Upload folder using huggingface_hub

README.md

@@ -17,7 +17,7 @@ Fine-tuned **DistilBERT** for text classification on a classmate's HW1 dataset.
 
 - Dataset: `ddecosmo/hw_text_dataset`
 - Text column: `Text`
-- Label column: `label` (classes: ['alumni house', 'armstrong hall', 'art museum', 'assumption hall', 'baker hall', 'barrett honors hall', 'bertrand library', 'brickyard', 'bucknell golf club', 'bucknell pickleball', 'cemetery', 'cereus hall', 'christy mathewson stadium', 'cmu tennis courts', 'college of fine art', 'cooper fieldhouse', 'coor', 'cyert hall', 'dept. of visual arts', 'design school north/south', 'dfa', 'donner house', 'douglas hall', 'duq med school', 'duq union', 'durham', 'emmitt field', 'engineering center', "farmer's education building", 'frankling antonio hall', 'galbraith', 'gallery 100', 'gammage', 'gates hillman', 'geisel library', 'gesling stadium', 'gumberg library', 'hall of the arts', 'hamerschlag', 'hassayampa academic village', 'hayden', 'highmark center', 'hunt library', 'institute of religion', 'istb7', 'john muir field', 'juniper hall', 'main gym', 'malesardi quadrangle', 'margaret morrison', 'mcgill house', 'mcginley hall', 'mclintock', 'memorial union', 'mullet arena', 'music building', 'neeb hall', 'newell simon', 'noble library', 'observatory', 'old main', 'physical science', 'porter hall', 'posner', 'power center', 'psych north', 'purnell', 'rcic', 'rec tennis courts', 'revelle plaza', 'rooney field', 'san diego supercomputer center', 'scaife', 'school of engineering', 'school of pharmacy', 'scob', 'scott hall', 'scripps pier', 'sdfc', 'simulator building', 'smith hall', 'soccer field', 'sojka pavilion', 'st ann hall', 'student pavilion', 'student service', 'sun devil bookstore', 'sun devil stadium', 'sun devil tennis court', 'tepper', 'towers hall', 'triton arena', 'university center', 'university farm', 'vaughan hall', 'vickroy hall', 'vista del sol', 'warner hall', 'wean', 'weis center'])
+- Label column: `label` (classes: ['asu', 'bucknell', 'cmu', 'duq', 'ucsd', 'uscd'])
 - Train/Eval split: 80/20 (stratified if available)
 
 ## Training
@@ -27,14 +27,14 @@ Fine-tuned **DistilBERT** for text classification on a classmate's HW1 dataset.
 - Best model by: F1 (macro)
 
 ## Results (Test)
-- Accuracy: 0.0000
-- F1 (macro): 0.0000
+- Accuracy: 0.4000
+- F1 (macro): 0.1231
 - Precision (macro): nan
 - Recall (macro): nan
 
