clean up TODOs a bit, they are stale

train.py

@@ -103,7 +103,7 @@
 ptdtype = {'float32': torch.float32, 'bfloat16': torch.bfloat16, 'float16': torch.float16}[dtype]
 ctx = nullcontext() if device_type == 'cpu' else torch.amp.autocast(device_type=device_type, dtype=ptdtype)
 
-# poor man's data loader, TODO evaluate need for actual DataLoader
+# poor man's data loader
 data_dir = os.path.join('data', dataset)
 train_data = np.memmap(os.path.join(data_dir, 'train.bin'), dtype=np.uint16, mode='r')
 val_data = np.memmap(os.path.join(data_dir, 'val.bin'), dtype=np.uint16, mode='r')
@@ -302,7 +302,7 @@ def get_lr(it):
     dt = t1 - t0
     t0 = t1
     if iter_num % log_interval == 0 and master_process:
-        lossf = loss.item() # loss as float. TODO note CPU-GPU sync! profile, make sure not too slow
+        lossf = loss.item() # loss as float. note: this is a CPU-GPU sync point
         print(f"iter {iter_num}: loss {lossf:.4f}, time {dt*1000:.2f}ms")
     iter_num += 1