Update train.py

train.py

@@ -17,16 +17,16 @@ INPUT_DATASET = "HuggingFaceTB/smollm-corpus"
 INSTRUCT_DATASET = "nroggendorff/elephant"
 OUTPUT_REPO = "smallama"
 FP16 = False
-WARMUP_STEPS = 20
+WARMUP_STEPS = 0
 DECAY = 0
-GRADIENT_ACCUMULATION_STEPS = 128
+GRADIENT_ACCUMULATION_STEPS = 256
 PUSH_TO_HUB = True
 
 def load_data():
     pretrain = load_dataset(INPUT_DATASET, "cosmopedia-v2", split="train", streaming=True)
-    pretrain = Dataset.from_generator(lambda: pretrain.take(int(3e+5)))
+    pretrain = Dataset.from_generator(lambda: pretrain.take(int(3e+4)))
     instruct = load_dataset(INSTRUCT_DATASET, split="train", streaming=True)
-    instruct = Dataset.from_generator(lambda: instruct.take(int(5e+5)))
+    instruct = Dataset.from_generator(lambda: instruct.take(int(5e+4)))
     dataset_dict = DatasetDict({
         'pretrain': pretrain,
         'instruct': instruct