Refactor Qwen3 decode program to improve score handling

examples/qwen3/qwen3_32b_decode.py

@@ -243,16 +243,19 @@ def qwen3_decode_layer(
                             scores = pl.mul(pl.matmul(q_rot_bf16, k_tile, b_trans=True), ATTN_SCALE)
                             # TODO(valid_shape): once the compiler propagates valid_shape
                             # from k_tile, scores will auto-get vs=[1, valid_len] and the
-                            # manual scores_valid view + exp_pad can be removed.
-                            scores_valid = pl.slice(scores, [1, valid_len], [0, 0])
-                            cur_mi = pl.cast(pl.row_max(scores_valid), target_type=pl.FP32)
-                            exp_scores = pl.exp(pl.row_expand_sub(scores_valid, cur_mi))
+                            # manual scores_valid view can be removed.
+                            scores_valid = pl.slice(
+                                scores,
+                                [1, SEQ_TILE],
+                                [0, 0],
+                                valid_shape=[1, valid_len],
+                            )
+                            scores_padded = pl.fillpad(scores_valid, pad_value=pl.PadValue.min)
+                            cur_mi = pl.cast(pl.row_max(scores_padded), target_type=pl.FP32)
+                            exp_scores = pl.exp(pl.row_expand_sub(scores_padded, cur_mi))
                             cur_li = pl.cast(pl.row_sum(exp_scores), target_type=pl.FP32)
-                            exp_pad = pl.create_tensor([1, SEQ_TILE], dtype=pl.FP32)
-                            exp_pad = pl.mul(exp_pad, 0.0)
-                            exp_pad = pl.assemble(exp_pad, exp_scores, [0, 0])
                             oi_tmp = pl.matmul(
-                                pl.cast(exp_pad, target_type=pl.BF16),
+                                pl.cast(exp_scores, target_type=pl.BF16),
                                 v_tile,
                                 out_dtype=pl.FP32,
                             )