Spaces:

sotirios-slv
/

scrambled-snake

Running

App Files Files Community

sotirios-slv commited on Dec 22, 2024

Commit

dd7931c

verified ·

1 Parent(s): 07c7fb5

Synced repo using 'sync_with_huggingface' Github Action

Browse files

Files changed (10) hide show

Pipfile +1 -0
Pipfile.lock +1015 -191
__pycache__/alma_harvester.cpython-311.pyc +0 -0
__pycache__/image_analysis.cpython-311.pyc +0 -0
__pycache__/text_analysis.cpython-311.pyc +0 -0
alma_harvester.py +79 -0
dataset_wrangler.py +4 -13
image_analysis.py +50 -20
requirements.txt +103 -108
text_analysis.py +73 -0

Pipfile CHANGED Viewed

@@ -12,6 +12,7 @@ scikit-learn = "*"
 matplotlib = "*"
 streamlit = "*"
 bokeh = "*"
 [dev-packages]

 matplotlib = "*"
 streamlit = "*"
 bokeh = "*"
+flair = "*"
 [dev-packages]

Pipfile.lock CHANGED Viewed

@@ -1,7 +1,7 @@
 {
     "_meta": {
         "hash": {
-            "sha256": "1798ba32767e70e63866b22f70defa1d9564846e0e4a234b66fa5c137bb2ad76"
         },
         "pipfile-spec": 6,
         "requires": {
@@ -16,6 +16,13 @@
         ]
     },
     "default": {
         "altair": {
             "hashes": [
                 "sha256:91a310b926508d560fe0148d02a194f38b824122641ef528113d029fcd129f8c",
@@ -26,11 +33,27 @@
         },
         "attrs": {
             "hashes": [
-                "sha256:5cfb1b9148b5b086569baec03f20d7b6bf3bcacc9a42bebf87ffaaca362f6346",
-                "sha256:81921eb96de3191c8258c199618104dd27ac608d9366f5e35d011eae1867ede2"
             ],
-            "markers": "python_version >= '3.7'",
-            "version": "==24.2.0"
         },
         "blinker": {
             "hashes": [
@@ -48,6 +71,22 @@
             "index": "pypi",
             "version": "==3.6.2"
         },
         "cachetools": {
             "hashes": [
                 "sha256:02134e8439cdc2ffb62023ce1debca2944c3f289d66bb17ead3ab3dede74b292",
@@ -58,11 +97,11 @@
         },
         "certifi": {
             "hashes": [
-                "sha256:922820b53db7a7257ffbda3f597266d435245903d80737e34f8a45ff3e3230d8",
-                "sha256:bec941d2aa8195e248a60b31ff9f0558284cf01a52591ceda73ea9afffd69fd9"
             ],
             "markers": "python_version >= '3.6'",
-            "version": "==2024.8.30"
         },
         "charset-normalizer": {
             "hashes": [
@@ -191,6 +230,14 @@
             "markers": "platform_system == 'Windows'",
             "version": "==0.4.6"
         },
         "contourpy": {
             "hashes": [
                 "sha256:041b640d4ec01922083645a94bb3b2e777e6b626788f4095cf21abbe266413c1",
@@ -259,61 +306,115 @@
             "markers": "python_version >= '3.8'",
             "version": "==0.12.1"
         },
         "fonttools": {
             "hashes": [
-                "sha256:09fe922a3eff181fd07dd724cdb441fb6b9fc355fd1c0f1aa79aca60faf1fbdd",
-                "sha256:0ecd1c2b1c2ec46bb73685bc5473c72e16ed0930ef79bc2919ccadc43a99fb16",
-                "sha256:10aff204e2edee1d312fa595c06f201adf8d528a3b659cfb34cd47eceaaa6a26",
-                "sha256:131591ac8d7a47043aaf29581aba755ae151d46e49d2bf49608601efd71e8b4d",
-                "sha256:18f082445b8fe5e91c53e6184f4c1c73f3f965c8bcc614c6cd6effd573ce6c1a",
-                "sha256:22ef222740eb89d189bf0612eb98fbae592c61d7efeac51bfbc2a1592d469557",
-                "sha256:25062b6ca03464dd5179fc2040fb19e03391b7cc49b9cc4f879312e638605c5c",
-                "sha256:27c0f91adbbd706e8acd1db73e3e510118e62d0ffb651864567dccc5b2339f90",
-                "sha256:2df61d9fc15199cc86dad29f64dd686874a3a52dda0c2d8597d21f509f95c332",
-                "sha256:3d8ccce035320d63dba0c35f52499322f5531dbe85bba1514c7cea26297e4c54",
-                "sha256:3d9bbc1e380fdaf04ad9eabd8e3e6a4301eaf3487940893e9fd98537ea2e283b",
-                "sha256:42a9afedff07b6f75aa0f39b5e49922ac764580ef3efce035ca30284b2ee65c8",
-                "sha256:42aca564b575252fd9954ed0d91d97a24de24289a16ce8ff74ed0bdf5ecebf11",
-                "sha256:44cf2a98aa661dbdeb8c03f5e405b074e2935196780bb729888639f5276067d9",
-                "sha256:45947e7b3f9673f91df125d375eb57b9a23f2a603f438a1aebf3171bffa7a205",
-                "sha256:487e1e8b524143a799bda0169c48b44a23a6027c1bb1957d5a172a7d3a1dd704",
-                "sha256:4c83381c3e3e3d9caa25527c4300543578341f21aae89e4fbbb4debdda8d82a2",
-                "sha256:508ebb42956a7a931c4092dfa2d9b4ffd4f94cea09b8211199090d2bd082506b",
-                "sha256:5b1a6e576db0c83c1b91925bf1363478c4bb968dbe8433147332fb5782ce6190",
-                "sha256:5cfa67414d7414442a5635ff634384101c54f53bb7b0e04aa6a61b013fcce194",
-                "sha256:616368b15716781bc84df5c2191dc0540137aaef56c2771eb4b89b90933f347a",
-                "sha256:627cf10d6f5af5bec6324c18a2670f134c29e1b7dce3fb62e8ef88baa6cba7a9",
-                "sha256:663eba5615d6abaaf616432354eb7ce951d518e43404371bcc2b0694ef21e8d6",
-                "sha256:6b5917ef79cac8300b88fd6113003fd01bbbbea2ea060a27b95d8f77cb4c65c2",
-                "sha256:6fc88cfb58b0cd7b48718c3e61dd0d0a3ee8e2c86b973342967ce09fbf1db6d4",
-                "sha256:7bbae4f3915225c2c37670da68e2bf18a21206060ad31dfb95fec91ef641caa7",
-                "sha256:803d5cef5fc47f44f5084d154aa3d6f069bb1b60e32390c225f897fa19b0f939",
-                "sha256:81ccd2b3a420b8050c7d9db3be0555d71662973b3ef2a1d921a2880b58957db8",
-                "sha256:8b02b10648d69d67a7eb055f4d3eedf4a85deb22fb7a19fbd9acbae7c7538199",
-                "sha256:8bc5f100de0173cc39102c0399bd6c3bd544bbdf224957933f10ee442d43cddd",
-                "sha256:8e2d89fbe9b08d96e22c7a81ec04a4e8d8439c31223e2dc6f2f9fc8ff14bdf9f",
-                "sha256:9008438ad59e5a8e403a62fbefef2b2ff377eb3857d90a3f2a5f4d674ff441b2",
-                "sha256:93f439ca27e55f585e7aaa04a74990acd983b5f2245e41d6b79f0a8b44e684d8",
-                "sha256:944228b86d472612d3b48bcc83b31c25c2271e63fdc74539adfcfa7a96d487fb",
-                "sha256:96e126df9615df214ec7f04bebcf60076297fbc10b75c777ce58b702d7708ffb",
-                "sha256:9b1726872e09268bbedb14dc02e58b7ea31ecdd1204c6073eda4911746b44797",
-                "sha256:9f0e55f5da594b85f269cfbecd2f6bd3e07d0abba68870bc3f34854de4fa4678",
-                "sha256:bbea0ab841113ac8e8edde067e099b7288ffc6ac2dded538b131c2c0595d5f77",
-                "sha256:bef0f8603834643b1a6419d57902f18e7d950ec1a998fb70410635c598dc1a1e",
-                "sha256:c1b9de46ef7b683d50400abf9f1578eaceee271ff51c36bf4b7366f2be29f498",
-                "sha256:c6457f650ebe15baa17fc06e256227f0a47f46f80f27ec5a0b00160de8dc2c13",
-                "sha256:d0bf24d2b02dbc9376d795a63062632ff73e3e9e60c0229373f500aed7e86dd7",
-                "sha256:d1100d8e665fe386a79cab59446992de881ea74d0d6c191bb988642692aa2421",
-                "sha256:d337ec087da8216a828574aa0525d869df0a2ac217a2efc1890974ddd1fbc5b9",
-                "sha256:d34525e8141286fa976e14806639d32294bfb38d28bbdb5f6be9f46a1cd695a6",
-                "sha256:d4ff250ed4ff05015dfd9cf2adf7570c7a383ca80f4d9732ac484a5ed0d8453c",
-                "sha256:d559eb1744c7dcfa90ae60cb1a4b3595e898e48f4198738c321468c01180cd83",
-                "sha256:dbdc251c5e472e5ae6bc816f9b82718b8e93ff7992e7331d6cf3562b96aa268e",
-                "sha256:e857fe1859901ad8c5cab32e0eebc920adb09f413d2d73b74b677cf47b28590c",
-                "sha256:f1c76f423f1a241df08f87614364dff6e0b7ce23c962c1b74bd995ec7c0dad13"
             ],
             "markers": "python_version >= '3.8'",
-            "version": "==4.55.2"
         },
         "gitdb": {
             "hashes": [
@@ -331,6 +432,14 @@
             "markers": "python_version >= '3.7'",
             "version": "==3.1.43"
         },
         "idna": {
             "hashes": [
                 "sha256:12f65c9b470abda6dc35cf8e63cc574b1c52b11df2c86030af0ac09b01b13ea9",
@@ -339,6 +448,12 @@
             "markers": "python_version >= '3.6'",
             "version": "==3.10"
         },
         "jinja2": {
             "hashes": [
                 "sha256:4a3aee7acbbe7303aede8e9648d13b8bf88a429282aa6122a993f0ac800cb369",
@@ -347,6 +462,14 @@
             "markers": "python_version >= '3.7'",
             "version": "==3.1.4"
         },
         "joblib": {
             "hashes": [
                 "sha256:06d478d5674cbc267e7496a410ee875abd68e4340feff4490bcb7afb88060ae6",
@@ -355,6 +478,14 @@
             "markers": "python_version >= '3.8'",
             "version": "==1.4.2"
         },
         "jsonschema": {
             "hashes": [
                 "sha256:d71497fef26351a33265337fa77ffeb82423f3ea21283cd9467bb03999266bc4",
@@ -491,6 +622,157 @@
             "markers": "python_version >= '3.8'",
             "version": "==1.4.7"
         },
         "markdown-it-py": {
             "hashes": [
                 "sha256:355216845c60bd96232cd8d8c40e8f9765cc86f46880e43a8fd22dc1a1a8cab1",
@@ -568,50 +850,43 @@
         },
         "matplotlib": {
             "hashes": [
-                "sha256:026bdf3137ab6022c866efa4813b6bbeddc2ed4c9e7e02f0e323a7bca380dfa0",
-                "sha256:031b7f5b8e595cc07def77ec5b58464e9bb67dc5760be5d6f26d9da24892481d",
-                "sha256:0a0a63cb8404d1d1f94968ef35738900038137dab8af836b6c21bb6f03d75465",
-                "sha256:0a361bd5583bf0bcc08841df3c10269617ee2a36b99ac39d455a767da908bbbc",
-                "sha256:10d3e5c7a99bd28afb957e1ae661323b0800d75b419f24d041ed1cc5d844a764",
-                "sha256:1c40c244221a1adbb1256692b1133c6fb89418df27bf759a31a333e7912a4010",
-                "sha256:203d18df84f5288973b2d56de63d4678cc748250026ca9e1ad8f8a0fd8a75d83",
-                "sha256:213d6dc25ce686516208d8a3e91120c6a4fdae4a3e06b8505ced5b716b50cc04",
-                "sha256:3119b2f16de7f7b9212ba76d8fe6a0e9f90b27a1e04683cd89833a991682f639",
-                "sha256:3fb0b37c896172899a4a93d9442ffdc6f870165f59e05ce2e07c6fded1c15749",
-                "sha256:41b016e3be4e740b66c79a031a0a6e145728dbc248142e751e8dab4f3188ca1d",
-                "sha256:4a8d279f78844aad213c4935c18f8292a9432d51af2d88bca99072c903948045",
-                "sha256:4e6eefae6effa0c35bbbc18c25ee6e0b1da44d2359c3cd526eb0c9e703cf055d",
-                "sha256:5f2a4ea08e6876206d511365b0bc234edc813d90b930be72c3011bbd7898796f",
-                "sha256:66d7b171fecf96940ce069923a08ba3df33ef542de82c2ff4fe8caa8346fa95a",
-                "sha256:687df7ceff57b8f070d02b4db66f75566370e7ae182a0782b6d3d21b0d6917dc",
-                "sha256:6be0ba61f6ff2e6b68e4270fb63b6813c9e7dec3d15fc3a93f47480444fd72f0",
-                "sha256:6e9de2b390d253a508dd497e9b5579f3a851f208763ed67fdca5dc0c3ea6849c",
-                "sha256:760a5e89ebbb172989e8273024a1024b0f084510b9105261b3b00c15e9c9f006",
-                "sha256:816a966d5d376bf24c92af8f379e78e67278833e4c7cbc9fa41872eec629a060",
-                "sha256:87ad73763d93add1b6c1f9fcd33af662fd62ed70e620c52fcb79f3ac427cf3a6",
-                "sha256:896774766fd6be4571a43bc2fcbcb1dcca0807e53cab4a5bf88c4aa861a08e12",
-                "sha256:8e0143975fc2a6d7136c97e19c637321288371e8f09cff2564ecd73e865ea0b9",
-                "sha256:90a85a004fefed9e583597478420bf904bb1a065b0b0ee5b9d8d31b04b0f3f70",
-                "sha256:9b081dac96ab19c54fd8558fac17c9d2c9cb5cc4656e7ed3261ddc927ba3e2c5",
-                "sha256:9d6b2e8856dec3a6db1ae51aec85c82223e834b228c1d3228aede87eee2b34f9",
-                "sha256:9f459c8ee2c086455744723628264e43c884be0c7d7b45d84b8cd981310b4815",
-                "sha256:9fa6e193c14d6944e0685cdb527cb6b38b0e4a518043e7212f214113af7391da",
-                "sha256:a42b9dc42de2cfe357efa27d9c50c7833fc5ab9b2eb7252ccd5d5f836a84e1e4",
-                "sha256:b651b0d3642991259109dc0351fc33ad44c624801367bb8307be9bfc35e427ad",
-                "sha256:b6c12514329ac0d03128cf1dcceb335f4fbf7c11da98bca68dca8dcb983153a9",
-                "sha256:c52f48eb75fcc119a4fdb68ba83eb5f71656999420375df7c94cc68e0e14686e",
-                "sha256:c96eeeb8c68b662c7747f91a385688d4b449687d29b691eff7068a4602fe6dc4",
-                "sha256:cd1077b9a09b16d8c3c7075a8add5ffbfe6a69156a57e290c800ed4d435bef1d",
-                "sha256:cd5dbbc8e25cad5f706845c4d100e2c8b34691b412b93717ce38d8ae803bcfa5",
-                "sha256:cf2a60daf6cecff6828bc608df00dbc794380e7234d2411c0ec612811f01969d",
