[Bug]: Invalid tool arguments generated in v0.6.5

[marcelodiaz558]

Your current environment

Collecting environment information...
PyTorch version: 2.5.1+cu124
Is debug build: False
CUDA used to build PyTorch: 12.4
ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.4 LTS (x86_64)
GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version: Could not collect
CMake version: Could not collect
Libc version: glibc-2.35

Python version: 3.12.7 (main, Oct 1 2024, 08:52:12) [GCC 11.4.0] (64-bit runtime)
Python platform: Linux-6.5.0-35-generic-x86_64-with-glibc2.35
Is CUDA available: True
CUDA runtime version: Could not collect
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration:
GPU 0: NVIDIA GeForce RTX 4090
GPU 1: NVIDIA GeForce RTX 4090
GPU 2: NVIDIA GeForce RTX 4090
GPU 3: NVIDIA GeForce RTX 4090

Nvidia driver version: 550.90.07
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 43 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 96
On-line CPU(s) list: 0-95
Vendor ID: AuthenticAMD
Model name: AMD EPYC 7K62 48-Core Processor
CPU family: 23
Model: 49
Thread(s) per core: 2
Core(s) per socket: 48
Socket(s): 1
Stepping: 0
Frequency boost: enabled
CPU max MHz: 2600.0000
CPU min MHz: 1500.0000
BogoMIPS: 5200.40
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 hw_pstate ssbd mba ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 cqm rdt_a rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif v_spec_ctrl umip rdpid overflow_recov succor smca sev sev_es
Virtualization: AMD-V
L1d cache: 1.5 MiB (48 instances)
L1i cache: 1.5 MiB (48 instances)
L2 cache: 24 MiB (48 instances)
L3 cache: 192 MiB (12 instances)
NUMA node(s): 1
NUMA node0 CPU(s): 0-95
Vulnerability Gather data sampling: 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 Retbleed: Mitigation; untrained return thunk; SMT enabled with STIBP protection
Vulnerability Spec rstack overflow: Mitigation; Safe RET
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; Retpolines; IBPB conditional; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected

Versions of relevant libraries:
[pip3] flashinfer==0.1.6+cu121torch2.4
[pip3] numpy==1.26.4
[pip3] nvidia-cublas-cu12==12.4.5.8
[pip3] nvidia-cuda-cupti-cu12==12.4.127
[pip3] nvidia-cuda-nvrtc-cu12==12.4.127
[pip3] nvidia-cuda-runtime-cu12==12.4.127
[pip3] nvidia-cudnn-cu12==9.1.0.70
[pip3] nvidia-cufft-cu12==11.2.1.3
[pip3] nvidia-curand-cu12==10.3.5.147
[pip3] nvidia-cusolver-cu12==11.6.1.9
[pip3] nvidia-cusparse-cu12==12.3.1.170
[pip3] nvidia-ml-py==12.560.30
[pip3] nvidia-nccl-cu12==2.21.5
[pip3] nvidia-nvjitlink-cu12==12.4.127
[pip3] nvidia-nvtx-cu12==12.4.127
[pip3] pyzmq==26.2.0
[pip3] torch==2.5.1
[pip3] torchvision==0.20.1
[pip3] transformers==4.46.2
[pip3] triton==3.1.0
[conda] Could not collect
ROCM Version: Could not collect
Neuron SDK Version: N/A
vLLM Version: 0.6.4.post1
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
GPU0 GPU1 GPU2 GPU3 CPU Affinity NUMA Affinity GPU NUMA ID
GPU0 X SYS SYS SYS 0-95 0 N/A
GPU1 SYS X PHB SYS 0-95 0 N/A
GPU2 SYS PHB X SYS 0-95 0 N/A
GPU3 SYS SYS SYS X 0-95 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

