Merge pull request vllm-project#2 from Bellk17/main

Merge latest

.buildkite/run-neuron-test.sh

@@ -0,0 +1,37 @@
+# This script build the Neuron docker image and run the API server inside the container.
+# It serves a sanity check for compilation and basic model usage.
+set -e
+
+# Try building the docker image
+aws ecr get-login-password --region us-west-2 | docker login --username AWS --password-stdin 763104351884.dkr.ecr.us-west-2.amazonaws.com
+docker build -t neuron -f Dockerfile.neuron .
+
+# Setup cleanup
+remove_docker_container() { docker rm -f neuron || true; }
+trap remove_docker_container EXIT
+remove_docker_container
+
+# Run the image
+docker run --device=/dev/neuron0 --device=/dev/neuron1 --network host --name neuron neuron python3 -m vllm.entrypoints.api_server \
+       --model TinyLlama/TinyLlama-1.1B-Chat-v1.0 --max-num-seqs 8 --max-model-len 128 --block-size 128 --device neuron --tensor-parallel-size 2 &
+
+# Wait for the server to start
+wait_for_server_to_start() {
+    timeout=300
+    counter=0
+
+    while [ "$(curl -s -o /dev/null -w ''%{http_code}'' localhost:8000/health)" != "200" ]; do
+        sleep 1
+        counter=$((counter + 1))
+        if [ $counter -ge $timeout ]; then
+            echo "Timeout after $timeout seconds"
+            break
+        fi
+    done
+}
+wait_for_server_to_start
+
+# Test a simple prompt
+curl -X POST -H "Content-Type: application/json" \
+    localhost:8000/generate \
+    -d '{"prompt": "San Francisco is a"}'

.buildkite/test-pipeline.yaml

@@ -12,7 +12,13 @@ steps:
   command: pytest -v -s async_engine
 
 - label: Basic Correctness Test
-  command: pytest -v -s basic_correctness
+  commands:
+  - VLLM_ATTENTION_BACKEND=XFORMERS pytest -v -s basic_correctness/test_basic_correctness.py
+  - VLLM_ATTENTION_BACKEND=FLASH_ATTN pytest -v -s basic_correctness/test_basic_correctness.py
+  - VLLM_ATTENTION_BACKEND=ROCM_FLASH pytest -v -s basic_correctness/test_basic_correctness.py
+  - VLLM_ATTENTION_BACKEND=XFORMERS pytest -v -s basic_correctness/test_chunked_prefill.py
+  - VLLM_ATTENTION_BACKEND=FLASH_ATTN pytest -v -s basic_correctness/test_chunked_prefill.py
+  - VLLM_ATTENTION_BACKEND=ROCM_FLASH pytest -v -s basic_correctness/test_chunked_prefill.py
 
 - label: Core Test
   command: pytest -v -s core
@@ -27,13 +33,14 @@ steps:
   num_gpus: 2 # only support 1 or 2 for now.
   commands:
   - pytest -v -s test_pynccl.py
+  - pytest -v -s test_pynccl_library.py
   - TEST_DIST_MODEL=facebook/opt-125m pytest -v -s test_basic_distributed_correctness.py
   - TEST_DIST_MODEL=meta-llama/Llama-2-7b-hf pytest -v -s test_basic_distributed_correctness.py
   - TEST_DIST_MODEL=facebook/opt-125m pytest -v -s test_chunked_prefill_distributed.py
   - TEST_DIST_MODEL=meta-llama/Llama-2-7b-hf pytest -v -s test_chunked_prefill_distributed.py
 
 - label: Engine Test
-  command: pytest -v -s engine tokenization test_sequence.py test_config.py
+  command: pytest -v -s engine tokenization test_sequence.py test_config.py test_logger.py
 
 - label: Entrypoints Test
   commands:
@@ -85,6 +92,9 @@ steps:
   command: pytest -v -s lora --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
   parallelism: 4
 
+- label: Tensorizer Test
+  command: apt-get install curl libsodium23 && pytest -v -s tensorizer_loader
+
 - label: Metrics Test
   command: pytest -v -s metrics
 

.buildkite/test-template.j2

@@ -3,13 +3,6 @@
 {% set default_working_dir = "/vllm-workspace/tests" %}
 
 steps:
-  - label: "AMD Test"
-    agents:
-      queue: amd
-    command: bash .buildkite/run-amd-test.sh
-
-  - label: "CPU Test"
-    command: bash .buildkite/run-cpu-test.sh
 
