[lm examples] fix overflow in perplexity calc (#11855)

* fix overflow in perplexity calc

* use inf

* fix

examples/pytorch/language-modeling/run_clm.py

@@ -440,7 +440,10 @@ def main():
 
         max_eval_samples = data_args.max_eval_samples if data_args.max_eval_samples is not None else len(eval_dataset)
         metrics["eval_samples"] = min(max_eval_samples, len(eval_dataset))
+        try:
             perplexity = math.exp(metrics["eval_loss"])
+        except OverflowError:
+            perplexity = float("inf")
         metrics["perplexity"] = perplexity
 
         trainer.log_metrics("eval", metrics)

examples/pytorch/language-modeling/run_clm_no_trainer.py

@@ -442,7 +442,10 @@ def main():
 
         losses = torch.cat(losses)
         losses = losses[: len(eval_dataset)]
+        try:
             perplexity = math.exp(torch.mean(losses))
+        except OverflowError:
+            perplexity = float("inf")
 
         logger.info(f"epoch {epoch}: perplexity: {perplexity}")
 

examples/pytorch/language-modeling/run_mlm.py

@@ -469,7 +469,10 @@ def main():
 
         max_eval_samples = data_args.max_eval_samples if data_args.max_eval_samples is not None else len(eval_dataset)
         metrics["eval_samples"] = min(max_eval_samples, len(eval_dataset))
+        try:
             perplexity = math.exp(metrics["eval_loss"])
+        except OverflowError:
+            perplexity = float("inf")
         metrics["perplexity"] = perplexity
 
         trainer.log_metrics("eval", metrics)

examples/pytorch/language-modeling/run_mlm_no_trainer.py

@@ -486,7 +486,10 @@ def main():
 
         losses = torch.cat(losses)
         losses = losses[: len(eval_dataset)]
+        try:
             perplexity = math.exp(torch.mean(losses))
+        except OverflowError:
+            perplexity = float("inf")
 
         logger.info(f"epoch {epoch}: perplexity: {perplexity}")
 

examples/pytorch/language-modeling/run_plm.py

@@ -445,7 +445,10 @@ def main():
 
         max_eval_samples = data_args.max_eval_samples if data_args.max_eval_samples is not None else len(eval_dataset)
         metrics["eval_samples"] = min(max_eval_samples, len(eval_dataset))
+        try:
             perplexity = math.exp(metrics["eval_loss"])
+        except OverflowError:
+            perplexity = float("inf")
         metrics["perplexity"] = perplexity
 
         trainer.log_metrics("eval", metrics)