def convert_text_to_embeddings()

in backend/matching-engine/services/text_to_image_match_service.py [0:0]

• 8 lines of code
• 2 McCabe index (conditional complexity)

    def convert_text_to_embeddings(self, target: str) -> Optional[List[float]]:
        # create transformer-readable tokens
        inputs = self.tokenizer(target, return_tensors="pt").to(self.device)

        # use CLIP to encode tokens into a meaningful embedding
        text_emb = self.model.get_text_features(**inputs)
        text_emb = text_emb.cpu().detach().numpy()

        if np.any(text_emb):
            return text_emb[0].tolist()
        else:
            return None