[Bug]: Error when loading mistral and gemma model using VLLM docker

[Adevils]

Your current environment

PyTorch version: N/A Is debug build: N/A CUDA used to build PyTorch: N/A ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.3 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: 3.10.12 (main, Mar 22 2024, 16:50:05) [GCC 11.4.0] (64-bit runtime) Python platform: Linux-5.15.0-112-generic-x86_64-with-glibc2.35 Is CUDA available: N/A CUDA runtime version: 11.5.119 CUDA_MODULE_LOADING set to: N/A GPU models and configuration: GPU 0: NVIDIA H100 PCIe MIG 3g.40gb Device 0: MIG 3g.40gb Device 1:

Nvidia driver version: 555.42.02 cuDNN version: Probably one of the following: /usr/lib/x86_64-linux-gnu/libcudnn.so.8.9.7 /usr/lib/x86_64-linux-gnu/libcudnn_adv_infer.so.8.9.7 /usr/lib/x86_64-linux-gnu/libcudnn_adv_train.so.8.9.7 /usr/lib/x86_64-linux-gnu/libcudnn_cnn_infer.so.8.9.7 /usr/lib/x86_64-linux-gnu/libcudnn_cnn_train.so.8.9.7 /usr/lib/x86_64-linux-gnu/libcudnn_ops_infer.so.8.9.7 /usr/lib/x86_64-linux-gnu/libcudnn_ops_train.so.8.9.7 HIP runtime version: N/A MIOpen runtime version: N/A Is XNNPACK available: N/A

CPU: Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 46 bits physical, 57 bits virtual Byte Order: Little Endian CPU(s): 96 On-line CPU(s) list: 0-95 Vendor ID: GenuineIntel Model name: Intel(R) Xeon(R) Gold 5318N CPU @ 2.10GHz CPU family: 6 Model: 106 Thread(s) per core: 2 Core(s) per socket: 24 Socket(s): 2 Stepping: 6 CPU max MHz: 3400.0000 CPU min MHz: 800.0000 BogoMIPS: 4200.00 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 pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 invpcid_single intel_ppin ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect wbnoinvd dtherm ida arat pln pts avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid fsrm md_clear pconfig flush_l1d arch_capabilities Virtualization: VT-x L1d cache: 2.3 MiB (48 instances) L1i cache: 1.5 MiB (48 instances) L2 cache: 60 MiB (48 instances) L3 cache: 72 MiB (2 instances) NUMA node(s): 2 NUMA node0 CPU(s): 0,2,4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34,36,38,40,42,44,46,48,50,52,54,56,58,60,62,64,66,68,70,72,74,76,78,80,82,84,86,88,90,92,94 NUMA node1 CPU(s): 1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63,65,67,69,71,73,75,77,79,81,83,85,87,89,91,93,95 Vulnerability Gather data sampling: Mitigation; Microcode Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable Vulnerability Retbleed: Not affected Vulnerability Spec rstack overflow: Not affected Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization Vulnerability Spectre v2: Mitigation; Enhanced IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI Syscall hardening, KVM SW loop Vulnerability Srbds: Not affected Vulnerability Tsx async abort: Not affected

Versions of relevant libraries: [pip3] numpy==1.26.4 [pip3] transformers==4.42.4 [conda] Could not collect ROCM Version: Could not collect Neuron SDK Version: N/A vLLM Version: N/A vLLM Build Flags: CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled GPU Topology: GPU0 CPU Affinity NUMA Affinity GPU NUMA ID GPU0 X 1,3,5,7,9,11 1 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

🐛 Describe the bug

Mistral: sudo docker run --runtime nvidia --gpus all \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HUGGING_FACE_HUB_TOKEN=secret key" \ -p 8000:8000 \ --ipc=host \ vllm/vllm-openai:latest \ --model mistralai/Mistral-7B-Instruct-v0.1\ --gpu-memory-utilization 0.2 \ --max-model-len 2048

   **The above code throws me :**
   INFO 07-20 06:26:49 api_server.py:719] args: Namespace(host=None, port=8000, allow_credentials=False, allowed_origins=['*'], allowed_methods=['*'], allowed_headers=['*'], served_model_name=None, chat_template=None, response_role='assistant', model='mistralai/Mistral-7B-Instruct-v0.1', tokenizer=None, revision=None, tokenizer_revision=None, tokenizer_mode='auto', trust_remote_code=False, download_dir=None, load_format='auto', dtype='auto', max_model_len=2048, worker_use_ray=False, pipeline_parallel_size=1, tensor_parallel_size=1, max_parallel_loading_workers=None, block_size=16, seed=0, swap_space=4, gpu_memory_utilization=0.2, max_num_batched_tokens=None, max_num_seqs=256, max_paddings=256, disable_log_stats=False, quantization=None, enforce_eager=False, max_context_len_to_capture=8192, engine_use_ray=False, disable_log_requests=False, max_log_len=None)

