From ce8b77f54cf5576108d06309706cab34fe94f630 Mon Sep 17 00:00:00 2001
From: Daniel Garvey <34486624+dan-garvey@users.noreply.github.com>
Date: Sat, 2 Nov 2024 12:35:18 -0500
Subject: [PATCH] Revert "temporary decompose for decode" (#409)

Reverts nod-ai/SHARK-Platform#353
---
 sharktank/sharktank/models/llama/llama.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/sharktank/sharktank/models/llama/llama.py b/sharktank/sharktank/models/llama/llama.py
index 656b4432b..ef3c4800d 100644
--- a/sharktank/sharktank/models/llama/llama.py
+++ b/sharktank/sharktank/models/llama/llama.py
@@ -269,7 +269,6 @@ def decode(
         for block_idx, block in enumerate(self.attn_blocks):
             if block_idx == 0:
                 self.trace_tensor(f"llama.attn_block.{block_idx}.input", h)
-            block.attn.attention_kernel = "decomposed"
             h = block(
                 h,
                 start_positions=start_positions,