-                "sha256:d3c93796b44fa111049b88a24105e947f03c01966b5c0cc782e2ee3887b790a3",
-                "sha256:d796272408f8567ff7eaa00eb2856b3a00524490e47ad505b0b4ca6bb8a7411f",
-                "sha256:e0fcb7da73fbf67b5f4bdaa57d85bb585a4e913d4a10f3e15b32baea56a67f0a",
-                "sha256:e14485bb1b83eeb3d55b6878f9560240981e7bbc7a8d4e1e8c38b9bd6ec8d2de",
-                "sha256:edd14cf733fdc4f6e6fe3f705af97676a7e52859bf0044aa2c84e55be739241c"
             ],
             "index": "pypi",
-            "version": "==3.9.3"
         },
         "mdurl": {
             "hashes": [
@@ -621,74 +896,85 @@
             "markers": "python_version >= '3.7'",
             "version": "==0.1.2"
         },
         "narwhals": {
             "hashes": [
-                "sha256:1ea3ac269df8126f1f79c1ba7b8f78c73b032a86861b2cc2f08c48010ef9f6f8",
-                "sha256:e5b764b1b571e25b08bb87db6feb89ff824d76f5d7b7e12e6d837c1c503763ae"
             ],
             "markers": "python_version >= '3.8'",
-            "version": "==1.16.0"
         },
-        "numpy": {
             "hashes": [
-                "sha256:0557eebc699c1c34cccdd8c3778c9294e8196df27d713706895edc6f57d29608",
-                "sha256:0798b138c291d792f8ea40fe3768610f3c7dd2574389e37c3f26573757c8f7ef",
-                "sha256:0da8495970f6b101ddd0c38ace92edea30e7e12b9a926b57f5fabb1ecc25bb90",
-                "sha256:0f0986e917aca18f7a567b812ef7ca9391288e2acb7a4308aa9d265bd724bdae",
-                "sha256:122fd2fcfafdefc889c64ad99c228d5a1f9692c3a83f56c292618a59aa60ae83",
-                "sha256:140dd80ff8981a583a60980be1a655068f8adebf7a45a06a6858c873fcdcd4a0",
-                "sha256:16757cf28621e43e252c560d25b15f18a2f11da94fea344bf26c599b9cf54b73",
-                "sha256:18142b497d70a34b01642b9feabb70156311b326fdddd875a9981f34a369b671",
-                "sha256:1c92113619f7b272838b8d6702a7f8ebe5edea0df48166c47929611d0b4dea69",
-                "sha256:1e25507d85da11ff5066269d0bd25d06e0a0f2e908415534f3e603d2a78e4ffa",
-                "sha256:30bf971c12e4365153afb31fc73f441d4da157153f3400b82db32d04de1e4066",
-                "sha256:3579eaeb5e07f3ded59298ce22b65f877a86ba8e9fe701f5576c99bb17c283da",
-                "sha256:36b2b43146f646642b425dd2027730f99bac962618ec2052932157e213a040e9",
-                "sha256:3905a5fffcc23e597ee4d9fb3fcd209bd658c352657548db7316e810ca80458e",
-                "sha256:3a4199f519e57d517ebd48cb76b36c82da0360781c6a0353e64c0cac30ecaad3",
-                "sha256:3f2f5cddeaa4424a0a118924b988746db6ffa8565e5829b1841a8a3bd73eb59a",
-                "sha256:40deb10198bbaa531509aad0cd2f9fadb26c8b94070831e2208e7df543562b74",
-                "sha256:440cfb3db4c5029775803794f8638fbdbf71ec702caf32735f53b008e1eaece3",
-                "sha256:4723a50e1523e1de4fccd1b9a6dcea750c2102461e9a02b2ac55ffeae09a4410",
-                "sha256:4bddbaa30d78c86329b26bd6aaaea06b1e47444da99eddac7bf1e2fab717bd72",
-                "sha256:4e58666988605e251d42c2818c7d3d8991555381be26399303053b58a5bbf30d",
-                "sha256:54dc1d6d66f8d37843ed281773c7174f03bf7ad826523f73435deb88ba60d2d4",
-                "sha256:57fcc997ffc0bef234b8875a54d4058afa92b0b0c4223fc1f62f24b3b5e86038",
-                "sha256:58b92a5828bd4d9aa0952492b7de803135038de47343b2aa3cc23f3b71a3dc4e",
-                "sha256:5a145e956b374e72ad1dff82779177d4a3c62bc8248f41b80cb5122e68f22d13",
-                "sha256:6ab153263a7c5ccaf6dfe7e53447b74f77789f28ecb278c3b5d49db7ece10d6d",
-                "sha256:7832f9e8eb00be32f15fdfb9a981d6955ea9adc8574c521d48710171b6c55e95",
-                "sha256:7fe4bb0695fe986a9e4deec3b6857003b4cfe5c5e4aac0b95f6a658c14635e31",
-                "sha256:7fe8f3583e0607ad4e43a954e35c1748b553bfe9fdac8635c02058023277d1b3",
-                "sha256:85ad7d11b309bd132d74397fcf2920933c9d1dc865487128f5c03d580f2c3d03",
-                "sha256:9874bc2ff574c40ab7a5cbb7464bf9b045d617e36754a7bc93f933d52bd9ffc6",
-                "sha256:a184288538e6ad699cbe6b24859206e38ce5fba28f3bcfa51c90d0502c1582b2",
-                "sha256:a222d764352c773aa5ebde02dd84dba3279c81c6db2e482d62a3fa54e5ece69b",
-                "sha256:a50aeff71d0f97b6450d33940c7181b08be1441c6c193e678211bff11aa725e7",
-                "sha256:a55dc7a7f0b6198b07ec0cd445fbb98b05234e8b00c5ac4874a63372ba98d4ab",
-                "sha256:a62eb442011776e4036af5c8b1a00b706c5bc02dc15eb5344b0c750428c94219",
-                "sha256:a7d41d1612c1a82b64697e894b75db6758d4f21c3ec069d841e60ebe54b5b571",
-                "sha256:a98f6f20465e7618c83252c02041517bd2f7ea29be5378f09667a8f654a5918d",
-                "sha256:afe8fb968743d40435c3827632fd36c5fbde633b0423da7692e426529b1759b1",
-                "sha256:b0b227dcff8cdc3efbce66d4e50891f04d0a387cce282fe1e66199146a6a8fca",
-                "sha256:b30042fe92dbd79f1ba7f6898fada10bdaad1847c44f2dff9a16147e00a93661",
-                "sha256:b606b1aaf802e6468c2608c65ff7ece53eae1a6874b3765f69b8ceb20c5fa78e",
-                "sha256:b6207dc8fb3c8cb5668e885cef9ec7f70189bec4e276f0ff70d5aa078d32c88e",
-                "sha256:c2aed8fcf8abc3020d6a9ccb31dbc9e7d7819c56a348cc88fd44be269b37427e",
-                "sha256:cb24cca1968b21355cc6f3da1a20cd1cebd8a023e3c5b09b432444617949085a",
-                "sha256:cff210198bb4cae3f3c100444c5eaa573a823f05c253e7188e1362a5555235b3",
-                "sha256:d35717333b39d1b6bb8433fa758a55f1081543de527171543a2b710551d40881",
-                "sha256:df12a1f99b99f569a7c2ae59aa2d31724e8d835fc7f33e14f4792e3071d11221",
-                "sha256:e09d40edfdb4e260cb1567d8ae770ccf3b8b7e9f0d9b5c2a9992696b30ce2742",
-                "sha256:e12c6c1ce84628c52d6367863773f7c8c8241be554e8b79686e91a43f1733773",
-                "sha256:e2b8cd48a9942ed3f85b95ca4105c45758438c7ed28fff1e4ce3e57c3b589d8e",
-                "sha256:e500aba968a48e9019e42c0c199b7ec0696a97fa69037bea163b55398e390529",
-                "sha256:ebe5e59545401fbb1b24da76f006ab19734ae71e703cdb4a8b347e84a0cece67",
-                "sha256:f0dd071b95bbca244f4cb7f70b77d2ff3aaaba7fa16dc41f58d14854a6204e6c",
-                "sha256:f8c8b141ef9699ae777c6278b52c706b653bf15d135d302754f6b2e90eb30367"
             ],
             "markers": "python_version >= '3.10'",
-            "version": "==2.2.0"
         },
         "opencv-python": {
             "hashes": [
@@ -840,6 +1126,12 @@
             "index": "pypi",
             "version": "==11.0.0"
         },
         "protobuf": {
             "hashes": [
                 "sha256:012ce28d862ff417fd629285aca5d9772807f15ceb1a0dbd15b88f58c776c98c",
@@ -857,6 +1149,29 @@
             "markers": "python_version >= '3.8'",
             "version": "==5.29.1"
         },
         "pyarrow": {
             "hashes": [
                 "sha256:01c034b576ce0eef554f7c3d8c341714954be9b3f5d5bc7117006b85fcf302fe",
@@ -929,6 +1244,14 @@
             "markers": "python_version >= '3.9'",
             "version": "==3.2.0"
         },
         "python-dateutil": {
             "hashes": [
                 "sha256:37dd54208da7e1cd875388217d5e00ebd4179249f90fb72437e91a35459a0ad3",
@@ -937,6 +1260,14 @@
             "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'",
             "version": "==2.9.0.post0"
         },
         "pytz": {
             "hashes": [
                 "sha256:2aa355083c50a0f93fa581709deac0c9ad65cca8a9e9beac660adcbd493c798a",
@@ -1011,6 +1342,106 @@
             "markers": "python_version >= '3.8'",
             "version": "==0.35.1"
         },
         "requests": {
             "hashes": [
                 "sha256:55365417734eb18255590a9ff9eb97e9e1da868d4ccd6402399eaf68af20a760",
@@ -1136,37 +1567,165 @@
             "markers": "python_version >= '3.9'",
             "version": "==0.22.3"
         },
         "scikit-learn": {
             "hashes": [
-                "sha256:03b6158efa3faaf1feea3faa884c840ebd61b6484167c711548fce208ea09445",
-                "sha256:178ddd0a5cb0044464fc1bfc4cca5b1833bfc7bb022d70b05db8530da4bb3dd3",
-                "sha256:1ff45e26928d3b4eb767a8f14a9a6efbf1cbff7c05d1fb0f95f211a89fd4f5de",
-                "sha256:299406827fb9a4f862626d0fe6c122f5f87f8910b86fe5daa4c32dcd742139b6",
-                "sha256:2d4cad1119c77930b235579ad0dc25e65c917e756fe80cab96aa3b9428bd3fb0",
-                "sha256:394397841449853c2290a32050382edaec3da89e35b3e03d6cc966aebc6a8ae6",
-                "sha256:3a686885a4b3818d9e62904d91b57fa757fc2bed3e465c8b177be652f4dd37c8",
-                "sha256:3b923d119d65b7bd555c73be5423bf06c0105678ce7e1f558cb4b40b0a5502b1",
-                "sha256:3bed4909ba187aca80580fe2ef370d9180dcf18e621a27c4cf2ef10d279a7efe",
-                "sha256:52788f48b5d8bca5c0736c175fa6bdaab2ef00a8f536cda698db61bd89c551c1",
-                "sha256:57cc1786cfd6bd118220a92ede80270132aa353647684efa385a74244a41e3b1",
-                "sha256:643964678f4b5fbdc95cbf8aec638acc7aa70f5f79ee2cdad1eec3df4ba6ead8",
-                "sha256:6c16d84a0d45e4894832b3c4d0bf73050939e21b99b01b6fd59cbb0cf39163b6",
-                "sha256:757c7d514ddb00ae249832fe87100d9c73c6ea91423802872d9e74970a0e40b9",
-                "sha256:8c412ccc2ad9bf3755915e3908e677b367ebc8d010acbb3f182814524f2e5540",
-                "sha256:b0768ad641981f5d3a198430a1d31c3e044ed2e8a6f22166b4d546a5116d7908",
-                "sha256:b4237ed7b3fdd0a4882792e68ef2545d5baa50aca3bb45aa7df468138ad8f94d",
-                "sha256:b7b0f9a0b1040830d38c39b91b3a44e1b643f4b36e36567b80b7c6bd2202a27f",
-                "sha256:c15b1ca23d7c5f33cc2cb0a0d6aaacf893792271cddff0edbd6a40e8319bc113",
-                "sha256:ca64b3089a6d9b9363cd3546f8978229dcbb737aceb2c12144ee3f70f95684b7",
-                "sha256:e9a702e2de732bbb20d3bad29ebd77fc05a6b427dc49964300340e4c9328b3f5",
-                "sha256:f60021ec1574e56632be2a36b946f8143bf4e5e6af4a06d85281adc22938e0dd",
-                "sha256:f7284ade780084d94505632241bf78c44ab3b6f1e8ccab3d2af58e0e950f9c12",
-                "sha256:f763897fe92d0e903aa4847b0aec0e68cadfff77e8a0687cabd946c89d17e675",
-                "sha256:f8b0ccd4a902836493e026c03256e8b206656f91fbcc4fde28c57a5b752561f1",
-                "sha256:f932a02c3f4956dfb981391ab24bda1dbd90fe3d628e4b42caef3e041c67707a"
             ],
             "index": "pypi",
-            "version": "==1.5.2"
         },
         "scipy": {
             "hashes": [
@@ -1207,6 +1766,79 @@
             "markers": "python_version >= '3.10'",
             "version": "==1.14.1"
         },
         "six": {
             "hashes": [
                 "sha256:4721f391ed90541fddacab5acf947aa0d3dc7d27b2e1e8eda2be8970586c3274",
@@ -1223,13 +1855,50 @@
             "markers": "python_version >= '3.7'",
             "version": "==5.0.1"
         },
         "streamlit": {
             "hashes": [
-                "sha256:0cc131fc9b18065feaff8f6f241c81164ad37d8d9e3a85499a0240aaaf6a6a61",
-                "sha256:7f6d1379a590f9625a6aee79ca73ceccff03cd2e05a3acbe5fe98915c27a7ffe"
             ],
             "index": "pypi",
-            "version": "==1.40.2"
         },
         "tenacity": {
             "hashes": [
@@ -1247,6 +1916,27 @@
             "markers": "python_version >= '3.8'",
             "version": "==3.5.0"
         },
         "toml": {
             "hashes": [
                 "sha256:806143ae5bfb6a3c6e736a764057db0e6a0e05e338b5630894a5f779cabb4f9b",
@@ -1255,6 +1945,29 @@
             "markers": "python_version >= '2.6' and python_version not in '3.0, 3.1, 3.2, 3.3'",
             "version": "==0.10.2"
         },
         "tornado": {
             "hashes": [
                 "sha256:072ce12ada169c5b00b7d92a99ba089447ccc993ea2143c9ede887e0937aa803",
@@ -1272,6 +1985,33 @@
             "markers": "python_version >= '3.8'",
             "version": "==6.4.2"
         },
         "typing-extensions": {
             "hashes": [
                 "sha256:04e5ca0351e0f3f85c6853954072df659d0d13fac324d0072316b67d7794700d",
@@ -1332,6 +2072,90 @@
             "markers": "platform_system != 'Darwin'",
             "version": "==6.0.0"
         },
         "xyzservices": {
             "hashes": [
                 "sha256:68fb8353c9dbba4f1ff6c0f2e5e4e596bb9e1db7f94f4f7dfbcb26e25aa66fde",

