Update sdp_on_bf16 option for ST example

examples/sentence-transformers-training/nli/training_nli.py

@@ -103,6 +103,7 @@ def main():
         warmup_ratio=0.1,
         # fp16=True,  # Set to False if you get an error that your GPU can't run on FP16
         bf16=args.bf16,  # Set to True if you have a GPU that supports BF16
+        # sdp_on_bf16=True, #Set to True for better performance (but this setting can affect accuracy)
         # Optional tracking/debugging parameters:
         evaluation_strategy="steps",
         eval_steps=100,

examples/sentence-transformers-training/nli/training_nli_v2.py

@@ -81,6 +81,7 @@ def main():
         warmup_ratio=0.1,
         # fp16=True,  # Set to False if you get an error that your GPU can't run on FP16
         # bf16=False,  # Set to True if you have a GPU that supports BF16
+        # sdp_on_bf16=True, #Set to True for better performance (but this setting can affect accuracy)
         batch_sampler=BatchSamplers.NO_DUPLICATES,
         # Optional tracking/debugging parameters:
         evaluation_strategy="steps",

examples/sentence-transformers-training/nli/training_nli_v3.py

@@ -82,6 +82,7 @@ def main():
         warmup_ratio=0.1,
         # fp16=True,  # Set to False if you get an error that your GPU can't run on FP16
         # bf16=False,  # Set to True if you have a GPU that supports BF16
+        # sdp_on_bf16=True, #Set to True for better performance (but this setting can affect accuracy)
         batch_sampler=BatchSamplers.NO_DUPLICATES,
         # Optional tracking/debugging parameters:
         evaluation_strategy="steps",

examples/sentence-transformers-training/paraphrases/training_paraphrases.py

@@ -101,6 +101,7 @@
     warmup_ratio=0.1,
     # fp16=True,  # Set to False if you get an error that your GPU can't run on FP16
     # bf16=False,  # Set to True if you have a GPU that supports BF16
+    # sdp_on_bf16=True, #Set to True for better performance (but this setting can affect accuracy)
     batch_sampler=BatchSamplers.NO_DUPLICATES,  # MultipleNegativesRankingLoss benefits from no duplicate samples in a batch
     # We can use ROUND_ROBIN or PROPORTIONAL - to avoid focusing too much on one dataset, we will
     # use round robin, which samples the same amount of batches from each dataset, until one dataset is empty

examples/sentence-transformers-training/sts/training_stsbenchmark.py

@@ -100,6 +100,7 @@ def main():
         warmup_ratio=0.1,
         # fp16=True,  # Set to False if you get an error that your GPU can't run on FP16
         bf16=args.bf16,  # Set to True if you have a GPU that supports BF16
+        # sdp_on_bf16=True, #Set to True for better performance (but this setting can affect accuracy)
         # Optional tracking/debugging parameters:
         evaluation_strategy="steps",
         eval_steps=100,

examples/sentence-transformers-training/sts/training_stsbenchmark_continue_training.py

@@ -75,6 +75,7 @@ def main():
         warmup_ratio=0.1,
         # fp16=True,  # Set to False if you get an error that your GPU can't run on FP16
         # bf16=False,  # Set to True if you have a GPU that supports BF16
+        # sdp_on_bf16=True, #Set to True for better performance (but this setting can affect accuracy)
         # Optional tracking/debugging parameters:
         evaluation_strategy="steps",
         eval_steps=100,