Issue Station
Sign In
Tools hereToolbox

vllm - ✅(Solved) Fix [Bug]: Triton Attention AssertionError on supported kv_cache_dtype [1 pull requests, 1 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
vllm-project/vllm#43109•Fetched 2026-05-20 03:39:47
View on GitHub
Comments
0
Participants
1
Timeline
2
Reactions
0
Author
JamesLee-JonesJamesLee-Jones
Participants
JamesLee-JonesJamesLee-Jones
Timeline (top)
cross-referenced ×1labeled ×1

Error Message

(EngineCore pid=8748) File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/attention/ops/triton_reshape_and_cache_flash.py", line 353, in triton_reshape_and_cache_flash (EngineCore pid=8748) assert kv_cache_dtype == "auto" or is_quantized_kv_cache(kv_cache_dtype), ( (EngineCore pid=8748) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ (EngineCore pid=8748) AssertionError: unsupported kv_cache_dtype (str), got float16.

Fix Action

Fix / Workaround

============================== CPU Info

Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 46 bits physical, 48 bits virtual Byte Order: Little Endian CPU(s): 22 On-line CPU(s) list: 0-21 Vendor ID: GenuineIntel Model name: Intel(R) Core(TM) Ultra 9 185H CPU family: 6 Model: 170 Thread(s) per core: 2 Core(s) per socket: 16 Socket(s): 1 Stepping: 4 CPU(s) scaling MHz: 36% CPU max MHz: 5100.0000 CPU min MHz: 400.0000 BogoMIPS: 6144.00 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb ssbd ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid rdseed adx smap clflushopt clwb intel_pt sha_ni xsaveopt xsavec xgetbv1 xsaves split_lock_detect user_shstk avx_vnni dtherm ida arat pln pts hwp hwp_notify hwp_act_window hwp_epp hwp_pkg_req hfi vnmi umip pku ospke waitpkg gfni vaes vpclmulqdq rdpid bus_lock_detect movdiri movdir64b fsrm md_clear serialize pconfig arch_lbr ibt flush_l1d arch_capabilities Virtualization: VT-x L1d cache: 544 KiB (14 instances) L1i cache: 896 KiB (14 instances) L2 cache: 18 MiB (9 instances) L3 cache: 24 MiB (1 instance) NUMA node(s): 1 NUMA node0 CPU(s): 0-21 Vulnerability Gather data sampling: Not affected Vulnerability Ghostwrite: Not affected Vulnerability Indirect target selection: Not affected Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Not affected Vulnerability Old microcode: Not affected Vulnerability Reg file data sampling: Not affected Vulnerability Retbleed: Not affected Vulnerability Spec rstack overflow: Not affected Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; PBRSB-eIBRS Not affected; BHI BHI_DIS_S Vulnerability Srbds: Not affected Vulnerability Tsa: Not affected Vulnerability Tsx async abort: Not affected Vulnerability Vmscape: Mitigation; IBPB before exit to userspace

PR fix notes

PR #43138: [Bugfix] Allow float16/bfloat16 as kv_cache_dtype in Triton attention ops

Description (problem / solution / changelog)

Summary

Fix an AssertionError raised when kv_cache_dtype is "float16" or "bfloat16" in the Triton attention backend (reshape_and_cache_flash and reshape_and_cache_flash_mla).

Root cause: The assert in both functions only accepted "auto" or quantized types (fp8*, nvfp4, etc.):

# Before (too strict)
assert kv_cache_dtype == "auto" or is_quantized_kv_cache(kv_cache_dtype), (
    f"unsupported kv_cache_dtype (str), got {kv_cache_dtype}."
)

But flash_attn.py explicitly lists "float16" and "bfloat16" as valid kv_cache_dtype values (line 193: return kv_cache_dtype in ["auto", "float16", "bfloat16"]). When passed, the downstream code handles them correctly - is_quantized_kv_cache("float16") returns False, so kv_cache_torch_dtype = key_cache.dtype (same as the "auto" path, no casting).

Fix: Expand the assert to include "float16" and "bfloat16" alongside "auto".

Fixes #43109

Test Plan

  • Manually verified the downstream code path for "float16"/"bfloat16" is safe (falls through as key_cache.dtype, same as "auto")
  • Confirmed flash_attn.py already accepts these values as valid
  • Run existing Triton attention tests to verify no regressions

Note: Parts of this fix were developed with AI (Claude) assistance.

