updates

dvc.lock

@@ -4,45 +4,45 @@ stages:
     cmd: python src/models/train_model.py
     deps:
     - path: data/processed/train.csv
-      md5: 707c5ed455a15ec48965daf92fed7df6
-      size: 12653913
+      md5: d98ef1caf1d0d8c5bdb0b1f2b3b08391
+      size: 62359655
     - path: data/processed/validation.csv
-      md5: 1e021dc163cc87a32cef74a98e4a0d51
-      size: 558403
+      md5: b684e7d369e0851ffa1167787bc6a488
+      size: 2861403
     - path: params.yml
-      md5: 1c81ba3474bff51c4e36e59af83ba4eb
-      size: 188
+      md5: 40a4e5557f78126afbd7800f61ef1d11
+      size: 189
     - path: src/models/train_model.py
       md5: fca8acf70f09cecd679ca1ddb2eef6a9
       size: 1198
     outs:
     - path: models
-      md5: 889c9e429c8c0d7aeb66c6204c983645.dir
+      md5: 468fdfb51858db2f7253b6940afabcd8.dir
       size: 243476333
       nfiles: 5
     - path: reports/training_metrics.txt
-      md5: 35f2808d4380b428f3de8f4c6b2c119e
-      size: 124
+      md5: f7950559cd2021be3651a424a612d2d3
+      size: 162
   eval:
     cmd: python src/models/evaluate_model.py
     deps:
     - path: data/processed/test.csv
-      md5: 5f2bfb37d55a13ead3c81564dbee2fd5
-      size: 508508
+      md5: 2ae4ba38c85f7eb36c9b1b067cfa5051
+      size: 2449346
     - path: models
-      md5: 889c9e429c8c0d7aeb66c6204c983645.dir
+      md5: 468fdfb51858db2f7253b6940afabcd8.dir
       size: 243476333
       nfiles: 5
     - path: params.yml
-      md5: 1c81ba3474bff51c4e36e59af83ba4eb
-      size: 188
+      md5: 40a4e5557f78126afbd7800f61ef1d11
+      size: 189
     - path: src/models/evaluate_model.py
       md5: aa01b1564d737fef54ae45d25c5018d1
       size: 615
     outs:
     - path: reports/metrics.txt
-      md5: d9557204b6599a6b0db7d5f1ea80559d
-      size: 1604
+      md5: 6cfc8de10905de3c6dda7d19fdb004d0
+      size: 1605
   process_data:
     cmd: python src/data/process_data.py
     deps:

reports/metrics.txt

@@ -1 +1 @@
-{"Rouge 1": {"Rouge_1 Low Precision": 0.35879412940695404, "Rouge_1 Low recall": 0.273733478679664, "Rouge_1 Low F1": 0.30242065341410873, "Rouge_1 Mid Precision": 0.4338533395974997, "Rouge_1 Mid recall": 0.32042622316817226, "Rouge_1 Mid F1": 0.358115556137624, "Rouge_1 High Precision": 0.5280148017181922, "Rouge_1 High recall": 0.37164701807593853, "Rouge_1 High F1": 0.4196716488587748}, "Rouge 2": {"Rouge_2 Low Precision": 0.14079866776770866, "Rouge_2 Low recall": 0.10234397527752646, "Rouge_2 Low F1": 0.11454339253852734, "Rouge_2 Mid Precision": 0.20278273142043576, "Rouge_2 Mid recall": 0.1437386638831592, "Rouge_2 Mid F1": 0.1626240666356349, "Rouge_2 High Precision": 0.2834076711211276, "Rouge_2 High recall": 0.19964227594691578, "Rouge_2 High F1": 0.22508874620349306}, "Rouge L": {"Rouge_L Low Precision": 0.24715297672294737, "Rouge_L Low recall": 0.19726701009989336, "Rouge_L Low F1": 0.21458962512439578, "Rouge_L Mid Precision": 0.31881722334892215, "Rouge_L Mid recall": 0.23988574061697504, "Rouge_L Mid F1": 0.26443860973621613, "Rouge_L High Precision": 0.40068438045602844, "Rouge_L High recall": 0.2924354094810886, "Rouge_L High F1": 0.32317934985902896}, "rougeLsum": {"rougeLsum Low Precision": 0.313922368347619, "rougeLsum Low recall": 0.24053636172149256, "rougeLsum Low F1": 0.26759225635037753, "rougeLsum Mid Precision": 0.38675681843219567, "rougeLsum Mid recall": 0.28576071040262474, "rougeLsum Mid F1": 0.319041888317666, "rougeLsum High Precision": 0.46780971284574424, "rougeLsum High recall": 0.3329834309090794, "rougeLsum High F1": 0.37448259921289817}}
+{"Rouge 1": {"Rouge_1 Low Precision": 0.3101147803216913, "Rouge_1 Low recall": 0.26667179616787196, "Rouge_1 Low F1": 0.28102521213390536, "Rouge_1 Mid Precision": 0.37467224911406116, "Rouge_1 Mid recall": 0.3231162755643715, "Rouge_1 Mid F1": 0.33590674547536586, "Rouge_1 High Precision": 0.44552759888935306, "Rouge_1 High recall": 0.3833114041719603, "Rouge_1 High F1": 0.3915885794585388}, "Rouge 2": {"Rouge_2 Low Precision": 0.0979809815200762, "Rouge_2 Low recall": 0.09066363889125824, "Rouge_2 Low F1": 0.09147076435057812, "Rouge_2 Mid Precision": 0.14213812497089268, "Rouge_2 Mid recall": 0.1279409901288579, "Rouge_2 Mid F1": 0.12999327841435968, "Rouge_2 High Precision": 0.19643857982972704, "Rouge_2 High recall": 0.17330384956888412, "Rouge_2 High F1": 0.17464491307113122}, "Rouge L": {"Rouge_L Low Precision": 0.20022594334256222, "Rouge_L Low recall": 0.17455575689075598, "Rouge_L Low F1": 0.18405218677519644, "Rouge_L Mid Precision": 0.24431417960034996, "Rouge_L Mid recall": 0.21178120691756747, "Rouge_L Mid F1": 0.2187251307495292, "Rouge_L High Precision": 0.2969701241821441, "Rouge_L High recall": 0.25125195047639504, "Rouge_L High F1": 0.25666465900458324}, "rougeLsum": {"rougeLsum Low Precision": 0.2569235361439761, "rougeLsum Low recall": 0.22179714311010448, "rougeLsum Low F1": 0.23051340704076662, "rougeLsum Mid Precision": 0.3092598345428193, "rougeLsum Mid recall": 0.26713930485044357, "rougeLsum Mid F1": 0.27680670227990145, "rougeLsum High Precision": 0.3646127530821929, "rougeLsum High recall": 0.323567582903172, "rougeLsum High F1": 0.330886027811868}}

reports/training_metrics.txt

@@ -0,0 +1 @@
+{"train_loss": 1.8116508722305298, "epoch": 9, "trainer/global_step": 3589, "_runtime": 800, "_timestamp": 1627352142, "_step": 80, "val_loss": 1.923620581626892}