 {
     "_meta": {
         "hash": {
+            "sha256": "dd0bb1225c6e079e1823db0d99980f4b735bd5f63f2e014ac9761ac6597cd4ed"
         },
         "pipfile-spec": 6,
         "requires": {
         ]
     },
     "default": {
+        "accelerate": {
+            "hashes": [
+                "sha256:03e161fc69d495daf2b9b5c8d5b43d06e2145520c04727b5bda56d49f1a43ab5",
+                "sha256:be1cbb958cf837e7cdfbde46b812964b1b8ae94c9c7d94d921540beafcee8ddf"
+            ],
+            "version": "==1.2.1"
+        },
         "altair": {
             "hashes": [
                 "sha256:91a310b926508d560fe0148d02a194f38b824122641ef528113d029fcd129f8c",
         },
         "attrs": {
             "hashes": [
+                "sha256:8f5c07333d543103541ba7be0e2ce16eeee8130cb0b3f9238ab904ce1e85baff",
+                "sha256:ac96cd038792094f438ad1f6ff80837353805ac950cd2aa0e0625ef19850c308"
             ],
+            "markers": "python_version >= '3.8'",
+            "version": "==24.3.0"
+        },
+        "beautifulsoup4": {
+            "hashes": [
+                "sha256:74e3d1928edc070d21748185c46e3fb33490f22f52a3addee9aee0f4f7781051",
+                "sha256:b80878c9f40111313e55da8ba20bdba06d8fa3969fc68304167741bbf9e082ed"
+            ],
+            "markers": "python_full_version >= '3.6.0'",
+            "version": "==4.12.3"
+        },
+        "bioc": {
+            "hashes": [
+                "sha256:7d9248198bdae291b0ebb218de2dc653c96d32a7eac2fa0ef4ed0c74ce45aaaa",
+                "sha256:f1730d26821330f9f625058841612d6cfb616efb906fc682297fe04dd5d9398a"
+            ],
+            "markers": "python_version >= '3.6'",
+            "version": "==2.1"
         },
         "blinker": {
             "hashes": [
             "index": "pypi",
             "version": "==3.6.2"
         },
+        "boto3": {
+            "hashes": [
+                "sha256:2bbaf1551b1ed55770cb437d7040f1abe6742601103695057b30ce6328eef286",
+                "sha256:c422b68ae76959b9e23b77eb79e41c3483332f7e1de918d2b083c456d8cf234c"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==1.35.82"
+        },
+        "botocore": {
+            "hashes": [
+                "sha256:78dd7bf8f49616d00073698d7bbaf5a115208fe730b7b7afae4456adddb3552e",
+                "sha256:e43b97d8cbf19d35ce3a177f144bd97cc370f0a67d0984c7d7cf105ac198748f"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==1.35.82"
+        },
         "cachetools": {
             "hashes": [
                 "sha256:02134e8439cdc2ffb62023ce1debca2944c3f289d66bb17ead3ab3dede74b292",
         },
         "certifi": {
             "hashes": [
+                "sha256:1275f7a45be9464efc1173084eaa30f866fe2e47d389406136d332ed4967ec56",
+                "sha256:b650d30f370c2b724812bee08008be0c4163b163ddaec3f2546c1caf65f191db"
             ],
             "markers": "python_version >= '3.6'",
+            "version": "==2024.12.14"
         },
         "charset-normalizer": {
             "hashes": [
             "markers": "platform_system == 'Windows'",
             "version": "==0.4.6"
         },
+        "conllu": {
+            "hashes": [
+                "sha256:2b962b315c3c575e429105d045c888726df780b87a6dfe7609367e861990902d",
+                "sha256:a016cf77e203b2e3ace82fcf0cba2874530d1458e874521640eba36e19546acc"
+            ],
+            "markers": "python_version >= '3.6'",
+            "version": "==4.5.3"
+        },
         "contourpy": {
             "hashes": [
                 "sha256:041b640d4ec01922083645a94bb3b2e777e6b626788f4095cf21abbe266413c1",
             "markers": "python_version >= '3.8'",
             "version": "==0.12.1"
         },
+        "deprecated": {
+            "hashes": [
+                "sha256:353bc4a8ac4bfc96800ddab349d89c25dec1079f65fd53acdcc1e0b975b21320",
+                "sha256:683e561a90de76239796e6b6feac66b99030d2dd3fcf61ef996330f14bbb9b0d"
+            ],
+            "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'",
+            "version": "==1.2.15"
+        },
+        "docopt": {
+            "hashes": [
+                "sha256:49b3a825280bd66b3aa83585ef59c4a8c82f2c8a522dbe754a8bc8d08c85c491"
+            ],
+            "version": "==0.6.2"
+        },
+        "filelock": {
+            "hashes": [
+                "sha256:2082e5703d51fbf98ea75855d9d5527e33d8ff23099bec374a134febee6946b0",
+                "sha256:c249fbfcd5db47e5e2d6d62198e565475ee65e4831e2561c8e313fa7eb961435"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==3.16.1"
+        },
+        "flair": {
+            "hashes": [
+                "sha256:10735065ff462c05d0ea06ff01bc90e647f822a287858f9a6d0dabc7e402c754",
+                "sha256:dc14b58e93a52141f204d9995191aa3b7e0463a661a41faa8f8db30745a188a4"
+            ],
+            "index": "pypi",
+            "version": "==0.14.0"
+        },
         "fonttools": {
             "hashes": [
+                "sha256:07f8288aacf0a38d174445fc78377a97fb0b83cfe352a90c9d9c1400571963c7",
+                "sha256:11e5de1ee0d95af4ae23c1a138b184b7f06e0b6abacabf1d0db41c90b03d834b",
+                "sha256:1bc7ad24ff98846282eef1cbeac05d013c2154f977a79886bb943015d2b1b261",
+                "sha256:1dcc07934a2165ccdc3a5a608db56fb3c24b609658a5b340aee4ecf3ba679dc0",
+                "sha256:22f38464daa6cdb7b6aebd14ab06609328fe1e9705bb0fcc7d1e69de7109ee02",
+                "sha256:27e4ae3592e62eba83cd2c4ccd9462dcfa603ff78e09110680a5444c6925d841",
+                "sha256:3983313c2a04d6cc1fe9251f8fc647754cf49a61dac6cb1e7249ae67afaafc45",
+                "sha256:529cef2ce91dc44f8e407cc567fae6e49a1786f2fefefa73a294704c415322a4",
+                "sha256:5323a22eabddf4b24f66d26894f1229261021dacd9d29e89f7872dd8c63f0b8b",
+                "sha256:54153c49913f45065c8d9e6d0c101396725c5621c8aee744719300f79771d75a",
+                "sha256:546565028e244a701f73df6d8dd6be489d01617863ec0c6a42fa25bf45d43048",
+                "sha256:5480673f599ad410695ca2ddef2dfefe9df779a9a5cda89503881e503c9c7d90",
+                "sha256:5e8d657cd7326eeaba27de2740e847c6b39dde2f8d7cd7cc56f6aad404ddf0bd",
+                "sha256:62d65a3022c35e404d19ca14f291c89cc5890032ff04f6c17af0bd1927299674",
+                "sha256:6314bf82c54c53c71805318fcf6786d986461622dd926d92a465199ff54b1b72",
+                "sha256:7a8aa2c5e5b8b3bcb2e4538d929f6589a5c6bdb84fd16e2ed92649fb5454f11c",
+                "sha256:827e95fdbbd3e51f8b459af5ea10ecb4e30af50221ca103bea68218e9615de07",
+                "sha256:859c358ebf41db18fb72342d3080bce67c02b39e86b9fbcf1610cca14984841b",
+                "sha256:86721fbc389ef5cc1e2f477019e5069e8e4421e8d9576e9c26f840dbb04678de",
+                "sha256:89bdc5d88bdeec1b15af790810e267e8332d92561dce4f0748c2b95c9bdf3926",
+                "sha256:8c4491699bad88efe95772543cd49870cf756b019ad56294f6498982408ab03e",
+                "sha256:8c5ec45428edaa7022f1c949a632a6f298edc7b481312fc7dc258921e9399628",
+                "sha256:8e75f12c82127486fac2d8bfbf5bf058202f54bf4f158d367e41647b972342ca",
+                "sha256:a430178ad3e650e695167cb53242dae3477b35c95bef6525b074d87493c4bf29",
+                "sha256:a8c2794ded89399cc2169c4d0bf7941247b8d5932b2659e09834adfbb01589aa",
+                "sha256:aca318b77f23523309eec4475d1fbbb00a6b133eb766a8bdc401faba91261abe",
+                "sha256:ae3b6600565b2d80b7c05acb8e24d2b26ac407b27a3f2e078229721ba5698427",
+                "sha256:aedbeb1db64496d098e6be92b2e63b5fac4e53b1b92032dfc6988e1ea9134a4d",
+                "sha256:aee3b57643827e237ff6ec6d28d9ff9766bd8b21e08cd13bff479e13d4b14765",
+                "sha256:b54baf65c52952db65df39fcd4820668d0ef4766c0ccdf32879b77f7c804d5c5",
+                "sha256:b586ab5b15b6097f2fb71cafa3c98edfd0dba1ad8027229e7b1e204a58b0e09d",
+                "sha256:b8d5e8916c0970fbc0f6f1bece0063363bb5857a7f170121a4493e31c3db3314",
+                "sha256:bc5dbb4685e51235ef487e4bd501ddfc49be5aede5e40f4cefcccabc6e60fb4b",
+                "sha256:bdcc9f04b36c6c20978d3f060e5323a43f6222accc4e7fcbef3f428e216d96af",
+                "sha256:c3ca99e0d460eff46e033cd3992a969658c3169ffcd533e0a39c63a38beb6831",
+                "sha256:caf8230f3e10f8f5d7593eb6d252a37caf58c480b19a17e250a63dad63834cf3",
+                "sha256:cd70de1a52a8ee2d1877b6293af8a2484ac82514f10b1c67c1c5762d38073e56",
+                "sha256:cf4fe7c124aa3f4e4c1940880156e13f2f4d98170d35c749e6b4f119a872551e",
+                "sha256:d342e88764fb201286d185093781bf6628bbe380a913c24adf772d901baa8276",
+                "sha256:da9da6d65cd7aa6b0f806556f4985bcbf603bf0c5c590e61b43aa3e5a0f822d0",
+                "sha256:dc5294a3d5c84226e3dbba1b6f61d7ad813a8c0238fceea4e09aa04848c3d851",
+                "sha256:dd68c87a2bfe37c5b33bcda0fba39b65a353876d3b9006fde3adae31f97b3ef5",
+                "sha256:e6e8766eeeb2de759e862004aa11a9ea3d6f6d5ec710551a88b476192b64fd54",
+                "sha256:e894b5bd60d9f473bed7a8f506515549cc194de08064d829464088d23097331b",
+                "sha256:eb6ca911c4c17eb51853143624d8dc87cdcdf12a711fc38bf5bd21521e79715f",
+                "sha256:ed63959d00b61959b035c7d47f9313c2c1ece090ff63afea702fe86de00dbed4",
+                "sha256:f412604ccbeee81b091b420272841e5ec5ef68967a9790e80bffd0e30b8e2977",
+                "sha256:f7d66c15ba875432a2d2fb419523f5d3d347f91f48f57b8b08a2dfc3c39b8a3f",
+                "sha256:f9e736f60f4911061235603a6119e72053073a12c6d7904011df2d8fad2c0e35",
+                "sha256:fb594b5a99943042c702c550d5494bdd7577f6ef19b0bc73877c948a63184a32"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==4.55.3"
+        },
+        "fsspec": {
+            "hashes": [
+                "sha256:03b9a6785766a4de40368b88906366755e2819e758b83705c88cd7cb5fe81871",
+                "sha256:eda2d8a4116d4f2429db8550f2457da57279247dd930bb12f821b58391359493"
             ],
             "markers": "python_version >= '3.8'",
+            "version": "==2024.10.0"
+        },
+        "ftfy": {
+            "hashes": [
+                "sha256:7c70eb532015cd2f9adb53f101fb6c7945988d023a085d127d1573dc49dd0083",
+                "sha256:9b3c3d90f84fb267fe64d375a07b7f8912d817cf86009ae134aa03e1819506ec"
+            ],
+            "markers": "python_version >= '3.9'",
+            "version": "==6.3.1"
+        },
+        "gdown": {
+            "hashes": [
+                "sha256:2145165062d85520a3cd98b356c9ed522c5e7984d408535409fd46f94defc787",
+                "sha256:33083832d82b1101bdd0e9df3edd0fbc0e1c5f14c9d8c38d2a35bf1683b526d6"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==5.2.0"
         },
         "gitdb": {
             "hashes": [
             "markers": "python_version >= '3.7'",
             "version": "==3.1.43"
         },
+        "huggingface-hub": {
+            "hashes": [
+                "sha256:8f2e834517f1f1ddf1ecc716f91b120d7333011b7485f665a9a412eacb1a2a81",
+                "sha256:902cce1a1be5739f5589e560198a65a8edcfd3b830b1666f36e4b961f0454fac"
+            ],
+            "markers": "python_full_version >= '3.8.0'",
+            "version": "==0.27.0"
+        },
         "idna": {
             "hashes": [
                 "sha256:12f65c9b470abda6dc35cf8e63cc574b1c52b11df2c86030af0ac09b01b13ea9",
             "markers": "python_version >= '3.6'",
             "version": "==3.10"
         },
+        "intervaltree": {
+            "hashes": [
+                "sha256:902b1b88936918f9b2a19e0e5eb7ccb430ae45cde4f39ea4b36932920d33952d"
+            ],
+            "version": "==3.1.0"
+        },
         "jinja2": {
             "hashes": [
                 "sha256:4a3aee7acbbe7303aede8e9648d13b8bf88a429282aa6122a993f0ac800cb369",
             "markers": "python_version >= '3.7'",
             "version": "==3.1.4"
         },
+        "jmespath": {
+            "hashes": [
+                "sha256:02e2e4cc71b5bcab88332eebf907519190dd9e6e82107fa7f83b1003a6252980",
+                "sha256:90261b206d6defd58fdd5e85f478bf633a2901798906be2ad389150c5c60edbe"
+            ],
+            "markers": "python_version >= '3.7'",
+            "version": "==1.0.1"
+        },
         "joblib": {
             "hashes": [
                 "sha256:06d478d5674cbc267e7496a410ee875abd68e4340feff4490bcb7afb88060ae6",
             "markers": "python_version >= '3.8'",
             "version": "==1.4.2"
         },
+        "jsonlines": {
+            "hashes": [
+                "sha256:0c6d2c09117550c089995247f605ae4cf77dd1533041d366351f6f298822ea74",
+                "sha256:185b334ff2ca5a91362993f42e83588a360cf95ce4b71a73548502bda52a7c55"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==4.0.0"
+        },
         "jsonschema": {
             "hashes": [
                 "sha256:d71497fef26351a33265337fa77ffeb82423f3ea21283cd9467bb03999266bc4",
             "markers": "python_version >= '3.8'",
             "version": "==1.4.7"
         },
+        "langdetect": {
+            "hashes": [
+                "sha256:7cbc0746252f19e76f77c0b1690aadf01963be835ef0cd4b56dddf2a8f1dfc2a",
+                "sha256:cbc1fef89f8d062739774bd51eda3da3274006b3661d199c2655f6b3f6d605a0"
+            ],
+            "version": "==1.0.9"
+        },
+        "lxml": {
+            "hashes": [
+                "sha256:01220dca0d066d1349bd6a1726856a78f7929f3878f7e2ee83c296c69495309e",
+                "sha256:02ced472497b8362c8e902ade23e3300479f4f43e45f4105c85ef43b8db85229",
+                "sha256:052d99051e77a4f3e8482c65014cf6372e61b0a6f4fe9edb98503bb5364cfee3",
+                "sha256:07da23d7ee08577760f0a71d67a861019103e4812c87e2fab26b039054594cc5",
+                "sha256:094cb601ba9f55296774c2d57ad68730daa0b13dc260e1f941b4d13678239e70",
+                "sha256:0a7056921edbdd7560746f4221dca89bb7a3fe457d3d74267995253f46343f15",
+                "sha256:0c120f43553ec759f8de1fee2f4794452b0946773299d44c36bfe18e83caf002",
+                "sha256:0d7b36afa46c97875303a94e8f3ad932bf78bace9e18e603f2085b652422edcd",
+                "sha256:0fdf3a3059611f7585a78ee10399a15566356116a4288380921a4b598d807a22",
+                "sha256:109fa6fede314cc50eed29e6e56c540075e63d922455346f11e4d7a036d2b8cf",
+                "sha256:146173654d79eb1fc97498b4280c1d3e1e5d58c398fa530905c9ea50ea849b22",
+                "sha256:1473427aff3d66a3fa2199004c3e601e6c4500ab86696edffdbc84954c72d832",
+                "sha256:1483fd3358963cc5c1c9b122c80606a3a79ee0875bcac0204149fa09d6ff2727",
+                "sha256:168f2dfcfdedf611eb285efac1516c8454c8c99caf271dccda8943576b67552e",
+                "sha256:17e8d968d04a37c50ad9c456a286b525d78c4a1c15dd53aa46c1d8e06bf6fa30",
+                "sha256:18feb4b93302091b1541221196a2155aa296c363fd233814fa11e181adebc52f",
+                "sha256:1afe0a8c353746e610bd9031a630a95bcfb1a720684c3f2b36c4710a0a96528f",
+                "sha256:1d04f064bebdfef9240478f7a779e8c5dc32b8b7b0b2fc6a62e39b928d428e51",
+                "sha256:1fdc9fae8dd4c763e8a31e7630afef517eab9f5d5d31a278df087f307bf601f4",
+                "sha256:1ffc23010330c2ab67fac02781df60998ca8fe759e8efde6f8b756a20599c5de",
+                "sha256:20094fc3f21ea0a8669dc4c61ed7fa8263bd37d97d93b90f28fc613371e7a875",
+                "sha256:213261f168c5e1d9b7535a67e68b1f59f92398dd17a56d934550837143f79c42",
+                "sha256:218c1b2e17a710e363855594230f44060e2025b05c80d1f0661258142b2add2e",
+                "sha256:23e0553b8055600b3bf4a00b255ec5c92e1e4aebf8c2c09334f8368e8bd174d6",
+                "sha256:25f1b69d41656b05885aa185f5fdf822cb01a586d1b32739633679699f220391",
+                "sha256:2b3778cb38212f52fac9fe913017deea2fdf4eb1a4f8e4cfc6b009a13a6d3fcc",
+                "sha256:2bc9fd5ca4729af796f9f59cd8ff160fe06a474da40aca03fcc79655ddee1a8b",
+                "sha256:2c226a06ecb8cdef28845ae976da407917542c5e6e75dcac7cc33eb04aaeb237",
+                "sha256:2c3406b63232fc7e9b8783ab0b765d7c59e7c59ff96759d8ef9632fca27c7ee4",
+                "sha256:2c86bf781b12ba417f64f3422cfc302523ac9cd1d8ae8c0f92a1c66e56ef2e86",
+                "sha256:2d9b8d9177afaef80c53c0a9e30fa252ff3036fb1c6494d427c066a4ce6a282f",
+                "sha256:2dec2d1130a9cda5b904696cec33b2cfb451304ba9081eeda7f90f724097300a",
