(anna) D:\Local\Anna>python -c "import torch; print(torch.__version__); print(torch.xpu.is_available()); print(torch.xpu.get_device_name(0) if torch.xpu.is_available() else None)"
2.8.0+xpu
True
Intel(R) Arc(TM) B580 Graphics
(anna) D:\Local\Anna>set ANNA_DPCPP=C:\Program Files (x86)\Intel\oneAPI\compiler\latest\bin\dpcpp.exe
(anna) D:\Local\Anna>set ANNA_VCVARS64=C:\Program Files\Microsoft Visual Studio\2022\Professional\VC\Auxiliary\Build\vcvars64.bat
(anna) D:\Local\Anna>python tools\build_gated_delta_fused_op.py
Compiling Anna fused XPU/SYCL ops...
C:\Program Files (x86)\Intel\oneAPI\compiler\2026.1\bin\dpcpp.exe -fsycl -fsycl-targets=spir64 -std=c++17 -shared -O2 -Wno-ignored-attributes -Wno-deprecated-declarations D:\Local\Anna\src\anna\model\custom_ops\gated_delta_fused_op.cpp -ID:\miniconda3\envs\anna\Lib\site-packages\torch\include -ID:\miniconda3\envs\anna\Lib\site-packages\torch\include\torch\csrc\api\include -ID:\miniconda3\envs\anna\include -LD:\miniconda3\envs\anna\Lib\site-packages\torch\lib -LC:\Program Files (x86)\Intel\oneAPI\compiler\2026.1\bin -LC:\Program Files (x86)\Intel\oneAPI\compiler\2026.1\lib -lc10 -lc10_xpu -ltorch_cpu -ltorch_xpu -ltorch -o D:\Local\Anna\.build\anna_gated_delta_fused\anna_gated_delta_fused.pyd
icpx: warning: use of 'dpcpp' is deprecated and will be removed in a future release. Use 'icx -fsycl' [-Wdeprecated]
Creando biblioteca D:\Local\Anna\.build\anna_gated_delta_fused\anna_gated_delta_fused.lib y objeto D:\Local\Anna\.build\anna_gated_delta_fused\anna_gated_delta_fused.exp
library_path=D:\Local\Anna\.build\anna_gated_delta_fused\anna_gated_delta_fused.pyd
gqa_decode_registered=True
gqa_decode_splitkv_registered=True
gqa_decode_splitkv_out_registered=True
gqa_decode_splitkv_turboquant_out_registered=True
paged_gqa_decode_registered=True
moe_router_registered=True
moe_dispatch_registered=True
moe_scatter_registered=True
moe_grouped_int4_mlp_registered=True
rmsnorm_registered=True
rmsnorm_ex_registered=True
rmsnorm_gated_registered=True
qk_norm_rotary_registered=True
qk_norm_rotary_ex_registered=True
gated_delta_registered=True
flashqla_gated_delta_registered=True
flashqla_chunk_local_cumsum_registered=True
flashqla_cumsum_kkt_build_registered=True
flashqla_kkt_build_registered=True
flashqla_kkt_solve_registered=True
flashqla_wu_build_registered=True
flashqla_solve_wu_build_registered=True
flashqla_chunk_gdr_fwd_registered=True
causal_conv1d_registered=True
(anna) D:\Local\Anna>set PYTHONPATH = "D:\Local\Anna\src"
(anna) D:\Local\Anna>set ANNA_GATED_DELTA_OP_LIB = "D:\Local\Anna\.build\anna_gated_delta_fused\anna_gated_delta_fused.pyd"
(anna) D:\Local\Anna>
(anna) D:\Local\Anna>anna-serve --model-dir D:\Model\Qwen3.5-4B --model-name qwen3.5 --device xpu --dtype bf16 --host 12
7.0.0.1 --port 8000
2026-07-09 17:08:08,883 INFO anna.cli.xpu_env: Configured XPU environment: {'UR_L0_ENABLE_RELAXED_ALLOCATION_LIMITS': '1', 'ZES_ENABLE_SYSMAN': '1'}
2026-07-09 17:08:09,667 INFO anna.runtime.device: Resolved XPU device info: {'device_index': 0, 'name': 'Intel(R) Arc(TM) B580 Graphics', 'total_memory': None, 'free_memory': None, 'allocated_memory': 0, 'reserved_memory': 0, 'runtime': 'torch.xpu', 'device_type': 'arc', 'is_arc_alchemist': True, 'is_acm_g10': False, 'is_arc_a770_or_a750': False}
Traceback (most recent call last):
File "<frozen runpy>", line 198, in _run_module_as_main
File "<frozen runpy>", line 88, in _run_code
File "D:\miniconda3\envs\anna\Scripts\anna-serve.exe\__main__.py", line 5, in <module>
File "D:\Local\Anna\src\anna\cli\serve.py", line 502, in main
engine = load_model_runtime_from_model_dir(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Local\Anna\src\anna\runtime\model_runtime_loader.py", line 157, in load_model_runtime_from_model_dir
return AnnaQwen3_5TextEngine.from_model_dir(model_dir, **shared)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Local\Anna\src\anna\runtime\qwen3_5_text_engine.py", line 497, in from_model_dir
resolved_offload_mode = cls._resolve_offload_mode(
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Local\Anna\src\anna\runtime\qwen3_5_text_engine.py", line 820, in _resolve_offload_mode
memory_info = device_context.get_memory_info()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Local\Anna\src\anna\runtime\device.py", line 354, in get_memory_info
free_bytes, total_bytes = xpu.mem_get_info()
^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\xpu\memory.py", line 194, in mem_get_info
return torch._C._xpu_getMemoryInfo(device)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: The device (Intel(R) Arc(TM) B580 Graphics) doesn't support querying the available free memory. You can file an issue at https://github.com/pytorch/pytorch/issues to help us prioritize its implementation.