INFO 07-20 06:26:50 llm_engine.py:73] Initializing an LLM engine with config: model='mistralai/Mistral-7B-Instruct-v0.1', tokenizer='mistralai/Mistral-7B-Instruct-v0.1', tokenizer_mode=auto, revision=None, tokenizer_revision=None, trust_remote_code=False, dtype=torch.bfloat16, max_seq_len=2048, download_dir=None, load_format=auto, tensor_parallel_size=1, quantization=None, enforce_eager=False, seed=0) Traceback (most recent call last): File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "/usr/lib/python3.10/runpy.py", line 86, in _run_code exec(code, run_globals) File "/workspace/vllm/entrypoints/openai/api_server.py", line 729, in engine = AsyncLLMEngine.from_engine_args(engine_args) File "/workspace/vllm/engine/async_llm_engine.py", line 496, in from_engine_args engine = cls(parallel_config.worker_use_ray, File "/workspace/vllm/engine/async_llm_engine.py", line 269, in init self.engine = self._init_engine(*args, kwargs) File "/workspace/vllm/engine/async_llm_engine.py", line 314, in _init_engine return engine_class(*args, *kwargs) File "/workspace/vllm/engine/llm_engine.py", line 98, in init self.tokenizer = get_tokenizer( File "/workspace/vllm/transformers_utils/tokenizer.py", line 28, in get_tokenizer tokenizer = AutoTokenizer.from_pretrained( File "/usr/local/lib/python3.10/dist-packages/transformers/models/auto/tokenization_auto.py", line 787, in from_pretrained return tokenizer_class.from_pretrained(pretrained_model_name_or_path, inputs, kwargs) File "/usr/local/lib/python3.10/dist-packages/transformers/tokenization_utils_base.py", line 2028, in from_pretrained return cls._from_pretrained( File "/usr/local/lib/python3.10/dist-packages/transformers/tokenization_utils_base.py", line 2260, in _from_pretrained tokenizer = cls(*init_inputs, **init_kwargs) File "/usr/local/lib/python3.10/dist-packages/transformers/models/llama/tokenization_llama_fast.py", line 124, in init super().init( File "/usr/local/lib/python3.10/dist-packages/transformers/tokenization_utils_fast.py", line 111, in init fast_tokenizer = TokenizerFast.from_file(fast_tokenizer_file) Exception: data did not match any variant of untagged enum PyPreTokenizerTypeWrapper at line 40 column 3

Gemma: sudo docker run --runtime nvidia --gpus all \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HUGGING_FACE_HUB_TOKEN=Secret key " \ -p 8000:8000 \ --ipc=host \ vllm/vllm-openai:latest \ --model google/gemma-2b\ --gpu-memory-utilization 0.2 \ --max-model-len 2048

The above code throws me :

INFO 07-20 06:31:40 api_server.py:719] args: Namespace(host=None, port=8000, allow_credentials=False, allowed_origins=[''], allowed_methods=[''], allowed_headers=['*'], served_model_name=None, chat_template=None, response_role='assistant', model='google/gemma-2b', tokenizer=None, revision=None, tokenizer_revision=None, tokenizer_mode='auto', trust_remote_code=False, download_dir=None, load_format='auto', dtype='auto', max_model_len=2048, worker_use_ray=False, pipeline_parallel_size=1, tensor_parallel_size=1, max_parallel_loading_workers=None, block_size=16, seed=0, swap_space=4, gpu_memory_utilization=0.2, max_num_batched_tokens=None, max_num_seqs=256, max_paddings=256, disable_log_stats=False, quantization=None, enforce_eager=False, max_context_len_to_capture=8192, engine_use_ray=False, disable_log_requests=False, max_log_len=None) Traceback (most recent call last): File "/usr/lib/python3.10/runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "/usr/lib/python3.10/runpy.py", line 86, in _run_code exec(code, run_globals) File "/workspace/vllm/entrypoints/openai/api_server.py", line 729, in engine = AsyncLLMEngine.from_engine_args(engine_args) File "/workspace/vllm/engine/async_llm_engine.py", line 490, in from_engine_args engine_configs = engine_args.create_engine_configs() File "/workspace/vllm/engine/arg_utils.py", line 216, in create_engine_configs model_config = ModelConfig(self.model, self.tokenizer, File "/workspace/vllm/config.py", line 101, in init self.hf_config = get_config(self.model, trust_remote_code, revision) File "/workspace/vllm/transformers_utils/config.py", line 23, in get_config config = AutoConfig.from_pretrained( File "/usr/local/lib/python3.10/dist-packages/transformers/models/auto/configuration_auto.py", line 1098, in from_pretrained config_class = CONFIG_MAPPING[config_dict["model_type"]] File "/usr/local/lib/python3.10/dist-packages/transformers/models/auto/configuration_auto.py", line 795, in getitem raise KeyError(key) KeyError: 'gemma'

[youkaichao]

is it because of transformers version inside the docker?

[Adevils]

After facing this issue i upgraded transformers to newer version using pip again , but unfortunately error persists, I am not sure how to resolve this. If Transformer version is issue with the docker then what version do you suggest for it?

[youkaichao]

might be related https://huggingface.co/google/gemma-2-27b-it/discussions/6

[Adevils]

What about the mistral model then?? Also how can i specify mig UUID when running the docker container?? So that i can specify that it should run in 2nd gpu? So what is the efficient way to use async with VLLM then? I have tried the fast api method and it throws error, you have suggested me earlier to use Docker with openai chat template, I now have tried with Docker vllm and it throws error. So what is the way to do it correctly for async req handling??

[youkaichao]

chances are your docker image is not the newest one. you need to check your docker registry to make sure it pulls the latest image.

[Adevils]

Thanks, I have used vllm/vllm-openai:latest. What is the stable version ? Also how can i specify mig UUID when running the docker container?? So that i can specify that it should run in 2nd gpu?

[youkaichao]

I'm not an expert on MIG in any way. If you choose to use MIG, the general advice is to reach out to your admin to learn about the usage, or search the web to find something like https://codeyarns.com/tech/2020-12-15-how-to-use-mig.html .

[youkaichao]

the latest version of docker image, is docker pull vllm/vllm-openai:v0.5.2

[Adevils]

Thank you for the response, Concurrently how many requests it can handle if we are doing it through vllm docker and openai chat end point??