[Bug]: Llama3 tool calling broken with non-ascii characters #17703
Description
Your current environment
The output of python collect_env.py
PyTorch version: 2.6.0+cu124
Is debug build: False
CUDA used to build PyTorch: 12.4
ROCM used to build PyTorch: N/A
OS: Ubuntu 20.04.6 LTS (x86_64)
GCC version: (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0
Clang version: Could not collect
CMake version: version 3.28.0
Libc version: glibc-2.31
Python version: 3.10.15 (main, Oct 3 2024, 07:27:34) [GCC 11.2.0] (64-bit runtime)
Python platform: Linux-5.15.0-1073-azure-x86_64-with-glibc2.31
Is CUDA available: True
CUDA runtime version: 12.1.105
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration: GPU 0: NVIDIA A100 80GB PCIe
Nvidia driver version: 535.183.01
cuDNN version: Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.4.0
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.4.0
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.4.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.4.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.4.0
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.4.0
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.4.0
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.4.0
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
Byte Order: Little Endian
Address sizes: 48 bits physical, 48 bits virtual
CPU(s): 24
On-line CPU(s) list: 0-23
Thread(s) per core: 1
Core(s) per socket: 24
Socket(s): 1
NUMA node(s): 1
Vendor ID: AuthenticAMD
CPU family: 25
Model: 1
Model name: AMD EPYC 7V13 64-Core Processor
Stepping: 1
CPU MHz: 2445.438
BogoMIPS: 4890.87
Hypervisor vendor: Microsoft
Virtualization type: full
L1d cache: 768 KiB
L1i cache: 768 KiB
L2 cache: 12 MiB
L3 cache: 96 MiB
NUMA node0 CPU(s): 0-23
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 Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Mitigation; safe RET, no microcode
Vulnerability Spec store bypass: Vulnerable
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Retpolines; STIBP disabled; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
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 tsc_reliable nonstop_tsc cpuid extd_apicid aperfmperf pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw topoext perfctr_core invpcid_single vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves clzero xsaveerptr rdpru arat umip vaes vpclmulqdq rdpid fsrm
Versions of relevant libraries:
[pip3] mypy==1.15.0
[pip3] mypy-extensions==1.0.0
[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-cusparselt-cu12==0.6.2
[pip3] nvidia-nccl-cu12==2.21.5
[pip3] nvidia-nvjitlink-cu12==12.4.127
[pip3] nvidia-nvtx-cu12==12.4.127
[pip3] onnxruntime-gpu==1.20.1
[pip3] pyzmq==26.2.0
[pip3] torch==2.6.0
[pip3] torchaudio==2.6.0
[pip3] torchvision==0.21.0
[pip3] transformers==4.51.0
[pip3] triton==3.2.0
[conda] No relevant packages
ROCM Version: Could not collect
Neuron SDK Version: N/A
vLLM Version: 0.8.3
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
GPU0 CPU Affinity NUMA Affinity GPU NUMA ID
GPU0 X 0-23 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
LD_LIBRARY_PATH=/usr/local/cuda-12.1/lib64
MKL_THREADING_LAYER=GNU
NCCL_CUMEM_ENABLE=0
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
CUDA_MODULE_LOADING=LAZY
🐛 Describe the bug
When using the llama3_json tool call parser in streaming mode with non-ascii characters in the tool arguments an additional DeltaToolMessage is yielded (containing the full argument dict) after the arguments were already streamed.
For brevity I will omit the tool definition. The example uses the example weather tool provided here.
import openai
client = openai.OpenAI(...)
