Update batch size and configs

.gitignore

@@ -2,3 +2,4 @@
 __pycache__
 colloquial_data/
 wandb/
+README.md

11 Indobert-base-p1_without_law.py

@@ -77,7 +77,7 @@ import wandb
 MODEL_NAME = "indobenchmark/indobert-base-p1"
 OUTPUT_DIR = "./11 Indobert-base-p1_without_law"
 MAX_LENGTH = 512
-BATCH_SIZE = 64
+BATCH_SIZE = 512
 GRADIENT_ACCUMULATION_STEPS = 3
 NUM_EPOCHS = 30
 SEED = 42

12 Indobert-base-p1_with_law.py

@@ -77,7 +77,7 @@ import wandb
 MODEL_NAME = "indobenchmark/indobert-base-p1"
 OUTPUT_DIR = "./12 Indobert-base-p1_with_law"
 MAX_LENGTH = 512
-BATCH_SIZE = 64
+BATCH_SIZE = 512
 GRADIENT_ACCUMULATION_STEPS = 3
 NUM_EPOCHS = 30
 SEED = 42

21 Indobert-IndoLegalBERT_without_law.py

@@ -77,7 +77,7 @@ import wandb
 MODEL_NAME = "archi-ai/Indo-LegalBERT"
 OUTPUT_DIR = "./21 Indobert-IndoLegalBERT_without_law"
 MAX_LENGTH = 512
-BATCH_SIZE = 64
+BATCH_SIZE = 512
 GRADIENT_ACCUMULATION_STEPS = 3
 NUM_EPOCHS = 30
 SEED = 42

22 Indobert-IndoLegalBERT_with_law.py

@@ -77,7 +77,7 @@ import wandb
 MODEL_NAME = "archi-ai/Indo-LegalBERT"
 OUTPUT_DIR = "./22 Indobert-IndoLegalBERT_with_law"
 MAX_LENGTH = 512
-BATCH_SIZE = 64
+BATCH_SIZE = 512
 GRADIENT_ACCUMULATION_STEPS = 3
 NUM_EPOCHS = 30
 SEED = 42

31 Indobert-base-p1_without_law_with_svr.py

@@ -77,7 +77,7 @@ import wandb
 MODEL_NAME = "indobenchmark/indobert-base-p1"
 OUTPUT_DIR = "./31 Indobert-base-p1_without_law_with_svr"
 MAX_LENGTH = 512
-BATCH_SIZE = 64
+BATCH_SIZE = 512
 GRADIENT_ACCUMULATION_STEPS = 3
 NUM_EPOCHS = 30
 SEED = 42

32 Indobert-base-p1_with_law_with_svr.py

@@ -77,7 +77,7 @@ import wandb
 MODEL_NAME = "indobenchmark/indobert-base-p1"
 OUTPUT_DIR = "./32 Indobert-base-p1_with_law_with_svr"
 MAX_LENGTH = 512
-BATCH_SIZE = 64
+BATCH_SIZE = 512
 GRADIENT_ACCUMULATION_STEPS = 3
 NUM_EPOCHS = 30
 SEED = 42

41 Indobert-base-p1_without_law_mlp.py

@@ -79,7 +79,7 @@ import wandb
 MODEL_NAME = "indobenchmark/indobert-base-p1"
 OUTPUT_DIR = "./41 Indobert-base-p1_without_law_mlp"
 MAX_LENGTH = 512
-BATCH_SIZE = 64
+BATCH_SIZE = 512
 GRADIENT_ACCUMULATION_STEPS = 3
 NUM_EPOCHS = 30
 SEED = 42

42 Indobert-base-p1_with_law_mlp.py

@@ -79,7 +79,7 @@ import wandb
 MODEL_NAME = "indobenchmark/indobert-base-p1"
 OUTPUT_DIR = "./42 Indobert-base-p1_with_law_mlp"
 MAX_LENGTH = 512
-BATCH_SIZE = 64
+BATCH_SIZE = 512
 GRADIENT_ACCUMULATION_STEPS = 3
 NUM_EPOCHS = 30
 SEED = 42

51 Indobert-base-p1_without_law_with_xgboost.py

@@ -34,7 +34,7 @@ import xgboost as xgb
 MODEL_NAME = "indobenchmark/indobert-base-p1"
 OUTPUT_DIR = "./51 Indobert-base-p1_without_law_with_xgboost"
 MAX_LENGTH = 512
-BATCH_SIZE = 64
+BATCH_SIZE = 512
 SEED = 42
 
 def set_seed(seed: int):

52 Indobert-base-p1_with_law_with_xgboost.py

@@ -34,7 +34,7 @@ import xgboost as xgb
 MODEL_NAME = "indobenchmark/indobert-base-p1"
 OUTPUT_DIR = "./52 Indobert-base-p1_with_law_with_xgboost"
 MAX_LENGTH = 512
-BATCH_SIZE = 64
+BATCH_SIZE = 512
 SEED = 42
 
 def set_seed(seed: int):

README.md

@@ -5,6 +5,9 @@ Panduan singkat ini menunjukkan cara menjalankan proses training di server lewat
 install git xet
 curl --proto '=https' --tlsv1.2 -sSf https://raw.githubusercontent.com/huggingface/xet-core/refs/heads/main/git_xet/install.sh | sh
 
+ssh
+ssh -i "c:\Users\patri\Documents\Skripsi\Regression\id_rsa_primeintellect" root@31.22.104.199 -p 22
+
 ## Persiapan (prasyarat)
 
 - Sistem target: Linux server (atau WSL di Windows).
@@ -22,6 +25,9 @@ tmux new -s mysession
 # atau beri nama: tmux new -s nama_sesi
 ```
 
+```bash
+```
+
 ---
 
 ## Clone repository