Trainer evaluation delay (#16356)

* Initial commit

* Reversed signs, adjusted log entery.

* Check only when

* Cleanup checks

* Only trigger if we want to eval

* Run

* Move changes to callback

src/transformers/trainer_callback.py

@@ -416,7 +416,7 @@ class DefaultFlowCallback(TrainerCallback):
             control.should_log = True
 
         # Evaluate
-        if args.evaluation_strategy == IntervalStrategy.STEPS and state.global_step % args.eval_steps == 0:
+        if args.evaluation_strategy == IntervalStrategy.STEPS and state.global_step % args.eval_steps == 0 and args.eval_delay > state.global_step:
             control.should_evaluate = True
 
         # Save
@@ -439,7 +439,7 @@ class DefaultFlowCallback(TrainerCallback):
             control.should_log = True
 
         # Evaluate
-        if args.evaluation_strategy == IntervalStrategy.EPOCH:
+        if args.evaluation_strategy == IntervalStrategy.EPOCH and args.eval_delay > state.epoch:
             control.should_evaluate = True
 
         # Save

src/transformers/training_args.py

@@ -138,6 +138,8 @@ class TrainingArguments:
             Number of predictions steps to accumulate the output tensors for, before moving the results to the CPU. If
             left unset, the whole predictions are accumulated on GPU/TPU before being moved to the CPU (faster but
             requires more memory).
+        eval_delay (`float`, *optional*):
+            Number of epochs or steps to wait for before the first evaluation can be performed, depending on the evaluation_strategy.
         learning_rate (`float`, *optional*, defaults to 5e-5):
             The initial learning rate for [`AdamW`] optimizer.
         weight_decay (`float`, *optional*, defaults to 0):
@@ -472,6 +474,11 @@ class TrainingArguments:
         metadata={"help": "Number of predictions steps to accumulate before moving the tensors to the CPU."},
     )
 
+    eval_delay: Optional[float] = field(
+        default=0,
+        metadata={"help": "Number of epochs or steps to wait for before the first evaluation can be performed, depending on the evaluation_strategy."},
+    )            
+
     learning_rate: float = field(default=5e-5, metadata={"help": "The initial learning rate for AdamW."})
     weight_decay: float = field(default=0.0, metadata={"help": "Weight decay for AdamW if we apply some."})
     adam_beta1: float = field(default=0.9, metadata={"help": "Beta1 for AdamW optimizer"})