[Enhance] rename consumed_tokens to step_* in other_log for consistency

Author: nil0x9

tests/train/test_trainer.py

@@ -57,7 +57,7 @@ def train_step(self, *args, **kwargs):
         self.train_step_calls += 1
         return (
             {"local_loss": 1.0, "reduced_llm_loss": 0.8},
-            {"consumed_tokens": 100, "grad_norm": torch.tensor(1.0), "efficient_attn_ratio": 0.5}
+            {"step_consumed_tokens": 100, "grad_norm": torch.tensor(1.0), "efficient_attn_ratio": 0.5}
         )
 
     def save_hf(self, hf_path):

xtuner/v1/engine/train_engine.py

@@ -53,8 +53,8 @@ class LossLog(TypedDict):
 class OtherLog(TypedDict):
     __pydantic_config__ = ConfigDict(arbitrary_types_allowed=True)  # type: ignore[misc]
     maxvio: NotRequired[float]
-    consumed_tokens: int
-    consumed_img_tokens: NotRequired[float]
+    step_consumed_tokens: int
+    step_consumed_img_tokens: NotRequired[float]
     extra_info: ModelForwardExtraLogInfo
     efficient_attn_ratio: float
 
@@ -347,7 +347,7 @@ def train_step(self, data_batches: list[ModelItem]) -> tuple[LossLog, OtherLog]:
             reduced_z_loss = step_z_loss
             dist.all_reduce(reduced_z_loss.div_(dist.get_world_size()))
             loss_log["reduced_z_loss"] = reduced_z_loss.item()
-        other_log["consumed_tokens"] = cast(int, step_consumed_tokens.item())
+        other_log["step_consumed_tokens"] = cast(int, step_consumed_tokens.item())
         other_log["extra_info"] = train_engine_extra_info
         other_log["efficient_attn_ratio"] = (efficient_forward_tokens / total_forward_tokens).item()
         return loss_log, other_log

xtuner/v1/engine/vision_compose_train_engine.py

@@ -220,8 +220,8 @@ def train_step(self, data_batches: List[ModelItem]) -> tuple[LossLog, OtherLog]:
             reduced_z_loss = step_z_loss
             dist.all_reduce(reduced_z_loss.div_(dist.get_world_size()))
             loss_log["reduced_z_loss"] = reduced_z_loss.item()
-        other_log["consumed_tokens"] = cast(int, step_consumed_tokens.item())
+        other_log["step_consumed_tokens"] = cast(int, step_consumed_tokens.item())
         other_log["extra_info"] = train_engine_extra_info  # type: ignore[assignment]
         other_log["efficient_attn_ratio"] = (efficient_forward_tokens / total_forward_tokens).item()
-        other_log["consumed_img_tokens"] = step_consumed_img_tokens
+        other_log["step_consumed_img_tokens"] = step_consumed_img_tokens
         return loss_log, other_log

xtuner/v1/train/trainer.py

@@ -760,8 +760,8 @@ def fit(self):
             time_after_train_step = time.time()
             ProberList.after_step()
             step_time = time_after_train_step - time_before_train_step
-            step_consumed_tokens = other_log["consumed_tokens"]
-            step_consumed_img_tokens = other_log.get("consumed_img_tokens", None)
+            step_consumed_tokens = other_log["step_consumed_tokens"]
+            step_consumed_img_tokens = other_log.get("step_consumed_img_tokens", None)
 
             extra_info = other_log.get("extra_info", {})
             if isinstance(extra_info, ModelForwardExtraLogInfo):