Comments (3)
Update: vllm 5.0 post, added ray to requirements-neuron.txt, still not working:
WARNING 06-14 08:15:40 _custom_ops.py:14] Failed to import from vllm._C with ModuleNotFoundError("No module named 'vllm._C'")
INFO 06-14 08:15:45 api_server.py:177] vLLM API server version 0.5.0.post1
INFO 06-14 08:15:45 api_server.py:178] args: Namespace(host=None, port=8000, uvicorn_log_level='info', allow_credentials=False, allowed_origins=['*'], allowed_methods=['*'], allowed_headers=['*'], api_key=None, lora_modules=None, chat_template=None, response_role='assistant', ssl_keyfile=None, ssl_certfile=None, ssl_ca_certs=None, ssl_cert_reqs=0, root_path=None, middleware=[], model='meta-llama/Meta-Llama-3-8B-Instruct', tokenizer=None, skip_tokenizer_init=False, revision=None, code_revision=None, tokenizer_revision=None, tokenizer_mode='auto', trust_remote_code=False, download_dir=None, load_format='auto', dtype='auto', kv_cache_dtype='auto', quantization_param_path=None, max_model_len=None, guided_decoding_backend='outlines', distributed_executor_backend=None, worker_use_ray=False, pipeline_parallel_size=1, tensor_parallel_size=2, max_parallel_loading_workers=None, ray_workers_use_nsight=False, block_size=16, enable_prefix_caching=True, disable_sliding_window=False, use_v2_block_manager=False, num_lookahead_slots=0, seed=0, swap_space=4, gpu_memory_utilization=0.9, num_gpu_blocks_override=None, max_num_batched_tokens=None, max_num_seqs=256, max_logprobs=20, disable_log_stats=False, quantization=None, rope_scaling=None, rope_theta=None, enforce_eager=False, max_context_len_to_capture=None, max_seq_len_to_capture=8192, disable_custom_all_reduce=False, tokenizer_pool_size=0, tokenizer_pool_type='ray', tokenizer_pool_extra_config=None, enable_lora=False, max_loras=1, max_lora_rank=16, lora_extra_vocab_size=256, lora_dtype='auto', long_lora_scaling_factors=None, max_cpu_loras=None, fully_sharded_loras=False, device='auto', image_input_type=None, image_token_id=None, image_input_shape=None, image_feature_size=None, image_processor=None, image_processor_revision=None, disable_image_processor=False, scheduler_delay_factor=0.0, enable_chunked_prefill=False, speculative_model=None, num_speculative_tokens=None, speculative_max_model_len=None, speculative_disable_by_batch_size=None, ngram_prompt_lookup_max=None, ngram_prompt_lookup_min=None, model_loader_extra_config=None, preemption_mode=None, served_model_name=None, qlora_adapter_name_or_path=None, engine_use_ray=False, disable_log_requests=True, max_log_len=None)
/opt/conda/lib/python3.10/site-packages/huggingface_hub/file_download.py:1132: FutureWarning: `resume_download` is deprecated and will be removed in version 1.0.0. Downloads always resume when possible. If you want to force a new download, use `force_download=True`.
warnings.warn(
INFO 06-14 08:15:47 config.py:623] Defaulting to use ray for distributed inference
WARNING 06-14 08:15:47 config.py:437] Possibly too large swap space. 8.00 GiB out of the 15.27 GiB total CPU memory is allocated for the swap space.
INFO 06-14 08:15:47 llm_engine.py:161] Initializing an LLM engine (v0.5.0.post1) with config: model='meta-llama/Meta-Llama-3-8B-Instruct', speculative_config=None, tokenizer='meta-llama/Meta-Llama-3-8B-Instruct', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, rope_scaling=None, rope_theta=None, tokenizer_revision=None, trust_remote_code=False, dtype=torch.bfloat16, max_seq_len=8192, download_dir=None, load_format=LoadFormat.AUTO, tensor_parallel_size=2, disable_custom_all_reduce=False, quantization=None, enforce_eager=False, kv_cache_dtype=auto, quantization_param_path=None, device_config=cpu, decoding_config=DecodingConfig(guided_decoding_backend='outlines'), seed=0, served_model_name=meta-llama/Meta-Llama-3-8B-Instruct)
Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
WARNING 06-14 08:15:48 utils.py:465] Pin memory is not supported on Neuron.
Loading checkpoint shards: 25%|██▌ | 1/4 [00:37<01:51, 37.24s/it]Traceback (most recent call last):
File "/usr/local/bin/dockerd-entrypoint.py", line 28, in <module>
subprocess.check_call(shlex.split(" ".join(sys.argv[1:])))
File "/opt/conda/lib/python3.10/subprocess.py", line 369, in check_call
raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['python3', '-m', 'vllm.entrypoints.openai.api_server', '--port=8000', '--model=meta-llama/Meta-Llama-3-8B-Instruct', '--tensor-parallel-size=2', '--disable-log-requests', '--enable-prefix-caching', '--gpu-memory-utilization=0.9']' died with <Signals.SIGKILL: 9>.
from vllm.
Ray is not required for Neuron device.
I see you are attaching only one core to the container when calling docker run
, for tp=2, at least 2 Neuron cores should be attached. Can you please modify the docker run command to include these two devices?
-device=/dev/neuron0 -device=/dev/neuron1
from vllm.
@ashrafMahgoub Reading the AWS doc, it seems that each neuron device should have two neuron cores. In this case, requesting for a single device should be enough? With EKS, I tried requesting for 2 devices on a inf2
instance that has a single Inferentia2 chip and it failed: Could not open the nd2
from vllm.
Related Issues (20)
- [Usage]: example/offline_inference_chat.py run error. HOT 1
- [Usage]: How to stop VLLM during generation ? HOT 2
- [RFC]: Pinned Caching with Automatic Prefix Caching (Related to Anthropic Prompt Caching API) HOT 12
- [Feature]: Official ROCm Binary to Speed Up vLLM Installation
- [Performance]: V0.6.0 version of the model, benchmarking found that the number of successful responses accounted for half of the number of requests, which is why HOT 7
- [Bug]: guided generation can't always finish generating the requested structure HOT 13
- [Bug]: vllm v0.6.0 profiler report GPUExecutorAsync object has no attribute '_run_workers' on ROCm and NV H20 HOT 2
- [Bug]: Kernel died while waiting for execute reply in Kaggle TPU VM v3-8 (2024-08-22) HOT 2
- [Usage]: how to shutdown vllm server HOT 1
- [Bug]: MistralTokenizer object has no attribute 'get_vocab' HOT 5
- [Feature]: Conditional Prompt Inclusion in `generate` Function for Streaming Efficiency
- [Bug]: internvl2 multi-prompt input with one image each get RuntimeError HOT 2
- [Misc]: vLLM v0.6.0 CUDA 12 missing wheel file HOT 1
- [Bug]: vLLM 0.6.0 produces CUDA error when loading quantized models (FP8) HOT 1
- [Usage]: Execution speed of non-Lora requests HOT 6
- [Bug]: Crash after few multi image calls HOT 2
- [Performance]: INFO 09-11 12:41:50 spec_decode_worker.py:790] SpecDecodeWorker scoring_time_ms is slow HOT 3
- [Bug]: Can't load any gemma 2 model
- [Bug]: OOM when running llama3.1-8B-Instruct HOT 1
- [Bug]: OOM When trying to start server on Sagemaker
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from vllm.