Task-290 Integrate use of models within the application for use in inference

Pipfile

@@ -8,6 +8,8 @@ streamlit = "*"
 pandas = "*"
 numpy = "*"
 st-annotated-text = "*"
+transformers = "*"
+torch = "*"
 
 [dev-packages]
 

Pipfile.lock

@@ -1,7 +1,7 @@
 {
     "_meta": {
         "hash": {
-            "sha256": "b114c11edb911ee8c918ef91522f2a4e6895019b452b37a052da3b4deb4c456e"
+            "sha256": "d44f8f17557914a1bc97b5e9ce219979a85e81b74eb603b3c0c6920cac065c91"
         },
         "pipfile-spec": 6,
         "requires": {
@@ -170,6 +170,22 @@
             "markers": "platform_system == 'Windows'",
             "version": "==0.4.6"
         },
+        "filelock": {
+            "hashes": [
+                "sha256:533dc2f7ba78dc2f0f531fc6c4940addf7b70a481e269a5a3b93be94ffbe8338",
+                "sha256:ee4e77401ef576ebb38cd7f13b9b28893194acc20a8e68e18730ba9c0e54660e"
+            ],
+            "markers": "python_version >= '3.9'",
+            "version": "==3.17.0"
+        },
+        "fsspec": {
+            "hashes": [
+                "sha256:1c24b16eaa0a1798afa0337aa0db9b256718ab2a89c425371f5628d22c3b6afd",
+                "sha256:9de2ad9ce1f85e1931858535bc882543171d197001a0a5eb2ddc04f1781ab95b"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==2025.2.0"
+        },
         "gitdb": {
             "hashes": [
                 "sha256:5ef71f855d191a3326fcfbc0d5da835f26b13fbcba60c32c21091c349ffdb571",
@@ -193,6 +209,14 @@
             "markers": "python_version >= '3.7'",
             "version": "==0.9.0"
         },
+        "huggingface-hub": {
+            "hashes": [
+                "sha256:352f69caf16566c7b6de84b54a822f6238e17ddd8ae3da4f8f2272aea5b198d5",
+                "sha256:9524eae42077b8ff4fc459ceb7a514eca1c1232b775276b009709fe2a084f250"
+            ],
+            "markers": "python_full_version >= '3.8.0'",
+            "version": "==0.29.1"
+        },
         "idna": {
             "hashes": [
                 "sha256:12f65c9b470abda6dc35cf8e63cc574b1c52b11df2c86030af0ac09b01b13ea9",
@@ -308,13 +332,28 @@
             "markers": "python_version >= '3.7'",
             "version": "==0.1.2"
         },
+        "mpmath": {
+            "hashes": [
+                "sha256:7a28eb2a9774d00c7bc92411c19a89209d5da7c4c9a9e227be8330a23a25b91f",
+                "sha256:a0b2b9fe80bbcd81a6647ff13108738cfb482d481d826cc0e02f5b35e5c88d2c"
+            ],
+            "version": "==1.3.0"
+        },
         "narwhals": {
             "hashes": [
-                "sha256:45d909ad6240944d447b0dae38074c5a919830dff3868d57b05a5526c1f06fe4",
-                "sha256:a2213fa44a039f724278fb15609889319e7c240403413f2606cc856c8d8f708d"
+                "sha256:1021c345d56c66ff0cc8e6d03ca8c543d01ffc411630973a5cb69ee86824d823",
+                "sha256:653aa8e5eb435816e7b50c8def17e7e5e3324c2ffd8a3eec03fef85792e9cf5e"
             ],
             "markers": "python_version >= '3.8'",
-            "version": "==1.28.0"
+            "version": "==1.29.0"
+        },
+        "networkx": {
+            "hashes": [