+                "sha256:2dfab5fa6a28a0b60a20638dc48e6343c02ea9933e3279ccb132f555a62323d8",
+                "sha256:2ecdd78ab768f844c7a1d4a03595038c166b609f6395e25af9b0f3f26ae1230f",
+                "sha256:315f9542011b2c4e1d280e4a20ddcca1761993dda3afc7a73b01235f8641e903",
+                "sha256:36aef61a1678cb778097b4a6eeae96a69875d51d1e8f4d4b491ab3cfb54b5a03",
+                "sha256:384aacddf2e5813a36495233b64cb96b1949da72bef933918ba5c84e06af8f0e",
+                "sha256:3879cc6ce938ff4eb4900d901ed63555c778731a96365e53fadb36437a131a99",
+                "sha256:3c174dc350d3ec52deb77f2faf05c439331d6ed5e702fc247ccb4e6b62d884b7",
+                "sha256:3eb44520c4724c2e1a57c0af33a379eee41792595023f367ba3952a2d96c2aab",
+                "sha256:406246b96d552e0503e17a1006fd27edac678b3fcc9f1be71a2f94b4ff61528d",
+                "sha256:41ce1f1e2c7755abfc7e759dc34d7d05fd221723ff822947132dc934d122fe22",
+                "sha256:423b121f7e6fa514ba0c7918e56955a1d4470ed35faa03e3d9f0e3baa4c7e492",
+                "sha256:44264ecae91b30e5633013fb66f6ddd05c006d3e0e884f75ce0b4755b3e3847b",
+                "sha256:482c2f67761868f0108b1743098640fbb2a28a8e15bf3f47ada9fa59d9fe08c3",
+                "sha256:4b0c7a688944891086ba192e21c5229dea54382f4836a209ff8d0a660fac06be",
+                "sha256:4c1fefd7e3d00921c44dc9ca80a775af49698bbfd92ea84498e56acffd4c5469",
+                "sha256:4e109ca30d1edec1ac60cdbe341905dc3b8f55b16855e03a54aaf59e51ec8c6f",
+                "sha256:501d0d7e26b4d261fca8132854d845e4988097611ba2531408ec91cf3fd9d20a",
+                "sha256:516f491c834eb320d6c843156440fe7fc0d50b33e44387fcec5b02f0bc118a4c",
+                "sha256:51806cfe0279e06ed8500ce19479d757db42a30fd509940b1701be9c86a5ff9a",
+                "sha256:562e7494778a69086f0312ec9689f6b6ac1c6b65670ed7d0267e49f57ffa08c4",
+                "sha256:56b9861a71575f5795bde89256e7467ece3d339c9b43141dbdd54544566b3b94",
+                "sha256:5b8f5db71b28b8c404956ddf79575ea77aa8b1538e8b2ef9ec877945b3f46442",
+                "sha256:5c2fb570d7823c2bbaf8b419ba6e5662137f8166e364a8b2b91051a1fb40ab8b",
+                "sha256:5c54afdcbb0182d06836cc3d1be921e540be3ebdf8b8a51ee3ef987537455f84",
+                "sha256:5d6a6972b93c426ace71e0be9a6f4b2cfae9b1baed2eed2006076a746692288c",
+                "sha256:609251a0ca4770e5a8768ff902aa02bf636339c5a93f9349b48eb1f606f7f3e9",
+                "sha256:62d172f358f33a26d6b41b28c170c63886742f5b6772a42b59b4f0fa10526cb1",
+                "sha256:62f7fdb0d1ed2065451f086519865b4c90aa19aed51081979ecd05a21eb4d1be",
+                "sha256:658f2aa69d31e09699705949b5fc4719cbecbd4a97f9656a232e7d6c7be1a367",
+                "sha256:65ab5685d56914b9a2a34d67dd5488b83213d680b0c5d10b47f81da5a16b0b0e",
+                "sha256:68934b242c51eb02907c5b81d138cb977b2129a0a75a8f8b60b01cb8586c7b21",
+                "sha256:68b87753c784d6acb8a25b05cb526c3406913c9d988d51f80adecc2b0775d6aa",
+                "sha256:69959bd3167b993e6e710b99051265654133a98f20cec1d9b493b931942e9c16",
+                "sha256:6a7095eeec6f89111d03dabfe5883a1fd54da319c94e0fb104ee8f23616b572d",
+                "sha256:6b038cc86b285e4f9fea2ba5ee76e89f21ed1ea898e287dc277a25884f3a7dfe",
+                "sha256:6ba0d3dcac281aad8a0e5b14c7ed6f9fa89c8612b47939fc94f80b16e2e9bc83",
+                "sha256:6e91cf736959057f7aac7adfc83481e03615a8e8dd5758aa1d95ea69e8931dba",
+                "sha256:6ee8c39582d2652dcd516d1b879451500f8db3fe3607ce45d7c5957ab2596040",
+                "sha256:6f651ebd0b21ec65dfca93aa629610a0dbc13dbc13554f19b0113da2e61a4763",
+                "sha256:71a8dd38fbd2f2319136d4ae855a7078c69c9a38ae06e0c17c73fd70fc6caad8",
+                "sha256:74068c601baff6ff021c70f0935b0c7bc528baa8ea210c202e03757c68c5a4ff",
+                "sha256:7437237c6a66b7ca341e868cda48be24b8701862757426852c9b3186de1da8a2",
+                "sha256:747a3d3e98e24597981ca0be0fd922aebd471fa99d0043a3842d00cdcad7ad6a",
+                "sha256:74bcb423462233bc5d6066e4e98b0264e7c1bed7541fff2f4e34fe6b21563c8b",
+                "sha256:78d9b952e07aed35fe2e1a7ad26e929595412db48535921c5013edc8aa4a35ce",
+                "sha256:7b1cd427cb0d5f7393c31b7496419da594fe600e6fdc4b105a54f82405e6626c",
+                "sha256:7d3d1ca42870cdb6d0d29939630dbe48fa511c203724820fc0fd507b2fb46577",
+                "sha256:7e2f58095acc211eb9d8b5771bf04df9ff37d6b87618d1cbf85f92399c98dae8",
+                "sha256:7f41026c1d64043a36fda21d64c5026762d53a77043e73e94b71f0521939cc71",
+                "sha256:81b4e48da4c69313192d8c8d4311e5d818b8be1afe68ee20f6385d0e96fc9512",
+                "sha256:86a6b24b19eaebc448dc56b87c4865527855145d851f9fc3891673ff97950540",
+                "sha256:874a216bf6afaf97c263b56371434e47e2c652d215788396f60477540298218f",
+                "sha256:89e043f1d9d341c52bf2af6d02e6adde62e0a46e6755d5eb60dc6e4f0b8aeca2",
+                "sha256:8c72e9563347c7395910de6a3100a4840a75a6f60e05af5e58566868d5eb2d6a",
+                "sha256:8dc2c0395bea8254d8daebc76dcf8eb3a95ec2a46fa6fae5eaccee366bfe02ce",
+                "sha256:8f0de2d390af441fe8b2c12626d103540b5d850d585b18fcada58d972b74a74e",
+                "sha256:92e67a0be1639c251d21e35fe74df6bcc40cba445c2cda7c4a967656733249e2",
+                "sha256:94d6c3782907b5e40e21cadf94b13b0842ac421192f26b84c45f13f3c9d5dc27",
+                "sha256:97acf1e1fd66ab53dacd2c35b319d7e548380c2e9e8c54525c6e76d21b1ae3b1",
+                "sha256:9ada35dd21dc6c039259596b358caab6b13f4db4d4a7f8665764d616daf9cc1d",
+                "sha256:9c52100e2c2dbb0649b90467935c4b0de5528833c76a35ea1a2691ec9f1ee7a1",
+                "sha256:9e41506fec7a7f9405b14aa2d5c8abbb4dbbd09d88f9496958b6d00cb4d45330",
+                "sha256:9e4b47ac0f5e749cfc618efdf4726269441014ae1d5583e047b452a32e221920",
+                "sha256:9fb81d2824dff4f2e297a276297e9031f46d2682cafc484f49de182aa5e5df99",
+                "sha256:a0eabd0a81625049c5df745209dc7fcef6e2aea7793e5f003ba363610aa0a3ff",
+                "sha256:a3d819eb6f9b8677f57f9664265d0a10dd6551d227afb4af2b9cd7bdc2ccbf18",
+                "sha256:a87de7dd873bf9a792bf1e58b1c3887b9264036629a5bf2d2e6579fe8e73edff",
+                "sha256:aa617107a410245b8660028a7483b68e7914304a6d4882b5ff3d2d3eb5948d8c",
+                "sha256:aac0bbd3e8dd2d9c45ceb82249e8bdd3ac99131a32b4d35c8af3cc9db1657179",
+                "sha256:ab6dd83b970dc97c2d10bc71aa925b84788c7c05de30241b9e96f9b6d9ea3080",
+                "sha256:ace2c2326a319a0bb8a8b0e5b570c764962e95818de9f259ce814ee666603f19",
+                "sha256:ae5fe5c4b525aa82b8076c1a59d642c17b6e8739ecf852522c6321852178119d",
+                "sha256:b11a5d918a6216e521c715b02749240fb07ae5a1fefd4b7bf12f833bc8b4fe70",
+                "sha256:b1c8c20847b9f34e98080da785bb2336ea982e7f913eed5809e5a3c872900f32",
+                "sha256:b369d3db3c22ed14c75ccd5af429086f166a19627e84a8fdade3f8f31426e52a",
+                "sha256:b710bc2b8292966b23a6a0121f7a6c51d45d2347edcc75f016ac123b8054d3f2",
+                "sha256:bd96517ef76c8654446fc3db9242d019a1bb5fe8b751ba414765d59f99210b79",
+                "sha256:c00f323cc00576df6165cc9d21a4c21285fa6b9989c5c39830c3903dc4303ef3",
+                "sha256:c162b216070f280fa7da844531169be0baf9ccb17263cf5a8bf876fcd3117fa5",
+                "sha256:c1a69e58a6bb2de65902051d57fde951febad631a20a64572677a1052690482f",
+                "sha256:c1f794c02903c2824fccce5b20c339a1a14b114e83b306ff11b597c5f71a1c8d",
+                "sha256:c24037349665434f375645fa9d1f5304800cec574d0310f618490c871fd902b3",
+                "sha256:c300306673aa0f3ed5ed9372b21867690a17dba38c68c44b287437c362ce486b",
+                "sha256:c56a1d43b2f9ee4786e4658c7903f05da35b923fb53c11025712562d5cc02753",
+                "sha256:c6379f35350b655fd817cd0d6cbeef7f265f3ae5fedb1caae2eb442bbeae9ab9",
+                "sha256:c802e1c2ed9f0c06a65bc4ed0189d000ada8049312cfeab6ca635e39c9608957",
+                "sha256:cb83f8a875b3d9b458cada4f880fa498646874ba4011dc974e071a0a84a1b033",
+                "sha256:cf120cce539453ae086eacc0130a324e7026113510efa83ab42ef3fcfccac7fb",
+                "sha256:dd36439be765e2dde7660212b5275641edbc813e7b24668831a5c8ac91180656",
+                "sha256:dd5350b55f9fecddc51385463a4f67a5da829bc741e38cf689f38ec9023f54ab",
+                "sha256:df5c7333167b9674aa8ae1d4008fa4bc17a313cc490b2cca27838bbdcc6bb15b",
+                "sha256:e63601ad5cd8f860aa99d109889b5ac34de571c7ee902d6812d5d9ddcc77fa7d",
+                "sha256:e92ce66cd919d18d14b3856906a61d3f6b6a8500e0794142338da644260595cd",
+                "sha256:e99f5507401436fdcc85036a2e7dc2e28d962550afe1cbfc07c40e454256a859",
+                "sha256:ea2e2f6f801696ad7de8aec061044d6c8c0dd4037608c7cab38a9a4d316bfb11",
+                "sha256:eafa2c8658f4e560b098fe9fc54539f86528651f61849b22111a9b107d18910c",
+                "sha256:ecd4ad8453ac17bc7ba3868371bffb46f628161ad0eefbd0a855d2c8c32dd81a",
+                "sha256:ee70d08fd60c9565ba8190f41a46a54096afa0eeb8f76bd66f2c25d3b1b83005",
+                "sha256:eec1bb8cdbba2925bedc887bc0609a80e599c75b12d87ae42ac23fd199445654",
+                "sha256:ef0c1fe22171dd7c7c27147f2e9c3e86f8bdf473fed75f16b0c2e84a5030ce80",
+                "sha256:f2901429da1e645ce548bf9171784c0f74f0718c3f6150ce166be39e4dd66c3e",
+                "sha256:f422a209d2455c56849442ae42f25dbaaba1c6c3f501d58761c619c7836642ec",
+                "sha256:f65e5120863c2b266dbcc927b306c5b78e502c71edf3295dfcb9501ec96e5fc7",
+                "sha256:f7d4a670107d75dfe5ad080bed6c341d18c4442f9378c9f58e5851e86eb79965",
+                "sha256:f914c03e6a31deb632e2daa881fe198461f4d06e57ac3d0e05bbcab8eae01945",
+                "sha256:fb66442c2546446944437df74379e9cf9e9db353e61301d1a0e26482f43f0dd8"
+            ],
+            "markers": "python_version >= '3.6'",
+            "version": "==5.3.0"
+        },
         "markdown-it-py": {
             "hashes": [
                 "sha256:355216845c60bd96232cd8d8c40e8f9765cc86f46880e43a8fd22dc1a1a8cab1",
         },
         "matplotlib": {
             "hashes": [
+                "sha256:01d2b19f13aeec2e759414d3bfe19ddfb16b13a1250add08d46d5ff6f9be83c6",
+                "sha256:12eaf48463b472c3c0f8dbacdbf906e573013df81a0ab82f0616ea4b11281908",
+                "sha256:2c5829a5a1dd5a71f0e31e6e8bb449bc0ee9dbfb05ad28fc0c6b55101b3a4be6",
+                "sha256:2fbbabc82fde51391c4da5006f965e36d86d95f6ee83fb594b279564a4c5d0d2",
+                "sha256:3547d153d70233a8496859097ef0312212e2689cdf8d7ed764441c77604095ae",
+                "sha256:359f87baedb1f836ce307f0e850d12bb5f1936f70d035561f90d41d305fdacea",
+                "sha256:3b427392354d10975c1d0f4ee18aa5844640b512d5311ef32efd4dd7db106ede",
+                "sha256:4659665bc7c9b58f8c00317c3c2a299f7f258eeae5a5d56b4c64226fca2f7c59",
+                "sha256:4673ff67a36152c48ddeaf1135e74ce0d4bce1bbf836ae40ed39c29edf7e2765",
+                "sha256:503feb23bd8c8acc75541548a1d709c059b7184cde26314896e10a9f14df5f12",
+                "sha256:5439f4c5a3e2e8eab18e2f8c3ef929772fd5641876db71f08127eed95ab64683",
+                "sha256:5cdbaf909887373c3e094b0318d7ff230b2ad9dcb64da7ade654182872ab2593",
+                "sha256:5e6c6461e1fc63df30bf6f80f0b93f5b6784299f721bc28530477acd51bfc3d1",
+                "sha256:5fd41b0ec7ee45cd960a8e71aea7c946a28a0b8a4dcee47d2856b2af051f334c",
+                "sha256:607b16c8a73943df110f99ee2e940b8a1cbf9714b65307c040d422558397dac5",
+                "sha256:7e8632baebb058555ac0cde75db885c61f1212e47723d63921879806b40bec6a",
+                "sha256:81713dd0d103b379de4516b861d964b1d789a144103277769238c732229d7f03",
+                "sha256:845d96568ec873be63f25fa80e9e7fae4be854a66a7e2f0c8ccc99e94a8bd4ef",
+                "sha256:95b710fea129c76d30be72c3b38f330269363fbc6e570a5dd43580487380b5ff",
+                "sha256:96f2886f5c1e466f21cc41b70c5a0cd47bfa0015eb2d5793c88ebce658600e25",
+                "sha256:994c07b9d9fe8d25951e3202a68c17900679274dadfc1248738dcfa1bd40d7f3",
+                "sha256:9ade1003376731a971e398cc4ef38bb83ee8caf0aee46ac6daa4b0506db1fd06",
+                "sha256:9b0558bae37f154fffda54d779a592bc97ca8b4701f1c710055b609a3bac44c8",
+                "sha256:a2a43cbefe22d653ab34bb55d42384ed30f611bcbdea1f8d7f431011a2e1c62e",
+                "sha256:a994f29e968ca002b50982b27168addfd65f0105610b6be7fa515ca4b5307c95",
+                "sha256:ad2e15300530c1a94c63cfa546e3b7864bd18ea2901317bae8bbf06a5ade6dcf",
+                "sha256:ae80dc3a4add4665cf2faa90138384a7ffe2a4e37c58d83e115b54287c4f06ef",
+                "sha256:b886d02a581b96704c9d1ffe55709e49b4d2d52709ccebc4be42db856e511278",
+                "sha256:c40ba2eb08b3f5de88152c2333c58cee7edcead0a2a0d60fcafa116b17117adc",
+                "sha256:c55b20591ced744aa04e8c3e4b7543ea4d650b6c3c4b208c08a05b4010e8b442",
+                "sha256:c58a9622d5dbeb668f407f35f4e6bfac34bb9ecdcc81680c04d0258169747997",
+                "sha256:d44cb942af1693cced2604c33a9abcef6205601c445f6d0dc531d813af8a2f5a",
+                "sha256:d907fddb39f923d011875452ff1eca29a9e7f21722b873e90db32e5d8ddff12e",
+                "sha256:fd44fc75522f58612ec4a33958a7e5552562b7705b42ef1b4f8c0818e304a363"
             ],
             "index": "pypi",
+            "version": "==3.10.0"
         },
         "mdurl": {
             "hashes": [
             "markers": "python_version >= '3.7'",
             "version": "==0.1.2"
         },
+        "more-itertools": {
+            "hashes": [
+                "sha256:037b0d3203ce90cca8ab1defbbdac29d5f993fc20131f3664dc8d6acfa872aef",
+                "sha256:5482bfef7849c25dc3c6dd53a6173ae4795da2a41a80faea6700d9f5846c5da6"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==10.5.0"
+        },
+        "mpld3": {
+            "hashes": [
+                "sha256:80877acce87ea447380fad7374668737505c8c0684aab05238e7c5dc1fab38c1",
+                "sha256:a478eb404fa5212505c59133cf272cd9a94105872e605597720e7f84de38fbc7"
+            ],
+            "version": "==0.5.10"
+        },
+        "mpmath": {
+            "hashes": [
+                "sha256:7a28eb2a9774d00c7bc92411c19a89209d5da7c4c9a9e227be8330a23a25b91f",
+                "sha256:a0b2b9fe80bbcd81a6647ff13108738cfb482d481d826cc0e02f5b35e5c88d2c"
+            ],
+            "version": "==1.3.0"
+        },
         "narwhals": {
             "hashes": [
+                "sha256:b1da4e2e4ab185824781760319ac1ec8ee2944a929795064c3a64ffff16b00c4",
+                "sha256:c6bb6b6fba59caeab28a7d6ec1e79ab0040c75baef2e4152199ad1a9c266ef96"
             ],
             "markers": "python_version >= '3.8'",
+            "version": "==1.18.4"
         },
+        "networkx": {
             "hashes": [
+                "sha256:307c3669428c5362aab27c8a1260aa8f47c4e91d3891f48be0141738d8d053e1",
+                "sha256:df5d4365b724cf81b8c6a7312509d0c22386097011ad1abe274afd5e9d3bbc5f"
             ],
             "markers": "python_version >= '3.10'",
+            "version": "==3.4.2"
+        },
+        "numpy": {
+            "hashes": [
+                "sha256:03a8c78d01d9781b28a6989f6fa1bb2c4f2d51201cf99d3dd875df6fbd96b23b",
+                "sha256:08beddf13648eb95f8d867350f6a018a4be2e5ad54c8d8caed89ebca558b2818",
+                "sha256:1af303d6b2210eb850fcf03064d364652b7120803a0b872f5211f5234b399f20",
+                "sha256:1dda2e7b4ec9dd512f84935c5f126c8bd8b9f2fc001e9f54af255e8c5f16b0e0",
+                "sha256:2a02aba9ed12e4ac4eb3ea9421c420301a0c6460d9830d74a9df87efa4912010",
+                "sha256:2e4ee3380d6de9c9ec04745830fd9e2eccb3e6cf790d39d7b98ffd19b0dd754a",
+                "sha256:3373d5d70a5fe74a2c1bb6d2cfd9609ecf686d47a2d7b1d37a8f3b6bf6003aea",
+                "sha256:47711010ad8555514b434df65f7d7b076bb8261df1ca9bb78f53d3b2db02e95c",
+                "sha256:4c66707fabe114439db9068ee468c26bbdf909cac0fb58686a42a24de1760c71",
+                "sha256:50193e430acfc1346175fcbdaa28ffec49947a06918b7b92130744e81e640110",
+                "sha256:52b8b60467cd7dd1e9ed082188b4e6bb35aa5cdd01777621a1658910745b90be",
+                "sha256:60dedbb91afcbfdc9bc0b1f3f402804070deed7392c23eb7a7f07fa857868e8a",
