diff --git a/autoparallel/optimize_sharding.py b/autoparallel/optimize_sharding.py
index 49f6d020..045b8dff 100644
--- a/autoparallel/optimize_sharding.py
+++ b/autoparallel/optimize_sharding.py
@@ -628,23 +628,45 @@ def get_log(self, colored=False):
             attr_color = _identity
         l_id = 1
         plc_txt = txt_color("# placement=")
-        cost_txt = txt_color(", cost=")
+        cost_txt = txt_color(", compute_cost=")
         for node in nodes:
             if node.op == "output":
                 continue
             d = opt[node]
+            d_inputs = [opt[n] for n in self._all_input_nodes(node)]
+            redistributions = []
+            for i, x in enumerate(d_inputs):
+                src_strat = x[0]["out_strat"]
+                if not isinstance(src_strat, DTensorSpec):
+                    src_strat = src_strat[i]
+                dst_strat = d[i]["inp_strat"]
+                if src_strat != dst_strat:
+                    redistributions.append(
+                        (
+                            i,
+                            self._all_input_nodes(node)[i],
+                            src_strat,
+                            dst_strat,
+                        )
+                    )
+            # this is an annoying special case
+            if node.target == operator.getitem:
+                redistributions = []
+            preline = ""
+            if redistributions:
+                preline += f"\n    # Redistributing for node {str(node)}:"
+                for i, n, src, dst in redistributions:
+                    src_s = "".join(str(p) for p in src.placements)
+                    dst_s = "".join(str(p) for p in dst.placements)
+                    comment = f"# shape={str(tuple(src.shape))}, comm_cost={d[i]['comm_cost']}"
+                    preline += f"\n    # {str(n)} = redistribute({str(n)}, src={src_s}, dst={dst_s})  {comment}"
             strat = str(d[0]["full_strat"])
-            costs = [
-                (x["comm_cost"], x["compute_cost"], x["sharding_transition_cost"])
-                for x in d
-            ]
+            costs = sum(x["compute_cost"] for x in d)
             device_order = getattr(node.meta, "device_order", None)
+            device_order_str = ""
             if device_order:
-                line = f"  {plc_txt}{attr_color(strat)} device_order: {device_order} {cost_txt}{attr_color(str(costs))}"
-            else:
-                line = (
-                    f"  {plc_txt}{attr_color(strat)} {cost_txt}{attr_color(str(costs))}"
-                )
+                device_order_str = f" device_order={device_order} "
+            line = f"  {plc_txt}{attr_color(strat)}{device_order_str} {cost_txt}{attr_color(str(costs))}"
             if node.op == "placeholder":
                 line = f"    # {node.name}: {line}"
                 code.insert(l_id, line)
@@ -654,6 +676,9 @@ def get_log(self, colored=False):
             while not code[l_id].lstrip().startswith(repr(node)):
                 l_id += 1
             code[l_id] += line
+            if preline:
+                code.insert(l_id, preline)
+                l_id += 1
             l_id += 1
         code = "\n".join(code)
         total_cost = sum(self.ds[x]["cost"] for x in self.res)