From 724fc621bfbb9bfa1db790e39163dc5f7edd1b93 Mon Sep 17 00:00:00 2001
From: Wuwei Lin <wuwei@apache.org>
Date: Fri, 15 Dec 2023 19:19:58 +0000
Subject: [PATCH] cleanup

---
 mlc_llm/relax_model/mixtral.py | 40 ----------------------------------
 1 file changed, 40 deletions(-)

diff --git a/mlc_llm/relax_model/mixtral.py b/mlc_llm/relax_model/mixtral.py
index 9b3fde0fae..16839178cd 100644
--- a/mlc_llm/relax_model/mixtral.py
+++ b/mlc_llm/relax_model/mixtral.py
@@ -150,46 +150,6 @@ def __init__(self, config: MixtralConfig):
         self.dtype = config.dtype
         self.hidden_size = config.hidden_size
 
-    def topk(self, x, is_ascend, index_dtype, k=-1):
-        if not is_ascend and k == 2:
-            # fast path
-            total_rows = x.struct_info.shape[0]
-            result = nn.emit(
-                relax.call_dps_packed(
-                    "top2",
-                    [x],
-                    out_sinfo=[
-                        relax.TensorStructInfo([total_rows, k], x.struct_info.dtype),
-                        relax.TensorStructInfo([total_rows, k], index_dtype),
-                    ],
-                )
-            )
-            return relax.TupleGetItem(result, 0), relax.TupleGetItem(result, 1)
-
-        # topk along axis -1
-        result = nn.emit(
-            relax.call_dps_packed(
-                "tvm.contrib.thrust.sort_dps",
-                [x, is_ascend],
-                out_sinfo=[
-                    x.struct_info,
-                    relax.TensorStructInfo(x.struct_info.shape, index_dtype),
-                ],
-            )
-        )
-        sorted_x = relax.TupleGetItem(result, 0)
-        indices = relax.TupleGetItem(result, 1)
-        if k != -1:
-            ndim = len(x.struct_info.shape)
-            beg = [0] * ndim
-            end = [x.struct_info.shape[i] for i in range(ndim - 1)] + [k]
-            axes = list(range(ndim))
-            sorted_x = nn.emit(
-                relax.op.strided_slice(sorted_x, axes, beg, end, assume_inbound=True)
-            )
-            indices = nn.emit(relax.op.strided_slice(indices, axes, beg, end, assume_inbound=True))
-        return sorted_x, indices
-
     def scatter(self, linear_out, indices):
         @T.prim_func
         def scatter_func(