feat: extended EPLB support for Mistral Large 3 and additional MoE backends (#48355)
Signed-off-by: jdebache <jdebache@nvidia.com>
This commit is contained in:
@@ -19,7 +19,12 @@ def test_shared_nvfp4_input_scales_have_writable_storage(monkeypatch):
|
||||
monkeypatch.setattr(flashinfer_fp4_moe, "swizzle_blockscale", lambda x: x)
|
||||
|
||||
num_experts = 3
|
||||
layer = SimpleNamespace(activation=SimpleNamespace(is_gated=False))
|
||||
layer = SimpleNamespace(
|
||||
activation=SimpleNamespace(is_gated=False),
|
||||
moe_config=SimpleNamespace(
|
||||
moe_parallel_config=SimpleNamespace(enable_eplb=False)
|
||||
),
|
||||
)
|
||||
w13 = torch.zeros((num_experts, 2, 1), dtype=torch.uint8)
|
||||
w2 = torch.zeros((num_experts, 2, 1), dtype=torch.uint8)
|
||||
w13_scale = torch.zeros((num_experts, 2, 1), dtype=torch.float8_e4m3fn)
|
||||
|
||||
Reference in New Issue
Block a user