combine_input = self.run_moe_core(
  File "/home/jobuser/zminglei/sglang/python/sglang/srt/layers/moe/fused_moe_triton/layer.py", line 980, in run_moe_core
    return self.quant_method.apply(
  File "/home/jobuser/zminglei/sglang/python/sglang/srt/layers/quantization/mxfp4.py", line 716, in apply
    return self.runner.run(dispatch_output, quant_info)
  File "/home/jobuser/zminglei/sglang/python/sglang/srt/layers/moe/moe_runner/runner.py", line 78, in run
    return self.fused_func(dispatch_output, quant_info, self.config)
  File "/home/jobuser/zminglei/sglang/python/sglang/srt/layers/moe/moe_runner/triton.py", line 339, in fused_experts_none_to_triton
    output = fused_experts(
  File "/home/jobuser/zminglei/sglang/python/sglang/srt/layers/moe/fused_moe_triton/fused_moe.py", line 213, in fused_experts
    inplace_fused_experts(
  File "/home/jobuser/zminglei/sglang/venv/lib/python3.10/site-packages/torch/_ops.py", line 1255, in __call__
    return self._op(*args, **kwargs)
  File "/home/jobuser/zminglei/sglang/python/sglang/srt/layers/moe/fused_moe_triton/fused_moe.py", line 91, in inplace_fused_experts
    fused_experts_impl(
  File "/home/jobuser/zminglei/sglang/python/sglang/srt/layers/moe/fused_moe_triton/fused_moe.py", line 526, in fused_experts_impl
    raise ValueError(f"Unsupported activation: {activation=}, with {is_gated=}")
ValueError: Unsupported activation: activation='npu_swiglu_oai', with is_gated=True

[2026-01-21 22:09:40] INFO:     Application startup complete.
[2026-01-21 22:09:40] INFO:     Uvicorn running on http://127.0.0.1:30000 (Press CTRL+C to quit)
[2026-01-21 22:09:41] INFO:     127.0.0.1:54606 - "GET /model_info HTTP/1.1" 200 OK
[2026-01-21 22:09:41 TP0] Prefill batch, #new-seq: 1, #new-token: 6, #cached-token: 0, full token usage: 0.00, swa token usage: 0.00, #running-req: 0, #queue-req: 0,
[2026-01-21 22:09:42] INFO:     127.0.0.1:54614 - "POST /generate HTTP/1.1" 200 OK
[2026-01-21 22:09:42] The server is fired up and ready to roll!