import torch
import torch.nn.functional as F
from transformers import AutoTokenizer, AutoModel, AutoImageProcessor
import gradio as gr
import spaces

model = AutoModel.from_pretrained('neuralmind/bert-base-portuguese-cased')
tokenizer = AutoTokenizer.from_pretrained('neuralmind/bert-base-portuguese-cased', do_lower_case=False)

# processor = AutoImageProcessor.from_pretrained("nomic-ai/nomic-embed-vision-v1.5")
# vision_model = AutoModel.from_pretrained("nomic-ai/nomic-embed-vision-v1.5", trust_remote_code=True)

# tokenizer = AutoTokenizer.from_pretrained('nomic-ai/nomic-embed-text-v1.5')
# text_model = AutoModel.from_pretrained('nomic-ai/nomic-embed-text-v1.5', trust_remote_code=True)
# text_model.eval()

def mean_pooling(model_output, attention_mask):
    token_embeddings = model_output[0]
    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)
        
@spaces.GPU
def TxtEmbed(text):
    import torch

    
    input_ids = tokenizer.encode(text, return_tensors='pt')
    
    with torch.no_grad():
        outs = model(input_ids)
        encoded = outs[0][0, 1:-1]  # Ignore [CLS] and [SEP] special tokens


    
   # sentences = [text]
   # encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
   # 
   # with torch.no_grad():
   #     model_output = text_model(**encoded_input)
   # 
   # text_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
   # text_embeddings = F.layer_norm(text_embeddings, normalized_shape=(text_embeddings.shape[1],))
   # text_embeddings = F.normalize(text_embeddings, p=2, dim=1)
    
    return (encoded.tolist())[0];




with gr.Blocks() as demo:      
       txt = gr.Text();
       out = gr.Text();
       
       btn = gr.Button("Gerar embeddings")
       btn.click(TxtEmbed, [txt], [out])
       
       
if __name__ == "__main__":
    demo.launch(show_api=True)