{ "cells": [ { "cell_type": "code", "execution_count": null, "id": "928261ae", "metadata": {}, "outputs": [], "source": [ "from vllm import LLM, SamplingParams" ] }, { "cell_type": "code", "execution_count": null, "id": "a35fc2b4", "metadata": {}, "outputs": [], "source": [ "model_id = \"./llms/OLMoE-1B-7B-0924-Instruct\"\n", "\n", "llm = LLM(\n", " model=model_id,\n", " # cpu_offload_gb=4,\n", " tensor_parallel_size=2,\n", " # gpu_memory_utilization=0.90,\n", " max_model_len=4096,\n", ")" ] }, { "cell_type": "code", "execution_count": null, "id": "6a708f11", "metadata": {}, "outputs": [], "source": [ "sampling_params = SamplingParams(\n", " temperature=0.6,\n", " top_p=0.95,\n", " top_k=20,\n", " max_tokens=1024,\n", ")\n", "\n", "# Prepare the input to the model\n", "prompt = \"Give me a short introduction to large language models.\"\n", "messages = [\n", " {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n", " {\"role\": \"user\", \"content\": prompt}\n", "]\n", "\n", "# Generate outputs\n", "outputs = llm.chat(\n", " messages, \n", " sampling_params=sampling_params,\n", " # chat_template_kwargs={\"enable_thinking\": True}, # Set to False to strictly disable thinking\n", ")\n", "\n", "# Print the outputs.\n", "for out in outputs:\n", " # out.prompt is the input prompt; out.outputs is a list of completion choices\n", " print(\"=== PROMPT ===\")\n", " print(out.prompt)\n", " print(\"=== COMPLETION ===\")\n", " print(out.outputs[0].text)\n", " print(\"\\n---\\n\")" ] } ], "metadata": { "kernelspec": { "display_name": "moe-explore", "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": "3.12.11" } }, "nbformat": 4, "nbformat_minor": 5 }