Thomas Jc tmasjc

➰

Strange Loop

Two-bit Analyst

tmasjc / analyze_span_threshold.py

Created April 23, 2026 11:56

	"""Span-level LOO threshold analysis for a converged session.

	Re-embeds each FIT chunk's ``span_text`` via BGE-M3 ``/embed-all`` and
	reports ``T = quantile(LOO_nearest_distances, q)`` at q=0.90 and q=0.95,
	plus the full sorted LOO distance distribution.

	Counterpart to the production chunk-level pipeline in
	``src/anchor/threshold.py`` — same math, different vectors.
	"""

tmasjc / bootstrap.sh

Last active April 23, 2026 14:38

gpu-provision-on-new-server

	#!/usr/bin/env bash
	set -euo pipefail

	# Ubuntu bootstrap: base tools + Docker + Python (deadsnakes) + optional GPU utils + Ollama
	# Usage:
	# ./bootstrap.sh
	# ./bootstrap.sh --with-nvidia --nvidia-version 535
	# ./bootstrap.sh --with-amd
	# ./bootstrap.sh --no-snap
	# ./bootstrap.sh --no-ollama

tmasjc / china_cities.html

Created May 8, 2025 11:10

tmasjc / find_unique_strings.R

Last active December 13, 2024 15:46

寻找最相似或最不相似的长字符串

	library(tidyverse)
	library(text2vec)
	library(stopwords)
	library(jsonlite)
	library(listviewer)

	raw <- read_csv(rstudioapi::selectFile())

	preprocess <- function(x, special_keywords) {
	x \|>

tmasjc / sankey.R

Created December 4, 2024 15:50

Transform data and draw a Sankey diagram.

	library(tidyverse)
	library(highcharter)

	raw <- readr::read_rds("sample.rds")
	weight = "XXXX" # replace with correct column

	vpouch <- raw \|>
	select(-matches(weight)) \|>
	names() \|>
	(\(x)

tmasjc / query.py

Created October 13, 2023 02:57

	#!/usr/bin/env python3
	import argparse
	import configparser
	import sqlite3
	import openai
	import pinecone
	config = configparser.ConfigParser()
	config.read("config.ini")

tmasjc / neo4j_full_example.py

Created October 4, 2023 08:44

	from getpass import getpass
	from pprint import pprint
	from neo4j import GraphDatabase
	from pymongo import MongoClient

	## get data from document database ----
	mongodb_password = getpass("Enter Atlas password:")
	client = MongoClient(
	f"mongodb+srv://admin:{mongodb_password}@serverlessinstance0.9vrdx.mongodb.net"
	)

tmasjc / fizz_buzz.swift

Created November 26, 2022 07:37

Fizz Buzz demo in Swift

	// if x is a multiple of 3 then output "fizz",
	// if x is a multiple of 5 then output "buzz"
	// if x is a multiple of 15 then output "fizz buzz"

	// basic
	func fizz_buzz_alpha(max: Int) -> () {
	for number in 1...max {
	if number % 15 == 0 {
	print("fizz buzz")
	} else if number % 3 == 0 {

tmasjc / survival_shiny.R

Created August 13, 2021 06:55

	library(tidyverse)
	library(hrbrthemes)
	library(janitor)
	library(survival)
	library(survminer)
	library(shiny)
	library(miniUI)

	theme_set(theme_ipsum())
	pals <- rownames(RColorBrewer::brewer.pal.info)

tmasjc / project_network.R

Created July 2, 2021 04:08

Network visualization of team projects

	library(tidyverse)
	library(networkD3)

	raw <- pins::pin_get("projects", "local")

	df <- raw %>%
	mutate(associate = str_split(associate, pattern = "，\|、")) %>%
	unnest_longer(associate) %>%
	filter(!is.na(associate)) %>%
	count(owner, associate)