(anna) D:\Local\Anna>
(anna) D:\Local\Anna>set UR_L0_ENABLE_SYSMAN_ENV_DEFAULT=0
(anna) D:\Local\Anna>
(anna) D:\Local\Anna>anna-serve --model-dir D:\Model\Qwen3.5-4B --model-name qwen3.5 --device xpu --dtype bf16 --host 127.0.0.1 --port 8000
2026-07-09 17:13:53,278 INFO anna.cli.xpu_env: Configured XPU environment: {'UR_L0_ENABLE_RELAXED_ALLOCATION_LIMITS': '1', 'ZES_ENABLE_SYSMAN': '1'}
2026-07-09 17:13:54,038 INFO anna.runtime.device: Resolved XPU device info: {'device_index': 0, 'name': 'Intel(R) Arc(TM) B580 Graphics', 'total_memory': 12526497792, 'free_memory': 10715209728, 'allocated_memory': 0, 'reserved_memory': 0, 'runtime': 'torch.xpu', 'device_type': 'arc', 'is_arc_alchemist': True, 'is_acm_g10': False, 'is_arc_a770_or_a750': False}
2026-07-09 17:13:54,081 INFO anna.runtime.qwen3_5_text_engine: Building Qwen3.5 runtime: model_dir=D:\Model\Qwen3.5-4B compute_device=xpu load_device=xpu offload=none expert_quant=none weight_quant=none resident_expert_layers=0 cached_experts_per_layer=0 kv_cache=none
2026-07-09 17:13:54,082 INFO anna.weights.qwen3_5_text_weight_loader: Constructing Qwen3.5 model skeleton on meta device.
2026-07-09 17:13:54,202 INFO anna.weights.qwen3_5_text_weight_loader: Replacing Qwen3.5 linear layers with quantized placeholders.
2026-07-09 17:13:54,202 INFO anna.weights.qwen3_5_text_weight_loader: Allocating empty Qwen3.5 tensors on xpu.
2026-07-09 17:13:54,831 INFO anna.weights.qwen3_5_text_weight_loader: Constructed Qwen3.5 model skeleton: quantized_placeholders=0 target_device=xpu
2026-07-09 17:13:54,853 INFO anna.weights.qwen3_5_text_weight_loader: Loading Qwen3.5 weights from 2 shard(s), total=9319828096 bytes, model_dir=D:\Model\Qwen3.5-4B, safetensors_device=xpu:0
2026-07-09 17:13:54,853 INFO anna.weights.qwen3_5_text_weight_loader: Loading Qwen3.5 weight shard 1/2: model.safetensors-00001-of-00002.safetensors (5329398688 bytes)
2026-07-09 17:13:59,988 INFO anna.weights.qwen3_5_text_weight_loader: Loaded Qwen3.5 weight shard 1/2: model.safetensors-00001-of-00002.safetensors (cumulative_bytes=5329398688/9319828096, tensors_loaded=84, tensors_skipped=3)
2026-07-09 17:13:59,989 INFO anna.weights.qwen3_5_text_weight_loader: Loading Qwen3.5 weight shard 2/2: model.safetensors-00002-of-00002.safetensors (3990429408 bytes)
2026-07-09 17:14:06,029 INFO anna.weights.qwen3_5_text_weight_loader: Loaded Qwen3.5 weight shard 2/2: model.safetensors-00002-of-00002.safetensors (cumulative_bytes=9319828096/9319828096, tensors_loaded=723, tensors_skipped=15)
2026-07-09 17:14:06,118 INFO anna.runtime.qwen3_5_text_engine: Finished loading Qwen3.5 weights: tensors_loaded=723 tensors_skipped=15 quantized_placeholders=0
2026-07-09 17:14:06,118 INFO anna.runtime.qwen3_5_text_engine: Configuring Qwen3.5 runtime placement on xpu: offload_experts=False offload_vision=False resident_expert_indices=[] cached_experts_per_layer=0
2026-07-09 17:14:06,132 INFO anna.runtime.qwen3_5_text_engine: Preparing loaded quantized Qwen3.5 modules for XPU execution.
