misc

jessicazhongeee · jessicazhongeee · commit a80b7e5b9aae · 2025-01-17T14:14:42.000-08:00
diff --git a/recipes/dev/generate_v2.py b/recipes/dev/generate_v2.py
@@ -113,10 +113,10 @@ def log_metrics(self, total_time: int, tokens_per_second: float) -> None:
             f"Time for inference: {total_time:.02f} sec total, {tokens_per_second:.02f} tokens/sec"
         )
         self._logger.info(
-            f"Bandwidth achieved: {model_size * tokens_per_second / 1024 / 1024 / 1024:.02f} GiB/s"
+            f"Bandwidth achieved: {model_size * tokens_per_second / (1024**3):.02f} GiB/s"
         )
         self._logger.info(
-            f"Max memory allocated: {torch.cuda.max_memory_allocated() / 1024 / 1024 / 1024:.02f} GiB"
+            f"Max memory allocated: {torch.cuda.max_memory_allocated() / (1024**3):.02f} GiB"
         )
 
     @torch.inference_mode()
diff --git a/recipes/dev/generate_v2_distributed.py b/recipes/dev/generate_v2_distributed.py
@@ -154,10 +154,10 @@ def log_metrics(self, total_time: int, tokens_per_second: float) -> None:
             f"Time for inference: {total_time:.02f} sec total, {tokens_per_second:.02f} tokens/sec"
         )
         self._logger.info(
-            f"Bandwidth achieved: {model_size * tokens_per_second / 1024 / 1024 / 1024:.02f} GiB/s"
+            f"Bandwidth achieved: {model_size * tokens_per_second / (1024**3):.02f} GiB/s"
         )
         self._logger.info(
-            f"Max memory allocated: {torch.cuda.max_memory_allocated() / 1024 / 1024 / 1024 :.02f} GiB"
+            f"Max memory allocated: {torch.cuda.max_memory_allocated() / (1024**3):.02f} GiB"
         )
 
     @torch.inference_mode()

Original file line number	Diff line number	Diff line change
`@@ -113,10 +113,10 @@ def log_metrics(self, total_time: int, tokens_per_second: float) -> None:`
`113`	`113`	`f"Time for inference: {total_time:.02f} sec total, {tokens_per_second:.02f} tokens/sec"`
`114`	`114`	`)`
`115`	`115`	`self._logger.info(`
`116`		`- f"Bandwidth achieved: {model_size * tokens_per_second / 1024 / 1024 / 1024:.02f} GiB/s"`
	`116`	`+ f"Bandwidth achieved: {model_size * tokens_per_second / (1024**3):.02f} GiB/s"`
`117`	`117`	`)`
`118`	`118`	`self._logger.info(`
`119`		`- f"Max memory allocated: {torch.cuda.max_memory_allocated() / 1024 / 1024 / 1024:.02f} GiB"`
	`119`	`+ f"Max memory allocated: {torch.cuda.max_memory_allocated() / (1024**3):.02f} GiB"`
`120`	`120`	`)`
`121`	`121`
`122`	`122`	`@torch.inference_mode()`
Original file line number	Diff line number	Diff line change
`@@ -154,10 +154,10 @@ def log_metrics(self, total_time: int, tokens_per_second: float) -> None:`
`154`	`154`	`f"Time for inference: {total_time:.02f} sec total, {tokens_per_second:.02f} tokens/sec"`
`155`	`155`	`)`
`156`	`156`	`self._logger.info(`
`157`		`- f"Bandwidth achieved: {model_size * tokens_per_second / 1024 / 1024 / 1024:.02f} GiB/s"`
	`157`	`+ f"Bandwidth achieved: {model_size * tokens_per_second / (1024**3):.02f} GiB/s"`
`158`	`158`	`)`
`159`	`159`	`self._logger.info(`
`160`		`- f"Max memory allocated: {torch.cuda.max_memory_allocated() / 1024 / 1024 / 1024 :.02f} GiB"`
	`160`	`+ f"Max memory allocated: {torch.cuda.max_memory_allocated() / (1024**3):.02f} GiB"`
`161`	`161`	`)`
`162`	`162`
`163`	`163`	`@torch.inference_mode()`