+                "sha256:62b8e4b1e28009ef2846b4c7852046736bab361f7aeadeb6a5b89ebec3c7055a",
+                "sha256:666dbfb6ec68962c033a450943ded891bed2d54e6755e35e5835d63f4f6931d5",
+                "sha256:675d61ffbfa78604709862923189bad94014bef562cc35cf61d3a07bba02a7ed",
+                "sha256:679b0076f67ecc0138fd2ede3a8fd196dddc2ad3254069bcb9faf9a79b1cebcd",
+                "sha256:7349ab0fa0c429c82442a27a9673fc802ffdb7c7775fad780226cb234965e53c",
+                "sha256:7ab55401287bfec946ced39700c053796e7cc0e3acbef09993a9ad2adba6ca6e",
+                "sha256:7e50d0a0cc3189f9cb0aeb3a6a6af18c16f59f004b866cd2be1c14b36134a4a0",
+                "sha256:95a7476c59002f2f6c590b9b7b998306fba6a5aa646b1e22ddfeaf8f78c3a29c",
+                "sha256:96ff0b2ad353d8f990b63294c8986f1ec3cb19d749234014f4e7eb0112ceba5a",
+                "sha256:9fad7dcb1aac3c7f0584a5a8133e3a43eeb2fe127f47e3632d43d677c66c102b",
+                "sha256:9ff0f4f29c51e2803569d7a51c2304de5554655a60c5d776e35b4a41413830d0",
+                "sha256:a354325ee03388678242a4d7ebcd08b5c727033fcff3b2f536aea978e15ee9e6",
+                "sha256:a4abb4f9001ad2858e7ac189089c42178fcce737e4169dc61321660f1a96c7d2",
+                "sha256:ab47dbe5cc8210f55aa58e4805fe224dac469cde56b9f731a4c098b91917159a",
+                "sha256:afedb719a9dcfc7eaf2287b839d8198e06dcd4cb5d276a3df279231138e83d30",
+                "sha256:b3ce300f3644fb06443ee2222c2201dd3a89ea6040541412b8fa189341847218",
+                "sha256:b97fe8060236edf3662adfc2c633f56a08ae30560c56310562cb4f95500022d5",
+                "sha256:bfe25acf8b437eb2a8b2d49d443800a5f18508cd811fea3181723922a8a82b07",
+                "sha256:cd25bcecc4974d09257ffcd1f098ee778f7834c3ad767fe5db785be9a4aa9cb2",
+                "sha256:d209d8969599b27ad20994c8e41936ee0964e6da07478d6c35016bc386b66ad4",
+                "sha256:d5241e0a80d808d70546c697135da2c613f30e28251ff8307eb72ba696945764",
+                "sha256:edd8b5fe47dab091176d21bb6de568acdd906d1887a4584a15a9a96a1dca06ef",
+                "sha256:f870204a840a60da0b12273ef34f7051e98c3b5961b61b0c2c1be6dfd64fbcd3",
+                "sha256:ffa75af20b44f8dba823498024771d5ac50620e6915abac414251bd971b4529f"
+            ],
+            "markers": "python_version >= '3.9'",
+            "version": "==1.26.4"
         },
         "opencv-python": {
             "hashes": [
             "index": "pypi",
             "version": "==11.0.0"
         },
+        "pptree": {
+            "hashes": [
+                "sha256:4dd0ba2f58000cbd29d68a5b64bac29bcb5a663642f79404877c0059668a69f6"
+            ],
+            "version": "==3.1"
+        },
         "protobuf": {
             "hashes": [
                 "sha256:012ce28d862ff417fd629285aca5d9772807f15ceb1a0dbd15b88f58c776c98c",
             "markers": "python_version >= '3.8'",
             "version": "==5.29.1"
         },
+        "psutil": {
+            "hashes": [
+                "sha256:000d1d1ebd634b4efb383f4034437384e44a6d455260aaee2eca1e9c1b55f047",
+                "sha256:045f00a43c737f960d273a83973b2511430d61f283a44c96bf13a6e829ba8fdc",
+                "sha256:0895b8414afafc526712c498bd9de2b063deaac4021a3b3c34566283464aff8e",
+                "sha256:1209036fbd0421afde505a4879dee3b2fd7b1e14fee81c0069807adcbbcca747",
+                "sha256:1ad45a1f5d0b608253b11508f80940985d1d0c8f6111b5cb637533a0e6ddc13e",
+                "sha256:353815f59a7f64cdaca1c0307ee13558a0512f6db064e92fe833784f08539c7a",
+                "sha256:498c6979f9c6637ebc3a73b3f87f9eb1ec24e1ce53a7c5173b8508981614a90b",
+                "sha256:5cd2bcdc75b452ba2e10f0e8ecc0b57b827dd5d7aaffbc6821b2a9a242823a76",
+                "sha256:6d3fbbc8d23fcdcb500d2c9f94e07b1342df8ed71b948a2649b5cb060a7c94ca",
+                "sha256:6e2dcd475ce8b80522e51d923d10c7871e45f20918e027ab682f94f1c6351688",
+                "sha256:9118f27452b70bb1d9ab3198c1f626c2499384935aaf55388211ad982611407e",
+                "sha256:9dcbfce5d89f1d1f2546a2090f4fcf87c7f669d1d90aacb7d7582addece9fb38",
+                "sha256:a8506f6119cff7015678e2bce904a4da21025cc70ad283a53b099e7620061d85",
+                "sha256:a8fb3752b491d246034fa4d279ff076501588ce8cbcdbb62c32fd7a377d996be",
+                "sha256:c0e0c00aa18ca2d3b2b991643b799a15fc8f0563d2ebb6040f64ce8dc027b942",
+                "sha256:d905186d647b16755a800e7263d43df08b790d709d575105d419f8b6ef65423a",
+                "sha256:ff34df86226c0227c52f38b919213157588a678d049688eded74c76c8ba4a5d0"
+            ],
+            "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4, 3.5'",
+            "version": "==6.1.0"
+        },
         "pyarrow": {
             "hashes": [
                 "sha256:01c034b576ce0eef554f7c3d8c341714954be9b3f5d5bc7117006b85fcf302fe",
             "markers": "python_version >= '3.9'",
             "version": "==3.2.0"
         },
+        "pysocks": {
+            "hashes": [
+                "sha256:08e69f092cc6dbe92a0fdd16eeb9b9ffbc13cadfe5ca4c7bd92ffb078b293299",
+                "sha256:2725bd0a9925919b9b51739eea5f9e2bae91e83288108a9ad338b2e3a4435ee5",
+                "sha256:3f8804571ebe159c380ac6de37643bb4685970655d3bba243530d6558b799aa0"
+            ],
+            "version": "==1.7.1"
+        },
         "python-dateutil": {
             "hashes": [
                 "sha256:37dd54208da7e1cd875388217d5e00ebd4179249f90fb72437e91a35459a0ad3",
             "markers": "python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'",
             "version": "==2.9.0.post0"
         },
+        "pytorch-revgrad": {
+            "hashes": [
+                "sha256:2276fb189b2ce26f756a97effe2a6bcf8f7fdc60542c5dfb45c53f09ef123aa7",
+                "sha256:9cf097a7d18cbadddeaec9fef74b258d70b6cb8d0c77f524baab18bffc7d7be9"
+            ],
+            "markers": "python_version >= '3.5'",
+            "version": "==0.2.0"
+        },
         "pytz": {
             "hashes": [
                 "sha256:2aa355083c50a0f93fa581709deac0c9ad65cca8a9e9beac660adcbd493c798a",
             "markers": "python_version >= '3.8'",
             "version": "==0.35.1"
         },
+        "regex": {
+            "hashes": [
+                "sha256:02a02d2bb04fec86ad61f3ea7f49c015a0681bf76abb9857f945d26159d2968c",
+                "sha256:02e28184be537f0e75c1f9b2f8847dc51e08e6e171c6bde130b2687e0c33cf60",
+                "sha256:040df6fe1a5504eb0f04f048e6d09cd7c7110fef851d7c567a6b6e09942feb7d",
+                "sha256:068376da5a7e4da51968ce4c122a7cd31afaaec4fccc7856c92f63876e57b51d",
+                "sha256:06eb1be98df10e81ebaded73fcd51989dcf534e3c753466e4b60c4697a003b67",
+                "sha256:072623554418a9911446278f16ecb398fb3b540147a7828c06e2011fa531e773",
+                "sha256:086a27a0b4ca227941700e0b31425e7a28ef1ae8e5e05a33826e17e47fbfdba0",
+                "sha256:08986dce1339bc932923e7d1232ce9881499a0e02925f7402fb7c982515419ef",
+                "sha256:0a86e7eeca091c09e021db8eb72d54751e527fa47b8d5787caf96d9831bd02ad",
+                "sha256:0c32f75920cf99fe6b6c539c399a4a128452eaf1af27f39bce8909c9a3fd8cbe",
+                "sha256:0d7f453dca13f40a02b79636a339c5b62b670141e63efd511d3f8f73fba162b3",
+                "sha256:1062b39a0a2b75a9c694f7a08e7183a80c63c0d62b301418ffd9c35f55aaa114",
+                "sha256:13291b39131e2d002a7940fb176e120bec5145f3aeb7621be6534e46251912c4",
+                "sha256:149f5008d286636e48cd0b1dd65018548944e495b0265b45e1bffecce1ef7f39",
+                "sha256:164d8b7b3b4bcb2068b97428060b2a53be050085ef94eca7f240e7947f1b080e",
+                "sha256:167ed4852351d8a750da48712c3930b031f6efdaa0f22fa1933716bfcd6bf4a3",
+                "sha256:1c4de13f06a0d54fa0d5ab1b7138bfa0d883220965a29616e3ea61b35d5f5fc7",
+                "sha256:202eb32e89f60fc147a41e55cb086db2a3f8cb82f9a9a88440dcfc5d37faae8d",
+                "sha256:220902c3c5cc6af55d4fe19ead504de80eb91f786dc102fbd74894b1551f095e",
+                "sha256:2b3361af3198667e99927da8b84c1b010752fa4b1115ee30beaa332cabc3ef1a",
+                "sha256:2c89a8cc122b25ce6945f0423dc1352cb9593c68abd19223eebbd4e56612c5b7",
+                "sha256:2d548dafee61f06ebdb584080621f3e0c23fff312f0de1afc776e2a2ba99a74f",
+                "sha256:2e34b51b650b23ed3354b5a07aab37034d9f923db2a40519139af34f485f77d0",
+                "sha256:32f9a4c643baad4efa81d549c2aadefaeba12249b2adc5af541759237eee1c54",
+                "sha256:3a51ccc315653ba012774efca4f23d1d2a8a8f278a6072e29c7147eee7da446b",
+                "sha256:3cde6e9f2580eb1665965ce9bf17ff4952f34f5b126beb509fee8f4e994f143c",
+                "sha256:40291b1b89ca6ad8d3f2b82782cc33807f1406cf68c8d440861da6304d8ffbbd",
+                "sha256:41758407fc32d5c3c5de163888068cfee69cb4c2be844e7ac517a52770f9af57",
+                "sha256:4181b814e56078e9b00427ca358ec44333765f5ca1b45597ec7446d3a1ef6e34",
+                "sha256:4f51f88c126370dcec4908576c5a627220da6c09d0bff31cfa89f2523843316d",
+                "sha256:50153825ee016b91549962f970d6a4442fa106832e14c918acd1c8e479916c4f",
+                "sha256:5056b185ca113c88e18223183aa1a50e66507769c9640a6ff75859619d73957b",
+                "sha256:5071b2093e793357c9d8b2929dfc13ac5f0a6c650559503bb81189d0a3814519",
+                "sha256:525eab0b789891ac3be914d36893bdf972d483fe66551f79d3e27146191a37d4",
+                "sha256:52fb28f528778f184f870b7cf8f225f5eef0a8f6e3778529bdd40c7b3920796a",
+                "sha256:5478c6962ad548b54a591778e93cd7c456a7a29f8eca9c49e4f9a806dcc5d638",
+                "sha256:5670bce7b200273eee1840ef307bfa07cda90b38ae56e9a6ebcc9f50da9c469b",
+                "sha256:5704e174f8ccab2026bd2f1ab6c510345ae8eac818b613d7d73e785f1310f839",
+                "sha256:59dfe1ed21aea057a65c6b586afd2a945de04fc7db3de0a6e3ed5397ad491b07",
+                "sha256:5e7e351589da0850c125f1600a4c4ba3c722efefe16b297de54300f08d734fbf",
+                "sha256:63b13cfd72e9601125027202cad74995ab26921d8cd935c25f09c630436348ff",
+                "sha256:658f90550f38270639e83ce492f27d2c8d2cd63805c65a13a14d36ca126753f0",
+                "sha256:684d7a212682996d21ca12ef3c17353c021fe9de6049e19ac8481ec35574a70f",
+                "sha256:69ab78f848845569401469da20df3e081e6b5a11cb086de3eed1d48f5ed57c95",
+                "sha256:6f44ec28b1f858c98d3036ad5d7d0bfc568bdd7a74f9c24e25f41ef1ebfd81a4",
+                "sha256:70b7fa6606c2881c1db9479b0eaa11ed5dfa11c8d60a474ff0e095099f39d98e",
+                "sha256:764e71f22ab3b305e7f4c21f1a97e1526a25ebdd22513e251cf376760213da13",
+                "sha256:7ab159b063c52a0333c884e4679f8d7a85112ee3078fe3d9004b2dd875585519",
+                "sha256:805e6b60c54bf766b251e94526ebad60b7de0c70f70a4e6210ee2891acb70bf2",
+                "sha256:8447d2d39b5abe381419319f942de20b7ecd60ce86f16a23b0698f22e1b70008",
+                "sha256:86fddba590aad9208e2fa8b43b4c098bb0ec74f15718bb6a704e3c63e2cef3e9",
+                "sha256:89d75e7293d2b3e674db7d4d9b1bee7f8f3d1609428e293771d1a962617150cc",
+                "sha256:93c0b12d3d3bc25af4ebbf38f9ee780a487e8bf6954c115b9f015822d3bb8e48",
+                "sha256:94d87b689cdd831934fa3ce16cc15cd65748e6d689f5d2b8f4f4df2065c9fa20",
+                "sha256:9714398225f299aa85267fd222f7142fcb5c769e73d7733344efc46f2ef5cf89",
+                "sha256:982e6d21414e78e1f51cf595d7f321dcd14de1f2881c5dc6a6e23bbbbd68435e",
+                "sha256:997d6a487ff00807ba810e0f8332c18b4eb8d29463cfb7c820dc4b6e7562d0cf",
+                "sha256:a03e02f48cd1abbd9f3b7e3586d97c8f7a9721c436f51a5245b3b9483044480b",
+                "sha256:a36fdf2af13c2b14738f6e973aba563623cb77d753bbbd8d414d18bfaa3105dd",
+                "sha256:a6ba92c0bcdf96cbf43a12c717eae4bc98325ca3730f6b130ffa2e3c3c723d84",
+                "sha256:a7c2155f790e2fb448faed6dd241386719802296ec588a8b9051c1f5c481bc29",
+                "sha256:a93c194e2df18f7d264092dc8539b8ffb86b45b899ab976aa15d48214138e81b",
+                "sha256:abfa5080c374a76a251ba60683242bc17eeb2c9818d0d30117b4486be10c59d3",
+                "sha256:ac10f2c4184420d881a3475fb2c6f4d95d53a8d50209a2500723d831036f7c45",
+                "sha256:ad182d02e40de7459b73155deb8996bbd8e96852267879396fb274e8700190e3",
+                "sha256:b2837718570f95dd41675328e111345f9b7095d821bac435aac173ac80b19983",
+                "sha256:b489578720afb782f6ccf2840920f3a32e31ba28a4b162e13900c3e6bd3f930e",
+                "sha256:b583904576650166b3d920d2bcce13971f6f9e9a396c673187f49811b2769dc7",
+                "sha256:b85c2530be953a890eaffde05485238f07029600e8f098cdf1848d414a8b45e4",
+                "sha256:b97c1e0bd37c5cd7902e65f410779d39eeda155800b65fc4d04cc432efa9bc6e",
+                "sha256:ba9b72e5643641b7d41fa1f6d5abda2c9a263ae835b917348fc3c928182ad467",
+                "sha256:bb26437975da7dc36b7efad18aa9dd4ea569d2357ae6b783bf1118dabd9ea577",
+                "sha256:bb8f74f2f10dbf13a0be8de623ba4f9491faf58c24064f32b65679b021ed0001",
+                "sha256:bde01f35767c4a7899b7eb6e823b125a64de314a8ee9791367c9a34d56af18d0",
+                "sha256:bec9931dfb61ddd8ef2ebc05646293812cb6b16b60cf7c9511a832b6f1854b55",
+                "sha256:c36f9b6f5f8649bb251a5f3f66564438977b7ef8386a52460ae77e6070d309d9",
+                "sha256:cdf58d0e516ee426a48f7b2c03a332a4114420716d55769ff7108c37a09951bf",
+                "sha256:d1cee317bfc014c2419a76bcc87f071405e3966da434e03e13beb45f8aced1a6",
+                "sha256:d22326fcdef5e08c154280b71163ced384b428343ae16a5ab2b3354aed12436e",
+                "sha256:d3660c82f209655a06b587d55e723f0b813d3a7db2e32e5e7dc64ac2a9e86fde",
+                "sha256:da8f5fc57d1933de22a9e23eec290a0d8a5927a5370d24bda9a6abe50683fe62",
+                "sha256:df951c5f4a1b1910f1a99ff42c473ff60f8225baa1cdd3539fe2819d9543e9df",
+                "sha256:e5364a4502efca094731680e80009632ad6624084aff9a23ce8c8c6820de3e51",
+                "sha256:ea1bfda2f7162605f6e8178223576856b3d791109f15ea99a9f95c16a7636fb5",
+                "sha256:f02f93b92358ee3f78660e43b4b0091229260c5d5c408d17d60bf26b6c900e86",
+                "sha256:f056bf21105c2515c32372bbc057f43eb02aae2fda61052e2f7622c801f0b4e2",
+                "sha256:f1ac758ef6aebfc8943560194e9fd0fa18bcb34d89fd8bd2af18183afd8da3a2",
+                "sha256:f2a19f302cd1ce5dd01a9099aaa19cae6173306d1302a43b627f62e21cf18ac0",
+                "sha256:f654882311409afb1d780b940234208a252322c24a93b442ca714d119e68086c",
+                "sha256:f65557897fc977a44ab205ea871b690adaef6b9da6afda4790a2484b04293a5f",
+                "sha256:f9d1e379028e0fc2ae3654bac3cbbef81bf3fd571272a42d56c24007979bafb6",
+                "sha256:fdabbfc59f2c6edba2a6622c647b716e34e8e3867e0ab975412c5c2f79b82da2",
+                "sha256:fdd6028445d2460f33136c55eeb1f601ab06d74cb3347132e1c24250187500d9",
+                "sha256:ff590880083d60acc0433f9c3f713c51f7ac6ebb9adf889c79a261ecf541aa91"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==2024.11.6"
+        },
         "requests": {
             "hashes": [
                 "sha256:55365417734eb18255590a9ff9eb97e9e1da868d4ccd6402399eaf68af20a760",
             "markers": "python_version >= '3.9'",
             "version": "==0.22.3"
         },
+        "s3transfer": {
+            "hashes": [
+                "sha256:244a76a24355363a68164241438de1b72f8781664920260c48465896b712a41e",
+                "sha256:29edc09801743c21eb5ecbc617a152df41d3c287f67b615f73e5f750583666a7"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==0.10.4"
+        },
+        "safetensors": {
+            "hashes": [
+                "sha256:01c8f00da537af711979e1b42a69a8ec9e1d7112f208e0e9b8a35d2c381085ef",
+                "sha256:023b6e5facda76989f4cba95a861b7e656b87e225f61811065d5c501f78cdb3f",
+                "sha256:09566792588d77b68abe53754c9f1308fadd35c9f87be939e22c623eaacbed6b",
+                "sha256:098923e2574ff237c517d6e840acada8e5b311cb1fa226019105ed82e9c3b62f",
+                "sha256:09dedf7c2fda934ee68143202acff6e9e8eb0ddeeb4cfc24182bef999efa9f42",
+                "sha256:133620f443450429322f238fda74d512c4008621227fccf2f8cf4a76206fea7c",
+                "sha256:139fbee92570ecea774e6344fee908907db79646d00b12c535f66bc78bd5ea2c",
+                "sha256:13ca0902d2648775089fa6a0c8fc9e6390c5f8ee576517d33f9261656f851e3f",