Changed files

  • vllm/v1/attention/ops/triton_reshape_and_cache_flash.py (modified, +12/-2)

Code Example

Collecting environment information...
uv is set
==============================
        System Info
==============================
OS                           : Ubuntu 25.10 (x86_64)
GCC version                  : (Ubuntu 14.3.0-8ubuntu1) 14.3.0
Clang version                : 20.1.8 (0ubuntu4)
CMake version                : version 3.31.6
Libc version                 : glibc-2.42

==============================
       PyTorch Info
==============================
PyTorch version              : 2.11.0+cu130
Is debug build               : False
CUDA used to build PyTorch   : 13.0
ROCM used to build PyTorch   : N/A
XPU used to build PyTorch    : N/A

==============================
      Python Environment
==============================
Python version               : 3.13.5 (main, Jun 12 2025, 12:40:22) [Clang 20.1.4 ] (64-bit runtime)
Python platform              : Linux-6.17.0-29-generic-x86_64-with-glibc2.42
    
==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 12.9.86
CUDA_MODULE_LOADING set to   : 
GPU models and configuration : GPU 0: NVIDIA GeForce RTX 4060 Laptop GPU
Nvidia driver version        : 590.48.01
cuDNN version                : Could not collect
HIP runtime version          : N/A
MIOpen runtime version       : N/A
Is XNNPACK available         : True

==============================
          CPU Info
==============================
Architecture:                            x86_64
CPU op-mode(s):                          32-bit, 64-bit
Address sizes:                           46 bits physical, 48 bits virtual
Byte Order:                              Little Endian
CPU(s):                                  22
On-line CPU(s) list:                     0-21
Vendor ID:                               GenuineIntel
Model name:                              Intel(R) Core(TM) Ultra 9 185H
CPU family:                              6
Model:                                   170
Thread(s) per core:                      2
Core(s) per socket:                      16
Socket(s):                               1
Stepping:                                4
CPU(s) scaling MHz:                      36%
CPU max MHz:                             5100.0000
CPU min MHz:                             400.0000
BogoMIPS:                                6144.00
Flags:                                   fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb ssbd ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid rdseed adx smap clflushopt clwb intel_pt sha_ni xsaveopt xsavec xgetbv1 xsaves split_lock_detect user_shstk avx_vnni dtherm ida arat pln pts hwp hwp_notify hwp_act_window hwp_epp hwp_pkg_req hfi vnmi umip pku ospke waitpkg gfni vaes vpclmulqdq rdpid bus_lock_detect movdiri movdir64b fsrm md_clear serialize pconfig arch_lbr ibt flush_l1d arch_capabilities
Virtualization:                          VT-x
L1d cache:                               544 KiB (14 instances)
L1i cache:                               896 KiB (14 instances)
L2 cache:                                18 MiB (9 instances)
L3 cache:                                24 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-21
Vulnerability Gather data sampling:      Not affected
Vulnerability Ghostwrite:                Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Old microcode:             Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; Enhanced / Automatic IBRS; IBPB conditional; PBRSB-eIBRS Not affected; BHI BHI_DIS_S
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Mitigation; IBPB before exit to userspace

==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.6.8.post1
[pip3] numpy==2.4.6
[pip3] nvidia-cublas==13.1.0.3
[pip3] nvidia-cuda-cupti==13.0.85
[pip3] nvidia-cuda-nvrtc==13.0.88
[pip3] nvidia-cuda-runtime==13.0.96
[pip3] nvidia-cudnn-cu13==9.19.0.56
[pip3] nvidia-cudnn-frontend==1.18.0
[pip3] nvidia-cufft==12.0.0.61
[pip3] nvidia-cufile==1.15.1.6
[pip3] nvidia-curand==10.4.0.35
[pip3] nvidia-cusolver==12.0.4.66
[pip3] nvidia-cusparse==12.6.3.3
[pip3] nvidia-cusparselt-cu13==0.8.0
[pip3] nvidia-cutlass-dsl==4.4.2
[pip3] nvidia-cutlass-dsl-libs-base==4.4.2
[pip3] nvidia-ml-py==13.595.45
[pip3] nvidia-nccl-cu13==2.28.9
[pip3] nvidia-nvjitlink==13.0.88
[pip3] nvidia-nvshmem-cu13==3.4.5
[pip3] nvidia-nvtx==13.0.85
[pip3] pyzmq==27.1.0
[pip3] tokenspeed-triton==3.7.10.post20260505
[pip3] torch==2.11.0
[pip3] torch-c-dlpack-ext==0.1.5
[pip3] torchaudio==2.11.0
[pip3] torchvision==0.26.0
[pip3] transformers==5.8.1
[pip3] triton==3.6.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
vLLM Version                 : 0.21.0
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled; XPU: Disabled
GPU Topology:
  	GPU0	CPU Affinity	NUMA Affinity	GPU NUMA ID
GPU0	 X 	0-21	0		N/A

