tensorflow/models · error · ValueError

Unknown `attn_type` {}.

Error message

Unknown `attn_type` {}.

What it means

Error "Unknown `attn_type` {}." thrown in tensorflow/models.

Source

Thrown at official/legacy/xlnet/xlnet_modeling.py:665

            tf.equal(cat_id, tf.constant([data_utils.SEG_ID_CLS]))[None, :])
        seg_mat = tf.equal(seg_id[:, None], cat_id[None, :])
        seg_mat = tf.logical_or(cls_mat, seg_mat)
      else:
        seg_mat = tf.logical_not(tf.equal(seg_id[:, None], cat_id[None, :]))
    else:
      seg_mat = None

    dtype = self.tf_float
    freq_seq = tf.range(0, self.d_model, 2.0)
    if dtype is not None and dtype != tf.float32:
      freq_seq = tf.cast(freq_seq, dtype=self.dtype)

    if self.attn_type == "bi":
      beg, end = klen, -qlen
    elif self.attn_type == "uni":
      beg, end = klen, -1
    else:
      raise ValueError("Unknown `attn_type` {}.".format(self.attn_type))

    if self.bi_data:
      fwd_pos_seq = tf.range(beg, end, -1.0)
      bwd_pos_seq = tf.range(-beg, -end, 1.0)

      if dtype is not None and dtype != tf.float32:
        fwd_pos_seq = tf.cast(fwd_pos_seq, dtype=dtype)
        bwd_pos_seq = tf.cast(bwd_pos_seq, dtype=dtype)

      if self.clamp_len > 0:
        fwd_pos_seq = tf.clip_by_value(fwd_pos_seq, -self.clamp_len,
                                       self.clamp_len)
        bwd_pos_seq = tf.clip_by_value(bwd_pos_seq, -self.clamp_len,
                                       self.clamp_len)

      if bsz is not None:
        fwd_pos_emb = self.fwd_position_embedding(fwd_pos_seq, bsz // 2)
        bwd_pos_emb = self.bwd_position_embedding(bwd_pos_seq, bsz // 2)

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/legacy/xlnet/xlnet_modeling.py:665 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of tensorflow/models@e006f5f0d5 (2026-08-24). Data as JSON: /api/errors/9532cde747ebe216. Report an issue: GitHub.