:hammer_and_pick: hyenadna ready to deploy

app.py

@@ -245,15 +245,18 @@ def start():
                                                                         trust_remote_code=True)
   args = {
     "output_dir": "output_hyena_dna-mqtl_classification",
-    "num_train_epochs": 1,
-    "max_steps": 10,
+    "num_train_epochs": 10,
+    "max_steps": 40_000,  # train 36k + val 4k = 40k
     # Set the number of steps you expect to train, originally 1000, takes too much time. So I set it to 10 to run faster and check my code/pipeline
     "run_name": "laptop_run_hyena_dna-mqtl_classification",  # Override run_name here
     "per_device_train_batch_size": 1,
-    "gradient_accumulation_steps": 4,
+    "gradient_accumulation_steps": 32,
     "gradient_checkpointing": True,
-    "learning_rate": 2e-5,
-    "save_safetensors": False  # I added it. this solves the runtime error!
+    "learning_rate": 1e-3,
+    "save_safetensors": False,  # I added it. this solves the runtime error!
+    # not sure if it is a good idea. sklearn may slow down training, causing time loss... if so, disable these 2 lines below
+    "evaluation_strategy": "epoch",  # To calculate metrics per epoch
+    "logging_strategy": "epoch"  # Extra: to log training data stats for loss
   }
 
   # """