Legend:

  X    = Self
  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
  PIX  = Connection traversing at most a single PCIe bridge
  NV#  = Connection traversing a bonded set of # NVLinks

==============================
     Environment Variables
==============================
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
TORCHINDUCTOR_CACHE_DIR=/tmp/torchinductor_jlj

---

(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/attention/ops/triton_reshape_and_cache_flash.py", line 353, in triton_reshape_and_cache_flash
(EngineCore pid=8748)     assert kv_cache_dtype == "auto" or is_quantized_kv_cache(kv_cache_dtype), (
(EngineCore pid=8748)            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) AssertionError: unsupported kv_cache_dtype (str), got float16.

---

from vllm import LLM
from vllm.config import AttentionConfig
from vllm.v1.attention.backends.registry import AttentionBackendEnum

llm = LLM(
    "inferno-project/vllm-gpt2-1",
    kv_cache_dtype="float16",
)
llm.generate("token_0")

---

(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140] Traceback (most recent call last):
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 1114, in run_engine_core
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     engine_core = EngineCoreProc(*args, engine_index=dp_rank, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/tracing/otel.py", line 178, in sync_wrapper
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 880, in __init__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     super().__init__(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ~~~~~~~~~~~~~~~~^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         vllm_config,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<3 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         internal_dp_balancing,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 128, in __init__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     kv_cache_config = self._initialize_kv_caches(vllm_config)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/tracing/otel.py", line 178, in sync_wrapper
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 250, in _initialize_kv_caches
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     available_gpu_memory = self.model_executor.determine_available_memory()
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/executor/abstract.py", line 147, in determine_available_memory
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self.collective_rpc("determine_available_memory")
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/executor/uniproc_executor.py", line 93, in collective_rpc
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     result = run_method(self.driver_worker, method, args, kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/serial_utils.py", line 510, in run_method
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_worker.py", line 407, in determine_available_memory
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     cudagraph_memory_estimate = self.model_runner.profile_cudagraph_memory()
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 6092, in profile_cudagraph_memory
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     self._warmup_and_capture(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ~~~~~~~~~~~~~~~~~~~~~~~~^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         desc,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<8 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ),
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 6262, in _warmup_and_capture
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     self._dummy_run(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ~~~~~~~~~~~~~~~^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         desc.num_tokens,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<7 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         profile_seq_lens=profile_seq_lens,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 5616, in _dummy_run
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     outputs = self.model(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         input_ids=input_ids,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<3 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         **model_kwargs,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/cuda_graph.py", line 254, in __call__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self.runnable(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1779, in _wrapped_call_impl
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self._call_impl(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1790, in _call_impl
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return forward_call(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/models/gpt2.py", line 306, in forward
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     hidden_states = self.transformer(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         input_ids, positions, intermediate_tensors, inputs_embeds
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/decorators.py", line 520, in __call__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self.aot_compiled_fn(self, *args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/_dynamo/aot_compile.py", line 224, in __call__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self.fn(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/models/gpt2.py", line 219, in forward
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     def forward(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/caching.py", line 217, in __call__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self.optimized_call(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "<string>", line 48, in execution_fn
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "<string>", line 5, in __vllm_inlined_submods__1
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/_ops.py", line 1269, in __call__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self._op(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/layers/attention/attention.py", line 678, in unified_kv_cache_update
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     attn_layer.impl.do_kv_cache_update(  # type: ignore[attr-defined]
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         attn_layer,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<3 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         layer_slot_mapping,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/attention/backends/triton_attn.py", line 724, in do_kv_cache_update
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     triton_reshape_and_cache_flash(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         key,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<6 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         layer._v_scale,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/attention/ops/triton_reshape_and_cache_flash.py", line 353, in triton_reshape_and_cache_flash
