers.json

@@ -1,6 +1,6 @@
 {
   "__version__": {
-    "sentence_transformers": "3.4.1",
+    "sentence_transformers": "4.1.0",
     "transformers": "4.51.3",
     "pytorch": "2.6.0+cu124"
   },

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:54c6e7bf16068c5948125e9b4f15700979f4bea31f2154d0a2041f950686427d
-size 438240096
+oid sha256:1ee2a7f419365e4f5aed12a0740460c5a4aa84365ced586546df0923664677ee
+size 438359904

tokenizer.json

@@ -902,6 +902,357 @@
       "rstrip": false,
       "normalized": true,
       "special": false
+    },
+    {
+      "id": 30616,
+      "content": "SP2010",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30617,
+      "content": "Statistik",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30618,
+      "content": "SUSENAS",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30619,
+      "content": "SHK",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30620,
+      "content": "SP",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30621,
+      "content": "SE",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30622,
+      "content": "ST",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30623,
+      "content": "Podes",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30624,
+      "content": "IBS",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30625,
+      "content": "SKHI",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30626,
+      "content": "GK",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30627,
+      "content": "METADATA",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30628,
+      "content": "PODES",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30629,
+      "content": "ihp",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30630,
+      "content": "ihpb",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30631,
+      "content": "sp2010",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30632,
+      "content": "susenas",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30633,
+      "content": "shk",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30634,
+      "content": "podes",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30635,
+      "content": "ibs",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30636,
+      "content": "skhi",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30637,
+      "content": "GDP",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30638,
+      "content": "GRDP",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30639,
+      "content": "CPI",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30640,
+      "content": "IPP",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30641,
+      "content": "HDI",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30642,
+      "content": "UHC",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30643,
+      "content": "MYS",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30644,
+      "content": "LFPR",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30645,
+      "content": "MDGs",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30646,
+      "content": "SDGs",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30647,
+      "content": "LFS",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30648,
+      "content": "BOP",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30649,
+      "content": "CMR",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30650,
+      "content": "IMR",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30651,
+      "content": "U5MR",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30652,
+      "content": "MMR",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30653,
+      "content": "TFR",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 30654,
+      "content": "ASFR",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
     }
   ],
   "normalizer": {

tokenizer_config.json

@@ -799,6 +799,318 @@
       "rstrip": false,
       "single_word": false,
       "special": false
+    },
+    "30616": {
+      "content": "SP2010",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30617": {
+      "content": "Statistik",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30618": {
+      "content": "SUSENAS",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30619": {
+      "content": "SHK",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30620": {
+      "content": "SP",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30621": {
+      "content": "SE",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30622": {
+      "content": "ST",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30623": {
+      "content": "Podes",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30624": {
+      "content": "IBS",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30625": {
+      "content": "SKHI",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30626": {
+      "content": "GK",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30627": {
+      "content": "METADATA",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30628": {
+      "content": "PODES",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30629": {
+      "content": "ihp",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30630": {
+      "content": "ihpb",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30631": {
+      "content": "sp2010",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30632": {
+      "content": "susenas",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30633": {
+      "content": "shk",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30634": {
+      "content": "podes",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30635": {
+      "content": "ibs",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30636": {
+      "content": "skhi",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30637": {
+      "content": "GDP",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30638": {
+      "content": "GRDP",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30639": {
+      "content": "CPI",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30640": {
+      "content": "IPP",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30641": {
+      "content": "HDI",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30642": {
+      "content": "UHC",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30643": {
+      "content": "MYS",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30644": {
+      "content": "LFPR",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30645": {
+      "content": "MDGs",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30646": {
+      "content": "SDGs",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30647": {
+      "content": "LFS",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30648": {
+      "content": "BOP",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30649": {
+      "content": "CMR",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30650": {
+      "content": "IMR",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30651": {
+      "content": "U5MR",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30652": {
+      "content": "MMR",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30653": {
+      "content": "TFR",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30654": {
+      "content": "ASFR",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
     }
   },
   "clean_up_tokenization_spaces": true,