2026-07-09 17:14:06,575 INFO anna.runtime.qwen3_5_text_engine: Post-load Qwen3.5 CPU tensor residency: 0.00 B
2026-07-09 17:14:07,301 INFO anna.model.fused_ops: Loaded Anna fused-op library from D:\Local\Anna\.build\anna_gated_delta_fused\anna_gated_delta_fused.pyd
2026-07-09 17:14:07,302 INFO anna.runtime.qwen3_5_text_engine: Loaded model qwen3.5 on xpu (compute=torch.bfloat16, requested=bfloat16, default_max_completion_tokens=None, default_temperature=0.7, default_top_p=0.8, default_top_k=20, default_min_p=0.0, default_presence_penalty=1.5, default_repetition_penalty=1.0, default_enable_thinking=True, reasoning_format=deepseek, offload=none, offload_vision=False, expert_quant=none, weight_quant=none, resident_expert_layers=0, resident_expert_layer_indices=[], cached_experts_per_layer=0, full_attention_cache_mirror=False, weight_load_device=xpu); tensors loaded=723 skipped=15 quantized=0
2026-07-09 17:14:07,304 INFO anna.runtime.qwen3_5_text_engine: Enabled auto prefill chunking on xpu: chunk_size=544 block_size=32 estimated_bytes_per_token=456.00 KiB target_budget=256.00 MiB
2026-07-09 17:14:07,304 INFO anna.runtime.qwen3_5_text_engine: compile_mode=auto resolved to reduce-overhead
2026-07-09 17:14:09,056 INFO anna.runtime.qwen3_5_text_engine: Enabled torch.compile for XPU text path: mode=reduce-overhead fullgraph=False
2026-07-09 17:14:09,056 INFO anna.runtime.qwen3_5_text_engine: Qwen3.5 KV cache runtime: mode=none turboquant_enabled=False turboquant_bits=None turboquant_residual_len=None full_attention_layers=8 turboquant_quantized_layers=0
W0709 17:14:10.426000 12928 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:14:10.426000 12928 site-packages\torch\_dynamo\exc.py:525] [2/0_1] Traceback:
W0709 17:14:10.426000 12928 site-packages\torch\_dynamo\exc.py:525] [2/0_1] File "D:\Local\Anna\src\anna\model\ops.py", line 416, in set_prompt_token_ids
W0709 17:14:10.426000 12928 site-packages\torch\_dynamo\exc.py:525] [2/0_1] self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:14:10.426000 12928 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:14:10.426000 12928 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
Traceback (most recent call last):
File "<frozen runpy>", line 198, in _run_module_as_main
File "<frozen runpy>", line 88, in _run_code
File "D:\miniconda3\envs\anna\Scripts\anna-serve.exe\__main__.py", line 5, in <module>
File "D:\Local\Anna\src\anna\cli\serve.py", line 537, in main
engine.warmup_inference_kernels(
File "D:\Local\Anna\src\anna\runtime\qwen3_5_text_engine.py", line 2072, in warmup_inference_kernels
outputs = self._forward_generation_model(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Local\Anna\src\anna\runtime\qwen3_5_text_engine.py", line 1979, in _forward_generation_model
return forward_fn(
^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_dynamo\eval_frame.py", line 749, in compile_wrapper
raise e.remove_dynamo_frames() from None # see TORCHDYNAMO_VERBOSE=1
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\compile_fx.py", line 923, in _compile_fx_inner
raise InductorError(e, currentframe()).with_traceback(
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\compile_fx.py", line 907, in _compile_fx_inner
mb_compiled_graph = fx_codegen_and_compile(
^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\compile_fx.py", line 1578, in fx_codegen_and_compile
return scheme.codegen_and_compile(gm, example_inputs, inputs_to_check, graph_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\compile_fx.py", line 1456, in codegen_and_compile
compiled_module = graph.compile_to_module()
^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\graph.py", line 2293, in compile_to_module
return self._compile_to_module()
^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\graph.py", line 2299, in _compile_to_module
self.codegen_with_cpp_wrapper() if self.cpp_wrapper else self.codegen()
^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\graph.py", line 2238, in codegen
self.scheduler.codegen()
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\scheduler.py", line 4598, in codegen
else self._codegen(self.nodes)
^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\scheduler.py", line 4750, in _codegen
self.get_backend(device).codegen_node(node)
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\codegen\simd.py", line 1371, in codegen_node
return self.codegen_node_schedule(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\codegen\simd.py", line 1424, in codegen_node_schedule
src_code = kernel.codegen_kernel()
^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\codegen\triton.py", line 3677, in codegen_kernel
**self.inductor_meta_common(),
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\codegen\triton.py", line 3501, in inductor_meta_common
"backend_hash": torch.utils._triton.triton_hash_with_backend(),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\utils\_triton.py", line 165, in triton_hash_with_backend
backend = triton_backend()
^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\utils\_triton.py", line 157, in triton_backend
target = driver.active.get_current_target()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\backends\intel\driver.py", line 747, in get_current_target
device = self.get_current_device()
^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\backends\intel\driver.py", line 739, in get_current_device
return self.utils.get_current_device()
^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\backends\intel\driver.py", line 733, in __getattr__
self.utils = XPUUtils()
^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\backends\intel\driver.py", line 309, in __init__
self.mod = compile_module_from_src(Path(os.path.join(dirname, "driver.c")).read_text(), "spirv_utils")
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\backends\intel\driver.py", line 272, in compile_module_from_src
so = _build(name, src_path, tmpdir, COMPILATION_HELPER.library_dir, COMPILATION_HELPER.include_dir,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\runtime\build.py", line 64, in _build
raise RuntimeError(
torch._inductor.exc.InductorError: RuntimeError: Failed to find C compiler. Please specify via CC environment variable or set triton.knobs.build.impl.
Set TORCHDYNAMO_VERBOSE=1 for the internal stack trace (please do this especially if you're reporting a bug to PyTorch). For even more developer context, set TORCH_LOGS="+dynamo"
(base) D:\Local\Anna>call "C:\Program Files\Microsoft Visual Studio\2022\Professional\VC\Auxiliary\Build\vcvars64.bat"
**********************************************************************
** Visual Studio 2022 Developer Command Prompt v17.12.3
** Copyright (c) 2022 Microsoft Corporation
**********************************************************************
[vcvarsall.bat] Environment initialized for: 'x64'
(base) D:\Local\Anna>conda activate anna
(anna) D:\Local\Anna>set PYTHONPATH=D:\Local\Anna\src
(anna) D:\Local\Anna>set ANNA_GATED_DELTA_OP_LIB=D:\Local\Anna\.build\anna_gated_delta_fused\anna_gated_delta_fused.pyd
(anna) D:\Local\Anna>set UR_L0_ENABLE_SYSMAN_ENV_DEFAULT=0
(anna) D:\Local\Anna>anna-serve --model-dir D:\Model\Qwen3.5-4B --model-name qwen3.5 --device xpu --dtype bf16 --host 12
7.0.0.1 --port 8000
2026-07-09 17:52:54,902 INFO anna.cli.xpu_env: Configured XPU environment: {'UR_L0_ENABLE_RELAXED_ALLOCATION_LIMITS': '1', 'ZES_ENABLE_SYSMAN': '1'}
2026-07-09 17:52:55,483 INFO anna.runtime.device: Resolved XPU device info: {'device_index': 0, 'name': 'Intel(R) Arc(TM) B580 Graphics', 'total_memory': 12526497792, 'free_memory': 10944131072, 'allocated_memory': 0, 'reserved_memory': 0, 'runtime': 'torch.xpu', 'device_type': 'arc', 'is_arc_alchemist': True, 'is_acm_g10': False, 'is_arc_a770_or_a750': False}
2026-07-09 17:52:55,512 INFO anna.runtime.qwen3_5_text_engine: Building Qwen3.5 runtime: model_dir=D:\Model\Qwen3.5-4B compute_device=xpu load_device=xpu offload=none expert_quant=none weight_quant=none resident_expert_layers=0 cached_experts_per_layer=0 kv_cache=none
2026-07-09 17:52:55,512 INFO anna.weights.qwen3_5_text_weight_loader: Constructing Qwen3.5 model skeleton on meta device.
