| How to quantize 70B model so it will fit on 2x4090 GPUs: | |
| I tried EXL2, AutoAWQ, and SqueezeLLM and they all failed for different reasons (issues opened). | |
| HQQ worked: | |
| I rented a 4x GPU 1TB RAM ($19/hr) instance on runpod with 1024GB container and 1024GB workspace disk space. | |
| I think you only need 2x GPU with 80GB VRAM and 512GB+ system RAM so probably overpaid. | |
| Note you need to fill in the form to get access to the 70B Meta weights. |
| [ | |
| "Amaranth", | |
| "Amber", | |
| "Amethyst", | |
| "Apricot", | |
| "Aquamarine", | |
| "Azure", | |
| "Beige", | |
| "Black", | |
| "Blue", |
| // Charles Proxy License | |
| // Registration code for any version of Charles, who would want to use a cracked version? | |
| // Charles 4.5.5 is currently the latest version and is available. | |
| Registered Name: https://zhile.io | |
| License Key: 48891cf209c6d32bf4 | |
| Author: Neo Peng |
This is not working complete code.
This is strictly a v0, scrapy, proof of concept for the first version of a personal AI Assistant working end to end in just ~322 LOC.
It's only a frame of reference for you to consume the core ideas of how to build a POC of a personal AI Assistant.
To see the high level of how this works check out the explanation video. To follow our agentic journey check out the @IndyDevDan channel.
Stay focused, keep building.
This is a tutorial on how to write a fuzzer for a non-trivial real-world library, namely Artem Amirkhanov's CDT. It is a library for computing Constrained Delaunay Triangulations (CDTs, hence the name of the library). We will be working from the 9d99b32ae56b26cd2781678dc4405c98b8679a9f commit, since that is what I originally wrote the fuzzer for, and that way, we will be able to rediscover the same bugs I found back then.
If you want to follow along, clone the library using
$ git clone https://github.com/artem-ogre/CDT
$ cd CDT
| // 1. Import everything | |
| import { Wallet, BigNumber, ethers, providers } from 'ethers' | |
| const { FlashbotsBundleProvider, FlashbotsBundleResolution } = require('@flashbots/ethers-provider-bundle') | |
| /* | |
| Mainnet | |
| const provider = new providers.JsonRpcProvider('https://eth-mainnet.g.alchemy.com/v2/cmHEQqWnoliAP0lgTieeUtwHi0KxEOlh') | |
| const wsProvider = new providers.WebSocketProvider('wss://eth-mainnet.g.alchemy.com/v2/cmHEQqWnoliAP0lgTieeUtwHi0KxEOlh') | |
| */ |
| using System.Collections.Generic; | |
| using UnityEngine; | |
| [RequireComponent(typeof(MeshFilter)), RequireComponent(typeof(MeshRenderer))] | |
| public class ProtoMesh : MonoBehaviour | |
| { | |
| #region Helper Properties | |
| // Simple helpers to cache the MeshFilter and MeshRenderer | |
| public MeshRenderer MyMeshRenderer | |
| { |
| <html> | |
| <head> | |
| <title>Mixpanel Tracking Proxy Demo</title> | |
| <script type="text/javascript"> | |
| /** | |
| * Configuration Variables - CHANGE THESE! | |
| */ | |
| const MIXPANEL_PROJECT_TOKEN = YOUR_MIXPANEL_PROJECT_TOKEN; // e.g. "67e8bfdec29d84ab2d36ae18c57b8535" | |
| const MIXPANEL_PROXY_DOMAIN = YOUR_PROXY_DOMAIN; // e.g. "https://proxy-eoca2pin3q-uc.a.run.app" | |
| # on os x use brew to get ffmpeg with libfdk_aac | |
| brew install ffmpeg --with-fdk-aac | |
| # convert and use m4a as file extension | |
| find . -name '*.flac' -exec sh -c 'ffmpeg -i "$1" -c:a libfdk_aac -b:a 320k "${1%.flac}.m4a"' _ {} \; |