+                "sha256:307c3669428c5362aab27c8a1260aa8f47c4e91d3891f48be0141738d8d053e1",
+                "sha256:df5d4365b724cf81b8c6a7312509d0c22386097011ad1abe274afd5e9d3bbc5f"
+            ],
+            "markers": "python_version >= '3.10'",
+            "version": "==3.4.2"
         },
         "numpy": {
             "hashes": [
@@ -608,6 +647,65 @@
             ],
             "version": "==2025.1"
         },
+        "pyyaml": {
+            "hashes": [
+                "sha256:01179a4a8559ab5de078078f37e5c1a30d76bb88519906844fd7bdea1b7729ff",
+                "sha256:0833f8694549e586547b576dcfaba4a6b55b9e96098b36cdc7ebefe667dfed48",
+                "sha256:0a9a2848a5b7feac301353437eb7d5957887edbf81d56e903999a75a3d743086",
+                "sha256:0b69e4ce7a131fe56b7e4d770c67429700908fc0752af059838b1cfb41960e4e",
+                "sha256:0ffe8360bab4910ef1b9e87fb812d8bc0a308b0d0eef8c8f44e0254ab3b07133",
+                "sha256:11d8f3dd2b9c1207dcaf2ee0bbbfd5991f571186ec9cc78427ba5bd32afae4b5",
+                "sha256:17e311b6c678207928d649faa7cb0d7b4c26a0ba73d41e99c4fff6b6c3276484",
+                "sha256:1e2120ef853f59c7419231f3bf4e7021f1b936f6ebd222406c3b60212205d2ee",
+                "sha256:1f71ea527786de97d1a0cc0eacd1defc0985dcf6b3f17bb77dcfc8c34bec4dc5",
+                "sha256:23502f431948090f597378482b4812b0caae32c22213aecf3b55325e049a6c68",
+                "sha256:24471b829b3bf607e04e88d79542a9d48bb037c2267d7927a874e6c205ca7e9a",
+                "sha256:29717114e51c84ddfba879543fb232a6ed60086602313ca38cce623c1d62cfbf",
+                "sha256:2e99c6826ffa974fe6e27cdb5ed0021786b03fc98e5ee3c5bfe1fd5015f42b99",
+                "sha256:39693e1f8320ae4f43943590b49779ffb98acb81f788220ea932a6b6c51004d8",
+                "sha256:3ad2a3decf9aaba3d29c8f537ac4b243e36bef957511b4766cb0057d32b0be85",
+                "sha256:3b1fdb9dc17f5a7677423d508ab4f243a726dea51fa5e70992e59a7411c89d19",
+                "sha256:41e4e3953a79407c794916fa277a82531dd93aad34e29c2a514c2c0c5fe971cc",
+                "sha256:43fa96a3ca0d6b1812e01ced1044a003533c47f6ee8aca31724f78e93ccc089a",
+                "sha256:50187695423ffe49e2deacb8cd10510bc361faac997de9efef88badc3bb9e2d1",
+                "sha256:5ac9328ec4831237bec75defaf839f7d4564be1e6b25ac710bd1a96321cc8317",
+                "sha256:5d225db5a45f21e78dd9358e58a98702a0302f2659a3c6cd320564b75b86f47c",
+                "sha256:6395c297d42274772abc367baaa79683958044e5d3835486c16da75d2a694631",
+                "sha256:688ba32a1cffef67fd2e9398a2efebaea461578b0923624778664cc1c914db5d",
+                "sha256:68ccc6023a3400877818152ad9a1033e3db8625d899c72eacb5a668902e4d652",
+                "sha256:70b189594dbe54f75ab3a1acec5f1e3faa7e8cf2f1e08d9b561cb41b845f69d5",
+                "sha256:797b4f722ffa07cc8d62053e4cff1486fa6dc094105d13fea7b1de7d8bf71c9e",
+                "sha256:7c36280e6fb8385e520936c3cb3b8042851904eba0e58d277dca80a5cfed590b",
+                "sha256:7e7401d0de89a9a855c839bc697c079a4af81cf878373abd7dc625847d25cbd8",
+                "sha256:80bab7bfc629882493af4aa31a4cfa43a4c57c83813253626916b8c7ada83476",