 ## Notes & Limitations
 - Small student dataset; results may vary with seeds.
-- Labels mapped as: {'alumni house': 0, 'armstrong hall': 1, 'art museum': 2, 'assumption hall': 3, 'baker hall': 4, 'barrett honors hall': 5, 'bertrand library': 6, 'brickyard': 7, 'bucknell golf club': 8, 'bucknell pickleball': 9, 'cemetery': 10, 'cereus hall': 11, 'christy mathewson stadium': 12, 'cmu tennis courts': 13, 'college of fine art': 14, 'cooper fieldhouse': 15, 'coor': 16, 'cyert hall': 17, 'dept. of visual arts': 18, 'design school north/south': 19, 'dfa': 20, 'donner house': 21, 'douglas hall': 22, 'duq med school': 23, 'duq union': 24, 'durham': 25, 'emmitt field': 26, 'engineering center': 27, "farmer's education building": 28, 'frankling antonio hall': 29, 'galbraith': 30, 'gallery 100': 31, 'gammage': 32, 'gates hillman': 33, 'geisel library': 34, 'gesling stadium': 35, 'gumberg library': 36, 'hall of the arts': 37, 'hamerschlag': 38, 'hassayampa academic village': 39, 'hayden': 40, 'highmark center': 41, 'hunt library': 42, 'institute of religion': 43, 'istb7': 44, 'john muir field': 45, 'juniper hall': 46, 'main gym': 47, 'malesardi quadrangle': 48, 'margaret morrison': 49, 'mcgill house': 50, 'mcginley hall': 51, 'mclintock': 52, 'memorial union': 53, 'mullet arena': 54, 'music building': 55, 'neeb hall': 56, 'newell simon': 57, 'noble library': 58, 'observatory': 59, 'old main': 60, 'physical science': 61, 'porter hall': 62, 'posner': 63, 'power center': 64, 'psych north': 65, 'purnell': 66, 'rcic': 67, 'rec tennis courts': 68, 'revelle plaza': 69, 'rooney field': 70, 'san diego supercomputer center': 71, 'scaife': 72, 'school of engineering': 73, 'school of pharmacy': 74, 'scob': 75, 'scott hall': 76, 'scripps pier': 77, 'sdfc': 78, 'simulator building': 79, 'smith hall': 80, 'soccer field': 81, 'sojka pavilion': 82, 'st ann hall': 83, 'student pavilion': 84, 'student service': 85, 'sun devil bookstore': 86, 'sun devil stadium': 87, 'sun devil tennis court': 88, 'tepper': 89, 'towers hall': 90, 'triton arena': 91, 'university center': 92, 'university farm': 93, 'vaughan hall': 94, 'vickroy hall': 95, 'vista del sol': 96, 'warner hall': 97, 'wean': 98, 'weis center': 99}
+- Labels mapped as: {'asu': 0, 'bucknell': 1, 'cmu': 2, 'duq': 3, 'ucsd': 4, 'uscd': 5}
 