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     assert kv_cache_dtype == "auto" or is_quantized_kv_cache(kv_cache_dtype), (
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140] AssertionError: unsupported kv_cache_dtype (str), got float16.
(EngineCore pid=8748) Process EngineCore:
(EngineCore pid=8748) Traceback (most recent call last):
(EngineCore pid=8748)   File "/home/jlj/.local/share/uv/python/cpython-3.13.5-linux-x86_64-gnu/lib/python3.13/multiprocessing/process.py", line 313, in _bootstrap
(EngineCore pid=8748)     self.run()
(EngineCore pid=8748)     ~~~~~~~~^^
(EngineCore pid=8748)   File "/home/jlj/.local/share/uv/python/cpython-3.13.5-linux-x86_64-gnu/lib/python3.13/multiprocessing/process.py", line 108, in run
(EngineCore pid=8748)     self._target(*self._args, **self._kwargs)
(EngineCore pid=8748)     ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 1144, in run_engine_core
(EngineCore pid=8748)     raise e
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 1114, in run_engine_core
(EngineCore pid=8748)     engine_core = EngineCoreProc(*args, engine_index=dp_rank, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/tracing/otel.py", line 178, in sync_wrapper
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 880, in __init__
(EngineCore pid=8748)     super().__init__(
(EngineCore pid=8748)     ~~~~~~~~~~~~~~~~^
(EngineCore pid=8748)         vllm_config,
(EngineCore pid=8748)         ^^^^^^^^^^^^
(EngineCore pid=8748)     ...<3 lines>...
(EngineCore pid=8748)         internal_dp_balancing,
(EngineCore pid=8748)         ^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)     )
(EngineCore pid=8748)     ^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 128, in __init__
(EngineCore pid=8748)     kv_cache_config = self._initialize_kv_caches(vllm_config)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/tracing/otel.py", line 178, in sync_wrapper
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 250, in _initialize_kv_caches
(EngineCore pid=8748)     available_gpu_memory = self.model_executor.determine_available_memory()
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/executor/abstract.py", line 147, in determine_available_memory
(EngineCore pid=8748)     return self.collective_rpc("determine_available_memory")
(EngineCore pid=8748)            ~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/executor/uniproc_executor.py", line 93, in collective_rpc
(EngineCore pid=8748)     result = run_method(self.driver_worker, method, args, kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/serial_utils.py", line 510, in run_method
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_worker.py", line 407, in determine_available_memory
(EngineCore pid=8748)     cudagraph_memory_estimate = self.model_runner.profile_cudagraph_memory()
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 6092, in profile_cudagraph_memory
(EngineCore pid=8748)     self._warmup_and_capture(
(EngineCore pid=8748)     ~~~~~~~~~~~~~~~~~~~~~~~~^
(EngineCore pid=8748)         desc,
(EngineCore pid=8748)         ^^^^^
(EngineCore pid=8748)     ...<8 lines>...
(EngineCore pid=8748)         ),
(EngineCore pid=8748)         ^^
(EngineCore pid=8748)     )
(EngineCore pid=8748)     ^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 6262, in _warmup_and_capture
(EngineCore pid=8748)     self._dummy_run(
(EngineCore pid=8748)     ~~~~~~~~~~~~~~~^
(EngineCore pid=8748)         desc.num_tokens,
(EngineCore pid=8748)         ^^^^^^^^^^^^^^^^
(EngineCore pid=8748)     ...<7 lines>...
(EngineCore pid=8748)         profile_seq_lens=profile_seq_lens,
(EngineCore pid=8748)         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)     )
(EngineCore pid=8748)     ^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 5616, in _dummy_run
(EngineCore pid=8748)     outputs = self.model(
(EngineCore pid=8748)         input_ids=input_ids,
(EngineCore pid=8748)     ...<3 lines>...
(EngineCore pid=8748)         **model_kwargs,
(EngineCore pid=8748)     )
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/cuda_graph.py", line 254, in __call__
(EngineCore pid=8748)     return self.runnable(*args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1779, in _wrapped_call_impl
(EngineCore pid=8748)     return self._call_impl(*args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1790, in _call_impl
(EngineCore pid=8748)     return forward_call(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/models/gpt2.py", line 306, in forward
(EngineCore pid=8748)     hidden_states = self.transformer(
(EngineCore pid=8748)         input_ids, positions, intermediate_tensors, inputs_embeds
(EngineCore pid=8748)     )
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/decorators.py", line 520, in __call__
(EngineCore pid=8748)     return self.aot_compiled_fn(self, *args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/_dynamo/aot_compile.py", line 224, in __call__
(EngineCore pid=8748)     return self.fn(*args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/models/gpt2.py", line 219, in forward
(EngineCore pid=8748)     def forward(
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/caching.py", line 217, in __call__
(EngineCore pid=8748)     return self.optimized_call(*args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "<string>", line 48, in execution_fn
(EngineCore pid=8748)   File "<string>", line 5, in __vllm_inlined_submods__1
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/_ops.py", line 1269, in __call__
(EngineCore pid=8748)     return self._op(*args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/layers/attention/attention.py", line 678, in unified_kv_cache_update
(EngineCore pid=8748)     attn_layer.impl.do_kv_cache_update(  # type: ignore[attr-defined]
(EngineCore pid=8748)     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)         attn_layer,
(EngineCore pid=8748)         ^^^^^^^^^^^
(EngineCore pid=8748)     ...<3 lines>...
(EngineCore pid=8748)         layer_slot_mapping,
(EngineCore pid=8748)         ^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)     )
(EngineCore pid=8748)     ^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/attention/backends/triton_attn.py", line 724, in do_kv_cache_update
(EngineCore pid=8748)     triton_reshape_and_cache_flash(
(EngineCore pid=8748)     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
(EngineCore pid=8748)         key,
(EngineCore pid=8748)         ^^^^
(EngineCore pid=8748)     ...<6 lines>...
(EngineCore pid=8748)         layer._v_scale,
(EngineCore pid=8748)         ^^^^^^^^^^^^^^^
(EngineCore pid=8748)     )
(EngineCore pid=8748)     ^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/attention/ops/triton_reshape_and_cache_flash.py", line 353, in triton_reshape_and_cache_flash
(EngineCore pid=8748)     assert kv_cache_dtype == "auto" or is_quantized_kv_cache(kv_cache_dtype), (
(EngineCore pid=8748)            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) AssertionError: unsupported kv_cache_dtype (str), got float16.
RAW_BUFFERClick to expand / collapse

