Update app.py

app.py

@@ -71,9 +71,7 @@ def mlp_exec(bs, h, n, d):
   return flop, nbytes, exec_time
   
 def print_kernel_execution(c1, c2, comp_flop, mem_bytes):
-  arith_int = comp_flop/mem_bytes
   exec_time = calc_exec_time(comp_flop, mem_bytes)
-
   comp_flop = round(comp_flop/1e9, 2)
   mem_bytes = round(mem_bytes/1e6, 2)
   
@@ -108,7 +106,7 @@ mqa_total_time = 0.
 mha_total_time = 0.
 
 for i in range(n_start, n):
-  shared_time = out_exec(bs, h, i, d)[2] + softmax_exec(bs, h, i , d)[2] + 2*ln_exec(bs, h, i, d)[2] \ 
+  shared_time = out_exec(bs, h, i, d)[2] + softmax_exec(bs, h, i , d)[2] + 2*ln_exec(bs, h, i, d)[2] \
                 + 2*mlp_exec(bs, h, i, d)[2] + 3*ln_exec(bs, h, i, d)
   mha_time = shared_time + qkv_mha_exec(bs, h, i, d)[2] + att1_mha_exec(bs, h, i, d)[2] + att2_mha_exec(bs, h, i, d)[2]
   mha_total_time += l*mha_time