jdlugo

#!/usr/bin/env python3
"""Embed episodes on RunPod, saving results to local binary files.

Reads episode IDs and text from a CSV file, embeds them using local vLLM
instances, and writes (id, embedding) pairs to binary output files.

The output files can be synced back and loaded into Postgres using
sync_embeddings.py.

Usage (on RunPod):

jdlugo

#!/bin/bash
# RunPod startup script for vLLM serving Qwen3-Embedding-4B
#
# Usage: SSH into RunPod pod and run:
#   bash runpod-vllm-startup.sh
#
# Template: RunPod PyTorch 2.4.0 (or any template with CUDA 12.x)
# Recommended: 6x RTX A5000 ($1.62/hr) — one vLLM instance per GPU
#
# This script:

jdlugo

from flask import Flask, jsonify, request
from typing import Dict, Union
import logging

# Initialize Flask app
app = Flask(__name__)

# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

jdlugo

Verifying my Blockstack ID is secured with the address 1PRj1gSNUWEvwZ1nsqTjHspwQuFdqjKmfY https://explorer.blockstack.org/address/1PRj1gSNUWEvwZ1nsqTjHspwQuFdqjKmfY