NVIDIA_VISIBLE_DEVICES=all
NVIDIA_REQUIRE_CUDA=cuda>=12.4 brand=tesla,driver>=470,driver<471 brand=unknown,driver>=470,driver<471 brand=nvidia,driver>=470,driver<471 brand=nvidiartx,driver>=470,driver<471 brand=geforce,driver>=470,driver<471 brand=geforcertx,driver>=470,driver<471 brand=quadro,driver>=470,driver<471 brand=quadrortx,driver>=470,driver<471 brand=titan,driver>=470,driver<471 brand=titanrtx,driver>=470,driver<471 brand=tesla,driver>=525,driver<526 brand=unknown,driver>=525,driver<526 brand=nvidia,driver>=525,driver<526 brand=nvidiartx,driver>=525,driver<526 brand=geforce,driver>=525,driver<526 brand=geforcertx,driver>=525,driver<526 brand=quadro,driver>=525,driver<526 brand=quadrortx,driver>=525,driver<526 brand=titan,driver>=525,driver<526 brand=titanrtx,driver>=525,driver<526 brand=tesla,driver>=535,driver<536 brand=unknown,driver>=535,driver<536 brand=nvidia,driver>=535,driver<536 brand=nvidiartx,driver>=535,driver<536 brand=geforce,driver>=535,driver<536 brand=geforcertx,driver>=535,driver<536 brand=quadro,driver>=535,driver<536 brand=quadrortx,driver>=535,driver<536 brand=titan,driver>=535,driver<536 brand=titanrtx,driver>=535,driver<536
NVIDIA_DRIVER_CAPABILITIES=compute,utility
VLLM_USAGE_SOURCE=production-docker-image
CUDA_VERSION=12.4.1
LD_LIBRARY_PATH=/usr/local/lib/python3.12/dist-packages/cv2/../../lib64:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
CUDA_MODULE_LOADING=LAZY

Model Input Dumps

No response

🐛 Describe the bug

I have been using vLLM v0.6.4.post1 for a while, running it with Docker and the following parameters:

--model unsloth/Qwen2.5-14B-Instruct
--max-model-len 32768
--enable-auto-tool-choice
--tool-call-parser hermes
--enable-prefix-caching
--tensor-parallel-size 4
--disable-log-requests

On v0.6.4.post1, everything works well, except for an occasional log message when using tools (although the tools themselves function correctly). The log contains the following error:

2024-12-17T22:29:48.448500613Z INFO:     52.22.254.35:43726 - "POST /v1/chat/completions HTTP/1.1" 200 OK
2024-12-17T22:29:48.560290478Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338] Error trying to handle streaming tool call.
2024-12-17T22:29:48.560314239Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338] Traceback (most recent call last):
2024-12-17T22:29:48.560320019Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338]   File "/usr/local/lib/python3.12/dist-packages/vllm/entrypoints/openai/tool_parsers/hermes_tool_parser.py", line 227, in extract_tool_calls_streaming
2024-12-17T22:29:48.560331980Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338]     function_name: Union[str, None] = current_tool_call.get("name")
2024-12-17T22:29:48.560369061Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338]                                       ^^^^^^^^^^^^^^^^^^^^^
2024-12-17T22:29:48.560375201Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338] AttributeError: 'NoneType' object has no attribute 'get'
2024-12-17T22:29:48.908843003Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338] Error trying to handle streaming tool call.
2024-12-17T22:29:48.908886274Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338] Traceback (most recent call last):
2024-12-17T22:29:48.908892084Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338]   File "/usr/local/lib/python3.12/dist-packages/vllm/entrypoints/openai/tool_parsers/hermes_tool_parser.py", line 291, in extract_tool_calls_streaming
2024-12-17T22:29:48.908898214Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338]     args_delta_start_loc = cur_arguments_json.index(delta_text) \
2024-12-17T22:29:48.908904145Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338]                            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2024-12-17T22:29:48.908909785Z ERROR 12-17 14:29:48 hermes_tool_parser.py:338] ValueError: substring not found

When testing the newer Docker image vllm/vllm-openai:v0.6.5, I noticed that the model frequently generates incorrect function arguments. Sometimes, it outputs invalid JSON with missing commas, colons, or formats that don't align with the tool definitions, leading to JSON decoding issues on the client side.

The finish_reason is correctly set to "tool_calls", but the arguments format is often incorrect. This issue never occurred with v0.6.4.post1. Bear in mind that I have only tested this with streaming, I haven't tried with stream=False yet.

The issue might be related to #10398 or #10979.

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.