 ## AI Tool Disclosure
 This notebook used ChatGPT for scaffolding code and documentation.

checkpoint-15/config.json

@@ -9,209 +9,21 @@
   "dtype": "float32",
   "hidden_dim": 3072,
   "id2label": {
-    "0": "alumni house",
-    "1": "armstrong hall",
-    "2": "art museum",
-    "3": "assumption hall",
-    "4": "baker hall",
-    "5": "barrett honors hall",
-    "6": "bertrand library",
-    "7": "brickyard",
-    "8": "bucknell golf club",
-    "9": "bucknell pickleball",
-    "10": "cemetery",
-    "11": "cereus hall",
-    "12": "christy mathewson stadium",
-    "13": "cmu tennis courts",
-    "14": "college of fine art",
-    "15": "cooper fieldhouse",
-    "16": "coor",
-    "17": "cyert hall",
-    "18": "dept. of visual arts",
-    "19": "design school north/south",
-    "20": "dfa",
-    "21": "donner house",
-    "22": "douglas hall",
-    "23": "duq med school",
-    "24": "duq union",
-    "25": "durham",
-    "26": "emmitt field",
-    "27": "engineering center",
-    "28": "farmer's education building",
-    "29": "frankling antonio hall",
-    "30": "galbraith",
-    "31": "gallery 100",
-    "32": "gammage",
-    "33": "gates hillman",
-    "34": "geisel library",
-    "35": "gesling stadium",
-    "36": "gumberg library",
-    "37": "hall of the arts",
-    "38": "hamerschlag",
-    "39": "hassayampa academic village",
-    "40": "hayden",
-    "41": "highmark center",
-    "42": "hunt library",
-    "43": "institute of religion",
-    "44": "istb7",
-    "45": "john muir field",
-    "46": "juniper hall",
-    "47": "main gym",
-    "48": "malesardi quadrangle",
-    "49": "margaret morrison",
-    "50": "mcgill house",
-    "51": "mcginley hall",
-    "52": "mclintock",
-    "53": "memorial union",
-    "54": "mullet arena",
-    "55": "music building",
-    "56": "neeb hall",
-    "57": "newell simon",
-    "58": "noble library",
-    "59": "observatory",
-    "60": "old main",
-    "61": "physical science",
-    "62": "porter hall",
-    "63": "posner",
-    "64": "power center",
-    "65": "psych north",
-    "66": "purnell",
-    "67": "rcic",
-    "68": "rec tennis courts",
-    "69": "revelle plaza",
-    "70": "rooney field",
-    "71": "san diego supercomputer center",
-    "72": "scaife",
-    "73": "school of engineering",
-    "74": "school of pharmacy",
-    "75": "scob",
-    "76": "scott hall",
-    "77": "scripps pier",
-    "78": "sdfc",
-    "79": "simulator building",
-    "80": "smith hall",
-    "81": "soccer field",
-    "82": "sojka pavilion",
-    "83": "st ann hall",
-    "84": "student pavilion",
-    "85": "student service",
-    "86": "sun devil bookstore",
-    "87": "sun devil stadium",
-    "88": "sun devil tennis court",
-    "89": "tepper",
-    "90": "towers hall",
-    "91": "triton arena",
-    "92": "university center",
-    "93": "university farm",
-    "94": "vaughan hall",
-    "95": "vickroy hall",
-    "96": "vista del sol",
-    "97": "warner hall",
-    "98": "wean",
-    "99": "weis center"
+    "0": "asu",
+    "1": "bucknell",
+    "2": "cmu",
+    "3": "duq",
+    "4": "ucsd",
+    "5": "uscd"
   },
   "initializer_range": 0.02,
   "label2id": {
-    "alumni house": 0,
-    "armstrong hall": 1,
-    "art museum": 2,
-    "assumption hall": 3,
-    "baker hall": 4,
-    "barrett honors hall": 5,
-    "bertrand library": 6,
-    "brickyard": 7,
-    "bucknell golf club": 8,
-    "bucknell pickleball": 9,
-    "cemetery": 10,
-    "cereus hall": 11,
-    "christy mathewson stadium": 12,
-    "cmu tennis courts": 13,
-    "college of fine art": 14,
-    "cooper fieldhouse": 15,
-    "coor": 16,
-    "cyert hall": 17,
-    "dept. of visual arts": 18,
-    "design school north/south": 19,
-    "dfa": 20,
-    "donner house": 21,
-    "douglas hall": 22,
-    "duq med school": 23,
-    "duq union": 24,
-    "durham": 25,
-    "emmitt field": 26,
-    "engineering center": 27,
-    "farmer's education building": 28,
-    "frankling antonio hall": 29,
-    "galbraith": 30,
-    "gallery 100": 31,
-    "gammage": 32,
-    "gates hillman": 33,
-    "geisel library": 34,
-    "gesling stadium": 35,
-    "gumberg library": 36,
-    "hall of the arts": 37,
-    "hamerschlag": 38,
-    "hassayampa academic village": 39,
-    "hayden": 40,
-    "highmark center": 41,
-    "hunt library": 42,
-    "institute of religion": 43,
-    "istb7": 44,
-    "john muir field": 45,
-    "juniper hall": 46,
-    "main gym": 47,
-    "malesardi quadrangle": 48,
-    "margaret morrison": 49,
-    "mcgill house": 50,
-    "mcginley hall": 51,
-    "mclintock": 52,
-    "memorial union": 53,
-    "mullet arena": 54,
-    "music building": 55,
-    "neeb hall": 56,
-    "newell simon": 57,
-    "noble library": 58,
-    "observatory": 59,
-    "old main": 60,
-    "physical science": 61,
-    "porter hall": 62,
-    "posner": 63,
-    "power center": 64,
-    "psych north": 65,
-    "purnell": 66,
-    "rcic": 67,
-    "rec tennis courts": 68,
-    "revelle plaza": 69,
-    "rooney field": 70,
-    "san diego supercomputer center": 71,
-    "scaife": 72,
-    "school of engineering": 73,
-    "school of pharmacy": 74,
-    "scob": 75,
-    "scott hall": 76,
-    "scripps pier": 77,
-    "sdfc": 78,
-    "simulator building": 79,
-    "smith hall": 80,
-    "soccer field": 81,
-    "sojka pavilion": 82,
-    "st ann hall": 83,
-    "student pavilion": 84,
-    "student service": 85,
-    "sun devil bookstore": 86,
-    "sun devil stadium": 87,
-    "sun devil tennis court": 88,
-    "tepper": 89,
-    "towers hall": 90,
-    "triton arena": 91,
-    "university center": 92,
-    "university farm": 93,
-    "vaughan hall": 94,
-    "vickroy hall": 95,
-    "vista del sol": 96,
-    "warner hall": 97,
-    "wean": 98,
-    "weis center": 99
+    "asu": 0,
+    "bucknell": 1,
+    "cmu": 2,
+    "duq": 3,
+    "ucsd": 4,
+    "uscd": 5
   },
   "max_position_embeddings": 512,
   "model_type": "distilbert",