+                "sha256:1500418454529d0ed5c1564bda376c4ddff43f30fce9517d9bee7bcce5a8ef50",
+                "sha256:1524b54246e422ad6fb6aea1ac71edeeb77666efa67230e1faf6999df9b2e27f",
+                "sha256:21742b391b859e67b26c0b2ac37f52c9c0944a879a25ad2f9f9f3cd61e7fda8f",
+                "sha256:21f848d7aebd5954f92538552d6d75f7c1b4500f51664078b5b49720d180e47c",
+                "sha256:23fc9b4ec7b602915cbb4ec1a7c1ad96d2743c322f20ab709e2c35d1b66dad27",
+                "sha256:25e5f8e2e92a74f05b4ca55686234c32aac19927903792b30ee6d7bd5653d54e",
+                "sha256:2783956926303dcfeb1de91a4d1204cd4089ab441e622e7caee0642281109db3",
+                "sha256:309aaec9b66cbf07ad3a2e5cb8a03205663324fea024ba391594423d0f00d9fe",
+                "sha256:313514b0b9b73ff4ddfb4edd71860696dbe3c1c9dc4d5cc13dbd74da283d2cbf",
+                "sha256:31fa33ee326f750a2f2134a6174773c281d9a266ccd000bd4686d8021f1f3dac",
+                "sha256:3685ce7ed036f916316b567152482b7e959dc754fcc4a8342333d222e05f407c",
+                "sha256:39371fc551c1072976073ab258c3119395294cf49cdc1f8476794627de3130df",
+                "sha256:3a6ba28118636a130ccbb968bc33d4684c48678695dba2590169d5ab03a45646",
+                "sha256:4037676c86365a721a8c9510323a51861d703b399b78a6b4486a54a65a975fca",
+                "sha256:473300314e026bd1043cef391bb16a8689453363381561b8a3e443870937cc1e",
+                "sha256:4b99fbf72e3faf0b2f5f16e5e3458b93b7d0a83984fe8d5364c60aa169f2da89",
+                "sha256:4fb3e0609ec12d2a77e882f07cced530b8262027f64b75d399f1504ffec0ba56",
+                "sha256:500cac01d50b301ab7bb192353317035011c5ceeef0fca652f9f43c000bb7f8d",
+                "sha256:52452fa5999dc50c4decaf0c53aa28371f7f1e0fe5c2dd9129059fbe1e1599c7",
+                "sha256:53946c5813b8f9e26103c5efff4a931cc45d874f45229edd68557ffb35ffb9f8",
+                "sha256:540ce6c4bf6b58cb0fd93fa5f143bc0ee341c93bb4f9287ccd92cf898cc1b0dd",
+                "sha256:585f1703a518b437f5103aa9cf70e9bd437cb78eea9c51024329e4fb8a3e3679",
+                "sha256:59b77e4b7a708988d84f26de3ebead61ef1659c73dcbc9946c18f3b1786d2688",
+                "sha256:5a2d68a523a4cefd791156a4174189a4114cf0bf9c50ceb89f261600f3b2b81a",
+                "sha256:5d3bc83e14d67adc2e9387e511097f254bd1b43c3020440e708858c684cbac68",
+                "sha256:5f0032bedc869c56f8d26259fe39cd21c5199cd57f2228d817a0e23e8370af25",
+                "sha256:60c828a27e852ded2c85fc0f87bf1ec20e464c5cd4d56ff0e0711855cc2e17f8",
+                "sha256:63bfd425e25f5c733f572e2246e08a1c38bd6f2e027d3f7c87e2e43f228d1345",
+                "sha256:65573dc35be9059770808e276b017256fa30058802c29e1038eb1c00028502ea",
+                "sha256:670e95fe34e0d591d0529e5e59fd9d3d72bc77b1444fcaa14dccda4f36b5a38b",
+                "sha256:67e1e7cb8678bb1b37ac48ec0df04faf689e2f4e9e81e566b5c63d9f23748523",
+                "sha256:68814d599d25ed2fdd045ed54d370d1d03cf35e02dce56de44c651f828fb9b7b",
+                "sha256:6885016f34bef80ea1085b7e99b3c1f92cb1be78a49839203060f67b40aee761",
+                "sha256:6ac85d9a8c1af0e3132371d9f2d134695a06a96993c2e2f0bbe25debb9e3f67a",
+                "sha256:6d3de65718b86c3eeaa8b73a9c3d123f9307a96bbd7be9698e21e76a56443af5",
+                "sha256:7389129c03fadd1ccc37fd1ebbc773f2b031483b04700923c3511d2a939252cc",
+                "sha256:73e7d408e9012cd17511b382b43547850969c7979efc2bc353f317abaf23c84c",
+                "sha256:7469d70d3de970b1698d47c11ebbf296a308702cbaae7fcb993944751cf985f4",
+                "sha256:75331c0c746f03158ded32465b7d0b0e24c5a22121743662a2393439c43a45cf",
+                "sha256:76ded72f69209c9780fdb23ea89e56d35c54ae6abcdec67ccb22af8e696e449a",
+                "sha256:775409ce0fcc58b10773fdb4221ed1eb007de10fe7adbdf8f5e8a56096b6f0bc",
+                "sha256:77d9b228da8374c7262046a36c1f656ba32a93df6cc51cd4453af932011e77f1",
+                "sha256:788ee7d04cc0e0e7f944c52ff05f52a4415b312f5efd2ee66389fb7685ee030c",
+                "sha256:78dd8adfb48716233c45f676d6e48534d34b4bceb50162c13d1f0bdf6f78590a",
+                "sha256:801183a0f76dc647f51a2d9141ad341f9665602a7899a693207a82fb102cc53e",
+                "sha256:8158938cf3324172df024da511839d373c40fbfaa83e9abf467174b2910d7b4c",
+                "sha256:81efb124b58af39fcd684254c645e35692fea81c51627259cdf6d67ff4458916",
+                "sha256:834001bed193e4440c4a3950a31059523ee5090605c907c66808664c932b549c",
+                "sha256:83c4f13a9e687335c3928f615cd63a37e3f8ef072a3f2a0599fa09f863fb06a2",
+                "sha256:868f9df9e99ad1e7f38c52194063a982bc88fedc7d05096f4f8160403aaf4bd6",
+                "sha256:87bc42bd04fd9ca31396d3ca0433db0be1411b6b53ac5a32b7845a85d01ffc2e",
+                "sha256:8e8deb16c4321d61ae72533b8451ec4a9af8656d1c61ff81aa49f966406e4b68",
+                "sha256:9483f42be3b6bc8ff77dd67302de8ae411c4db39f7224dec66b0eb95822e4163",
+                "sha256:951d2fcf1817f4fb0ef0b48f6696688a4e852a95922a042b3f96aaa67eedc920",
+                "sha256:9633b663393d5796f0b60249549371e392b75a0b955c07e9c6f8708a87fc841f",
+                "sha256:96f1d038c827cdc552d97e71f522e1049fef0542be575421f7684756a748e457",
+                "sha256:9cc9449bd0b0bc538bd5e268221f0c5590bc5c14c1934a6ae359d44410dc68c4",
+                "sha256:9d1a94b9d793ed8fe35ab6d5cea28d540a46559bafc6aae98f30ee0867000cab",
+                "sha256:9e347d77e2c77eb7624400ccd09bed69d35c0332f417ce8c048d404a096c593b",
+                "sha256:9f556eea3aec1d3d955403159fe2123ddd68e880f83954ee9b4a3f2e15e716b6",
+                "sha256:a01e232e6d3d5cf8b1667bc3b657a77bdab73f0743c26c1d3c5dd7ce86bd3a92",
+                "sha256:a0dd565f83b30f2ca79b5d35748d0d99dd4b3454f80e03dfb41f0038e3bdf180",
+                "sha256:a3a315a6d0054bc6889a17f5668a73f94f7fe55121ff59e0a199e3519c08565f",
+                "sha256:a63eaccd22243c67e4f2b1c3e258b257effc4acd78f3b9d397edc8cf8f1298a7",
+                "sha256:a659467495de201e2f282063808a41170448c78bada1e62707b07a27b05e6943",
+                "sha256:a6c19feda32b931cae0acd42748a670bdf56bee6476a046af20181ad3fee4090",
+                "sha256:adaa9c6dead67e2dd90d634f89131e43162012479d86e25618e821a03d1eb1dc",
+                "sha256:b17b299ca9966ca983ecda1c0791a3f07f9ca6ab5ded8ef3d283fff45f6bcd5f",
+                "sha256:b3139098e3e8b2ad7afbca96d30ad29157b50c90861084e69fcb80dec7430461",
+                "sha256:b4db6a61d968de73722b858038c616a1bebd4a86abe2688e46ca0cc2d17558f2",
+                "sha256:b5a8810ad6a6f933fff6c276eae92c1da217b39b4d8b1bc1c0b8af2d270dc532",
+                "sha256:b75a616e02f21b6f1d5785b20cecbab5e2bd3f6358a90e8925b813d557666ec1",
+                "sha256:b98d40a2ffa560653f6274e15b27b3544e8e3713a44627ce268f419f35c49478",
+                "sha256:bad5e4b2476949bcd638a89f71b6916fa9a5cae5c1ae7eede337aca2100435c0",
+                "sha256:bb07000b19d41e35eecef9a454f31a8b4718a185293f0d0b1c4b61d6e4487971",
+                "sha256:bfeaa1a699c6b9ed514bd15e6a91e74738b71125a9292159e3d6b7f0a53d2cde",
+                "sha256:c36302c1c69eebb383775a89645a32b9d266878fab619819ce660309d6176c9b",
+                "sha256:c6d156bdb26732feada84f9388a9f135528c1ef5b05fae153da365ad4319c4c5",
+                "sha256:c7db3006a4915151ce1913652e907cdede299b974641a83fbc092102ac41b644",
+                "sha256:c859c7ed90b0047f58ee27751c8e56951452ed36a67afee1b0a87847d065eec6",
+                "sha256:cbd39cae1ad3e3ef6f63a6f07296b080c951f24cec60188378e43d3713000c04",
+                "sha256:cf727bb1281d66699bef5683b04d98c894a2803442c490a8d45cd365abfbdeb2",
+                "sha256:d0f1dd769f064adc33831f5e97ad07babbd728427f98e3e1db6902e369122737",
+                "sha256:d42ffd4c2259f31832cb17ff866c111684c87bd930892a1ba53fed28370c918c",
+                "sha256:d5f23198821e227cfc52d50fa989813513db381255c6d100927b012f0cfec63d",
+                "sha256:d641f5b8149ea98deb5ffcf604d764aad1de38a8285f86771ce1abf8e74c4891",
+                "sha256:d73de19682deabb02524b3d5d1f8b3aaba94c72f1bbfc7911b9b9d5d391c0310",
+                "sha256:d94581aab8c6b204def4d7320f07534d6ee34cd4855688004a4354e63b639a35",
+                "sha256:dbd280b07e6054ea68b0cb4b16ad9703e7d63cd6890f577cb98acc5354780142",
+                "sha256:dd8a1f6d2063a92cd04145c7fd9e31a1c7d85fbec20113a14b487563fdbc0597",
+                "sha256:dde2bf390d25f67908278d6f5d59e46211ef98e44108727084d4637ee70ab4f1",
+                "sha256:e3cec4a29eb7fe8da0b1c7988bc3828183080439dd559f720414450de076fcab",
+                "sha256:e7a97058f96340850da0601a3309f3d29d6191b0702b2da201e54c6e3e44ccf0",
+                "sha256:e98ef5524f8b6620c8cdef97220c0b6a5c1cef69852fcd2f174bb96c2bb316b1",
+                "sha256:f0b6453c54c57c1781292c46593f8a37254b8b99004c68d6c3ce229688931a22",
+                "sha256:f3664ac565d0e809b0b929dae7ccd74e4d3273cd0c6d1220c6430035befb678e",
+                "sha256:f4b15f51b4f8f2a512341d9ce3475cacc19c5fdfc5db1f0e19449e75f95c7dc8",
+                "sha256:f4beb84b6073b1247a773141a6331117e35d07134b3bb0383003f39971d414bb",
+                "sha256:f6594d130d0ad933d885c6a7b75c5183cb0e8450f799b80a39eae2b8508955eb",
+                "sha256:f68bf99ea970960a237f416ea394e266e0361895753df06e3e06e6ea7907d98b",
+                "sha256:fd33da8e9407559f8779c82a0448e2133737f922d71f884da27184549416bfed",
+                "sha256:fdadf66b5a22ceb645d5435a0be7a0292ce59648ca1d46b352f13cff3ea80410"
+            ],
+            "markers": "python_version >= '3.7'",
+            "version": "==0.4.5"
+        },
         "scikit-learn": {
             "hashes": [
+                "sha256:04a5ba45c12a5ff81518aa4f1604e826a45d20e53da47b15871526cda4ff5174",
+                "sha256:0baa91eeb8c32632628874a5c91885eaedd23b71504d24227925080da075837a",
+                "sha256:1dad624cffe3062276a0881d4e441bc9e3b19d02d17757cd6ae79a9d192a0027",
+                "sha256:1f50b4f24cf12a81c3c09958ae3b864d7534934ca66ded3822de4996d25d7285",
+                "sha256:21fadfc2ad7a1ce8bd1d90f23d17875b84ec765eecbbfc924ff11fb73db582ce",
+                "sha256:2fce7950a3fad85e0a61dc403df0f9345b53432ac0e47c50da210d22c60b6d85",
+                "sha256:30f34bb5fde90e020653bb84dcb38b6c83f90c70680dbd8c38bd9becbad7a127",
+                "sha256:34e20bfac8ff0ebe0ff20fb16a4d6df5dc4cc9ce383e00c2ab67a526a3c67b18",
+                "sha256:366fb3fa47dce90afed3d6106183f4978d6f24cfd595c2373424171b915ee718",
+                "sha256:3c716d13ba0a2f8762d96ff78d3e0cde90bc9c9b5c13d6ab6bb9b2d6ca6705fd",
+                "sha256:59cd96a8d9f8dfd546f5d6e9787e1b989e981388d7803abbc9efdcde61e47460",
+                "sha256:5be4577769c5dde6e1b53de8e6520f9b664ab5861dd57acee47ad119fd7405d6",
+                "sha256:5c3fa7d3dd5a0ec2d0baba0d644916fa2ab180ee37850c5d536245df916946bd",
+                "sha256:5fe11794236fb83bead2af26a87ced5d26e3370b8487430818b915dafab1724e",
+                "sha256:61fe3dcec0d82ae280877a818ab652f4988371e32dd5451e75251bece79668b1",
+                "sha256:66b1cf721a9f07f518eb545098226796c399c64abdcbf91c2b95d625068363da",
+                "sha256:7b35b60cf4cd6564b636e4a40516b3c61a4fa7a8b1f7a3ce80c38ebe04750bc3",
+                "sha256:98717d3c152f6842d36a70f21e1468fb2f1a2f8f2624d9a3f382211798516426",
+                "sha256:9aafd94bafc841b626681e626be27bf1233d5a0f20f0a6fdb4bee1a1963c6643",
+                "sha256:9d58481f9f7499dff4196927aedd4285a0baec8caa3790efbe205f13de37dd6e",
+                "sha256:a17860a562bac54384454d40b3f6155200c1c737c9399e6a97962c63fce503ac",
+                "sha256:a46d3ca0f11a540b8eaddaf5e38172d8cd65a86cb3e3632161ec96c0cffb774c",
+                "sha256:a73b1c2038c93bc7f4bf21f6c9828d5116c5d2268f7a20cfbbd41d3074d52083",
+                "sha256:b44e3a51e181933bdf9a4953cc69c6025b40d2b49e238233f149b98849beb4bf",
+                "sha256:b6916d1cec1ff163c7d281e699d7a6a709da2f2c5ec7b10547e08cc788ddd3ae",
+                "sha256:df778486a32518cda33818b7e3ce48c78cef1d5f640a6bc9d97c6d2e71449a51",
+                "sha256:e5453b2e87ef8accedc5a8a4e6709f887ca01896cd7cc8a174fe39bd4bb00aef",
+                "sha256:eb9ae21f387826da14b0b9cb1034f5048ddb9182da429c689f5f4a87dc96930b",
+                "sha256:eba06d75815406091419e06dd650b91ebd1c5f836392a0d833ff36447c2b1bfa",
+                "sha256:efa7a579606c73a0b3d210e33ea410ea9e1af7933fe324cb7e6fbafae4ea5948"
             ],
             "index": "pypi",
+            "version": "==1.6.0"
         },
         "scipy": {
             "hashes": [
             "markers": "python_version >= '3.10'",
             "version": "==1.14.1"
         },
+        "segtok": {
+            "hashes": [
+                "sha256:8ab2dd44245bcbfec25b575dc4618473bbdf2af8c2649698cd5a370f42f3db23",
+                "sha256:910616b76198c3141b2772df530270d3b706e42ae69a5b30ef115c7bd5d1501a"
+            ],
+            "version": "==1.5.11"
+        },
+        "semver": {
+            "hashes": [
+                "sha256:6253adb39c70f6e51afed2fa7152bcd414c411286088fb4b9effb133885ab4cc",
+                "sha256:b1ea4686fe70b981f85359eda33199d60c53964284e0cfb4977d243e37cf4bf4"
+            ],
+            "markers": "python_version >= '3.7'",
+            "version": "==3.0.2"
+        },
+        "sentencepiece": {
+            "hashes": [
+                "sha256:0461324897735512a32d222e3d886e24ad6a499761952b6bda2a9ee6e4313ea5",
+                "sha256:0993dbc665f4113017892f1b87c3904a44d0640eda510abcacdfb07f74286d36",
+                "sha256:0a91aaa3c769b52440df56fafda683b3aa48e3f2169cf7ee5b8c8454a7f3ae9b",
+                "sha256:0f67eae0dbe6f2d7d6ba50a354623d787c99965f068b81e145d53240198021b0",
+                "sha256:1380ce6540a368de2ef6d7e6ba14ba8f3258df650d39ba7d833b79ee68a52040",
+                "sha256:17982700c4f6dbb55fa3594f3d7e5dd1c8659a274af3738e33c987d2a27c9d5c",
+                "sha256:188779e1298a1c8b8253c7d3ad729cb0a9891e5cef5e5d07ce4592c54869e227",
+                "sha256:1e0f9c4d0a6b0af59b613175f019916e28ade076e21242fd5be24340d8a2f64a",
+                "sha256:20813a68d4c221b1849c62c30e1281ea81687894d894b8d4a0f4677d9311e0f5",
+                "sha256:22e37bac44dd6603388cb598c64ff7a76e41ca774646f21c23aadfbf5a2228ab",
+                "sha256:27f90c55a65013cbb8f4d7aab0599bf925cde4adc67ae43a0d323677b5a1c6cb",
+                "sha256:298f21cc1366eb60311aedba3169d30f885c363ddbf44214b0a587d2908141ad",
+                "sha256:2a3149e3066c2a75e0d68a43eb632d7ae728c7925b517f4c05c40f6f7280ce08",
+                "sha256:2fde4b08cfe237be4484c6c7c2e2c75fb862cfeab6bd5449ce4caeafd97b767a",
+                "sha256:3212121805afc58d8b00ab4e7dd1f8f76c203ddb9dc94aa4079618a31cf5da0f",
+                "sha256:38aed822fb76435fa1f12185f10465a94ab9e51d5e8a9159e9a540ce926f0ffd",
+                "sha256:3f1ec95aa1e5dab11f37ac7eff190493fd87770f7a8b81ebc9dd768d1a3c8704",
+                "sha256:4547683f330289ec4f093027bfeb87f9ef023b2eb6f879fdc4a8187c7e0ffb90",
+                "sha256:4c378492056202d1c48a4979650981635fd97875a00eabb1f00c6a236b013b5e",
+                "sha256:536b934e244829e3fe6c4f198652cd82da48adb9aa145c9f00889542726dee3d",
+                "sha256:632f3594d3e7ac8b367bca204cb3fd05a01d5b21455acd097ea4c0e30e2f63d7",
+                "sha256:6cf333625234f247ab357b0bd9836638405ea9082e1543d5b8408f014979dcbf",
+                "sha256:7140d9e5a74a0908493bb4a13f1f16a401297bd755ada4c707e842fbf6f0f5bf",
+                "sha256:787e480ca4c1d08c9985a7eb1eae4345c107729c99e9b5a9a00f2575fc7d4b4b",
+                "sha256:7a673a72aab81fef5ebe755c6e0cc60087d1f3a4700835d40537183c1703a45f",
+                "sha256:7b06b70af54daa4b4904cbb90b4eb6d35c9f3252fdc86c9c32d5afd4d30118d8",
+                "sha256:7c867012c0e8bcd5bdad0f791609101cb5c66acb303ab3270218d6debc68a65e",
+                "sha256:7cd6175f7eaec7142d2bf6f6597ce7db4c9ac89acf93fcdb17410c3a8b781eeb",
+                "sha256:7fd6071249c74f779c5b27183295b9202f8dedb68034e716784364443879eaa6",
+                "sha256:859ba1acde782609a0910a26a60e16c191a82bf39b5621107552c0cd79fad00f",
+                "sha256:89f65f69636b7e9c015b79dff9c9985a9bc7d19ded6f79ef9f1ec920fdd73ecf",
+                "sha256:926ef920ae2e8182db31d3f5d081ada57804e3e1d3a8c4ef8b117f9d9fb5a945",
