new updates

data.dvc

@@ -1,7 +1,7 @@
 deps:
 - path: params.yml
-  md5: 40a4e5557f78126afbd7800f61ef1d11
-  size: 189
+  md5: d0f3e81bc9191e752a69761045a449d9
+  size: 196
 - path: src/data/make_dataset.py
   md5: 9de71de0f8df5d0a7beb235ef7c7777d
   size: 772
@@ -11,4 +11,4 @@ outs:
   nfiles: 3
   path: data/raw
   size: 1358833013
-md5: 62b3e4497b565a311300b2adbba6f479
+md5: ff502232006c7fbef1015b5aa5cc4bbb

dvc.lock

@@ -4,45 +4,45 @@ stages:
     cmd: python src/models/train_model.py
     deps:
     - path: data/processed/train.csv
-      md5: d98ef1caf1d0d8c5bdb0b1f2b3b08391
-      size: 62359655
+      md5: 51edd724b75a8e99a78b9138f8f37c60
+      size: 25012573
     - path: data/processed/validation.csv
-      md5: b684e7d369e0851ffa1167787bc6a488
-      size: 2861403
+      md5: 0900e2bb330df94cb045faddd0b945d1
+      size: 1138285
     - path: params.yml
-      md5: 40a4e5557f78126afbd7800f61ef1d11
-      size: 189
+      md5: d0f3e81bc9191e752a69761045a449d9
+      size: 196
     - path: src/models/train_model.py
       md5: fca8acf70f09cecd679ca1ddb2eef6a9
       size: 1198
     outs:
     - path: models
-      md5: 468fdfb51858db2f7253b6940afabcd8.dir
-      size: 243476333
-      nfiles: 5
+      md5: 688745a9fb1cc7c8580887bae3873a39.dir
+      size: 486952666
+      nfiles: 10
     - path: reports/training_metrics.txt
-      md5: f7950559cd2021be3651a424a612d2d3
-      size: 162
+      md5: 048a956b0eb431535d287bbc3322cf76
+      size: 158
   eval:
     cmd: python src/models/evaluate_model.py
     deps:
     - path: data/processed/test.csv
-      md5: 2ae4ba38c85f7eb36c9b1b067cfa5051
-      size: 2449346
+      md5: 3cb7b63891f12d53b3ef3e81a2e93f8e
+      size: 986944
     - path: models
-      md5: 468fdfb51858db2f7253b6940afabcd8.dir
-      size: 243476333
-      nfiles: 5
+      md5: 688745a9fb1cc7c8580887bae3873a39.dir
+      size: 486952666
+      nfiles: 10
     - path: params.yml
-      md5: 40a4e5557f78126afbd7800f61ef1d11
-      size: 189
+      md5: d0f3e81bc9191e752a69761045a449d9
+      size: 196
     - path: src/models/evaluate_model.py
       md5: aa01b1564d737fef54ae45d25c5018d1
       size: 615
     outs:
     - path: reports/metrics.txt
-      md5: 6cfc8de10905de3c6dda7d19fdb004d0
-      size: 1605
+      md5: 27d21366dca75caa1bb3777575cb126b
+      size: 1596
   process_data:
     cmd: python src/data/process_data.py
     deps:
@@ -51,18 +51,18 @@ stages:
       size: 0
       nfiles: 0
     - path: params.yml
-      md5: 40a4e5557f78126afbd7800f61ef1d11
-      size: 189
+      md5: d0f3e81bc9191e752a69761045a449d9
+      size: 196
     - path: src/data/process_data.py
       md5: ba3ba7b7c8a905b736b6b0a28d2334c4
       size: 623
     outs:
     - path: data/processed/test.csv
-      md5: 2ae4ba38c85f7eb36c9b1b067cfa5051
-      size: 2449346
+      md5: 3cb7b63891f12d53b3ef3e81a2e93f8e
+      size: 986944
     - path: data/processed/train.csv
-      md5: d98ef1caf1d0d8c5bdb0b1f2b3b08391
-      size: 62359655
+      md5: 51edd724b75a8e99a78b9138f8f37c60
+      size: 25012573
     - path: data/processed/validation.csv
-      md5: b684e7d369e0851ffa1167787bc6a488
-      size: 2861403
+      md5: 0900e2bb330df94cb045faddd0b945d1
+      size: 1138285

reports/metrics.txt

@@ -1 +1 @@
-{"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}}
+{"Rouge 1": {"Rouge_1 Low Precision": 0.34885388166790793, "Rouge_1 Low recall": 0.28871556132198656, "Rouge_1 Low F1": 0.31058637096822267, "Rouge_1 Mid Precision": 0.412435004251884, "Rouge_1 Mid recall": 0.3386352228897427, "Rouge_1 Mid F1": 0.3517931748124066, "Rouge_1 High Precision": 0.47625451117848977, "Rouge_1 High recall": 0.39086727645312935, "Rouge_1 High F1": 0.3959993953753958}, "Rouge 2": {"Rouge_2 Low Precision": 0.1259156300716482, "Rouge_2 Low recall": 0.10333119800163641, "Rouge_2 Low F1": 0.10992592662502373, "Rouge_2 Mid Precision": 0.16879303949162833, "Rouge_2 Mid recall": 0.13805319188028575, "Rouge_2 Mid F1": 0.14400796293585816, "Rouge_2 High Precision": 0.21844214485938712, "Rouge_2 High recall": 0.1777722350788, "Rouge_2 High F1": 0.18342627795315522}, "Rouge L": {"Rouge_L Low Precision": 0.2322041975032734, "Rouge_L Low recall": 0.194000575085051, "Rouge_L Low F1": 0.20468107864660212, "Rouge_L Mid Precision": 0.2797360675037497, "Rouge_L Mid recall": 0.22647774162854406, "Rouge_L Mid F1": 0.2361293941929179, "Rouge_L High Precision": 0.3357160682858357, "Rouge_L High recall": 0.2622222798536235, "Rouge_L High F1": 0.27267217209978356}, "rougeLsum": {"rougeLsum Low Precision": 0.29651536760563263, "rougeLsum Low recall": 0.2432094838451322, "rougeLsum Low F1": 0.26048483356867896, "rougeLsum Mid Precision": 0.35317671791338556, "rougeLsum Mid recall": 0.286187817596869, "rougeLsum Mid F1": 0.2985727815225495, "rougeLsum High Precision": 0.4134539668577922, "rougeLsum High recall": 0.3365998852405162, "rougeLsum High F1": 0.3454898564714797}}

reports/training_metrics.txt

@@ -1 +1 @@
-{"train_loss": 1.8116508722305298, "epoch": 9, "trainer/global_step": 3589, "_runtime": 800, "_timestamp": 1627352142, "_step": 80, "val_loss": 1.923620581626892}
+{"train_loss": 2.785480260848999, "epoch": 4, "trainer/global_step": 289, "_runtime": 88, "_timestamp": 1627353229, "_step": 9, "val_loss": 2.181020975112915}