response = client.chat.completions.create(
model="SOME_LLAMA3_MODEL_WITH_LLAMA3_JSON_TOOL_CALL_PARSER,
messages=[{"role": "user", "content": "Wie ist das Wetter in Münster?"}],
tools=tools,
tool_choice="auto",
stream=True,
)
for chunk in response:
print(chunk)The expected output would be something along the lines of
ChatCompletionChunk(id='chatcmpl-9fcd4cff278d4cfe8038f985c8088262', choices=[Choice(delta=ChoiceDelta(content='', function_call=None, refusal=None, role='assistant', tool_calls=None), finish_reason=None, index=0, logprobs=None)], created=1746517938, model='llama3', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None)
ChatCompletionChunk(id='chatcmpl-9fcd4cff278d4cfe8038f985c8088262', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role=None, tool_calls=[ChoiceDeltaToolCall(index=0, id='chatcmpl-tool-8b06a11fa7f1439b9de4acebf172dfb7', function=ChoiceDeltaToolCallFunction(arguments=None, name='get_weather'), type='function')]), finish_reason=None, index=0, logprobs=None)], created=1746517938, model='llama3', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None)
ChatCompletionChunk(id='chatcmpl-9fcd4cff278d4cfe8038f985c8088262', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role=None, tool_calls=[ChoiceDeltaToolCall(index=0, id=None, function=ChoiceDeltaToolCallFunction(arguments='{"location": "', name=None), type=None)]), finish_reason=None, index=0, logprobs=None)], created=1746517938, model='llama3', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None)
ChatCompletionChunk(id='chatcmpl-9fcd4cff278d4cfe8038f985c8088262', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role=None, tool_calls=[ChoiceDeltaToolCall(index=0, id=None, function=ChoiceDeltaToolCallFunction(arguments='M', name=None), type=None)]), finish_reason=None, index=0, logprobs=None)], created=1746517938, model='llama3', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None)
ChatCompletionChunk(id='chatcmpl-9fcd4cff278d4cfe8038f985c8088262', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role=None, tool_calls=[ChoiceDeltaToolCall(index=0, id=None, function=ChoiceDeltaToolCallFunction(arguments='ün', name=None), type=None)]), finish_reason=None, index=0, logprobs=None)], created=1746517938, model='llama3', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None)
ChatCompletionChunk(id='chatcmpl-9fcd4cff278d4cfe8038f985c8088262', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role=None, tool_calls=[ChoiceDeltaToolCall(index=0, id=None, function=ChoiceDeltaToolCallFunction(arguments='ster"', name=None), type=None)]), finish_reason=None, index=0, logprobs=None)], created=1746517938, model='llama3', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None)
ChatCompletionChunk(id='chatcmpl-9fcd4cff278d4cfe8038f985c8088262', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role=None, tool_calls=[ChoiceDeltaToolCall(index=0, id=None, function=ChoiceDeltaToolCallFunction(arguments=', "unit": "', name=None), type=None)]), finish_reason=None, index=0, logprobs=None)], created=1746517938, model='llama3', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None)
ChatCompletionChunk(id='chatcmpl-9fcd4cff278d4cfe8038f985c8088262', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role=None, tool_calls=[ChoiceDeltaToolCall(index=0, id=None, function=ChoiceDeltaToolCallFunction(arguments='c', name=None), type=None)]), finish_reason=None, index=0, logprobs=None)], created=1746517938, model='llama3', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None)
ChatCompletionChunk(id='chatcmpl-9fcd4cff278d4cfe8038f985c8088262', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role=None, tool_calls=[ChoiceDeltaToolCall(index=0, id=None, function=ChoiceDeltaToolCallFunction(arguments='elsius"}', name=None), type=None)]), finish_reason=None, index=0, logprobs=None)], created=1746517938, model='llama3', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None)
ChatCompletionChunk(id='chatcmpl-9fcd4cff278d4cfe8038f985c8088262', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role=None, tool_calls=[ChoiceDeltaToolCall(index=0, id=None, function=ChoiceDeltaToolCallFunction(arguments='', name=None), type=None)]), finish_reason='tool_calls', index=0, logprobs=None, stop_reason=128008)], created=1746517938, model='llama3', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None)
However, what actually happens is that instead of the final message having arguments='' we get
ChatCompletionChunk(id='chatcmpl-e6395e070c814db88997057248282911', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role=None, tool_calls=[ChoiceDeltaToolCall(index=0, id=None, function=ChoiceDeltaToolCallFunction(arguments='{"location": "Münster", "unit": "celsius"}', name=None), type=None)]), finish_reason='tool_calls', index=0, logprobs=None, stop_reason=128008)], created=1746511089, model='llama3', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None)
i.e. the final message contains the full tool call arguments dict.
The reason for this is that the llama3_json tool call parser dumps the arguments using json.dumps with ensure_ascii=True (implicitly since this is the default) while OpenAIServingChat compares already sent arguments using ensure_ascii=False (which is compatible with e.g. the hermes and mistral tool call parsers). This leads to OpenAIServingChat erroneously thinking it still has to sent something in the end when there are non-ascii characters in the arguments.
I will open a PR to make the llama3_json tool call parser consistent with this usage.
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.