checkpoint-15/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e2deb523368cd05cde1bb4540f2e4ea9e46d41f207729b6134dec27397d2ba87
-size 268134024
+oid sha256:0a4bfa05bdc65ddfd8fe96f606014876cc6265d61f8db45ca288976c152ad930
+size 267844872

checkpoint-15/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8f09f091019c39adcb1edcc4c79d5a3afdddc0dbfa7b77f8e6910aa31e653747
-size 536327755
+oid sha256:1d92fd44ab979d8340d671d93e2948c6da687f7c537d72322cfd8f31d1cc5270
+size 535749451

checkpoint-15/trainer_state.json

@@ -26,7 +26,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 3640926362880.0,
+  "total_flos": 3634833520512.0,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null

checkpoint-15/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:87bdfca1dacb611df92cc74e5dbdc79d769792158a5cdda5f4c81624cdaa46b6
+oid sha256:9a8637fbaa7caac2416de4850d157fa12bf2bc42aec121fecff1fcd5a14478cd
 size 5777

config.json

@@ -9,209 +9,21 @@
   "dtype": "float32",
   "hidden_dim": 3072,
   "id2label": {
-    "0": "alumni house",
-    "1": "armstrong hall",
-    "2": "art museum",
-    "3": "assumption hall",
-    "4": "baker hall",
-    "5": "barrett honors hall",
-    "6": "bertrand library",
-    "7": "brickyard",
-    "8": "bucknell golf club",
-    "9": "bucknell pickleball",
-    "10": "cemetery",
-    "11": "cereus hall",
-    "12": "christy mathewson stadium",
-    "13": "cmu tennis courts",
-    "14": "college of fine art",
-    "15": "cooper fieldhouse",
-    "16": "coor",
-    "17": "cyert hall",
-    "18": "dept. of visual arts",
-    "19": "design school north/south",
-    "20": "dfa",
-    "21": "donner house",
-    "22": "douglas hall",
-    "23": "duq med school",
-    "24": "duq union",
-    "25": "durham",
-    "26": "emmitt field",
-    "27": "engineering center",
-    "28": "farmer's education building",
-    "29": "frankling antonio hall",
-    "30": "galbraith",
-    "31": "gallery 100",
-    "32": "gammage",
-    "33": "gates hillman",
-    "34": "geisel library",
-    "35": "gesling stadium",
-    "36": "gumberg library",
-    "37": "hall of the arts",
-    "38": "hamerschlag",
-    "39": "hassayampa academic village",
-    "40": "hayden",
-    "41": "highmark center",
-    "42": "hunt library",
-    "43": "institute of religion",
-    "44": "istb7",
-    "45": "john muir field",
-    "46": "juniper hall",
-    "47": "main gym",
-    "48": "malesardi quadrangle",
-    "49": "margaret morrison",
-    "50": "mcgill house",
-    "51": "mcginley hall",
-    "52": "mclintock",
-    "53": "memorial union",
-    "54": "mullet arena",
-    "55": "music building",
-    "56": "neeb hall",
-    "57": "newell simon",
-    "58": "noble library",
-    "59": "observatory",
-    "60": "old main",
-    "61": "physical science",
-    "62": "porter hall",
-    "63": "posner",
-    "64": "power center",
-    "65": "psych north",
-    "66": "purnell",
-    "67": "rcic",
-    "68": "rec tennis courts",
-    "69": "revelle plaza",
-    "70": "rooney field",
-    "71": "san diego supercomputer center",
-    "72": "scaife",
-    "73": "school of engineering",
-    "74": "school of pharmacy",
-    "75": "scob",
-    "76": "scott hall",
-    "77": "scripps pier",
-    "78": "sdfc",
-    "79": "simulator building",
-    "80": "smith hall",
-    "81": "soccer field",
-    "82": "sojka pavilion",
-    "83": "st ann hall",
-    "84": "student pavilion",
-    "85": "student service",
-    "86": "sun devil bookstore",
-    "87": "sun devil stadium",
-    "88": "sun devil tennis court",
-    "89": "tepper",
-    "90": "towers hall",
-    "91": "triton arena",