2026-07-09 17:52:55,614 INFO anna.weights.qwen3_5_text_weight_loader: Replacing Qwen3.5 linear layers with quantized placeholders.
2026-07-09 17:52:55,614 INFO anna.weights.qwen3_5_text_weight_loader: Allocating empty Qwen3.5 tensors on xpu.
2026-07-09 17:52:56,144 INFO anna.weights.qwen3_5_text_weight_loader: Constructed Qwen3.5 model skeleton: quantized_placeholders=0 target_device=xpu
2026-07-09 17:52:56,166 INFO anna.weights.qwen3_5_text_weight_loader: Loading Qwen3.5 weights from 2 shard(s), total=9319828096 bytes, model_dir=D:\Model\Qwen3.5-4B, safetensors_device=xpu:0
2026-07-09 17:52:56,166 INFO anna.weights.qwen3_5_text_weight_loader: Loading Qwen3.5 weight shard 1/2: model.safetensors-00001-of-00002.safetensors (5329398688 bytes)
2026-07-09 17:53:01,419 INFO anna.weights.qwen3_5_text_weight_loader: Loaded Qwen3.5 weight shard 1/2: model.safetensors-00001-of-00002.safetensors (cumulative_bytes=5329398688/9319828096, tensors_loaded=84, tensors_skipped=3)
2026-07-09 17:53:01,420 INFO anna.weights.qwen3_5_text_weight_loader: Loading Qwen3.5 weight shard 2/2: model.safetensors-00002-of-00002.safetensors (3990429408 bytes)
2026-07-09 17:53:06,586 INFO anna.weights.qwen3_5_text_weight_loader: Loaded Qwen3.5 weight shard 2/2: model.safetensors-00002-of-00002.safetensors (cumulative_bytes=9319828096/9319828096, tensors_loaded=723, tensors_skipped=15)
2026-07-09 17:53:06,690 INFO anna.runtime.qwen3_5_text_engine: Finished loading Qwen3.5 weights: tensors_loaded=723 tensors_skipped=15 quantized_placeholders=0
2026-07-09 17:53:06,690 INFO anna.runtime.qwen3_5_text_engine: Configuring Qwen3.5 runtime placement on xpu: offload_experts=False offload_vision=False resident_expert_indices=[] cached_experts_per_layer=0
2026-07-09 17:53:06,700 INFO anna.runtime.qwen3_5_text_engine: Preparing loaded quantized Qwen3.5 modules for XPU execution.
2026-07-09 17:53:07,176 INFO anna.runtime.qwen3_5_text_engine: Post-load Qwen3.5 CPU tensor residency: 0.00 B
2026-07-09 17:53:07,914 INFO anna.model.fused_ops: Loaded Anna fused-op library from D:\Local\Anna\.build\anna_gated_delta_fused\anna_gated_delta_fused.pyd
2026-07-09 17:53:07,915 INFO anna.runtime.qwen3_5_text_engine: Loaded model qwen3.5 on xpu (compute=torch.bfloat16, requested=bfloat16, default_max_completion_tokens=None, default_temperature=0.7, default_top_p=0.8, default_top_k=20, default_min_p=0.0, default_presence_penalty=1.5, default_repetition_penalty=1.0, default_enable_thinking=True, reasoning_format=deepseek, offload=none, offload_vision=False, expert_quant=none, weight_quant=none, resident_expert_layers=0, resident_expert_layer_indices=[], cached_experts_per_layer=0, full_attention_cache_mirror=False, weight_load_device=xpu); tensors loaded=723 skipped=15 quantized=0
2026-07-09 17:53:07,917 INFO anna.runtime.qwen3_5_text_engine: Enabled auto prefill chunking on xpu: chunk_size=544 block_size=32 estimated_bytes_per_token=456.00 KiB target_budget=256.00 MiB
2026-07-09 17:53:07,917 INFO anna.runtime.qwen3_5_text_engine: compile_mode=auto resolved to reduce-overhead
2026-07-09 17:53:09,601 INFO anna.runtime.qwen3_5_text_engine: Enabled torch.compile for XPU text path: mode=reduce-overhead fullgraph=False
2026-07-09 17:53:09,601 INFO anna.runtime.qwen3_5_text_engine: Qwen3.5 KV cache runtime: mode=none turboquant_enabled=False turboquant_bits=None turboquant_residual_len=None full_attention_layers=8 turboquant_quantized_layers=0
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] Backend compiler exception
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] Explanation: Backend compiler `inductor` failed with aten._local_scalar_dense.default
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] While executing %item : [num_users=1] = call_method[target=item](args = (%getitem_1,), kwargs = {})
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] GraphModule: class GraphModule(torch.nn.Module):
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] def forward(self, L_input_ids_: "i64[1, 2][2, 1]", L_self_modules_embed_tokens_parameters_weight_: "bf16[248320, 2560][2560, 1]", L_self_modules_rotary_emb_buffers_inv_freq_: "f32[32][1]"):
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] l_input_ids_ = L_input_ids_
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] l_self_modules_embed_tokens_parameters_weight_ = L_self_modules_embed_tokens_parameters_weight_
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] l_self_modules_rotary_emb_buffers_inv_freq_ = L_self_modules_rotary_emb_buffers_inv_freq_