Your current environment

<details> <summary>The output of <code>python collect_env.py</code></summary> 
Collecting environment information...
uv is set
==============================
        System Info
==============================
OS                           : Ubuntu 25.10 (x86_64)
GCC version                  : (Ubuntu 14.3.0-8ubuntu1) 14.3.0
Clang version                : 20.1.8 (0ubuntu4)
CMake version                : version 3.31.6
Libc version                 : glibc-2.42

==============================
       PyTorch Info
==============================
PyTorch version              : 2.11.0+cu130
Is debug build               : False
CUDA used to build PyTorch   : 13.0
ROCM used to build PyTorch   : N/A
XPU used to build PyTorch    : N/A

==============================
      Python Environment
==============================
Python version               : 3.13.5 (main, Jun 12 2025, 12:40:22) [Clang 20.1.4 ] (64-bit runtime)
Python platform              : Linux-6.17.0-29-generic-x86_64-with-glibc2.42
    
==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 12.9.86
CUDA_MODULE_LOADING set to   : 
GPU models and configuration : GPU 0: NVIDIA GeForce RTX 4060 Laptop GPU
Nvidia driver version        : 590.48.01
cuDNN version                : Could not collect
HIP runtime version          : N/A
MIOpen runtime version       : N/A
Is XNNPACK available         : True

==============================
          CPU Info
==============================
Architecture:                            x86_64
CPU op-mode(s):                          32-bit, 64-bit
Address sizes:                           46 bits physical, 48 bits virtual
Byte Order:                              Little Endian
CPU(s):                                  22
On-line CPU(s) list:                     0-21
Vendor ID:                               GenuineIntel
Model name:                              Intel(R) Core(TM) Ultra 9 185H
CPU family:                              6
Model:                                   170
Thread(s) per core:                      2
Core(s) per socket:                      16
Socket(s):                               1
Stepping:                                4
CPU(s) scaling MHz:                      36%
CPU max MHz:                             5100.0000
CPU min MHz:                             400.0000
BogoMIPS:                                6144.00
Flags:                                   fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb ssbd ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid rdseed adx smap clflushopt clwb intel_pt sha_ni xsaveopt xsavec xgetbv1 xsaves split_lock_detect user_shstk avx_vnni dtherm ida arat pln pts hwp hwp_notify hwp_act_window hwp_epp hwp_pkg_req hfi vnmi umip pku ospke waitpkg gfni vaes vpclmulqdq rdpid bus_lock_detect movdiri movdir64b fsrm md_clear serialize pconfig arch_lbr ibt flush_l1d arch_capabilities
Virtualization:                          VT-x
L1d cache:                               544 KiB (14 instances)
L1i cache:                               896 KiB (14 instances)
L2 cache:                                18 MiB (9 instances)
L3 cache:                                24 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-21
Vulnerability Gather data sampling:      Not affected
Vulnerability Ghostwrite:                Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Old microcode:             Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; Enhanced / Automatic IBRS; IBPB conditional; PBRSB-eIBRS Not affected; BHI BHI_DIS_S
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Mitigation; IBPB before exit to userspace