   - label: ":docker: build image"
     commands:
@@ -23,6 +16,19 @@ steps:
           limit: 5
   - wait
 
+  - label: "AMD Test"
+    agents:
+      queue: amd
+    command: bash .buildkite/run-amd-test.sh
+
+  - label: "Neuron Test"
+    agents:
+      queue: neuron
+    command: bash .buildkite/run-neuron-test.sh
+
+  - label: "CPU Test"
+    command: bash .buildkite/run-cpu-test.sh
+
   {% for step in steps %}
   - label: "{{ step.label }}"
     agents:

.github/ISSUE_TEMPLATE/200-installation.yml

@@ -18,6 +18,7 @@ body:
       # For security purposes, please feel free to check the contents of collect_env.py before running it.
       python collect_env.py
       ```
+      It is suggested to download and execute the latest script, as vllm might frequently update the diagnosis information needed for accurately and quickly responding to issues.
     value: |
       ```text
       The output of `python collect_env.py`

.github/ISSUE_TEMPLATE/300-usage.yml

@@ -18,6 +18,7 @@ body:
       # For security purposes, please feel free to check the contents of collect_env.py before running it.
       python collect_env.py
       ```
+      It is suggested to download and execute the latest script, as vllm might frequently update the diagnosis information needed for accurately and quickly responding to issues.
     value: |
       ```text
       The output of `python collect_env.py`

.github/ISSUE_TEMPLATE/400-bug report.yml

@@ -18,6 +18,7 @@ body:
       # For security purposes, please feel free to check the contents of collect_env.py before running it.
       python collect_env.py
       ```
+      It is suggested to download and execute the latest script, as vllm might frequently update the diagnosis information needed for accurately and quickly responding to issues.
     value: |
       ```text
       The output of `python collect_env.py`
@@ -57,6 +58,8 @@ body:
       If the code is too long (hopefully, it isn't), feel free to put it in a public gist and link it in the issue: https://gist.github.com.
 
       Please also paste or describe the results you observe instead of the expected results. If you observe an error, please paste the error message including the **full** traceback of the exception. It may be relevant to wrap error messages in ```` ```triple quotes blocks``` ````.
+
+      If you experienced crashes or hangs, it would be helpful to run vllm with `export VLLM_TRACE_FUNCTION=1` . All the function calls in vllm will be recorded. Inspect these log files, and tell which function crashes or hangs.
     placeholder: |
       A clear and concise description of what the bug is.
 

.github/ISSUE_TEMPLATE/700-performance discussion.yml

@@ -39,6 +39,7 @@ body:
       # For security purposes, please feel free to check the contents of collect_env.py before running it.
       python collect_env.py
       ```
+      It is suggested to download and execute the latest script, as vllm might frequently update the diagnosis information needed for accurately and quickly responding to issues.
     value: |
       ```text
       The output of `python collect_env.py`

.github/workflows/mypy.yaml

@@ -0,0 +1,50 @@
+name: mypy
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  push:
+    branches:
+      - main
+  pull_request:
+    branches:
+      - main
+
+jobs:
+  ruff:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.8", "3.9", "3.10", "3.11"]
+    steps:
+    - uses: actions/checkout@v2
+    - name: Set up Python ${{ matrix.python-version }}
+      uses: actions/setup-python@v2
+      with:
+        python-version: ${{ matrix.python-version }}
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install mypy==1.9.0
+        pip install types-setuptools
+        pip install types-PyYAML
+        pip install types-requests
+        pip install types-setuptools
+    - name: Mypy
+      run: |
+        mypy vllm/attention/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/core/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/distributed/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/entrypoints/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/executor/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/usage/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/transformers_utils/*.py --follow-imports=skip --config-file pyproject.toml
+
+        mypy vllm/engine/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/worker/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/spec_decode/*.py --follow-imports=skip --config-file pyproject.toml
+        mypy vllm/model_executor/*.py --follow-imports=skip --config-file pyproject.toml
+        # TODO(sang): Follow up
+        # mypy vllm/lora/*.py --follow-imports=skip --config-file pyproject.toml
+

.github/workflows/ruff.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.10"]
+        python-version: ["3.8", "3.9", "3.10", "3.11"]
     steps:
     - uses: actions/checkout@v2
     - name: Set up Python ${{ matrix.python-version }}

.github/workflows/yapf.yml

@@ -14,7 +14,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.10"]
+        python-version: ["3.8", "3.9", "3.10", "3.11"]
     steps:
     - uses: actions/checkout@v2
     - name: Set up Python ${{ matrix.python-version }}

Dockerfile.neuron

@@ -0,0 +1,36 @@
+# default base image
+ARG BASE_IMAGE="763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-inference-neuronx:2.1.1-neuronx-py310-sdk2.17.0-ubuntu20.04"
+
+FROM $BASE_IMAGE
+
+RUN echo "Base image is $BASE_IMAGE"
+
+# Install some basic utilities
+RUN apt-get update && apt-get install python3 python3-pip -y
+
+### Mount Point ###
+# When launching the container, mount the code directory to /app
+ARG APP_MOUNT=/app
+VOLUME [ ${APP_MOUNT} ]
+WORKDIR ${APP_MOUNT}
+
+RUN python3 -m pip install --upgrade pip
+RUN python3 -m pip install --no-cache-dir fastapi ninja tokenizers pandas
+RUN python3 -m pip install sentencepiece transformers==4.36.2 -U
+RUN python3 -m pip install transformers-neuronx --extra-index-url=https://pip.repos.neuron.amazonaws.com -U
+RUN python3 -m pip install --pre neuronx-cc==2.12.* --extra-index-url=https://pip.repos.neuron.amazonaws.com -U
+
+COPY ./vllm /app/vllm/vllm
+COPY ./setup.py /app/vllm/setup.py
+COPY ./requirements-common.txt /app/vllm/requirements-common.txt
+COPY ./requirements-neuron.txt /app/vllm/requirements-neuron.txt
+
+RUN cd /app/vllm \
+    && python3 -m pip install -U -r requirements-neuron.txt
+
+ENV VLLM_BUILD_WITH_NEURON 1
+RUN cd /app/vllm \
+    && pip install -e . \
+    && cd ..
+
+CMD ["/bin/bash"]

README.md

@@ -69,10 +69,10 @@ vLLM seamlessly supports many Hugging Face models, including the following archi
 - InternLM (`internlm/internlm-7b`, `internlm/internlm-chat-7b`, etc.)
 - InternLM2 (`internlm/internlm2-7b`, `internlm/internlm2-chat-7b`, etc.)
 - Jais (`core42/jais-13b`, `core42/jais-13b-chat`, `core42/jais-30b-v3`, `core42/jais-30b-chat-v3`, etc.)
-- LLaMA & LLaMA-2 (`meta-llama/Llama-2-70b-hf`, `lmsys/vicuna-13b-v1.3`, `young-geng/koala`, `openlm-research/open_llama_13b`, etc.)
+- LLaMA, Llama 2, and Meta Llama 3 (`meta-llama/Meta-Llama-3-8B-Instruct`, `meta-llama/Meta-Llama-3-70B-Instruct`, `meta-llama/Llama-2-70b-hf`, `lmsys/vicuna-13b-v1.3`, `young-geng/koala`, `openlm-research/open_llama_13b`, etc.)
 - MiniCPM (`openbmb/MiniCPM-2B-sft-bf16`, `openbmb/MiniCPM-2B-dpo-bf16`, etc.)
 - Mistral (`mistralai/Mistral-7B-v0.1`, `mistralai/Mistral-7B-Instruct-v0.1`, etc.)
-- Mixtral (`mistralai/Mixtral-8x7B-v0.1`, `mistralai/Mixtral-8x7B-Instruct-v0.1`, etc.)
+- Mixtral (`mistralai/Mixtral-8x7B-v0.1`, `mistralai/Mixtral-8x7B-Instruct-v0.1`, `mistral-community/Mixtral-8x22B-v0.1`, etc.)
 - MPT (`mosaicml/mpt-7b`, `mosaicml/mpt-30b`, etc.)
 - OLMo (`allenai/OLMo-1B`, `allenai/OLMo-7B`, etc.)
 - OPT (`facebook/opt-66b`, `facebook/opt-iml-max-30b`, etc.)

benchmarks/backend_request_func.py

@@ -135,6 +135,7 @@ async def async_request_trt_llm(
                                               "data:")
 
                         data = json.loads(chunk)
+                        output.generated_text += data["text_output"]
                         timestamp = time.perf_counter()
                         # First token
                         if ttft == 0.0:
@@ -149,7 +150,6 @@ async def async_request_trt_llm(
                         most_recent_timestamp = timestamp
 
                     output.latency = most_recent_timestamp - st
-                    output.generated_text = json.loads(data)["text_output"]
                     output.success = True
 
                 else:

benchmarks/benchmark_latency.py

@@ -9,6 +9,7 @@
 from tqdm import tqdm
 
 from vllm import LLM, SamplingParams
+from vllm.model_executor.layers.quantization import QUANTIZATION_METHODS
 
 
 def main(args: argparse.Namespace):
@@ -101,7 +102,7 @@ def run_to_completion(profile_dir: Optional[str] = None):
     parser.add_argument('--tokenizer', type=str, default=None)
     parser.add_argument('--quantization',
                         '-q',
-                        choices=['awq', 'gptq', 'squeezellm', None],
+                        choices=[*QUANTIZATION_METHODS, None],
                         default=None)
     parser.add_argument('--tensor-parallel-size', '-tp', type=int, default=1)
     parser.add_argument('--input-len', type=int, default=32)

benchmarks/benchmark_throughput.py

@@ -10,6 +10,8 @@
 from transformers import (AutoModelForCausalLM, AutoTokenizer,
                           PreTrainedTokenizerBase)
 
+from vllm.model_executor.layers.quantization import QUANTIZATION_METHODS
+
 
 def sample_requests(
     dataset_path: str,
@@ -267,7 +269,7 @@ def main(args: argparse.Namespace):
     parser.add_argument("--tokenizer", type=str, default=None)
     parser.add_argument('--quantization',
                         '-q',
-                        choices=['awq', 'gptq', 'squeezellm', None],
+                        choices=[*QUANTIZATION_METHODS, None],
                         default=None)
     parser.add_argument("--tensor-parallel-size", "-tp", type=int, default=1)
     parser.add_argument("--n",

collect_env.py

@@ -63,6 +63,7 @@
     "magma",
     "triton",
     "optree",
+    "nccl",
 }
 
 DEFAULT_PIP_PATTERNS = {
@@ -73,6 +74,7 @@
     "triton",
     "optree",
     "onnx",
+    "nccl",
 }
 
 

csrc/punica/bgmv/bgmv_config.h

@@ -47,6 +47,7 @@ void bgmv_kernel(out_T *__restrict__ Y, const in_T *__restrict__ X,
     f(in_T, out_T, W_T, narrow, 13696) \
     f(in_T, out_T, W_T, narrow, 13824) \
     f(in_T, out_T, W_T, narrow, 14336) \
+    f(in_T, out_T, W_T, narrow, 15360) \
     f(in_T, out_T, W_T, narrow, 16384) \
     f(in_T, out_T, W_T, narrow, 20480) \
     f(in_T, out_T, W_T, narrow, 22016) \
@@ -59,6 +60,7 @@ void bgmv_kernel(out_T *__restrict__ Y, const in_T *__restrict__ X,
     f(in_T, out_T, W_T, narrow, 32768) \
     f(in_T, out_T, W_T, narrow, 33024) \
     f(in_T, out_T, W_T, narrow, 36864) \
+    f(in_T, out_T, W_T, narrow, 43264) \
     f(in_T, out_T, W_T, narrow, 49152) \
     f(in_T, out_T, W_T, narrow, 64000) \
     f(in_T, out_T, W_T, narrow, 64256) \

docs/source/conf.py

@@ -83,6 +83,7 @@
     "vllm._C",
     "numpy",
     "tqdm",
+    "tensorizer",
 ]
 
 for mock_target in autodoc_mock_imports:

docs/source/models/engine_args.rst

@@ -36,7 +36,7 @@ Below, you can find an explanation of every engine argument for vLLM:
 
     Directory to download and load the weights, default to the default cache dir of huggingface.
 
-.. option:: --load-format {auto,pt,safetensors,npcache,dummy}
+.. option:: --load-format {auto,pt,safetensors,npcache,dummy,tensorizer}
 
     The format of the model weights to load.
 
@@ -45,6 +45,7 @@ Below, you can find an explanation of every engine argument for vLLM:
     * "safetensors" will load the weights in the safetensors format.
     * "npcache" will load the weights in pytorch format and store a numpy cache to speed up the loading.
     * "dummy" will initialize the weights with random values, mainly for profiling.
+    * "tensorizer" will load serialized weights using `CoreWeave's Tensorizer model deserializer. <https://github.com/coreweave/tensorizer>`_ See `examples/tensorize_vllm_model.py <https://github.com/vllm-project/vllm/blob/main/examples/tensorize_vllm_model.py>`_ to serialize a vLLM model, and for more information.
 
 .. option:: --dtype {auto,half,float16,bfloat16,float,float32}