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\miniconda3\envs\anna\Lib\site-packages\torch\nn\modules\sparse.py:192 in forward, code: return F.embedding(
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] inputs_embeds: "bf16[1, 2, 2560][5120, 2560, 1]" = torch.nn.functional.embedding(l_input_ids_, l_self_modules_embed_tokens_parameters_weight_, 248044, None, 2.0, False, False); l_self_modules_embed_tokens_parameters_weight_ = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:416 in set_prompt_token_ids, code: self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem: "i64[2][1]" = l_input_ids_[0]; l_input_ids_ = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_1 = getitem[0]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] item: "Sym(u0)" = getitem_1.item(); getitem_1 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_2 = getitem[1]; getitem = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] item_1: "Sym(u1)" = getitem_2.item(); getitem_2 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\qwen3_5_text_model.py:211 in forward, code: inputs_embeds = inputs_embeds.to(device=execution_device)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] inputs_embeds_1: "bf16[1, 2, 2560][5120, 2560, 1]" = inputs_embeds.to(device = device(type='xpu')); inputs_embeds = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\qwen3_5_text_model.py:221 in forward, code: past_seen_tokens = torch.zeros(batch_size, device=inputs_embeds.device, dtype=torch.long)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] past_seen_tokens: "i64[1][1]" = torch.zeros(1, device = device(type='xpu', index=0), dtype = torch.int64)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\qwen3_5_text_model.py:224 in forward, code: position_ids = torch.arange(seq_len, device=inputs_embeds.device).view(1, -1)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] arange: "i64[2][1]" = torch.arange(2, device = device(type='xpu', index=0))
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] position_ids: "i64[1, 2][2, 1]" = arange.view(1, -1); arange = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\qwen3_5_text_model.py:225 in forward, code: position_ids = position_ids + past_seen_tokens.view(-1, 1)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] view_1: "i64[1, 1][1, 1]" = past_seen_tokens.view(-1, 1); past_seen_tokens = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] position_ids_1: "i64[1, 2][2, 1]" = position_ids + view_1; position_ids = view_1 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1510 in forward, code: position_ids = position_ids[None, ...].expand(3, position_ids.shape[0], -1)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_3: "i64[1, 1, 2][2, 2, 1]" = position_ids_1[(None, Ellipsis)]; position_ids_1 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] position_ids_2: "i64[3, 1, 2][0, 2, 1]" = getitem_3.expand(3, 1, -1); getitem_3 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1511 in forward, code: inv_freq_expanded = self.inv_freq[None, None, :, None].float().expand(3, position_ids.shape[1], -1, 1)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_4: "f32[1, 1, 32, 1][32, 32, 1, 1]" = l_self_modules_rotary_emb_buffers_inv_freq_[(None, None, slice(None, None, None), None)]; l_self_modules_rotary_emb_buffers_inv_freq_ = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] float_1: "f32[1, 1, 32, 1][32, 32, 1, 1]" = getitem_4.float(); getitem_4 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] inv_freq_expanded: "f32[3, 1, 32, 1][0, 32, 1, 1]" = float_1.expand(3, 1, -1, 1); float_1 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1512 in forward, code: position_ids_expanded = position_ids[:, :, None, :].float()
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_5: "i64[3, 1, 1, 2][0, 2, 2, 1]" = position_ids_2[(slice(None, None, None), slice(None, None, None), None, slice(None, None, None))]; position_ids_2 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] Original traceback:
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] File "D:\Local\Anna\src\anna\model\qwen3_5_text_model.py", line 207, in forward
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] past_key_values.set_prompt_token_ids(input_ids)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] File "D:\Local\Anna\src\anna\model\ops.py", line 416, in set_prompt_token_ids
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] . Adding a graph break.
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] Hint: Report an issue to the backend compiler repo.
