Update app.py

app.py

@@ -13,6 +13,8 @@ You can also use 🐣e5-mistral🛌🏻 by cloning this space. 🧬🔬🔍 Simp
 Join us : 🌟TeamTonic🌟 is always making cool demos! Join our active builder's🛠️community on 👻Discord:  [![Let's build the future of AI together! 🚀🤖](https://discordapp.com/api/guilds/1109943800132010065/widget.png)](https://discord.gg/GWpVpekp) On 🤗Huggingface: [TeamTonic](https://huggingface.co/TeamTonic) & [MultiTransformer](https://huggingface.co/MultiTransformer) On 🌐Github: [Polytonic](https://github.com/tonic-ai) & contribute to 🌟 [Poly](https://github.com/tonic-ai/poly)
 """
 
+os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb:50'
+
 def last_token_pool(last_hidden_states: Tensor, attention_mask: Tensor) -> Tensor:
     left_padding = (attention_mask[:, -1].sum() == attention_mask.shape[0])
     if left_padding:
@@ -28,6 +30,9 @@ def get_detailed_instruct(task_description: str, query: str) -> str:
 
 @spaces.GPU
 def compute_embeddings(*input_texts):
+    torch.backends.cuda.matmul.allow_tf32 = True
+    torch.backends.cudnn.allow_tf32 = True
+    torch.backends.cudnn.benchmark = True
     tokenizer = AutoTokenizer.from_pretrained('intfloat/e5-mistral-7b-instruct')
     model = AutoModel.from_pretrained('intfloat/e5-mistral-7b-instruct')
     device = torch.device("cuda" if torch.cuda.is_available() else "cpu")