+                "sha256:98501e075f35dd1a1d5a20f65be26839fcb1938752ec61539af008a5aa6f510b",
+                "sha256:a1151d6a6dd4b43e552394aed0edfe9292820272f0194bd56c7c1660a0c06c3d",
+                "sha256:a52c19171daaf2e697dc6cbe67684e0fa341b1248966f6aebb541de654d15843",
+                "sha256:b293734059ef656dcd65be62ff771507bea8fed0a711b6733976e1ed3add4553",
+                "sha256:b99a308a2e5e569031ab164b74e6fab0b6f37dfb493c32f7816225f4d411a6dd",
+                "sha256:bcbbef6cc277f8f18f36959e305f10b1c620442d75addc79c21d7073ae581b50",
+                "sha256:bed9cf85b296fa2b76fc2547b9cbb691a523864cebaee86304c43a7b4cb1b452",
+                "sha256:c581258cf346b327c62c4f1cebd32691826306f6a41d8c4bec43b010dee08e75",
+                "sha256:cdb701eec783d3ec86b7cd4c763adad8eaf6b46db37ee1c36e5e6c44b3fe1b5f",
+                "sha256:d0cb51f53b6aae3c36bafe41e86167c71af8370a039f542c43b0cce5ef24a68c",
+                "sha256:d1e5ca43013e8935f25457a4fca47e315780172c3e821b4b13a890668911c792",
+                "sha256:d490142b0521ef22bc1085f061d922a2a6666175bb6b42e588ff95c0db6819b2",
+                "sha256:d7b67e724bead13f18db6e1d10b6bbdc454af574d70efbb36f27d90387be1ca3",
+                "sha256:d8cf876516548b5a1d6ac4745d8b554f5c07891d55da557925e5c13ff0b4e6ad",
+                "sha256:e3d1d2cc4882e8d6a1adf9d5927d7716f80617fc693385661caff21888972269",
+                "sha256:e58b47f933aca74c6a60a79dcb21d5b9e47416256c795c2d58d55cec27f9551d",
+                "sha256:ea5f536e32ea8ec96086ee00d7a4a131ce583a1b18d130711707c10e69601cb2",
+                "sha256:f295105c6bdbb05bd5e1b0cafbd78ff95036f5d3641e7949455a3f4e5e7c3109",
+                "sha256:f4d158189eb2ecffea3a51edf6d25e110b3678ec47f1a40f2d541eafbd8f6250",
+                "sha256:fb89f811e5efd18bab141afc3fea3de141c3f69f3fe9e898f710ae7fe3aab251",
+                "sha256:ff88712338b01031910e8e61e7239aff3ce8869ee31a47df63cb38aadd591bea"
+            ],
+            "version": "==0.2.0"
+        },
         "six": {
             "hashes": [
                 "sha256:4721f391ed90541fddacab5acf947aa0d3dc7d27b2e1e8eda2be8970586c3274",
             "markers": "python_version >= '3.7'",
             "version": "==5.0.1"
         },
+        "sortedcontainers": {
+            "hashes": [
+                "sha256:25caa5a06cc30b6b83d11423433f65d1f9d76c4c6a0c90e3379eaa43b9bfdb88",
+                "sha256:a163dcaede0f1c021485e957a39245190e74249897e2ae4b2aa38595db237ee0"
+            ],
+            "version": "==2.4.0"
+        },
+        "soupsieve": {
+            "hashes": [
+                "sha256:e2e68417777af359ec65daac1057404a3c8a5455bb8abc36f1a9866ab1a51abb",
+                "sha256:e72c4ff06e4fb6e4b5a9f0f55fe6e81514581fca1515028625d0f299c602ccc9"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==2.6"
+        },
+        "sqlitedict": {
+            "hashes": [
+                "sha256:03d9cfb96d602996f1d4c2db2856f1224b96a9c431bdd16e78032a72940f9e8c"
+            ],
+            "version": "==2.1.0"
+        },
         "streamlit": {
             "hashes": [
+                "sha256:0def00822480071d642e6df36cd63c089f991da3a69fd9eb4ab8f65ce27de4e0",
+                "sha256:6626d32b098ba1458b71eebdd634c62af2dd876380e59c4b6a1e828a39d62d69"
             ],
             "index": "pypi",
+            "version": "==1.41.1"
+        },
+        "sympy": {
+            "hashes": [
+                "sha256:9cebf7e04ff162015ce31c9c6c9144daa34a93bd082f54fd8f12deca4f47515f",
+                "sha256:db36cdc64bf61b9b24578b6f7bab1ecdd2452cf008f34faa33776680c26d66f8"
+            ],
+            "markers": "python_version >= '3.9'",
+            "version": "==1.13.1"
+        },
+        "tabulate": {
+            "hashes": [
+                "sha256:0095b12bf5966de529c0feb1fa08671671b3368eec77d7ef7ab114be2c068b3c",
+                "sha256:024ca478df22e9340661486f85298cff5f6dcdba14f3813e8830015b9ed1948f"
+            ],
+            "markers": "python_version >= '3.7'",
+            "version": "==0.9.0"
         },
         "tenacity": {
             "hashes": [
             "markers": "python_version >= '3.8'",
             "version": "==3.5.0"
         },
+        "tokenizers": {
+            "hashes": [
+                "sha256:089d56db6782a73a27fd8abf3ba21779f5b85d4a9f35e3b493c7bbcbbf0d539b",
+                "sha256:3c4c93eae637e7d2aaae3d376f06085164e1660f89304c0ab2b1d08a406636b2",
+                "sha256:400832c0904f77ce87c40f1a8a27493071282f785724ae62144324f171377273",
+                "sha256:4145505a973116f91bc3ac45988a92e618a6f83eb458f49ea0790df94ee243ff",
+                "sha256:6b177fb54c4702ef611de0c069d9169f0004233890e0c4c5bd5508ae05abf193",
+                "sha256:6b43779a269f4629bebb114e19c3fca0223296ae9fea8bb9a7a6c6fb0657ff8e",
+                "sha256:87841da5a25a3a5f70c102de371db120f41873b854ba65e52bccd57df5a3780c",
+                "sha256:9aeb255802be90acfd363626753fda0064a8df06031012fe7d52fd9a905eb00e",
+                "sha256:c87ca3dc48b9b1222d984b6b7490355a6fdb411a2d810f6f05977258400ddb74",
+                "sha256:d8b09dbeb7a8d73ee204a70f94fc06ea0f17dcf0844f16102b9f414f0b7463ba",
+                "sha256:e84ca973b3a96894d1707e189c14a774b701596d579ffc7e69debfc036a61a04",
+                "sha256:eb1702c2f27d25d9dd5b389cc1f2f51813e99f8ca30d9e25348db6585a97e24a",
+                "sha256:eb7202d231b273c34ec67767378cd04c767e967fda12d4a9e36208a34e2f137e",
+                "sha256:ee0894bf311b75b0c03079f33859ae4b2334d675d4e93f5a4132e1eae2834fe4",
+                "sha256:f53ea537c925422a2e0e92a24cce96f6bc5046bbef24a1652a5edc8ba975f62e"
+            ],
+            "markers": "python_version >= '3.7'",
+            "version": "==0.21.0"
+        },
         "toml": {
             "hashes": [
                 "sha256:806143ae5bfb6a3c6e736a764057db0e6a0e05e338b5630894a5f779cabb4f9b",
             "markers": "python_version >= '2.6' and python_version not in '3.0, 3.1, 3.2, 3.3'",
             "version": "==0.10.2"
         },
+        "torch": {
+            "hashes": [
+                "sha256:1f3b7fb3cf7ab97fae52161423f81be8c6b8afac8d9760823fd623994581e1a3",
+                "sha256:23d062bf70776a3d04dbe74db950db2a5245e1ba4f27208a87f0d743b0d06e86",
+                "sha256:31f8c39660962f9ae4eeec995e3049b5492eb7360dd4f07377658ef4d728fa4c",
+                "sha256:32a037bd98a241df6c93e4c789b683335da76a2ac142c0973675b715102dc5fa",
+                "sha256:340ce0432cad0d37f5a31be666896e16788f1adf8ad7be481196b503dad675b9",
+                "sha256:34bfa1a852e5714cbfa17f27c49d8ce35e1b7af5608c4bc6e81392c352dbc601",
+                "sha256:3f4b7f10a247e0dcd7ea97dc2d3bfbfc90302ed36d7f3952b0008d0df264e697",
+                "sha256:46c817d3ea33696ad3b9df5e774dba2257e9a4cd3c4a3afbf92f6bb13ac5ce2d",
+                "sha256:603c52d2fe06433c18b747d25f5c333f9c1d58615620578c326d66f258686f9a",
+                "sha256:71328e1bbe39d213b8721678f9dcac30dfc452a46d586f1d514a6aa0a99d4744",
+                "sha256:73e58e78f7d220917c5dbfad1a40e09df9929d3b95d25e57d9f8558f84c9a11c",
+                "sha256:7974e3dce28b5a21fb554b73e1bc9072c25dde873fa00d54280861e7a009d7dc",
+                "sha256:8046768b7f6d35b85d101b4b38cba8aa2f3cd51952bc4c06a49580f2ce682291",
+                "sha256:8c712df61101964eb11910a846514011f0b6f5920c55dbf567bff8a34163d5b1",
+                "sha256:9b61edf3b4f6e3b0e0adda8b3960266b9009d02b37555971f4d1c8f7a05afed7",
+                "sha256:de5b7d6740c4b636ef4db92be922f0edc425b65ed78c5076c43c42d362a45457",
+                "sha256:ed231a4b3a5952177fafb661213d690a72caaad97d5824dd4fc17ab9e15cec03"
+            ],
+            "markers": "python_full_version >= '3.8.0'",
+            "version": "==2.5.1"
+        },
         "tornado": {
             "hashes": [
                 "sha256:072ce12ada169c5b00b7d92a99ba089447ccc993ea2143c9ede887e0937aa803",
             "markers": "python_version >= '3.8'",
             "version": "==6.4.2"
         },
+        "tqdm": {
+            "hashes": [
+                "sha256:26445eca388f82e72884e0d580d5464cd801a3ea01e63e5601bdff9ba6a48de2",
+                "sha256:f8aef9c52c08c13a65f30ea34f4e5aac3fd1a34959879d7e59e63027286627f2"
+            ],
+            "markers": "python_version >= '3.7'",
+            "version": "==4.67.1"
+        },
+        "transformer-smaller-training-vocab": {
+            "hashes": [
+                "sha256:01cb3d8f4818121172e1591a06c3149bf49bc18d6f6f269eb42d2c4ed155cfcc",
+                "sha256:d7360ac084786f66f99ef16d621f34acbb0dce6d9a624525d1f7dc8b6c3a49f7"
+            ],
+            "markers": "python_version >= '3.8' and python_version < '4.0'",
+            "version": "==0.4.0"
+        },
+        "transformers": {
+            "extras": [
+                "sentencepiece"
+            ],
+            "hashes": [
+                "sha256:a8e1bafdaae69abdda3cad638fe392e37c86d2ce0ecfcae11d60abb8f949ff4d",
+                "sha256:f8ead7a5a4f6937bb507e66508e5e002dc5930f7b6122a9259c37b099d0f3b19"
+            ],
+            "markers": "python_full_version >= '3.9.0'",
+            "version": "==4.47.0"
+        },
         "typing-extensions": {
             "hashes": [
                 "sha256:04e5ca0351e0f3f85c6853954072df659d0d13fac324d0072316b67d7794700d",
             "markers": "platform_system != 'Darwin'",
             "version": "==6.0.0"
         },
+        "wcwidth": {
+            "hashes": [
+                "sha256:3da69048e4540d84af32131829ff948f1e022c1c6bdb8d6102117aac784f6859",
+                "sha256:72ea0c06399eb286d978fdedb6923a9eb47e1c486ce63e9b4e64fc18303972b5"
+            ],
+            "version": "==0.2.13"
+        },
+        "wikipedia-api": {
+            "hashes": [
+                "sha256:c72296adc0a6d8166d353a501f907add5db140179df462fc1cc0171c1c1bca00"
+            ],
+            "version": "==0.7.1"
+        },
+        "wrapt": {
+            "hashes": [
+                "sha256:0229b247b0fc7dee0d36176cbb79dbaf2a9eb7ecc50ec3121f40ef443155fb1d",
+                "sha256:0698d3a86f68abc894d537887b9bbf84d29bcfbc759e23f4644be27acf6da301",
+                "sha256:0a0a1a1ec28b641f2a3a2c35cbe86c00051c04fffcfcc577ffcdd707df3f8635",
+                "sha256:0b48554952f0f387984da81ccfa73b62e52817a4386d070c75e4db7d43a28c4a",
+                "sha256:0f2a28eb35cf99d5f5bd12f5dd44a0f41d206db226535b37b0c60e9da162c3ed",
+                "sha256:140ea00c87fafc42739bd74a94a5a9003f8e72c27c47cd4f61d8e05e6dec8721",
+                "sha256:16187aa2317c731170a88ef35e8937ae0f533c402872c1ee5e6d079fcf320801",
+                "sha256:17fcf043d0b4724858f25b8826c36e08f9fb2e475410bece0ec44a22d533da9b",
+                "sha256:18b956061b8db634120b58f668592a772e87e2e78bc1f6a906cfcaa0cc7991c1",
+                "sha256:2399408ac33ffd5b200480ee858baa58d77dd30e0dd0cab6a8a9547135f30a88",
+                "sha256:2a0c23b8319848426f305f9cb0c98a6e32ee68a36264f45948ccf8e7d2b941f8",
+                "sha256:2dfb7cff84e72e7bf975b06b4989477873dcf160b2fd89959c629535df53d4e0",
+                "sha256:2f495b6754358979379f84534f8dd7a43ff8cff2558dcdea4a148a6e713a758f",
+                "sha256:33539c6f5b96cf0b1105a0ff4cf5db9332e773bb521cc804a90e58dc49b10578",
+                "sha256:3c34f6896a01b84bab196f7119770fd8466c8ae3dfa73c59c0bb281e7b588ce7",
+                "sha256:498fec8da10e3e62edd1e7368f4b24aa362ac0ad931e678332d1b209aec93045",
+                "sha256:4d63f4d446e10ad19ed01188d6c1e1bb134cde8c18b0aa2acfd973d41fcc5ada",
+                "sha256:4e4b4385363de9052dac1a67bfb535c376f3d19c238b5f36bddc95efae15e12d",
+                "sha256:4e547b447073fc0dbfcbff15154c1be8823d10dab4ad401bdb1575e3fdedff1b",
+                "sha256:4f643df3d4419ea3f856c5c3f40fec1d65ea2e89ec812c83f7767c8730f9827a",
+                "sha256:4f763a29ee6a20c529496a20a7bcb16a73de27f5da6a843249c7047daf135977",
+                "sha256:5ae271862b2142f4bc687bdbfcc942e2473a89999a54231aa1c2c676e28f29ea",
+                "sha256:5d8fd17635b262448ab8f99230fe4dac991af1dabdbb92f7a70a6afac8a7e346",
+                "sha256:69c40d4655e078ede067a7095544bcec5a963566e17503e75a3a3e0fe2803b13",
+                "sha256:69d093792dc34a9c4c8a70e4973a3361c7a7578e9cd86961b2bbf38ca71e4e22",
+                "sha256:6a9653131bda68a1f029c52157fd81e11f07d485df55410401f745007bd6d339",
+                "sha256:6ff02a91c4fc9b6a94e1c9c20f62ea06a7e375f42fe57587f004d1078ac86ca9",
+                "sha256:714c12485aa52efbc0fc0ade1e9ab3a70343db82627f90f2ecbc898fdf0bb181",
+                "sha256:7264cbb4a18dc4acfd73b63e4bcfec9c9802614572025bdd44d0721983fc1d9c",
+                "sha256:73a96fd11d2b2e77d623a7f26e004cc31f131a365add1ce1ce9a19e55a1eef90",
+                "sha256:74bf625b1b4caaa7bad51d9003f8b07a468a704e0644a700e936c357c17dd45a",
+                "sha256:81b1289e99cf4bad07c23393ab447e5e96db0ab50974a280f7954b071d41b489",
+                "sha256:8425cfce27b8b20c9b89d77fb50e368d8306a90bf2b6eef2cdf5cd5083adf83f",
+                "sha256:875d240fdbdbe9e11f9831901fb8719da0bd4e6131f83aa9f69b96d18fae7504",
+                "sha256:879591c2b5ab0a7184258274c42a126b74a2c3d5a329df16d69f9cee07bba6ea",
+                "sha256:89fc28495896097622c3fc238915c79365dd0ede02f9a82ce436b13bd0ab7569",
+                "sha256:8a5e7cc39a45fc430af1aefc4d77ee6bad72c5bcdb1322cfde852c15192b8bd4",
+                "sha256:8f8909cdb9f1b237786c09a810e24ee5e15ef17019f7cecb207ce205b9b5fcce",
+                "sha256:914f66f3b6fc7b915d46c1cc424bc2441841083de01b90f9e81109c9759e43ab",
+                "sha256:92a3d214d5e53cb1db8b015f30d544bc9d3f7179a05feb8f16df713cecc2620a",
+                "sha256:948a9bd0fb2c5120457b07e59c8d7210cbc8703243225dbd78f4dfc13c8d2d1f",
+                "sha256:9c900108df470060174108012de06d45f514aa4ec21a191e7ab42988ff42a86c",
+                "sha256:9f2939cd4a2a52ca32bc0b359015718472d7f6de870760342e7ba295be9ebaf9",
+                "sha256:a4192b45dff127c7d69b3bdfb4d3e47b64179a0b9900b6351859f3001397dabf",
+                "sha256:a8fc931382e56627ec4acb01e09ce66e5c03c384ca52606111cee50d931a342d",
+                "sha256:ad47b095f0bdc5585bced35bd088cbfe4177236c7df9984b3cc46b391cc60627",
+                "sha256:b1ca5f060e205f72bec57faae5bd817a1560fcfc4af03f414b08fa29106b7e2d",
+                "sha256:ba1739fb38441a27a676f4de4123d3e858e494fac05868b7a281c0a383c098f4",
+                "sha256:baa7ef4e0886a6f482e00d1d5bcd37c201b383f1d314643dfb0367169f94f04c",
+                "sha256:bb90765dd91aed05b53cd7a87bd7f5c188fcd95960914bae0d32c5e7f899719d",
+                "sha256:bc7f729a72b16ee21795a943f85c6244971724819819a41ddbaeb691b2dd85ad",
+                "sha256:bdf62d25234290db1837875d4dceb2151e4ea7f9fff2ed41c0fde23ed542eb5b",
+                "sha256:c30970bdee1cad6a8da2044febd824ef6dc4cc0b19e39af3085c763fdec7de33",
+                "sha256:d2c63b93548eda58abf5188e505ffed0229bf675f7c3090f8e36ad55b8cbc371",
+                "sha256:d751300b94e35b6016d4b1e7d0e7bbc3b5e1751e2405ef908316c2a9024008a1",
+                "sha256:da427d311782324a376cacb47c1a4adc43f99fd9d996ffc1b3e8529c4074d393",
+                "sha256:daba396199399ccabafbfc509037ac635a6bc18510ad1add8fd16d4739cdd106",
+                "sha256:e185ec6060e301a7e5f8461c86fb3640a7beb1a0f0208ffde7a65ec4074931df",
+                "sha256:e4a557d97f12813dc5e18dad9fa765ae44ddd56a672bb5de4825527c847d6379",
+                "sha256:e5ed16d95fd142e9c72b6c10b06514ad30e846a0d0917ab406186541fe68b451",
+                "sha256:e711fc1acc7468463bc084d1b68561e40d1eaa135d8c509a65dd534403d83d7b",
+                "sha256:f28b29dc158ca5d6ac396c8e0a2ef45c4e97bb7e65522bfc04c989e6fe814575",
+                "sha256:f335579a1b485c834849e9075191c9898e0731af45705c2ebf70e0cd5d58beed",
+                "sha256:fce6fee67c318fdfb7f285c29a82d84782ae2579c0e1b385b7f36c6e8074fffb",
+                "sha256:fd136bb85f4568fffca995bd3c8d52080b1e5b225dbf1c2b17b66b4c5fa02838"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==1.17.0"
+        },
         "xyzservices": {
             "hashes": [
                 "sha256:68fb8353c9dbba4f1ff6c0f2e5e4e596bb9e1db7f94f4f7dfbcb26e25aa66fde",

