Alessandro Pappalardo volcacius

This worked on 14/May/23. The instructions will probably require updating in the future.

llama is a text prediction model similar to GPT-2, and the version of GPT-3 that has not been fine tuned yet. It is also possible to run fine tuned versions (like alpaca or vicuna with this. I think. Those versions are more focused on answering questions)

Note: I have been told that this does not support multiple GPUs. It can only use a single GPU.

It is possible to run LLama 13B with a 6GB graphics card now! (e.g. a RTX 2060). Thanks to the amazing work involved in llama.cpp. The latest change is CUDA/cuBLAS which allows you pick an arbitrary number of the transformer layers to be run on the GPU. This is perfect for low VRAM.

Clone llama.cpp from git, I am on commit 08737ef720f0510c7ec2aa84d7f70c691073c35d.

Reinforcement Learning for Language Models

Yoav Goldberg, April 2023.

Why RL?

With the release of the ChatGPT model and followup large language models (LLMs), there was a lot of discussion of the importance of "RLHF training", that is, "reinforcement learning from human feedback". I was puzzled for a while as to why RL (Reinforcement Learning) is better than learning from demonstrations (a.k.a supervised learning) for training language models. Shouldn't learning from demonstrations (or, in language model terminology "instruction fine tuning", learning to immitate human written answers) be sufficient? I came up with a theoretical argument that was somewhat convincing. But I came to realize there is an additional argumment which not only supports the case of RL training, but also requires it, in particular for models like ChatGPT. This additional argument is spelled out in (the first half of) a talk by John Schulman from OpenAI. This post pretty much

Stable Diffusion for AMD GPUs on Windows using DirectML

UPDATE: A faster (20x) approach for running Stable Diffusion using MLIR/Vulkan/IREE is available on Windows:

https://github.com/nod-ai/SHARK/blob/main/shark/examples/shark_inference/stable_diffusion/stable_diffusion_amd.md

Install 🤗 diffusers

conda create --name sd39 python=3.9 -y

Building Ray for aarch64

There are multiple ways to build ray on aarch64 processors. We will use the PYNQ image v2.4 with Ubuntu 18.04 OS as an example. In this document, we will introduce 2 approaches to build ray; we will always prefer the second approach (building on Amazon A1) since it is faster, cleaner, and easy to reproduce.

	# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved
	import os
	from multiprocessing import Queue
	import multiprocessing
	import logging
	from joblib import Parallel, delayed

	from omegaconf import open_dict

	from hydra._internal.pathlib import Path

	#lang rosette/safe

	(require rosette/lib/angelic
	rosette/lib/match)
	(current-bitwidth #f)

	(require (only-in racket list*))

	; bit operations
	(define (rotate-right i x)

	Mute these words in your settings here: https://twitter.com/settings/muted_keywords

	ActivityTweet
	generic_activity_highlights
	generic_activity_momentsbreaking
	RankedOrganicTweet
	suggest_activity
	suggest_activity_feed
	suggest_activity_highlights
	suggest_activity_tweet

	#! /bin/bash
	# Simple Utility Script for allowing debug of hardened macOS apps.
	# This is useful mostly for plug-in developer that would like keep developing without turning SIP off.
	# Credit for idea goes to (McMartin): https://forum.juce.com/t/apple-gatekeeper-notarised-distributables/29952/57?u=ttg
	# Update 2022-03-10: Based on Fabian's feedback, add capability to inject DYLD for sanitizers.
	#
	# Please note:
	# - Modern Logic (on M1s) uses `AUHostingService` which resides within the system thus not patchable and REQUIRES to turn-off SIP.
	# - Some hosts uses separate plug-in scanning or sandboxing.
	# if that's the case, it's required to patch those (if needed) and attach debugger to them instead.

	# Copyright (c) 2021 Zecong Hu
	#
	# Permission to use, copy, modify, and/or distribute this software for any
	# purpose with or without fee is hereby granted.
	#
	# THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH
	# REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY
	# AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT,
	# INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM
	# LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR