[AMD][ROCm] Enable DeepSeek model on ROCm (#12662)

Signed-off-by: Hongxia Yang <hongxia.yang@amd.com>
Co-authored-by: Matthew Wong <Matthew.Wong2@amd.com>
This commit is contained in:
Hongxia Yang
2025-02-04 08:24:11 +00:00
committed by GitHub
co-authored by Matthew Wong
parent 4896d0c2dd
commit c36ac98d01
5 changed files with 58 additions and 1 deletions
+9
View File
@@ -24,6 +24,15 @@ def _create_model_runner(model: str, *args, **kwargs) -> ModelRunner:
return model_runner
def test_deepseek_mla_attn_backend_module():
model_runner = _create_model_runner(
"deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct",
trust_remote_code=True,
enable_chunked_prefill=False,
)
assert model_runner.attn_backend.__name__ == "TritonMLABackend"
@pytest.mark.parametrize("batch_size", list(range(1, 257)))
def test_prepare_prompt(batch_size):
model_runner = _create_model_runner(