__pycache__/alma_harvester.cpython-311.pyc ADDED Viewed

Binary file (3.2 kB). View file

__pycache__/image_analysis.cpython-311.pyc ADDED Viewed

Binary file (5.38 kB). View file

__pycache__/text_analysis.cpython-311.pyc ADDED Viewed

Binary file (2.33 kB). View file

alma_harvester.py ADDED Viewed

	@@ -0,0 +1,79 @@

+import os
+import requests
+from xml.etree import ElementTree as ET
+def get_alma_information(endpoint: str, params={}):
+    alma_url = "https://api-ap.hosted.exlibrisgroup.com"
+    api_key = os.environ.get("ALMA_API_KEY", "l8xx313574e38e5c48f2af9ed8c3fa017926")
+    print("Key", api_key)
+    headers = {
+        "Accept": "application/json",
+        "Content-Type": "application/json",
+        "Authorization": f"apikey {api_key}",
+    }
+    r = requests.get(f"{alma_url}{endpoint}", headers=headers, params=params)
+    if r.status_code != 200:
+        print(
+            f"Could not retrieve info from ALMA, status {r.status_code} return with the following text {r.text}"
+        )
+        return False
+    return r.json()
+def retrieve_bib(mms_id: str):
+    endpoint = f"/almaws/v1/bibs/{mms_id}"
+    mms_data = get_alma_information(endpoint)
+    return mms_data
+def extract_mms_id_metadata(mms_id):
+    try:
+        bib_data = retrieve_bib(mms_id)
+        mms_id = bib_data.get("mms_id")
+        title = bib_data.get("title")
+        anies = bib_data["anies"]
+        tree = ET.fromstring(anies[0])
+        p_950 = "n/a"
+        summary = "Could not find"
+        title_statement = "n/a"
+        for node in tree:
+            if node.attrib.get("tag") == "520":
+                for n in node:
+                    summary = n.text
+            if node.attrib.get("tag") == "245":
+                for n in node:
+                    if n.attrib.get("code") == "a":
+                        title_statement = n.text
+            if node.attrib.get("tag") == "950":
+                for n in node:
+                    if n.attrib.get("code") == "p":
+                        p_950 = n.text
+        mms_id_metadata = [title, title_statement, summary, p_950]
+        print(mms_id, mms_id_metadata)
+        return mms_id_metadata
+    except Exception as e:
+        print(f"Couldn't retrieve metadata for {mms_id}. here's the error {e}")
+        return ["", "", "", ""]

dataset_wrangler.py CHANGED Viewed

@@ -1,23 +1,14 @@
 import pandas as pd
-dataset = "https://raw.githubusercontent.com/StateLibraryVictoria/public-domain-hack-2024/refs/heads/main/datasets/challenge-3-Image-Pool-2024-11-27.csv"
-columns = [
-    "IE PID",
-    "Title (DC)",
-    "ALMA _ MMS (Object Identifier - IE)",
-    "HANDLE (Object Identifier - IE)",
-    "Creator (DC)",
-    "Genre (DCTERMS)",
-    "Created (DCTERMS)",
-]
-def clean_df(columns=columns, dataset=dataset):
     df = pd.read_csv(dataset)
-    df = df[columns]
     df = df.dropna()
     return df

 import pandas as pd
+dataset = "https://raw.githubusercontent.com/StateLibraryVictoria/public-domain-hack-2024/refs/heads/ch4-data-viz/datasets/ch3_colour_data_viz_suggestions_set_2_augmented.csv"
+def clean_df(columns=None, dataset=dataset):
     df = pd.read_csv(dataset)
+    if columns:
+        df = df[columns]
     df = df.dropna()
     return df

image_analysis.py CHANGED Viewed

@@ -9,21 +9,29 @@ from sklearn.cluster import KMeans
 def get_iiif_image_urls(ie_pid: str):
-    manifest_url = f"https://rosetta.slv.vic.gov.au/delivery/iiif/presentation/2.1/{ie_pid}/manifest"
-    print(manifest_url)
-    session = requests.Session()
-    response = session.get(manifest_url)
-    manifest = response.json()
-    image_ids = [
-        canvas["images"][0]["resource"]["service"]["@id"]
-        for canvas in manifest["sequences"][0]["canvases"]
-    ]
-    image_urls = [f"{image_id}/full/600,/0/default.jpg" for image_id in image_ids]
     return image_urls
@@ -57,21 +65,43 @@ def get_palette_clusters(img, no_of_clusters=5):
     return clusters
-image_urls = get_iiif_image_urls("IE1267294")
-response = requests.get(image_urls[0])
-img = cv.imdecode(np.frombuffer(response.content, np.uint8), -1)
-img = cv.cvtColor(img, cv.COLOR_BGR2RGB)
-dim = (500, 300)
-img = cv.resize(img, dim, interpolation=cv.INTER_AREA)
-clt_1 = get_palette_clusters(img)
-img_palette = palette(clt_1)
-print(img_palette)
-show_img_compare(img, img_palette)

 def get_iiif_image_urls(ie_pid: str):
+    try:
+        manifest_url = f"https://rosetta.slv.vic.gov.au/delivery/iiif/presentation/2.1/{ie_pid}/manifest"
+        session = requests.Session()
+        response = session.get(manifest_url)
+        manifest = response.json()
+        image_ids = [
+            canvas["images"][0]["resource"]["service"]["@id"]
+            for canvas in manifest["sequences"][0]["canvases"]
+        ]
+        image_urls = [f"{image_id}/full/600,/0/default.jpg" for image_id in image_ids]
+    except Exception as e:
+        print(f"Could not get iiif image URLs for {ie_pid}, here's the error {e}")
+        print(f"Manifest url {manifest_url}")
+        image_urls = []
     return image_urls
     return clusters
+def get_colour_palette_iiif_image(ie_pid: str, dim=(500, 300)):
+    # get iiif image urls
+    image_urls = get_iiif_image_urls(ie_pid)
+    if not image_urls:
+        return False
+    response = requests.get(image_urls[0])
+    # decode image
+    img = cv.imdecode(np.frombuffer(response.content, np.uint8), -1)
+    img = cv.cvtColor(img, cv.COLOR_BGR2RGB)
+    img = cv.resize(img, dim, interpolation=cv.INTER_AREA)
+    palette_clusters = get_palette_clusters(img)
+    img_palette = palette(palette_clusters)
+    return img_palette
+def get_palette_rgbs(ie_pid: str):
+    print("Getting palette RGBs for", ie_pid)
+    img_palette = get_colour_palette_iiif_image(ie_pid)
+    if isinstance(img_palette, bool):
+        return ""
+    palette_rgbs = np.unique(img_palette[0], axis=0)
+    palette_rgbs = palette_rgbs.tolist()
+    return palette_rgbs
+# print(get_palette_rgbs("IE1391238"))
+# img, palette = get_colour_palette_iiif_image("IE1423319")
+# img, palette = get_colour_palette_iiif_image("IE1391238")
+# show_img_compare(img, palette)

requirements.txt CHANGED Viewed

@@ -1,108 +1,103 @@
--i https://pypi.org/simple
-altair==5.5.0 ; python_version >= '3.9'
-attrs==24.2.0 ; python_version >= '3.7'
-blinker==1.9.0 ; python_version >= '3.9'
-cachetools==5.5.0 ; python_version >= '3.7'
-certifi==2024.8.30 ; python_version >= '3.6'
-charset-normalizer==3.4.0 ; python_full_version >= '3.7.0'
-click==8.1.7 ; python_version >= '3.7'
-colorama==0.4.6 ; platform_system == 'Windows'
-contourpy==1.3.1 ; python_version >= '3.10'
-cycler==0.12.1 ; python_version >= '3.8'
-fonttools==4.55.2 ; python_version >= '3.8'
-gitdb==4.0.11 ; python_version >= '3.7'
-gitpython==3.1.43 ; python_version >= '3.7'
-idna==3.10 ; python_version >= '3.6'
-jinja2==3.1.4 ; python_version >= '3.7'
-joblib==1.4.2 ; python_version >= '3.8'
-jsonschema==4.23.0 ; python_version >= '3.8'
-jsonschema-specifications==2024.10.1 ; python_version >= '3.9'
-kiwisolver==1.4.7 ; python_version >= '3.8'
-markdown-it-py==3.0.0 ; python_version >= '3.8'
-markupsafe==3.0.2 ; python_version >= '3.9'
-matplotlib==3.9.3
-mdurl==0.1.2 ; python_version >= '3.7'
-narwhals==1.16.0 ; python_version >= '3.8'
-numpy==2.2.0 ; python_version >= '3.10'
-opencv-python==4.10.0.84
-packaging==24.2 ; python_version >= '3.8'
-pandas==2.2.3
-pillow==11.0.0
-protobuf==5.29.1 ; python_version >= '3.8'
-pyarrow==18.1.0 ; python_version >= '3.9'
-pydeck==0.9.1 ; python_version >= '3.8'
-pygments==2.18.0 ; python_version >= '3.8'
-pyparsing==3.2.0 ; python_version >= '3.9'
-python-dateutil==2.9.0.post0 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'
-pytz==2024.2
-referencing==0.35.1 ; python_version >= '3.8'
-requests==2.32.3
-rich==13.9.4 ; python_full_version >= '3.8.0'
-rpds-py==0.22.3 ; python_version >= '3.9'
-scikit-learn==1.5.2
-scipy==1.14.1 ; python_version >= '3.10'
-six==1.17.0 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'
-smmap==5.0.1 ; python_version >= '3.7'
-streamlit==1.40.2
-tenacity==9.0.0 ; python_version >= '3.8'
-threadpoolctl==3.5.0 ; python_version >= '3.8'
-toml==0.10.2 ; python_version >= '2.6' and python_version not in '3.0, 3.1, 3.2, 3.3'
-tornado==6.4.2 ; python_version >= '3.8'
-typing-extensions==4.12.2 ; python_version >= '3.8'
-tzdata==2024.2 ; python_version >= '2'
-urllib3==2.2.3 ; python_version >= '3.8'
-watchdog==6.0.0 ; platform_system != 'Darwin'

+-i https://pypi.org/simple
+accelerate==1.2.1
+altair==5.5.0 ; python_version >= '3.9'
+attrs==24.3.0 ; python_version >= '3.8'
+beautifulsoup4==4.12.3 ; python_full_version >= '3.6.0'
+bioc==2.1 ; python_version >= '3.6'
+blinker==1.9.0 ; python_version >= '3.9'
+bokeh==3.6.2
+boto3==1.35.82 ; python_version >= '3.8'
+botocore==1.35.82 ; python_version >= '3.8'
+cachetools==5.5.0 ; python_version >= '3.7'
+certifi==2024.12.14 ; python_version >= '3.6'
+charset-normalizer==3.4.0 ; python_full_version >= '3.7.0'
+click==8.1.7 ; python_version >= '3.7'
+colorama==0.4.6 ; platform_system == 'Windows'
+conllu==4.5.3 ; python_version >= '3.6'
+contourpy==1.3.1 ; python_version >= '3.10'
+cycler==0.12.1 ; python_version >= '3.8'
+deprecated==1.2.15 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'
+docopt==0.6.2
+filelock==3.16.1 ; python_version >= '3.8'
+flair==0.14.0
+fonttools==4.55.3 ; python_version >= '3.8'
+fsspec==2024.10.0 ; python_version >= '3.8'
+ftfy==6.3.1 ; python_version >= '3.9'
+gdown==5.2.0 ; python_version >= '3.8'
+gitdb==4.0.11 ; python_version >= '3.7'
+gitpython==3.1.43 ; python_version >= '3.7'
+huggingface-hub==0.27.0 ; python_full_version >= '3.8.0'
+idna==3.10 ; python_version >= '3.6'
+intervaltree==3.1.0
+jinja2==3.1.4 ; python_version >= '3.7'
+jmespath==1.0.1 ; python_version >= '3.7'
+joblib==1.4.2 ; python_version >= '3.8'
+jsonlines==4.0.0 ; python_version >= '3.8'
+jsonschema==4.23.0 ; python_version >= '3.8'
+jsonschema-specifications==2024.10.1 ; python_version >= '3.9'
+kiwisolver==1.4.7 ; python_version >= '3.8'
+langdetect==1.0.9
+lxml==5.3.0 ; python_version >= '3.6'
+markdown-it-py==3.0.0 ; python_version >= '3.8'
+markupsafe==3.0.2 ; python_version >= '3.9'
+matplotlib==3.10.0
+mdurl==0.1.2 ; python_version >= '3.7'
+more-itertools==10.5.0 ; python_version >= '3.8'
+mpld3==0.5.10
+mpmath==1.3.0
+narwhals==1.18.4 ; python_version >= '3.8'
+networkx==3.4.2 ; python_version >= '3.10'
+numpy==1.26.4 ; python_version >= '3.9'
+opencv-python==4.10.0.84
+packaging==24.2 ; python_version >= '3.8'
+pandas==2.2.3
+pillow==11.0.0
+pptree==3.1
+protobuf==5.29.1 ; python_version >= '3.8'
+psutil==6.1.0 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4, 3.5'
+pyarrow==18.1.0 ; python_version >= '3.9'
+pydeck==0.9.1 ; python_version >= '3.8'
+pygments==2.18.0 ; python_version >= '3.8'
+pyparsing==3.2.0 ; python_version >= '3.9'
+pysocks==1.7.1
+python-dateutil==2.9.0.post0 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'
+pytorch-revgrad==0.2.0 ; python_version >= '3.5'
+pytz==2024.2
+pyyaml==6.0.2 ; python_version >= '3.8'
+referencing==0.35.1 ; python_version >= '3.8'
+regex==2024.11.6 ; python_version >= '3.8'
+requests==2.32.3
+rich==13.9.4 ; python_full_version >= '3.8.0'
+rpds-py==0.22.3 ; python_version >= '3.9'
+s3transfer==0.10.4 ; python_version >= '3.8'
+safetensors==0.4.5 ; python_version >= '3.7'
+scikit-learn==1.6.0
+scipy==1.14.1 ; python_version >= '3.10'
+segtok==1.5.11
+semver==3.0.2 ; python_version >= '3.7'
+sentencepiece==0.2.0
+six==1.17.0 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'
+smmap==5.0.1 ; python_version >= '3.7'
+sortedcontainers==2.4.0
+soupsieve==2.6 ; python_version >= '3.8'
+sqlitedict==2.1.0
+streamlit==1.41.1
+sympy==1.13.1 ; python_version >= '3.9'
+tabulate==0.9.0 ; python_version >= '3.7'
+tenacity==9.0.0 ; python_version >= '3.8'
+threadpoolctl==3.5.0 ; python_version >= '3.8'
+tokenizers==0.21.0 ; python_version >= '3.7'
+toml==0.10.2 ; python_version >= '2.6' and python_version not in '3.0, 3.1, 3.2, 3.3'
+torch==2.5.1 ; python_full_version >= '3.8.0'
+tornado==6.4.2 ; python_version >= '3.8'
+tqdm==4.67.1 ; python_version >= '3.7'
+transformer-smaller-training-vocab==0.4.0 ; python_version >= '3.8' and python_version < '4.0'
+transformers[sentencepiece]==4.47.0 ; python_full_version >= '3.9.0'
+typing-extensions==4.12.2 ; python_version >= '3.8'
+tzdata==2024.2 ; python_version >= '2'
+urllib3==2.2.3 ; python_version >= '3.8'
+watchdog==6.0.0 ; platform_system != 'Darwin'
+wcwidth==0.2.13
+wikipedia-api==0.7.1
+wrapt==1.17.0 ; python_version >= '3.8'
+xyzservices==2024.9.0 ; python_version >= '3.8'

text_analysis.py ADDED Viewed

	@@ -0,0 +1,73 @@

+from flair.nn import Classifier
+from flair.data import Sentence
+from flair.models import SequenceTagger
+from flair.tokenization import Tokenizer
+from segtok.segmenter import split_single
+from collections import Counter
+import pandas as pd
+import os
+ner_tagger = SequenceTagger.load("flair/ner-english-ontonotes")
+pos_tagger = Classifier.load("pos")
+def get_named_entities(text: str, tagger=ner_tagger):
+    sentence = [Sentence(sent, use_tokenizer=True) for sent in split_single(text)]
+    tagger.predict(sentence)
+    entities = []
+    for token in sentence:
+        for entity in token.get_spans("ner"):
+            entity = str(entity)
+            entities.append(entity)
+    return entities
+def get_most_frequent_words(dataset: str, k=10):
+    split_str = dataset.split()
+    counter = Counter(split_str)
+    most_frequent = counter.most_common(k)
+    return most_frequent
+# POS categories https://huggingface.co/flair/pos-english
+def get_parts_of_sentence(text: str, tagger=pos_tagger):
+    sentence = Sentence(text)
+    tagger.predict(sentence)
+    return sentence
+# path_stem = os.path.join("datasets")
+# file_name = "ch3_colour_data_viz_suggestions_set_2"
+# ner_output_path = os.path.join(path_stem, f"{file_name}_ner.csv")
+# df = pd.read_csv(ner_output_path)
+# df = df.head(3)
+# ner_dataset = df["alma_metadata"].to_list()
+# ner_dataset = " ".join(ner_dataset)
+# tokenizer = Tokenizer()
+# tokens = Tokenizer.tokenize(Tokenizer, ner_dataset)
+# print(tokens)
+# # most_common = get_most_frequent_words(ner_dataset)
+# print(most_common)