Update hypernetwork.py

modules/hypernetworks/hypernetwork.py

@@ -16,6 +16,7 @@ from modules.textual_inversion import textual_inversion
 from modules.textual_inversion.learn_schedule import LearnRateScheduler
 from torch import einsum
 
+from statistics import stdev, mean
 
 class HypernetworkModule(torch.nn.Module):
     multiplier = 1.0
@@ -268,6 +269,32 @@ def stack_conds(conds):
     return torch.stack(conds)
 
 
+def log_statistics(loss_info:dict, key, value):
+    if key not in loss_info:
+        loss_info[key] = [value]
+    else:
+        loss_info[key].append(value)
+        if len(loss_info) > 1024:
+            loss_info.pop(0)
+
+
+def statistics(data):
+    total_information = f"loss:{mean(data):.3f}"+u"\u00B1"+f"({stdev(data)/ (len(data)**0.5):.3f})"
+    recent_data = data[-32:]
+    recent_information = f"recent 32 loss:{mean(recent_data):.3f}"+u"\u00B1"+f"({stdev(recent_data)/ (len(recent_data)**0.5):.3f})"
+    return total_information, recent_information
+
+
+def report_statistics(loss_info:dict):
+    keys = sorted(loss_info.keys(), key=lambda x: sum(loss_info[x]) / len(loss_info[x]))
+    for key in keys:
+        info, recent = statistics(loss_info[key])
+        print("Loss statistics for file " + key)
+        print(info)
+        print(recent)
+
+
+
 def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, data_root, log_directory, training_width, training_height, steps, create_image_every, save_hypernetwork_every, template_file, preview_from_txt2img, preview_prompt, preview_negative_prompt, preview_steps, preview_sampler_index, preview_cfg_scale, preview_seed, preview_width, preview_height):
     # images allows training previews to have infotext. Importing it at the top causes a circular import problem.
     from modules import images
@@ -310,7 +337,11 @@ def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, data_root, log
     for weight in weights:
         weight.requires_grad = True
 
-    losses = torch.zeros((32,))
+    size = len(ds.indexes)
+    loss_dict = {}
+    losses = torch.zeros((size,))
+    previous_mean_loss = 0
+    print("Mean loss of {} elements".format(size))
 
     last_saved_file = "<none>"
     last_saved_image = "<none>"
@@ -329,7 +360,9 @@ def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, data_root, log
     pbar = tqdm.tqdm(enumerate(ds), total=steps - ititial_step)
     for i, entries in pbar:
         hypernetwork.step = i + ititial_step
-
+        if loss_dict and i % size == 0:
+            previous_mean_loss = sum(i[-1] for i in loss_dict.values()) / len(loss_dict)
+            
         scheduler.apply(optimizer, hypernetwork.step)
         if scheduler.finished:
             break
@@ -346,7 +379,9 @@ def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, data_root, log
             del c
 
             losses[hypernetwork.step % losses.shape[0]] = loss.item()
-
+            for entry in entries:
+                log_statistics(loss_dict, entry.filename, loss.item())
+                
             optimizer.zero_grad()
             weights[0].grad = None
             loss.backward()
@@ -359,10 +394,9 @@ def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, data_root, log
 
             optimizer.step()
 
-        mean_loss = losses.mean()
-        if torch.isnan(mean_loss):
+        if torch.isnan(losses[hypernetwork.step % losses.shape[0]]):
             raise RuntimeError("Loss diverged.")
-        pbar.set_description(f"loss: {mean_loss:.7f}")
+        pbar.set_description(f"dataset loss: {previous_mean_loss:.7f}")
 
         if hypernetwork.step > 0 and hypernetwork_dir is not None and hypernetwork.step % save_hypernetwork_every == 0:
             # Before saving, change name to match current checkpoint.
@@ -371,7 +405,7 @@ def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, data_root, log
             hypernetwork.save(last_saved_file)
 
         textual_inversion.write_loss(log_directory, "hypernetwork_loss.csv", hypernetwork.step, len(ds), {
-            "loss": f"{mean_loss:.7f}",
+            "loss": f"{previous_mean_loss:.7f}",
             "learn_rate": scheduler.learn_rate
         })
 
@@ -420,14 +454,15 @@ def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, data_root, log
 
         shared.state.textinfo = f"""
 <p>
-Loss: {mean_loss:.7f}<br/>
+Loss: {previous_mean_loss:.7f}<br/>
 Step: {hypernetwork.step}<br/>
 Last prompt: {html.escape(entries[0].cond_text)}<br/>
 Last saved hypernetwork: {html.escape(last_saved_file)}<br/>
 Last saved image: {html.escape(last_saved_image)}<br/>
 </p>
 """
-
+        
+    report_statistics(loss_dict)
     checkpoint = sd_models.select_checkpoint()
 
     hypernetwork.sd_checkpoint = checkpoint.hash
@@ -438,5 +473,3 @@ Last saved image: {html.escape(last_saved_image)}<br/>
     hypernetwork.save(filename)
 
     return hypernetwork, filename
-
-