app.py

import gradio as gr
import peft
from peft import LoraConfig
from transformers import AutoTokenizer,BitsAndBytesConfig, AutoModelForCausalLM, CLIPVisionModel, AutoProcessor
import torch
from peft import PeftModel

clip_model_name = "openai/clip-vit-base-patch32"
phi_model_name  = "microsoft/phi-2"
tokenizer  = AutoTokenizer.from_pretrained(phi_model_name, trust_remote_code=True)
processor  = AutoProcessor.from_pretrained(clip_model_name)
tokenizer.pad_token = tokenizer.eos_token
IMAGE_TOKEN_ID = 23893 # token for word comment
device = "cuda" if torch.cuda.is_available() else "cpu"
clip_embed = 768
phi_embed  = 2560

class SimpleResBlock(nn.Module):
    def __init__(self, phi_embed):
        super().__init__()
        self.pre_norm = nn.LayerNorm(phi_embed)
        self.proj = nn.Sequential(
            nn.Linear(phi_embed, phi_embed),
            nn.GELU(),
            nn.Linear(phi_embed, phi_embed)
        )
    def forward(self, x):
        x = self.pre_norm(x)
        return x + self.proj(x)
        
# models
clip_model = CLIPVisionModel.from_pretrained(clip_model_name).to(device)
projection = torch.nn.Linear(clip_embed, phi_embed).to(device)
resblock = SimpleResBlock(phi_embed).to(device)

phi_model = AutoModelForCausalLM.from_pretrained(phi_model_name,trust_remote_code=True).to(device)

# load weights
model_to_merge = PeftModel.from_pretrained(phi_model,'./model_chkpt/lora_adaptor')
merged_model   = model_to_merge.merge_and_unload()
projection.load_state_dict(torch.load('./model_chkpt/step2_projection.pth',map_location=torch.device(device)))
resblock.load_state_dict(torch.load('./model_chkpt/step2_resblock.pth',map_location=torch.device(device)))

def model_generate_ans(img,val_q):

    max_generate_length = 100
    
    # image
    image_processed  = processor(images=img, return_tensors="pt").to(device)
    clip_val_outputs = clip_model(**image_processed).last_hidden_state[:,1:,:]
    val_image_embeds = projection(clip_val_outputs)
    val_image_embeds = resblock(val_image_embeds).to(torch.float16)

    img_token_tensor = torch.tensor(IMAGE_TOKEN_ID).to(device)
    img_token_embeds = merged_model.model.embed_tokens(img_token_tensor).unsqueeze(0).unsqueeze(0)

    val_q_tokenised = tokenizer(val_q, return_tensors="pt", return_attention_mask=False)['input_ids'].squeeze(0)
    val_q_embeds    = merged_model.model.embed_tokens(val_q_tokenised).unsqueeze(0)
    
    val_combined_embeds = torch.cat([val_image_embeds, img_token_embeds, val_q_embeds], dim=1) # 4, 69, 2560
    
    predicted_caption = torch.full((1,max_generate_length),50256)

    for g in range(max_generate_length):
        phi_output_logits = merged_model(inputs_embeds=val_combined_embeds)['logits'] # 4, 69, 51200
        predicted_word_token_logits = phi_output_logits[:, -1, :].unsqueeze(1) # 4,1,51200
        predicted_word_token = torch.argmax(predicted_word_token_logits, dim = -1) # 4,1
        predicted_caption[:,g] = predicted_word_token.view(1,-1).to(device)
        next_token_embeds = phi_model.model.embed_tokens(predicted_word_token) # 4,1,2560
        val_combined_embeds   = torch.cat([val_combined_embeds, next_token_embeds], dim=1)
        
    predicted_captions_decoded = tokenizer.batch_decode(predicted_caption,ignore_index = 50256)
    
    return predicted_captions_decoded
    

with gr.Blocks() as demo:

    gr.Markdown(
    """
    # Chat with MultiModal GPT !
    Build using combining clip model and phi-2 model.
    """
    )

    # app GUI
    with gr.Row():
        with gr.Column():
            img_input    = gr.Image(label='Image')
            img_question = gr.Text(label ='Question')
        with gr.Column():
            img_answer   = gr.Text(label ='Answer')

    section_btn = gr.Button("Submit")
    section_btn.click(model_generate_ans, inputs=[img_input,img_question], outputs=[img_answer])
    
if __name__ == "__main__":
    demo.launch()