+                "sha256:82d09873e40955485746739bcb8b4586983670466c23382c19cffecbf1fd8706",
+                "sha256:8388ee1976c416731879ac16da0aff3f63b286ffdd57cdeb95f3f2e085687563",
+                "sha256:8824b5a04a04a047e72eea5cec3bc266db09e35de6bdfe34c9436ac5ee27d237",
+                "sha256:8b9c7197f7cb2738065c481a0461e50ad02f18c78cd75775628afb4d7137fb3b",
+                "sha256:9056c1ecd25795207ad294bcf39f2db3d845767be0ea6e6a34d856f006006083",
+                "sha256:936d68689298c36b53b29f23c6dbb74de12b4ac12ca6cfe0e047bedceea56180",
+                "sha256:9b22676e8097e9e22e36d6b7bda33190d0d400f345f23d4065d48f4ca7ae0425",
+                "sha256:a4d3091415f010369ae4ed1fc6b79def9416358877534caf6a0fdd2146c87a3e",
+                "sha256:a8786accb172bd8afb8be14490a16625cbc387036876ab6ba70912730faf8e1f",
+                "sha256:a9f8c2e67970f13b16084e04f134610fd1d374bf477b17ec1599185cf611d725",
+                "sha256:bc2fa7c6b47d6bc618dd7fb02ef6fdedb1090ec036abab80d4681424b84c1183",
+                "sha256:c70c95198c015b85feafc136515252a261a84561b7b1d51e3384e0655ddf25ab",
+                "sha256:cc1c1159b3d456576af7a3e4d1ba7e6924cb39de8f67111c735f6fc832082774",
+                "sha256:ce826d6ef20b1bc864f0a68340c8b3287705cae2f8b4b1d932177dcc76721725",
+                "sha256:d584d9ec91ad65861cc08d42e834324ef890a082e591037abe114850ff7bbc3e",
+                "sha256:d7fded462629cfa4b685c5416b949ebad6cec74af5e2d42905d41e257e0869f5",
+                "sha256:d84a1718ee396f54f3a086ea0a66d8e552b2ab2017ef8b420e92edbc841c352d",
+                "sha256:d8e03406cac8513435335dbab54c0d385e4a49e4945d2909a581c83647ca0290",
+                "sha256:e10ce637b18caea04431ce14fabcf5c64a1c61ec9c56b071a4b7ca131ca52d44",
+                "sha256:ec031d5d2feb36d1d1a24380e4db6d43695f3748343d99434e6f5f9156aaa2ed",
+                "sha256:ef6107725bd54b262d6dedcc2af448a266975032bc85ef0172c5f059da6325b4",
+                "sha256:efdca5630322a10774e8e98e1af481aad470dd62c3170801852d752aa7a783ba",
+                "sha256:f753120cb8181e736c57ef7636e83f31b9c0d1722c516f7e86cf15b7aa57ff12",
+                "sha256:ff3824dc5261f50c9b0dfb3be22b4567a6f938ccce4587b38952d85fd9e9afe4"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==6.0.2"
+        },
         "referencing": {
             "hashes": [
                 "sha256:df2e89862cd09deabbdba16944cc3f10feb6b3e6f18e902f7cc25609a34775aa",
@@ -616,6 +714,14 @@
             "markers": "python_version >= '3.9'",
             "version": "==0.36.2"
         },
+        "regex": {
+            "hashes": [
+                "sha256:7ab159b063c52a0333c884e4679f8d7a85112ee3078fe3d9004b2dd875585519",
+                "sha256:a93c194e2df18f7d264092dc8539b8ffb86b45b899ab976aa15d48214138e81b"
+            ],
+            "markers": "python_version >= '3.8'",
+            "version": "==2024.11.6"
+        },
         "requests": {
             "hashes": [
                 "sha256:55365417734eb18255590a9ff9eb97e9e1da868d4ccd6402399eaf68af20a760",
@@ -741,6 +847,35 @@
             "markers": "python_version >= '3.9'",
