vLLM Ray Phi-3 bug

vllm-phi3-ray-bug.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-07-20 10:43:35,673\tINFO worker.py:1779 -- Started a local Ray instance. View the dashboard at \u001b[1m\u001b[32mhttp://127.0.0.1:8265 \u001b[39m\u001b[22m\n"
     ]
    }
   ],
   "source": [
    "from vllm import AsyncEngineArgs, AsyncLLMEngine\n",
    "\n",
    "# model source: https://huggingface.co/microsoft/Phi-3-mini-128k-instruct\n",
    "\n",
    "# this config works\n",
    "mp_args = AsyncEngineArgs(\n",
    "    model=\"../../models/microsoft/Phi-3-mini-128k-instruct\",\n",
    "    trust_remote_code=True,\n",
    "    max_model_len=8000, # limit mem utilization for this example\n",
    ")\n",
    "\n",
    "# this config does not work. it just sits at \n",
    "#   \"INFO worker.py:1779 -- Started a local Ray instance. View the dashboard at...\"\n",
    "# The actor dies with a `ray.exceptions.RaySystemError: System error: No module named 'transformers_modules'`\n",
    "ray_args = AsyncEngineArgs(\n",
    "    model=\"../../models/microsoft/Phi-3-mini-128k-instruct\",\n",
    "    trust_remote_code=True,\n",
    "    max_model_len=8000,\n",
    "    engine_use_ray=True,\n",
    "    distributed_executor_backend=\"ray\",\n",
    ")\n",
    "\n",
    "# engine = AsyncLLMEngine.from_engine_args(mp_args)\n",
    "engine = AsyncLLMEngine.from_engine_args(ray_args)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "infyrence_env",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}