shubham-kaushal

envs:
  model_name: llama-2
  tokenizer_path: /home/gcpuser/seed_workdir/ckpt/llama2-7b/original/tokenizer.model

run: |
  cd JetStream
  python benchmarks/benchmark_serving.py \
    --tokenizer=$tokenizer_path --num-prompts=100 \
    --dataset openorca --save-request-outputs \
    --warmup-mode=sampled --model=$model_name

shubham-kaushal

import time
from google.cloud import monitoring_v3
from prometheus_client import start_http_server, Gauge
import argparse

global project_id
global accelerator_utilization
global client
global project_name

shubham-kaushal

import argparse
import http.server
import socketserver


class MyHttpRequestHandler(http.server.SimpleHTTPRequestHandler):

    def do_GET(self):
        # Return 200 for all paths
        self.send_response(200)

shubham-kaushal

    /* Model mockup */
    var colors = [0xcceeee, 0xddeeff, 0xeeccee, 0x00cc00, 0x0000cc, 0xcc3333];

    for (i = 0; i < 10; ++i) {
        var w = 200 * Math.random();
        var h = 200 * Math.random();
        var d = 5 * Math.random() + 1;
        var dx = (Math.random() * 40 - 20);
        var dy = (Math.random() * 40 - 20);
        var dz = Math.random() * 10;