CI hotfix: IFU test update (#329)

Author: Micky774

ci/pytorch.sh

@@ -58,7 +58,7 @@ run_test_config(){
     run_default_fa 1 test_deferred_init.py
     run_default_fa 1 test_float8tensor.py
     run_default_fa 1 test_float8_current_scaling_exact.py
-    run_default_fa 1 test_cpu_offloading.py
+    test $_fus_attn = auto -o $_fus_attn = ck -o $_fus_attn = aotriton && NVTE_FLASH_ATTN=0 run 1 test_cpu_offloading.py
     run_default_fa 1 test_fused_rope.py
     run_default_fa 1 test_fusible_ops.py
     run_default_fa 3 test_gemm_autotune.py

tests/pytorch/test_cpu_offloading.py

@@ -29,7 +29,7 @@
 
 # Flash attention saves some internal tensor for the backward pass
 # that cannot be offloaded to CPU.
-assert os.getenv("NVTE_FLASH_ATTN") == "0"
+assert os.getenv("NVTE_FLASH_ATTN", "1") == "0"
 
 # Offloading is supported for attention only for fused and flash attention backends,
 # so the use of bfloat16 is required.