def play()

in src/model/self_play_all_vilbert.py [0:0]

• 105 lines of code
• 16 McCabe index (conditional complexity)

    def play(
        self, 
        qgen_img_feats,
        qgen_bboxs,
        image_features_rcnn_oracle, 
        bboxs_rcnn_oracle, 
        image_features_rcnn_gt_guesser,
        bboxs_rcnn_gt_guesser,
        tgt_cat, 
        tgt_bbox, 
        tgt_img_feat, 
        cats_guesser, 
        bboxs_mask, 
        sos_token, 
        pad_token, 
        eoq_token, 
        eod_token, 
        answer2id, 
        answer2token, 
        max_q_len, 
        greedy=True, 
        max_turns=8
        ):
        device = qgen_img_feats.device
        batch_size = qgen_img_feats.size(0)
        num_bboxs = qgen_img_feats.size(1)
        end_of_dialog = torch.zeros(batch_size).bool().to(device)
        last_wrd = torch.zeros(batch_size).fill_(sos_token).long().to(device)
        sos = torch.zeros(batch_size, 1).fill_(sos_token).long().to(device)
        last_state = None
        # pi = (torch.ones(batch_size, num_bboxs) / num_bboxs).to(device)
        pi = self.qgen.state_handler.init_state(batch_size, num_bboxs, device)

        dialog = [torch.LongTensor(0).to(device) for _ in range(batch_size)]
        q_log = [[] for _ in range(batch_size)]
        a_log = [[] for _ in range(batch_size)]
        a_conf_log = [[] for _ in range(batch_size)]

        guesser_state = self.guesser.init_state(
            batch_size, image_features_rcnn_gt_guesser.size(1), device)
        guesser_final_logits = torch.zeros_like(guesser_state)
        logits = torch.ones_like(guesser_final_logits)
        for turn in range(max_turns):

            q, q_len, state, end_of_dialog_next = self.qgen.generate_sentence(
                last_wrd, qgen_img_feats, qgen_bboxs, eoq_token, eod_token, end_of_dialog, 
                max_q_len=max_q_len, pi=pi, last_state=last_state, greedy=greedy
            )

            pad_q = pad_sequence(q, batch_first=True, padding_value=pad_token)
            q_plus_cls_token = torch.cat([sos, pad_q], dim=-1)
            # For oracle vilbert
            # +1 : [CLS] token
            txt_attn_mask = [[1] * (ql+1) + [0] * (q_len.max().item() - ql) for ql in q_len]
            txt_attn_mask = torch.tensor(txt_attn_mask).to(device)
            # a = self.oracle(pad_q, tgt_cat, tgt_bbox, tgt_img_feat, fake_q_len)
            a = self.oracle(
                q_plus_cls_token, 
                tgt_cat, 
                tgt_bbox, 
                tgt_img_feat, 
                bboxs_rcnn_oracle, 
                image_features_rcnn_oracle,
                attention_mask=txt_attn_mask
                )
            a_confidence = nn.functional.softmax(a, dim=-1)
            a_idx = a.argmax(dim=-1)


            for b in range(batch_size):
                if end_of_dialog_next[b]:
                    if guesser_final_logits[b].sum() == 0:
                        # last turn
                        guesser_final_logits[b] = logits[b]

            guesser_state, logits = self.guesser.forward_turn(
                q_plus_cls_token, 
                a_idx, 
                cats_guesser, 
                image_features_rcnn_gt_guesser,
                bboxs_rcnn_gt_guesser, 
                curr_state=guesser_state,
                bboxs_mask=bboxs_mask,
                attention_mask=txt_attn_mask,
                image_attention_mask=bboxs_mask.long(),
            )
            pi, _ = self.qgen.state_handler.forward_turn(
                q_plus_cls_token, 
                a_idx, 
                None, # cats 
                qgen_img_feats,
                qgen_bboxs, 
                curr_state=pi,
                attention_mask=txt_attn_mask,
                update_vilbert=False,
            )
            a = oracle_output_to_answer_token(a_idx, answer2id, answer2token)
            for b in range(batch_size):
                if not end_of_dialog[b]:
                    _q = q[b][:q_len[b]]
                    q_log[b].append(_q)
                    dialog[b] = torch.cat([dialog[b], _q])                    
                if not end_of_dialog_next[b]:
                    _a = a[b].view(-1)
                    a_log[b].append(_a)
                    a_conf_log[b].append(a_confidence[b, a_idx[b]])
                    dialog[b] = torch.cat([dialog[b], _a])

            if end_of_dialog_next.sum().item() == batch_size:
                break
            end_of_dialog = end_of_dialog_next
            last_wrd = a
            last_state = state
            # pi = self.qgen.refresh_pi(pi, a, last_state[0,0], obj_repr, input_token=True)
            
        for b in range(batch_size):
            if guesser_final_logits[b].sum() == 0:
                guesser_final_logits[b] = logits[b]
        # First one is global image token
        guess = guesser_final_logits[:, 1:]
        return guess, dialog, q_log, a_log, a_conf_log