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] Developer debug context: Backend: inductor
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] Exception:aten._local_scalar_dense.default
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] While executing %item : [num_users=1] = call_method[target=item](args = (%getitem_1,), kwargs = {})
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] GraphModule: class GraphModule(torch.nn.Module):
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] def forward(self, L_input_ids_: "i64[1, 2][2, 1]", L_self_modules_embed_tokens_parameters_weight_: "bf16[248320, 2560][2560, 1]", L_self_modules_rotary_emb_buffers_inv_freq_: "f32[32][1]"):
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] l_input_ids_ = L_input_ids_
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] l_self_modules_embed_tokens_parameters_weight_ = L_self_modules_embed_tokens_parameters_weight_
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] l_self_modules_rotary_emb_buffers_inv_freq_ = L_self_modules_rotary_emb_buffers_inv_freq_
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\miniconda3\envs\anna\Lib\site-packages\torch\nn\modules\sparse.py:192 in forward, code: return F.embedding(
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] inputs_embeds: "bf16[1, 2, 2560][5120, 2560, 1]" = torch.nn.functional.embedding(l_input_ids_, l_self_modules_embed_tokens_parameters_weight_, 248044, None, 2.0, False, False); l_self_modules_embed_tokens_parameters_weight_ = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:416 in set_prompt_token_ids, code: self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem: "i64[2][1]" = l_input_ids_[0]; l_input_ids_ = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_1 = getitem[0]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] item: "Sym(u0)" = getitem_1.item(); getitem_1 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_2 = getitem[1]; getitem = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] item_1: "Sym(u1)" = getitem_2.item(); getitem_2 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\qwen3_5_text_model.py:211 in forward, code: inputs_embeds = inputs_embeds.to(device=execution_device)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] inputs_embeds_1: "bf16[1, 2, 2560][5120, 2560, 1]" = inputs_embeds.to(device = device(type='xpu')); inputs_embeds = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\qwen3_5_text_model.py:221 in forward, code: past_seen_tokens = torch.zeros(batch_size, device=inputs_embeds.device, dtype=torch.long)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] past_seen_tokens: "i64[1][1]" = torch.zeros(1, device = device(type='xpu', index=0), dtype = torch.int64)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\qwen3_5_text_model.py:224 in forward, code: position_ids = torch.arange(seq_len, device=inputs_embeds.device).view(1, -1)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] arange: "i64[2][1]" = torch.arange(2, device = device(type='xpu', index=0))
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] position_ids: "i64[1, 2][2, 1]" = arange.view(1, -1); arange = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\qwen3_5_text_model.py:225 in forward, code: position_ids = position_ids + past_seen_tokens.view(-1, 1)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] view_1: "i64[1, 1][1, 1]" = past_seen_tokens.view(-1, 1); past_seen_tokens = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] position_ids_1: "i64[1, 2][2, 1]" = position_ids + view_1; position_ids = view_1 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1510 in forward, code: position_ids = position_ids[None, ...].expand(3, position_ids.shape[0], -1)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_3: "i64[1, 1, 2][2, 2, 1]" = position_ids_1[(None, Ellipsis)]; position_ids_1 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] position_ids_2: "i64[3, 1, 2][0, 2, 1]" = getitem_3.expand(3, 1, -1); getitem_3 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1511 in forward, code: inv_freq_expanded = self.inv_freq[None, None, :, None].float().expand(3, position_ids.shape[1], -1, 1)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_4: "f32[1, 1, 32, 1][32, 32, 1, 1]" = l_self_modules_rotary_emb_buffers_inv_freq_[(None, None, slice(None, None, None), None)]; l_self_modules_rotary_emb_buffers_inv_freq_ = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] float_1: "f32[1, 1, 32, 1][32, 32, 1, 1]" = getitem_4.float(); getitem_4 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] inv_freq_expanded: "f32[3, 1, 32, 1][0, 32, 1, 1]" = float_1.expand(3, 1, -1, 1); float_1 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1512 in forward, code: position_ids_expanded = position_ids[:, :, None, :].float()
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_5: "i64[3, 1, 1, 2][0, 2, 2, 1]" = position_ids_2[(slice(None, None, None), slice(None, None, None), None, slice(None, None, None))]; position_ids_2 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] position_ids_expanded: "f32[3, 1, 1, 2][2, 2, 2, 1]" = getitem_5.float(); getitem_5 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1513 in forward, code: freqs = (inv_freq_expanded @ position_ids_expanded).transpose(2, 3)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] matmul: "f32[3, 1, 32, 2][64, 64, 2, 1]" = inv_freq_expanded @ position_ids_expanded; inv_freq_expanded = position_ids_expanded = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] freqs: "f32[3, 1, 2, 32][64, 64, 1, 2]" = matmul.transpose(2, 3); matmul = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1498 in apply_interleaved_mrope, code: freqs_t = freqs[0].clone()
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_6: "f32[1, 2, 32][64, 1, 2]" = freqs[0]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] freqs_t: "f32[1, 2, 32][64, 1, 2]" = getitem_6.clone(); getitem_6 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1504 in apply_interleaved_mrope, code: idx = torch.arange(offset, length, 3, device=freqs.device)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] idx: "i64[11][1]" = torch.arange(1, 32, 3, device = device(type='xpu', index=0))
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1505 in apply_interleaved_mrope, code: freqs_t[..., idx] = freqs[dim, ..., idx]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_7: "f32[1, 2, 11][22, 11, 1]" = freqs[(1, Ellipsis, idx)]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] freqs_t[(Ellipsis, idx)] = getitem_7; setitem = freqs_t; idx = getitem_7 = setitem = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1504 in apply_interleaved_mrope, code: idx = torch.arange(offset, length, 3, device=freqs.device)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] idx_1: "i64[10][1]" = torch.arange(2, 30, 3, device = device(type='xpu', index=0))