             "version": "==0.23.1"
         },
+        "safetensors": {
+            "hashes": [
+                "sha256:1077f3e94182d72618357b04b5ced540ceb71c8a813d3319f1aba448e68a770d",
+                "sha256:11bce6164887cd491ca75c2326a113ba934be596e22b28b1742ce27b1d076467",
+                "sha256:21d01c14ff6c415c485616b8b0bf961c46b3b343ca59110d38d744e577f9cce7",
+                "sha256:32c3ef2d7af8b9f52ff685ed0bc43913cdcde135089ae322ee576de93eae5135",
+                "sha256:37f1521be045e56fc2b54c606d4455573e717b2d887c579ee1dbba5f868ece04",
+                "sha256:391ac8cab7c829452175f871fcaf414aa1e292b5448bd02620f675a7f3e7abb9",
+                "sha256:4a243be3590bc3301c821da7a18d87224ef35cbd3e5f5727e4e0728b8172411e",
+                "sha256:799021e78287bac619c7b3f3606730a22da4cda27759ddf55d37c8db7511c74b",
+                "sha256:836cbbc320b47e80acd40e44c8682db0e8ad7123209f69b093def21ec7cafd11",
+                "sha256:8bd84b12b1670a6f8e50f01e28156422a2bc07fb16fc4e98bded13039d688a0d",
+                "sha256:b6b0d6ecacec39a4fdd99cc19f4576f5219ce858e6fd8dbe7609df0b8dc56965",
+                "sha256:bd20eb133db8ed15b40110b7c00c6df51655a2998132193de2f75f72d99c7073",
+                "sha256:cead1fa41fc54b1e61089fa57452e8834f798cb1dc7a09ba3524f1eb08e0317a",
+                "sha256:cfc0ec0846dcf6763b0ed3d1846ff36008c6e7290683b61616c4b040f6a54ace",
+                "sha256:df26da01aaac504334644e1b7642fa000bfec820e7cef83aeac4e355e03195ff"
+            ],
+            "markers": "python_version >= '3.7'",
+            "version": "==0.5.3"
+        },
+        "setuptools": {
+            "hashes": [
+                "sha256:4880473a969e5f23f2a2be3646b2dfd84af9028716d398e46192f84bc36900d2",
+                "sha256:558e47c15f1811c1fa7adbd0096669bf76c1d3f433f58324df69f3f5ecac4e8f"
+            ],
+            "markers": "python_version >= '3.12'",
+            "version": "==75.8.2"
+        },
         "six": {
             "hashes": [
                 "sha256:4721f391ed90541fddacab5acf947aa0d3dc7d27b2e1e8eda2be8970586c3274",
@@ -775,6 +910,14 @@
             "markers": "python_version >= '3.9' and python_full_version != '3.9.7'",
             "version": "==1.42.2"
         },
+        "sympy": {
+            "hashes": [
+                "sha256:9cebf7e04ff162015ce31c9c6c9144daa34a93bd082f54fd8f12deca4f47515f",
+                "sha256:db36cdc64bf61b9b24578b6f7bab1ecdd2452cf008f34faa33776680c26d66f8"
+            ],
+            "markers": "python_version >= '3.9'",
+            "version": "==1.13.1"
+        },
         "tenacity": {
             "hashes": [
                 "sha256:807f37ca97d62aa361264d497b0e31e92b8027044942bfa756160d908320d73b",
@@ -783,6 +926,27 @@
             "markers": "python_version >= '3.8'",
             "version": "==9.0.0"
         },
+        "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",
@@ -791,6 +955,33 @@
             "markers": "python_version >= '2.6' and python_version not in '3.0, 3.1, 3.2'",
             "version": "==0.10.2"
         },
+        "torch": {
+            "hashes": [
