Fix naming in Train.example_train_result

bin/moons_benchmark.ml

@@ -104,8 +104,8 @@ let classify_moons ~seed ~on_device ~inlining_cutoff ~num_streams ~batch_size ~b
     outputs;
     model_result;
     infer_callback;
-    batch_losses;
-    epoch_losses;
+    rev_batch_losses;
+    rev_epoch_losses;
     learning_rates;
     used_memory;
   } =
@@ -148,7 +148,7 @@ let classify_moons ~seed ~on_device ~inlining_cutoff ~num_streams ~batch_size ~b
         Line_plot
           {
             points =
-              Array.of_list_rev_map batch_losses ~f:Float.(fun x -> max (log 0.00003) (log x));
+              Array.of_list_rev_map rev_batch_losses ~f:Float.(fun x -> max (log 0.00003) (log x));
             pixel = "-";
           };
       ]
@@ -158,7 +158,7 @@ let classify_moons ~seed ~on_device ~inlining_cutoff ~num_streams ~batch_size ~b
   let plot_loss =
     let open PrintBox_utils in
     plot ~size:(120, 30) ~x_label:"step" ~y_label:"epoch log loss"
-      [ Line_plot { points = Array.of_list_rev_map epoch_losses ~f:Float.log; pixel = "-" } ]
+      [ Line_plot { points = Array.of_list_rev_map rev_epoch_losses ~f:Float.log; pixel = "-" } ]
   in
   PrintBox_text.output Stdio.stdout plot_loss;
   Stdio.printf "\nLearning rate:\n%!";
@@ -186,7 +186,7 @@ let classify_moons ~seed ~on_device ~inlining_cutoff ~num_streams ~batch_size ~b
         result_label = "init time in sec, min loss, last loss";
         result =
           [%sexp_of: float * float * float]
-            (init_time_in_sec, List.reduce_exn epoch_losses ~f:Float.min, List.hd_exn epoch_losses);
+            (init_time_in_sec, List.reduce_exn rev_epoch_losses ~f:Float.min, List.hd_exn rev_epoch_losses);
       }
   in
   Stdio.printf "\n\n%!";

bin/moons_demo_parallel.ml

@@ -58,8 +58,8 @@ let experiment ~seed ~backend_name ~config () =
     outputs;
     model_result;
     infer_callback;
-    batch_losses;
-    epoch_losses;
+    rev_batch_losses;
+    rev_epoch_losses;
     learning_rates;
     used_memory;
   } =
@@ -91,14 +91,14 @@ let experiment ~seed ~backend_name ~config () =
   let plot_loss =
     let open PrintBox_utils in
     plot ~size:(120, 30) ~x_label:"step" ~y_label:"batch loss"
-      [ Line_plot { points = Array.of_list_rev batch_losses; pixel = "-" } ]
+      [ Line_plot { points = Array.of_list_rev rev_batch_losses; pixel = "-" } ]
   in
   PrintBox_text.output Stdio.stdout plot_loss;
   Stdio.printf "\nEpoch Loss:\n%!";
   let plot_loss =
     let open PrintBox_utils in
     plot ~size:(120, 30) ~x_label:"step" ~y_label:"epoch loss"
-      [ Line_plot { points = Array.of_list_rev epoch_losses; pixel = "-" } ]
+      [ Line_plot { points = Array.of_list_rev rev_epoch_losses; pixel = "-" } ]
   in
   PrintBox_text.output Stdio.stdout plot_loss;
   Stdio.printf "\nBatch Log-loss:\n%!";
@@ -109,7 +109,7 @@ let experiment ~seed ~backend_name ~config () =
         Line_plot
           {
             points =
-              Array.of_list_rev_map batch_losses ~f:Float.(fun x -> max (log 0.00003) (log x));
+              Array.of_list_rev_map rev_batch_losses ~f:Float.(fun x -> max (log 0.00003) (log x));
             pixel = "-";
           };
       ]
@@ -119,7 +119,7 @@ let experiment ~seed ~backend_name ~config () =
   let plot_loss =
     let open PrintBox_utils in
     plot ~size:(120, 30) ~x_label:"step" ~y_label:"epoch log loss"
-      [ Line_plot { points = Array.of_list_rev_map epoch_losses ~f:Float.log; pixel = "-" } ]
+      [ Line_plot { points = Array.of_list_rev_map rev_epoch_losses ~f:Float.log; pixel = "-" } ]
   in
   PrintBox_text.output Stdio.stdout plot_loss;
   Stdio.printf "\nLearning rate:\n%!";

lib/train.ml

@@ -450,8 +450,8 @@ type example_train_result = {
   model_result : Tensor.t;
   infer_callback : float array -> float array;
       (** Note: infer_callback is significantly less efficient than using the model via arrayjit. *)
-  batch_losses : float list;
-  epoch_losses : float list;
+  rev_batch_losses : float list;
+  rev_epoch_losses : float list;
   learning_rates : float list;
   used_memory : int;
 }
@@ -484,8 +484,8 @@ let example_train_loop ?(disable_rootness_check = false) ~seed ~batch_size ~init
   let step_n, bindings = IDX.get_static_symbol bindings in
   let%op input = inputs @| batch_n in
   let%op expectation = outputs @| batch_n in
-  let batch_losses = ref [] in
-  let epoch_losses = ref [] in
+  let rev_batch_losses = ref [] in
+  let rev_epoch_losses = ref [] in
   let learning_rates = ref [] in
   let%op loss_tensor = loss_fn ~output:(model input) ~expectation in
   let%op scalar_loss = (loss_tensor ++ "...|... => 0") /. !..batch_size in
@@ -524,7 +524,7 @@ let example_train_loop ?(disable_rootness_check = false) ~seed ~batch_size ~init
         Backend.await grad_updates.(0).context.stream;
         let batch_loss = scalar_loss.@[0] in
         epoch_loss := !epoch_loss +. batch_loss;
-        batch_losses := batch_loss :: !batch_losses;
+        rev_batch_losses := batch_loss :: !rev_batch_losses;
         Option.iter per_batch_callback ~f:(fun f ->
             f ~at_batch:!batch_ref ~at_step:!step_ref ~learning_rate:learning_rate.@[0] ~batch_loss
               ~epoch_loss:!epoch_loss))
@@ -538,7 +538,7 @@ let example_train_loop ?(disable_rootness_check = false) ~seed ~batch_size ~init
     (* Utils.capture_stdout_logs *)
      update ();
     learning_rates := learning_rate.@[0] :: !learning_rates;
-    epoch_losses := !epoch_loss :: !epoch_losses;
+    rev_epoch_losses := !epoch_loss :: !rev_epoch_losses;
     Option.iter per_epoch_callback ~f:(fun f ->
         f ~at_step:!step_ref ~at_epoch:epoch ~learning_rate:learning_rate.@[0]
           ~epoch_loss:!epoch_loss);
@@ -590,8 +590,8 @@ let example_train_loop ?(disable_rootness_check = false) ~seed ~batch_size ~init
     outputs;
     model_result;
     infer_callback;
-    batch_losses = !batch_losses;
-    epoch_losses = !epoch_losses;
+    rev_batch_losses = !rev_batch_losses;
+    rev_epoch_losses = !rev_epoch_losses;
     learning_rates = !learning_rates;
     used_memory;
   }