Ohad Eytan ohadeytan

Are multi-LLM-agent systems a thing? Yes they are. But.

Yoav Goldberg, Nov 24, 2024

This piece started with a pair of twitter and bluesky posts:

let's talk about "agents" (in the LLM sense). there's a lot of buzz around "multi-agent" systems where agents collaborate but... i don't really get how it differs from a thinking of a single agent with multiple modes of operation. what are the benefits of modeling as multi-agent?
— (((ل()(ل() 'yoav))))👾 (@yoavgo) November 23, 2024

The problem

Initial situation

Using Fedora 38, say you want to update the package albert from manuelschneid3r, hosted at https://software.opensuse.org/download.html?project=home:manuelschneid3r&package=albert.

And with dnf update, you get the following output:

~ sudo dnf update
Dependencies resolved.

Here is the best setup (I think so :D) for K-series Keychron keyboards on Linux.

Note: many newer Keychron keyboards use QMK as firmware and most tips here do not apply to them. Maybe the ones related to Bluetooth can be useful, but everything related to Apple's keyboard module (hid_apple) on Linux, won't work. As far as I know, all QMK-based boards use the hid_generic module instead. Examples of QMK-based boards are: Q, Q-Pro, V, K-Pro, etc.

Most of these commands have been tested on Ubuntu 20.04 and should also work on most Debian-based distributions. If a command happens not to work for you, take a look in the comment section.

Make Fn + F-keys work (NOT FOR QMK-BASED BOARDS)

Older Keychron keyboards (those not based on QMK) use the hid_apple driver on Linux, even in the Windows/Android mode, both in Bluetooth and Wired modes.

Install and use pip in a local directory without root/sudo access.

By: @saurabhshri

Why?

Many users when are given server access, do not have root (or sudo) privileges and can not simply do sudo apt-get install python-pip . Here's an easy way you can install and use pip without root (or sudo) access in a local directory. Note : This works without easy_install too.

How?

In your local clone of your forked repository, you can add the original GitHub repository as a "remote". ("Remotes" are like nicknames for the URLs of repositories - origin is one, for example.) Then you can fetch all the branches from that upstream repository, and rebase your work to continue working on the upstream version. In terms of commands that might look like:

Add the remote, call it "upstream":

git remote add upstream https://github.com/whoever/whatever.git

Fetch all the branches of that remote into remote-tracking branches, such as upstream/master:

git fetch upstream

	from langchain_huggingface import HuggingFaceEmbeddings
	from langchain_milvus.utils.sparse import BM25SparseEmbedding
	from langchain_milvus.vectorstores import Milvus

	# Using:
	# langchain_milvus==0.1.6
	# langchain-huggingface==0.1.0
	# pymilvus[model]==2.4.8

	DOCUMENTS = [

	""" Trains an agent with (stochastic) Policy Gradients on Pong. Uses OpenAI Gym. """
	import numpy as np
	import cPickle as pickle
	import gym

	# hyperparameters
	H = 200 # number of hidden layer neurons
	batch_size = 10 # every how many episodes to do a param update?
	learning_rate = 1e-4
	gamma = 0.99 # discount factor for reward

	# -- coding: utf-8 --
	"""
	Plot the convex hull around a set of points as a
	shaded polygon.

	@author: Nicolas Guarin Zapata
	@date: October 15, 2014
	"""
	import numpy as np
	from scipy.spatial import ConvexHull