Transformer-Explainability/baselines/ViT/ViT_explanation_generator.py

import argparse

import numpy as np
import torch
from numpy import *


# compute rollout between attention layers
def compute_rollout_attention(all_layer_matrices, start_layer=0):
    # adding residual consideration- code adapted from https://github.com/samiraabnar/attention_flow
    num_tokens = all_layer_matrices[0].shape[1]
    batch_size = all_layer_matrices[0].shape[0]
    eye = (
        torch.eye(num_tokens)
        .expand(batch_size, num_tokens, num_tokens)
        .to(all_layer_matrices[0].device)
    )
    all_layer_matrices = [
        all_layer_matrices[i] + eye for i in range(len(all_layer_matrices))
    ]
    matrices_aug = [
        all_layer_matrices[i] / all_layer_matrices[i].sum(dim=-1, keepdim=True)
        for i in range(len(all_layer_matrices))
    ]
    joint_attention = matrices_aug[start_layer]
    for i in range(start_layer + 1, len(matrices_aug)):
        joint_attention = matrices_aug[i].bmm(joint_attention)
    return joint_attention


class LRP:
    def __init__(self, model):
        self.model = model
        self.model.eval()

    def generate_LRP(
        self,
        input,
        index=None,
        method="transformer_attribution",
        is_ablation=False,
        start_layer=0,
    ):
        output = self.model(input)
        kwargs = {"alpha": 1}
        if index == None:
            index = np.argmax(output.cpu().data.numpy(), axis=-1)

        one_hot = np.zeros((1, output.size()[-1]), dtype=np.float32)
        one_hot[0, index] = 1
        one_hot_vector = one_hot
        one_hot = torch.from_numpy(one_hot).requires_grad_(True)
        one_hot = torch.sum(one_hot.cuda() * output)

        self.model.zero_grad()
        one_hot.backward(retain_graph=True)

        return self.model.relprop(
            torch.tensor(one_hot_vector).to(input.device),
            method=method,
            is_ablation=is_ablation,
            start_layer=start_layer,
            **kwargs
        )


class Baselines:
    def __init__(self, model):
        self.model = model
        self.model.eval()

    def generate_cam_attn(self, input, index=None):
        output = self.model(input.cuda(), register_hook=True)
        if index == None:
            index = np.argmax(output.cpu().data.numpy())

        one_hot = np.zeros((1, output.size()[-1]), dtype=np.float32)
        one_hot[0][index] = 1
        one_hot = torch.from_numpy(one_hot).requires_grad_(True)
        one_hot = torch.sum(one_hot.cuda() * output)

        self.model.zero_grad()
        one_hot.backward(retain_graph=True)
        #################### attn
        grad = self.model.blocks[-1].attn.get_attn_gradients()
        cam = self.model.blocks[-1].attn.get_attention_map()
        cam = cam[0, :, 0, 1:].reshape(-1, 14, 14)
        grad = grad[0, :, 0, 1:].reshape(-1, 14, 14)
        grad = grad.mean(dim=[1, 2], keepdim=True)
        cam = (cam * grad).mean(0).clamp(min=0)
        cam = (cam - cam.min()) / (cam.max() - cam.min())

        return cam
        #################### attn

    def generate_rollout(self, input, start_layer=0):
        self.model(input)
        blocks = self.model.blocks
        all_layer_attentions = []
        for blk in blocks:
            attn_heads = blk.attn.get_attention_map()
            avg_heads = (attn_heads.sum(dim=1) / attn_heads.shape[1]).detach()
            all_layer_attentions.append(avg_heads)
        rollout = compute_rollout_attention(
            all_layer_attentions, start_layer=start_layer
        )
        return rollout[:, 0, 1:]