==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.6.8.post1
[pip3] numpy==2.4.6
[pip3] nvidia-cublas==13.1.0.3
[pip3] nvidia-cuda-cupti==13.0.85
[pip3] nvidia-cuda-nvrtc==13.0.88
[pip3] nvidia-cuda-runtime==13.0.96
[pip3] nvidia-cudnn-cu13==9.19.0.56
[pip3] nvidia-cudnn-frontend==1.18.0
[pip3] nvidia-cufft==12.0.0.61
[pip3] nvidia-cufile==1.15.1.6
[pip3] nvidia-curand==10.4.0.35
[pip3] nvidia-cusolver==12.0.4.66
[pip3] nvidia-cusparse==12.6.3.3
[pip3] nvidia-cusparselt-cu13==0.8.0
[pip3] nvidia-cutlass-dsl==4.4.2
[pip3] nvidia-cutlass-dsl-libs-base==4.4.2
[pip3] nvidia-ml-py==13.595.45
[pip3] nvidia-nccl-cu13==2.28.9
[pip3] nvidia-nvjitlink==13.0.88
[pip3] nvidia-nvshmem-cu13==3.4.5
[pip3] nvidia-nvtx==13.0.85
[pip3] pyzmq==27.1.0
[pip3] tokenspeed-triton==3.7.10.post20260505
[pip3] torch==2.11.0
[pip3] torch-c-dlpack-ext==0.1.5
[pip3] torchaudio==2.11.0
[pip3] torchvision==0.26.0
[pip3] transformers==5.8.1
[pip3] triton==3.6.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
vLLM Version                 : 0.21.0
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled; XPU: Disabled
GPU Topology:
  	GPU0	CPU Affinity	NUMA Affinity	GPU NUMA ID
GPU0	 X 	0-21	0		N/A

Legend:

  X    = Self
  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
  PIX  = Connection traversing at most a single PCIe bridge
  NV#  = Connection traversing a bonded set of # NVLinks

==============================
     Environment Variables
==============================
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
TORCHINDUCTOR_CACHE_DIR=/tmp/torchinductor_jlj
</details>

🐛 Describe the bug

Trying to load the GPT2 model inferno-project/vllm-gpt2-1 leads to the following assertion error when kv_cache_dtype is one of float16, bfloat16:

(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/attention/ops/triton_reshape_and_cache_flash.py", line 353, in triton_reshape_and_cache_flash
(EngineCore pid=8748)     assert kv_cache_dtype == "auto" or is_quantized_kv_cache(kv_cache_dtype), (
(EngineCore pid=8748)            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) AssertionError: unsupported kv_cache_dtype (str), got float16.

Triton attention should support these dtypes and indeed this doesn't happen with the model openai-community/gpt2.

Reproducer

from vllm import LLM
from vllm.config import AttentionConfig
from vllm.v1.attention.backends.registry import AttentionBackendEnum

llm = LLM(
    "inferno-project/vllm-gpt2-1",
    kv_cache_dtype="float16",
)
llm.generate("token_0")

Full traceback

traceback.log

Here is the error traceback:

