Zach Williams zachwill

For the qX_0 variants, they are actually quite straight-forward so deepseek can come up with a correct result in 1 shot. It is already crazy enough, given that ChatGPT and Claude have never produced a working result for me.
For the qX_K it's more complicated, I would say most of the time I need to re-prompt it 4 to 8 more times.
The most difficult was q6_K, the code never works until I ask it to only optimize one specific part, while leaving the rest intact (so it does not mess up everything)

Around 3-5 minutes per response.

	You are an expert problem solver with deep analytical capabilities, guided by these core principles:

	1. Quantity Leads to Quality (95% confidence)
	- Rapid iteration beats theoretical perfection (like the ceramics class experiment where quantity produced better quality)
	- Getting "bad work" out quickly is essential (everyone has 100 bad projects in them)
	- Value learning rate over appearance of expertise
	- Early feedback loops are critical for improvement
	- Give yourself and others permission to be bad initially

	2. Essential vs Artificial Complexity (90% confidence)

	RANKED STYLE GUIDE (organized by category and importance)

	FUNDAMENTAL FORMATTING (95% confidence)
	1. Capitalization
	- Title case for headlines, sentence case for everything else
	- Job titles remain lowercase
	- Colon capitalization rules: capitalize independent clauses, lowercase dependent clauses

	2. Essential Punctuation (90% confidence)
	- Use Oxford comma (x, y, and z)

	import os
	import subprocess
	import webbrowser
	import time
	import pyperclip
	import tempfile
	from rich.console import Console
	from rich.panel import Panel
	from rich.prompt import Prompt

	import json
	import logging
	import re
	from typing import Any, AsyncGenerator, Optional, Union

	import aiohttp
	import openai
	from azure.search.documents.aio import SearchClient
	from azure.search.documents.models import QueryType

	# Best TL;DR prompt I’ve come up with

	---

	Role: Readable 60-word Content Summarizer

	Main Task:
	1. Read and understand the given article.
	2. Generate a single, dense and concise final summary, incorporating numerous specific entities relevant to the article. This summary should be 60 words and 4 sentences in length, while maintaining a Flesch-Kincaid Ease score of at least 60 to 70, ensuring it is neither too complex nor too simplistic.

	/**
	*Submitted for verification at Etherscan.io on 2021-04-16
	*/

	// SPDX-License-Identifier: MIT

	pragma solidity ^0.6.0;

	/**
	* @dev Interface of the ERC165 standard, as defined in the

	pragma solidity ^0.8.0;
	/**
	* @title KingFrog contract
	* @dev Extends ERC721Enumerable Non-Fungible Token Standard basic implementation
	*/

	/**
	* SPDX-License-Identifier: UNLICENSED
	*/

	date,domain,price,venue,length,dot_com,hyphen,numbers
	2020-05-30,sit.ai,12000,Sedo,3,0,0,0
	2020-05-30,zionism-israel.com,7801,GoDaddy,14,1,1,0
	2020-05-30,simplybetter.com,5500,DropCatch,12,1,0,0
	2020-05-30,libation.com,3656,NameJet,8,1,0,0
	2020-05-30,boots2business.org,3484,GoDaddy,14,0,0,1
	2020-05-30,mountainescape.com,3127,GoDaddy,14,1,0,0
	2020-05-30,cingulum.com,3000,BuyDomains,8,1,0,0
	2020-05-30,temperaturematters.com,2900,BuyDomains,18,1,0,0
	2020-05-30,landkind.com,2777,Sedo,8,1,0,0