-    "92": "university center",
-    "93": "university farm",
-    "94": "vaughan hall",
-    "95": "vickroy hall",
-    "96": "vista del sol",
-    "97": "warner hall",
-    "98": "wean",
-    "99": "weis center"
+    "0": "asu",
+    "1": "bucknell",
+    "2": "cmu",
+    "3": "duq",
+    "4": "ucsd",
+    "5": "uscd"
   },
   "initializer_range": 0.02,
   "label2id": {
-    "alumni house": 0,
-    "armstrong hall": 1,
-    "art museum": 2,
-    "assumption hall": 3,
-    "baker hall": 4,
-    "barrett honors hall": 5,
-    "bertrand library": 6,
-    "brickyard": 7,
-    "bucknell golf club": 8,
-    "bucknell pickleball": 9,
-    "cemetery": 10,
-    "cereus hall": 11,
-    "christy mathewson stadium": 12,
-    "cmu tennis courts": 13,
-    "college of fine art": 14,
-    "cooper fieldhouse": 15,
-    "coor": 16,
-    "cyert hall": 17,
-    "dept. of visual arts": 18,
-    "design school north/south": 19,
-    "dfa": 20,
-    "donner house": 21,
-    "douglas hall": 22,
-    "duq med school": 23,
-    "duq union": 24,
-    "durham": 25,
-    "emmitt field": 26,
-    "engineering center": 27,
-    "farmer's education building": 28,
-    "frankling antonio hall": 29,
-    "galbraith": 30,
-    "gallery 100": 31,
-    "gammage": 32,
-    "gates hillman": 33,
-    "geisel library": 34,
-    "gesling stadium": 35,
-    "gumberg library": 36,
-    "hall of the arts": 37,
-    "hamerschlag": 38,
-    "hassayampa academic village": 39,
-    "hayden": 40,
-    "highmark center": 41,
-    "hunt library": 42,
-    "institute of religion": 43,
-    "istb7": 44,
-    "john muir field": 45,
-    "juniper hall": 46,
-    "main gym": 47,
-    "malesardi quadrangle": 48,
-    "margaret morrison": 49,
-    "mcgill house": 50,
-    "mcginley hall": 51,
-    "mclintock": 52,
-    "memorial union": 53,
-    "mullet arena": 54,
-    "music building": 55,
-    "neeb hall": 56,
-    "newell simon": 57,
-    "noble library": 58,
-    "observatory": 59,
-    "old main": 60,
-    "physical science": 61,
-    "porter hall": 62,
-    "posner": 63,
-    "power center": 64,
-    "psych north": 65,
-    "purnell": 66,
-    "rcic": 67,
-    "rec tennis courts": 68,
-    "revelle plaza": 69,
-    "rooney field": 70,
-    "san diego supercomputer center": 71,
-    "scaife": 72,
-    "school of engineering": 73,
-    "school of pharmacy": 74,
-    "scob": 75,
-    "scott hall": 76,
-    "scripps pier": 77,
-    "sdfc": 78,
-    "simulator building": 79,
-    "smith hall": 80,
-    "soccer field": 81,
-    "sojka pavilion": 82,
-    "st ann hall": 83,
-    "student pavilion": 84,
-    "student service": 85,
-    "sun devil bookstore": 86,
-    "sun devil stadium": 87,
-    "sun devil tennis court": 88,
-    "tepper": 89,
-    "towers hall": 90,
-    "triton arena": 91,
-    "university center": 92,
-    "university farm": 93,
-    "vaughan hall": 94,
-    "vickroy hall": 95,
-    "vista del sol": 96,
-    "warner hall": 97,
-    "wean": 98,
-    "weis center": 99
+    "asu": 0,
+    "bucknell": 1,
+    "cmu": 2,
+    "duq": 3,
+    "ucsd": 4,
+    "uscd": 5
   },
   "max_position_embeddings": 512,
   "model_type": "distilbert",

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e2deb523368cd05cde1bb4540f2e4ea9e46d41f207729b6134dec27397d2ba87
-size 268134024
+oid sha256:0a4bfa05bdc65ddfd8fe96f606014876cc6265d61f8db45ca288976c152ad930
+size 267844872

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:87bdfca1dacb611df92cc74e5dbdc79d769792158a5cdda5f4c81624cdaa46b6
+oid sha256:9a8637fbaa7caac2416de4850d157fa12bf2bc42aec121fecff1fcd5a14478cd
 size 5777