(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140] Traceback (most recent call last):
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 1114, in run_engine_core
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     engine_core = EngineCoreProc(*args, engine_index=dp_rank, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/tracing/otel.py", line 178, in sync_wrapper
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 880, in __init__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     super().__init__(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ~~~~~~~~~~~~~~~~^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         vllm_config,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<3 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         internal_dp_balancing,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 128, in __init__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     kv_cache_config = self._initialize_kv_caches(vllm_config)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/tracing/otel.py", line 178, in sync_wrapper
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 250, in _initialize_kv_caches
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     available_gpu_memory = self.model_executor.determine_available_memory()
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/executor/abstract.py", line 147, in determine_available_memory
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self.collective_rpc("determine_available_memory")
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/executor/uniproc_executor.py", line 93, in collective_rpc
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     result = run_method(self.driver_worker, method, args, kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/serial_utils.py", line 510, in run_method
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_worker.py", line 407, in determine_available_memory
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     cudagraph_memory_estimate = self.model_runner.profile_cudagraph_memory()
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 6092, in profile_cudagraph_memory
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     self._warmup_and_capture(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ~~~~~~~~~~~~~~~~~~~~~~~~^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         desc,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<8 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ),
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 6262, in _warmup_and_capture
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     self._dummy_run(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ~~~~~~~~~~~~~~~^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         desc.num_tokens,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<7 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         profile_seq_lens=profile_seq_lens,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return func(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 5616, in _dummy_run
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     outputs = self.model(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         input_ids=input_ids,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<3 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         **model_kwargs,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/cuda_graph.py", line 254, in __call__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self.runnable(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1779, in _wrapped_call_impl
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self._call_impl(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1790, in _call_impl
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return forward_call(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/models/gpt2.py", line 306, in forward
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     hidden_states = self.transformer(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         input_ids, positions, intermediate_tensors, inputs_embeds
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/decorators.py", line 520, in __call__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self.aot_compiled_fn(self, *args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/_dynamo/aot_compile.py", line 224, in __call__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self.fn(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/models/gpt2.py", line 219, in forward
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     def forward(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/caching.py", line 217, in __call__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self.optimized_call(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "<string>", line 48, in execution_fn
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "<string>", line 5, in __vllm_inlined_submods__1
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/_ops.py", line 1269, in __call__
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     return self._op(*args, **kwargs)
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/layers/attention/attention.py", line 678, in unified_kv_cache_update
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     attn_layer.impl.do_kv_cache_update(  # type: ignore[attr-defined]
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         attn_layer,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<3 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         layer_slot_mapping,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/attention/backends/triton_attn.py", line 724, in do_kv_cache_update
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     triton_reshape_and_cache_flash(
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         key,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ...<6 lines>...
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         layer._v_scale,
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]         ^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     )
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     ^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/attention/ops/triton_reshape_and_cache_flash.py", line 353, in triton_reshape_and_cache_flash
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]     assert kv_cache_dtype == "auto" or is_quantized_kv_cache(kv_cache_dtype), (
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140]            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) ERROR 05-19 16:24:47 [v1/engine/core.py:1140] AssertionError: unsupported kv_cache_dtype (str), got float16.
(EngineCore pid=8748) Process EngineCore:
(EngineCore pid=8748) Traceback (most recent call last):
(EngineCore pid=8748)   File "/home/jlj/.local/share/uv/python/cpython-3.13.5-linux-x86_64-gnu/lib/python3.13/multiprocessing/process.py", line 313, in _bootstrap
(EngineCore pid=8748)     self.run()