+                "sha256:09e06f9949e1a0518c5b09fe95295bc9661f219d9ecb6f9893e5123e10696628",
+                "sha256:265f70de5fd45b864d924b64be1797f86e76c8e48a02c2a3a6fc7ec247d2226c",
+                "sha256:2bb8987f3bb1ef2675897034402373ddfc8f5ef0e156e2d8cfc47cacafdda4a9",
+                "sha256:46763dcb051180ce1ed23d1891d9b1598e07d051ce4c9d14307029809c4d64f7",
+                "sha256:4874a73507a300a5d089ceaff616a569e7bb7c613c56f37f63ec3ffac65259cf",
+                "sha256:510c73251bee9ba02ae1cb6c9d4ee0907b3ce6020e62784e2d7598e0cfa4d6cc",
+                "sha256:56eeaf2ecac90da5d9e35f7f35eb286da82673ec3c582e310a8d1631a1c02341",
+                "sha256:683410f97984103148e31b38a8631acf31c3034c020c0f4d26171e7626d8317a",
+                "sha256:6860df13d9911ac158f4c44031609700e1eba07916fff62e21e6ffa0a9e01961",
+                "sha256:7979834102cd5b7a43cc64e87f2f3b14bd0e1458f06e9f88ffa386d07c7446e1",
+                "sha256:7e1448426d0ba3620408218b50aa6ada88aeae34f7a239ba5431f6c8774b1239",
+                "sha256:94fc63b3b4bedd327af588696559f68c264440e2503cc9e6954019473d74ae21",
+                "sha256:9a610afe216a85a8b9bc9f8365ed561535c93e804c2a317ef7fabcc5deda0989",
+                "sha256:9ea955317cfcd3852b1402b62af258ce735c2edeee42ca9419b6bc889e5ae053",
+                "sha256:a0d5e1b9874c1a6c25556840ab8920569a7a4137afa8a63a32cee0bc7d89bd4b",
+                "sha256:b789069020c5588c70d5c2158ac0aa23fd24a028f34a8b4fcb8fcb4d7efcf5fb",
+                "sha256:bb2c6c3e65049f081940f5ab15c9136c7de40d3f01192541c920a07c7c585b7e",
+                "sha256:c4f103a49830ce4c7561ef4434cc7926e5a5fe4e5eb100c19ab36ea1e2b634ab",
+                "sha256:ccbd0320411fe1a3b3fec7b4d3185aa7d0c52adac94480ab024b5c8f74a0bf1d",
+                "sha256:ff96f4038f8af9f7ec4231710ed4549da1bdebad95923953a25045dcf6fd87e2"
+            ],
+            "index": "pypi",
+            "markers": "python_full_version >= '3.9.0'",
+            "version": "==2.6.0"
+        },
         "tornado": {
             "hashes": [
                 "sha256:072ce12ada169c5b00b7d92a99ba089447ccc993ea2143c9ede887e0937aa803",
@@ -808,6 +999,23 @@
             "markers": "python_version >= '3.8'",
             "version": "==6.4.2"
         },
+        "tqdm": {
+            "hashes": [
+                "sha256:26445eca388f82e72884e0d580d5464cd801a3ea01e63e5601bdff9ba6a48de2",
+                "sha256:f8aef9c52c08c13a65f30ea34f4e5aac3fd1a34959879d7e59e63027286627f2"
+            ],
+            "markers": "python_version >= '3.7'",
+            "version": "==4.67.1"
+        },
+        "transformers": {
+            "hashes": [
+                "sha256:6b4fded1c5fee04d384b1014495b4235a2b53c87503d7d592423c06128cbbe03",
+                "sha256:7e40e640b5b8dc3f48743f5f5adbdce3660c82baafbd3afdfc04143cdbd2089e"
+            ],
+            "index": "pypi",
+            "markers": "python_full_version >= '3.9.0'",
+            "version": "==4.49.0"
+        },
         "typing-extensions": {
             "hashes": [
                 "sha256:04e5ca0351e0f3f85c6853954072df659d0d13fac324d0072316b67d7794700d",

app.py

@@ -3,8 +3,17 @@ import pandas as pd
 from annotated_text import annotated_text, annotation
 import time
 from random import randint, uniform
+from scripts.predict import InferenceHandler
+from pathlib import Path
+
+ROOT = Path(__file__).resolve().parents[0]
+st.write(ROOT)
+MODELS_DIR = ROOT / 'models'
+BIN_MODEL_PATH = MODELS_DIR / 'binary_classification'
+ML_MODEL_PATH = MODELS_DIR / 'multilabel_regression'
 
 history_df = pd.DataFrame(data=[], columns=['Text', 'Classification', 'Gender', 'Race', 'Sexuality', 'Disability', 'Religion', 'Unspecified'])
+ih = InferenceHandler(BIN_MODEL_PATH, ML_MODEL_PATH)
 
 def extract_data(json_obj):
     row_data = []
@@ -40,8 +49,8 @@ def output_results(res):
             if res['numerical_sentiment'] == 1:
                 # st.markdown('##### Category Results:')
                 for entry in res['category_sentiments'].keys():
-                    if randint(0, 1) == 1:
-                        val = res['category_sentiments'][entry]
+                    val = res['category_sentiments'][entry]
+                    if val > 0.0:
                         perc = val * 100
                         at_list.append((entry, f'{perc:.2f}%', label_dict[entry]))
 
@@ -49,28 +58,25 @@ def output_results(res):
             st.markdown(f"#### Classification - {':red' if res['numerical_sentiment'] == 1 else ':green'}[{res['text_sentiment']}]")
 
             if len(at_list) > 0:
-                st.markdown('#### Categories: ')
-                cols = st.columns([1, 15])
-                with cols[1]:
-                    for cat in at_list:
-                        annotated_text(cat)
-
-def test_results(text):
-    test_val = int(randint(0, 1))
-    res_obj = {
-            'raw_text': text,
-            'text_sentiment': 'Discriminatory' if test_val == 1 else 'Non-Discriminatory',
-            'numerical_sentiment': test_val,
-            'category_sentiments': {
-                'Gender': None if test_val == 0 else uniform(0.0, 1.0),
-                'Race': None if test_val == 0 else uniform(0.0, 1.0),
-                'Sexuality': None if test_val == 0 else uniform(0.0, 1.0),  
-                'Disability': None if test_val == 0 else uniform(0.0, 1.0),
-                'Religion': None if test_val == 0 else uniform(0.0, 1.0),  
-                'Unspecified': None if test_val == 0 else uniform(0.0, 1.0)
-            }
-        }
-    return res_obj
+                annotated_text(at_list)
+                        
+
+# def test_results(text):
+#     test_val = int(randint(0, 1))
+#     res_obj = {
+#             'raw_text': text,
+#             'text_sentiment': 'Discriminatory' if test_val == 1 else 'Non-Discriminatory',
+#             'numerical_sentiment': test_val,
+#             'category_sentiments': {
+#                 'Gender': None if test_val == 0 else uniform(0.0, 1.0),
+#                 'Race': None if test_val == 0 else uniform(0.0, 1.0),
+#                 'Sexuality': None if test_val == 0 else uniform(0.0, 1.0),  
+#                 'Disability': None if test_val == 0 else uniform(0.0, 1.0),
+#                 'Religion': None if test_val == 0 else uniform(0.0, 1.0),  
+#                 'Unspecified': None if test_val == 0 else uniform(0.0, 1.0)
+#             }
+#         }
+#     return res_obj
 
 
 def analyze_text(text):
@@ -78,7 +84,7 @@ def analyze_text(text):
     with rc:
         with st.spinner("Processing...", show_time=True) as spnr:
             time.sleep(5)
-            res = test_results(text)
+            res = ih.classify_text(text)
             del spnr
 
     if res is not None:

scripts/predict.py

@@ -0,0 +1,105 @@
+"""
+Script file used for performing inference with an existing model.
+"""
+
+from pathlib import Path
+import torch
+import json
+
+from transformers import (
+    AutoTokenizer,
+    AutoModelForSequenceClassification
+)
+
+
+## Class used to encapsulate and handle the logic for inference
+class InferenceHandler:
+    def __init__(self, bin_model_path: Path, ml_regr_model_path: Path):
+        self.bin_tokenizer, self.bin_model = self.init_model_and_tokenizer(bin_model_path)
+        self.ml_regr_tokenizer, self.ml_regr_model = self.init_model_and_tokenizer(ml_regr_model_path)
+
+    ## Initializes a model and tokenizer for use in inference using the models path
+    def init_model_and_tokenizer(self, model_path: Path):
+        with open(model_path / 'config.json') as config_file:
+            config_json = json.load(config_file)
+        model_name = config_json['_name_or_path']
+        model_type = config_json['model_type']
+
+        tokenizer = AutoTokenizer.from_pretrained(model_name)
+        model = AutoModelForSequenceClassification.from_pretrained(model_path, model_type=model_type)
+        model.eval()
+
+        return tokenizer, model
+
+    ## Handles logic used to encode the text for use in binary classification
+    def encode_binary(self, text):
+        bin_tokenized_input = self.bin_tokenizer(text, return_tensors="pt", truncation=True, padding=True, max_length=512)
+        return bin_tokenized_input
+
+    ## Handles logic used to encode the text for use in multilabel regression
+    def encode_multilabel(self, text):
+        ml_tokenized_input = self.ml_regr_tokenizer(text, return_tensors="pt", truncation=True, padding=True, max_length=512)
+        return ml_tokenized_input
+
+    ## Handles text encoding for both binary classification and multilabel regression
+    def encode_input(self, text):
+        bin_inputs = self.encode_binary(text)
+        ml_inputs = self.encode_multilabel(text)
+        return bin_inputs, ml_inputs
+    
+    ## Handles performing the full sentiment analysis (binary classification and multilabel regression)
+    def classify_text(self, text):
+        res_obj = {
+            'raw_text': text,
+            'text_sentiment': None,
+            'numerical_sentiment': None,
+            'category_sentiments': {
+                'Gender': None,
+                'Race': None,
+                'Sexuality': None,  
+                'Disability': None,
+                'Religion': None,  
+                'Unspecified': None
+            }
+        }
+
+        text_prediction, pred_class = self.discriminatory_inference(text)
+        res_obj['text_sentiment'] = text_prediction
+        res_obj['numerical_sentiment'] = pred_class
+
+        if pred_class == 1:
+            ml_infer_results = self.category_inference(text)
+
+            for idx, key in enumerate(res_obj['category_sentiments'].keys()):
+                res_obj['category_sentiments'][key] = ml_infer_results[idx]
+
+        return res_obj
+    
+    ## Handles logic for checking the binary classfication of the text 
+    def discriminatory_inference(self, text):
+        bin_inputs = self.encode_binary(text)
+
+        with torch.no_grad():
+            bin_logits = self.bin_model(**bin_inputs).logits
+
+        probs = torch.nn.functional.softmax(bin_logits, dim=-1)
+        pred_class = torch.argmax(probs).item()
+        bin_label_map = {0: "Non-Discriminatory", 1: "Discriminatory"}
+        bin_text_pred = bin_label_map[pred_class]
+
+        return bin_text_pred, pred_class
+    
+    ## Handles logic for assessing the categories of discrimination
+    def category_inference(self, text):
+        ml_inputs = self.encode_multilabel(text)
+
+        with torch.no_grad():
+            ml_outputs = self.ml_regr_model(**ml_inputs).logits
+        
+        ml_op_list = ml_outputs.squeeze().tolist()
+
+        results = []
+        for item in ml_op_list:
+            results.append(max(0.0, item))
+
+        return results