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1505 in apply_interleaved_mrope, code: freqs_t[..., idx] = freqs[dim, ..., idx]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] getitem_8: "f32[1, 2, 10][20, 10, 1]" = freqs[(2, Ellipsis, idx_1)]; freqs = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] freqs_t[(Ellipsis, idx_1)] = getitem_8; setitem_1 = freqs_t; idx_1 = getitem_8 = setitem_1 = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1515 in forward, code: emb = torch.cat((freqs, freqs), dim=-1)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] emb: "f32[1, 2, 64][128, 64, 1]" = torch.cat((freqs_t, freqs_t), dim = -1); freqs_t = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1516 in forward, code: cos = emb.cos() * self.attention_scaling
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] cos: "f32[1, 2, 64][128, 64, 1]" = emb.cos()
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] cos_1: "f32[1, 2, 64][128, 64, 1]" = cos * 1.0; cos = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:1517 in forward, code: sin = emb.sin() * self.attention_scaling
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] sin: "f32[1, 2, 64][128, 64, 1]" = emb.sin(); emb = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] sin_1: "f32[1, 2, 64][128, 64, 1]" = sin * 1.0; sin = None
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] return (inputs_embeds_1, cos_1, sin_1, item, item_1)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] Original traceback:
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] File "D:\Local\Anna\src\anna\model\qwen3_5_text_model.py", line 207, in forward
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] past_key_values.set_prompt_token_ids(input_ids)
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] File "D:\Local\Anna\src\anna\model\ops.py", line 416, in set_prompt_token_ids
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] Traceback:
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] File "D:\Local\Anna\src\anna\model\qwen3_5_text_model.py", line 232, in forward
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1] try:
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.559000 3052 site-packages\torch\_dynamo\exc.py:525] [1/0_1]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] Backend compiler exception
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] Explanation: Backend compiler `inductor` failed with aten._local_scalar_dense.default
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] While executing %item : [num_users=1] = call_method[target=item](args = (%getitem_1,), kwargs = {})
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] GraphModule: class GraphModule(torch.nn.Module):
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] def forward(self, L_input_ids_: "i64[1, 2][2, 1]"):
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] l_input_ids_ = L_input_ids_
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] # File: D:\Local\Anna\src\anna\model\ops.py:416 in set_prompt_token_ids, code: self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] getitem: "i64[2][1]" = l_input_ids_[0]; l_input_ids_ = None
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] getitem_1 = getitem[0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] item: "Sym(u0)" = getitem_1.item(); getitem_1 = None
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] getitem_2 = getitem[1]; getitem = None
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] item_1: "Sym(u1)" = getitem_2.item(); getitem_2 = None
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] return (item, item_1)
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] Original traceback:
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] File "D:\Local\Anna\src\anna\model\ops.py", line 416, in set_prompt_token_ids
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] . Adding a graph break.
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] Hint: Report an issue to the backend compiler repo.
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] Developer debug context: Backend: inductor
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] Exception:aten._local_scalar_dense.default
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] While executing %item : [num_users=1] = call_method[target=item](args = (%getitem_1,), kwargs = {})
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] GraphModule: class GraphModule(torch.nn.Module):
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] def forward(self, L_input_ids_: "i64[1, 2][2, 1]"):
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] l_input_ids_ = L_input_ids_
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] # File: D:\Local\Anna\src\anna\model\ops.py:416 in set_prompt_token_ids, code: self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] getitem: "i64[2][1]" = l_input_ids_[0]; l_input_ids_ = None
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] getitem_1 = getitem[0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] item: "Sym(u0)" = getitem_1.item(); getitem_1 = None
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] getitem_2 = getitem[1]; getitem = None
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] item_1: "Sym(u1)" = getitem_2.item(); getitem_2 = None
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] return (item, item_1)
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] Original traceback:
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] File "D:\Local\Anna\src\anna\model\ops.py", line 416, in set_prompt_token_ids
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] Traceback:
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] File "D:\Local\Anna\src\anna\model\ops.py", line 416, in set_prompt_token_ids
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0] self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.828000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] Backend compiler exception
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] Explanation: Backend compiler `inductor` failed with aten._local_scalar_dense.default
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] While executing %item : [num_users=1] = call_method[target=item](args = (%getitem_1,), kwargs = {})
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] GraphModule: class GraphModule(torch.nn.Module):
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] def forward(self, L_input_ids_: "i64[1, 2][2, 1]"):
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] l_input_ids_ = L_input_ids_
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:416 in set_prompt_token_ids, code: self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] getitem: "i64[2][1]" = l_input_ids_[0]; l_input_ids_ = None
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] getitem_1 = getitem[0]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] item: "Sym(u0)" = getitem_1.item(); getitem_1 = None
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] getitem_2 = getitem[1]; getitem = None
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] item_1: "Sym(u1)" = getitem_2.item(); getitem_2 = None
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] return (item, item_1)
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] Original traceback:
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] File "D:\Local\Anna\src\anna\model\ops.py", line 416, in set_prompt_token_ids
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] . Adding a graph break.
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] Hint: Report an issue to the backend compiler repo.