(EngineCore pid=8748)     ~~~~~~~~^^
(EngineCore pid=8748)   File "/home/jlj/.local/share/uv/python/cpython-3.13.5-linux-x86_64-gnu/lib/python3.13/multiprocessing/process.py", line 108, in run
(EngineCore pid=8748)     self._target(*self._args, **self._kwargs)
(EngineCore pid=8748)     ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 1144, in run_engine_core
(EngineCore pid=8748)     raise e
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 1114, in run_engine_core
(EngineCore pid=8748)     engine_core = EngineCoreProc(*args, engine_index=dp_rank, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/tracing/otel.py", line 178, in sync_wrapper
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 880, in __init__
(EngineCore pid=8748)     super().__init__(
(EngineCore pid=8748)     ~~~~~~~~~~~~~~~~^
(EngineCore pid=8748)         vllm_config,
(EngineCore pid=8748)         ^^^^^^^^^^^^
(EngineCore pid=8748)     ...<3 lines>...
(EngineCore pid=8748)         internal_dp_balancing,
(EngineCore pid=8748)         ^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)     )
(EngineCore pid=8748)     ^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 128, in __init__
(EngineCore pid=8748)     kv_cache_config = self._initialize_kv_caches(vllm_config)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/tracing/otel.py", line 178, in sync_wrapper
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/engine/core.py", line 250, in _initialize_kv_caches
(EngineCore pid=8748)     available_gpu_memory = self.model_executor.determine_available_memory()
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/executor/abstract.py", line 147, in determine_available_memory
(EngineCore pid=8748)     return self.collective_rpc("determine_available_memory")
(EngineCore pid=8748)            ~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/executor/uniproc_executor.py", line 93, in collective_rpc
(EngineCore pid=8748)     result = run_method(self.driver_worker, method, args, kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/serial_utils.py", line 510, in run_method
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_worker.py", line 407, in determine_available_memory
(EngineCore pid=8748)     cudagraph_memory_estimate = self.model_runner.profile_cudagraph_memory()
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 6092, in profile_cudagraph_memory
(EngineCore pid=8748)     self._warmup_and_capture(
(EngineCore pid=8748)     ~~~~~~~~~~~~~~~~~~~~~~~~^
(EngineCore pid=8748)         desc,
(EngineCore pid=8748)         ^^^^^
(EngineCore pid=8748)     ...<8 lines>...
(EngineCore pid=8748)         ),
(EngineCore pid=8748)         ^^
(EngineCore pid=8748)     )
(EngineCore pid=8748)     ^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 6262, in _warmup_and_capture
(EngineCore pid=8748)     self._dummy_run(
(EngineCore pid=8748)     ~~~~~~~~~~~~~~~^
(EngineCore pid=8748)         desc.num_tokens,
(EngineCore pid=8748)         ^^^^^^^^^^^^^^^^
(EngineCore pid=8748)     ...<7 lines>...
(EngineCore pid=8748)         profile_seq_lens=profile_seq_lens,
(EngineCore pid=8748)         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)     )
(EngineCore pid=8748)     ^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=8748)     return func(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/worker/gpu_model_runner.py", line 5616, in _dummy_run
(EngineCore pid=8748)     outputs = self.model(
(EngineCore pid=8748)         input_ids=input_ids,
(EngineCore pid=8748)     ...<3 lines>...
(EngineCore pid=8748)         **model_kwargs,
(EngineCore pid=8748)     )
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/cuda_graph.py", line 254, in __call__
(EngineCore pid=8748)     return self.runnable(*args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1779, in _wrapped_call_impl
(EngineCore pid=8748)     return self._call_impl(*args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py", line 1790, in _call_impl
(EngineCore pid=8748)     return forward_call(*args, **kwargs)
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/models/gpt2.py", line 306, in forward
(EngineCore pid=8748)     hidden_states = self.transformer(
(EngineCore pid=8748)         input_ids, positions, intermediate_tensors, inputs_embeds
(EngineCore pid=8748)     )
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/decorators.py", line 520, in __call__
(EngineCore pid=8748)     return self.aot_compiled_fn(self, *args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/_dynamo/aot_compile.py", line 224, in __call__
(EngineCore pid=8748)     return self.fn(*args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/models/gpt2.py", line 219, in forward
(EngineCore pid=8748)     def forward(
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/compilation/caching.py", line 217, in __call__
(EngineCore pid=8748)     return self.optimized_call(*args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "<string>", line 48, in execution_fn
(EngineCore pid=8748)   File "<string>", line 5, in __vllm_inlined_submods__1
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/torch/_ops.py", line 1269, in __call__
(EngineCore pid=8748)     return self._op(*args, **kwargs)
(EngineCore pid=8748)            ~~~~~~~~^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/model_executor/layers/attention/attention.py", line 678, in unified_kv_cache_update
(EngineCore pid=8748)     attn_layer.impl.do_kv_cache_update(  # type: ignore[attr-defined]
(EngineCore pid=8748)     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)         attn_layer,
(EngineCore pid=8748)         ^^^^^^^^^^^
(EngineCore pid=8748)     ...<3 lines>...
(EngineCore pid=8748)         layer_slot_mapping,
(EngineCore pid=8748)         ^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748)     )
(EngineCore pid=8748)     ^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/attention/backends/triton_attn.py", line 724, in do_kv_cache_update
(EngineCore pid=8748)     triton_reshape_and_cache_flash(
(EngineCore pid=8748)     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
(EngineCore pid=8748)         key,
(EngineCore pid=8748)         ^^^^
(EngineCore pid=8748)     ...<6 lines>...
(EngineCore pid=8748)         layer._v_scale,
(EngineCore pid=8748)         ^^^^^^^^^^^^^^^
(EngineCore pid=8748)     )
(EngineCore pid=8748)     ^
(EngineCore pid=8748)   File "/home/jlj/dev/inferno-repro/.venv/lib/python3.13/site-packages/vllm/v1/attention/ops/triton_reshape_and_cache_flash.py", line 353, in triton_reshape_and_cache_flash
(EngineCore pid=8748)     assert kv_cache_dtype == "auto" or is_quantized_kv_cache(kv_cache_dtype), (
(EngineCore pid=8748)            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=8748) AssertionError: unsupported kv_cache_dtype (str), got float16.

Before submitting a new issue...

  • Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#api#dependency error#configuration error#environment variable#network issue

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING