Fix scheduler bugs (#177)

* fix scheduler bugs.

* fix

finetrainers/trainer.py

@@ -11,6 +11,7 @@ import diffusers
 import torch
 import torch.backends
 import transformers
+import wandb
 from accelerate import Accelerator, DistributedType
 from accelerate.logging import get_logger
 from accelerate.utils import (
@@ -29,8 +30,6 @@ from huggingface_hub import create_repo, upload_folder
 from peft import LoraConfig, get_peft_model_state_dict, set_peft_model_state_dict
 from tqdm import tqdm
 
-import wandb
-
 from .args import _INVERSE_DTYPE_MAP, Args, validate_args
 from .constants import (
     FINETRAINERS_LOG_LEVEL,
@@ -645,8 +644,14 @@ class Trainer:
             generator = generator.manual_seed(self.args.seed)
         self.state.generator = generator
 
-        scheduler_sigmas = get_scheduler_sigmas(self.scheduler).to(device=accelerator.device, dtype=torch.float32)
-        scheduler_alphas = get_scheduler_alphas(self.scheduler).to(device=accelerator.device, dtype=torch.float32)
+        scheduler_sigmas = get_scheduler_sigmas(self.scheduler)
+        scheduler_sigmas = (
+            scheduler_sigmas.to(device=accelerator.device, dtype=torch.float32) if scheduler_sigmas else None
+        )
+        scheduler_alphas = get_scheduler_alphas(self.scheduler)
+        scheduler_alphas = (
+            scheduler_alphas.to(device=accelerator.device, dtype=torch.float32) if scheduler_alphas else None
+        )
 
         for epoch in range(first_epoch, self.state.train_epochs):
             logger.debug(f"Starting epoch ({epoch + 1}/{self.state.train_epochs})")
@@ -751,6 +756,7 @@ class Trainer:
                     else:
                         # Default to flow-matching noise addition
                         noisy_latents = (1.0 - sigmas) * latent_conditions["latents"] + sigmas * noise
+                        noisy_latents = noisy_latents.to(latent_conditions["latents"].dtype)
 
                     latent_conditions.update({"noisy_latents": noisy_latents})
 

finetrainers/utils/diffusion_utils.py

@@ -121,7 +121,7 @@ def prepare_loss_weights(
     flow_weighting_scheme: str = "none",
 ) -> torch.Tensor:
     if isinstance(scheduler, FlowMatchEulerDiscreteScheduler):
-        return compute_loss_weighting_for_sd3(sigmas, weighting_scheme=flow_weighting_scheme)
+        return compute_loss_weighting_for_sd3(sigmas=sigmas, weighting_scheme=flow_weighting_scheme)
     elif isinstance(scheduler, CogVideoXDDIMScheduler):
         # SNR is computed as (alphas / (1 - alphas)), but for some reason CogVideoX uses 1 / (1 - alphas).
         # TODO(aryan): Experiment if using alphas / (1 - alphas) gives better results.