Update README.md

README.md

@@ -30,7 +30,6 @@ import mlx.core as mx
 
 model, tokenizer = load("mlx-community/embeddinggemma-300m-qat-q4_0-unquantized-bf16")
 
-
 # For text embedding
 sentences = [
     "task: sentence similarity | query: Nothing really matters.",
@@ -43,10 +42,15 @@ encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tenso
 # Compute token embeddings
 input_ids = encoded_input['input_ids']
 attention_mask = encoded_input['attention_mask']
-model_output = model(input_ids, attention_mask)
+output = model(input_ids, attention_mask)
+
+embeddings = output.text_embeds  # Normalized embeddings
 
-print("Sentence embeddings:")
-print(model_output.text_embeds)
+# Compute dot product between normalized embeddings
+similarity_matrix = mx.matmul(embeddings, embeddings.T)
+
+print("Similarity matrix between texts:")
+print(similarity_matrix)
 
 
 # You can use these task-specific prefixes for different tasks
@@ -66,5 +70,4 @@ task_prefixes = {
     "document": "title: none | text: "
 }
 
-
 ```