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] Developer debug context: Backend: inductor
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] Exception:aten._local_scalar_dense.default
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] While executing %item : [num_users=1] = call_method[target=item](args = (%getitem_1,), kwargs = {})
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] GraphModule: class GraphModule(torch.nn.Module):
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] def forward(self, L_input_ids_: "i64[1, 2][2, 1]"):
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] l_input_ids_ = L_input_ids_
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] # File: D:\Local\Anna\src\anna\model\ops.py:416 in set_prompt_token_ids, code: self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] getitem: "i64[2][1]" = l_input_ids_[0]; l_input_ids_ = None
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] getitem_1 = getitem[0]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] item: "Sym(u0)" = getitem_1.item(); getitem_1 = None
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] getitem_2 = getitem[1]; getitem = None
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] item_1: "Sym(u1)" = getitem_2.item(); getitem_2 = None
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] return (item, item_1)
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] Original traceback:
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] File "D:\Local\Anna\src\anna\model\ops.py", line 416, in set_prompt_token_ids
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] Traceback:
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] File "D:\Local\Anna\src\anna\model\ops.py", line 416, in set_prompt_token_ids
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1] self._prompt_token_ids = [int(t) for t in input_ids[0].tolist()]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
W0709 17:53:10.872000 3052 site-packages\torch\_dynamo\exc.py:525] [2/0_1]
icpx: error: linker command failed with exit code 1181 (use -v to see invocation)
Traceback (most recent call last):
File "<frozen runpy>", line 198, in _run_module_as_main
File "<frozen runpy>", line 88, in _run_code
File "D:\miniconda3\envs\anna\Scripts\anna-serve.exe\__main__.py", line 5, in <module>
File "D:\Local\Anna\src\anna\cli\serve.py", line 537, in main
engine.warmup_inference_kernels(
File "D:\Local\Anna\src\anna\runtime\qwen3_5_text_engine.py", line 2072, in warmup_inference_kernels
outputs = self._forward_generation_model(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Local\Anna\src\anna\runtime\qwen3_5_text_engine.py", line 1979, in _forward_generation_model
return forward_fn(
^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_dynamo\eval_frame.py", line 749, in compile_wrapper
raise e.remove_dynamo_frames() from None # see TORCHDYNAMO_VERBOSE=1
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\compile_fx.py", line 923, in _compile_fx_inner
raise InductorError(e, currentframe()).with_traceback(
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\compile_fx.py", line 907, in _compile_fx_inner
mb_compiled_graph = fx_codegen_and_compile(
^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\compile_fx.py", line 1578, in fx_codegen_and_compile
return scheme.codegen_and_compile(gm, example_inputs, inputs_to_check, graph_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\compile_fx.py", line 1456, in codegen_and_compile
compiled_module = graph.compile_to_module()
^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\graph.py", line 2293, in compile_to_module
return self._compile_to_module()
^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\graph.py", line 2299, in _compile_to_module
self.codegen_with_cpp_wrapper() if self.cpp_wrapper else self.codegen()
^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\graph.py", line 2238, in codegen
self.scheduler.codegen()
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\scheduler.py", line 4598, in codegen
else self._codegen(self.nodes)
^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\scheduler.py", line 4750, in _codegen
self.get_backend(device).codegen_node(node)
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\codegen\simd.py", line 1371, in codegen_node
return self.codegen_node_schedule(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\codegen\simd.py", line 1424, in codegen_node_schedule
src_code = kernel.codegen_kernel()
^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\codegen\triton.py", line 3677, in codegen_kernel
**self.inductor_meta_common(),
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\_inductor\codegen\triton.py", line 3501, in inductor_meta_common
"backend_hash": torch.utils._triton.triton_hash_with_backend(),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\utils\_triton.py", line 165, in triton_hash_with_backend
backend = triton_backend()
^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\torch\utils\_triton.py", line 157, in triton_backend
target = driver.active.get_current_target()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\backends\intel\driver.py", line 747, in get_current_target
device = self.get_current_device()
^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\backends\intel\driver.py", line 739, in get_current_device
return self.utils.get_current_device()
^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\backends\intel\driver.py", line 733, in __getattr__
self.utils = XPUUtils()
^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\backends\intel\driver.py", line 309, in __init__
self.mod = compile_module_from_src(Path(os.path.join(dirname, "driver.c")).read_text(), "spirv_utils")
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\backends\intel\driver.py", line 272, in compile_module_from_src
so = _build(name, src_path, tmpdir, COMPILATION_HELPER.library_dir, COMPILATION_HELPER.include_dir,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\anna\Lib\site-packages\triton\runtime\build.py", line 115, in _build
subprocess.check_call(cc_cmd, stdout=subprocess.DEVNULL)
File "D:\miniconda3\envs\anna\Lib\subprocess.py", line 413, in check_call
raise CalledProcessError(retcode, cmd)
torch._inductor.exc.InductorError: CalledProcessError: Command '['C:\\Program Files (x86)\\Intel\\oneAPI\\compiler\\2026.1\\bin\\icpx.EXE', 'C:\\Users\\Keb\\AppData\\Local\\Temp\\tmprv0rsf38\\main.cpp', '-O3', '-shared', '-Wno-psabi', '-Wno-deprecated-declarations', '-lze_loader', '-lsycl8', '-LC:\\Program Files (x86)\\Intel\\oneAPI\\compiler\\2026.1\\lib', '-LC:\\Program Files\\LevelZeroSDK\\1.28.2\\lib', '-LD:\\miniconda3\\envs\\anna\\Lib\\site-packages\\triton\\backends\\intel\\lib', '-LD:\\miniconda3\\envs\\anna\\libs', '-IC:\\Program Files\\LevelZeroSDK\\1.28.2\\include', '-IC:\\Program Files (x86)\\Intel\\oneAPI\\compiler\\2026.1\\include', '-IC:\\Program Files (x86)\\Intel\\oneAPI\\compiler\\2026.1\\include/sycl', '-ID:\\miniconda3\\envs\\anna\\Lib\\site-packages\\triton\\backends\\intel\\include', '-IC:\\Users\\Keb\\AppData\\Local\\Temp\\tmprv0rsf38', '-ID:\\miniconda3\\envs\\anna\\Include', '-ID:\\miniconda3\\envs\\anna\\Lib\\site-packages\\numpy\\_core\\include', '-o', 'C:\\Users\\Keb\\AppData\\Local\\Temp\\tmprv0rsf38\\spirv_utils.cp312-win_amd64.pyd', '/LIBPATH:C:\\Program Files (x86)\\Intel\\oneAPI\\compiler\\2026.1\\lib', '-fsycl']' returned non-zero exit status 1181.
Set TORCHDYNAMO_VERBOSE=1 for the internal stack trace (please do this especially if you're reporting a bug to PyTorch). For even more developer context, set TORCH_LOGS="+dynamo"
Hello,
Tried running Anna with a Intel Arc B580 in Windows 10, but did not work, This is only compatible for Alchemist Series?
Compiled the fused kernel and the model i can see that it loads into VRAM, but then it crashes
Full log: