Open-source LLMs have taken the world by storm in just a little over 2 months, ever since LLaMA’s weights were made available for anyone to tinker and play with. Just less than 2 weeks after the untrained LLaMA model was released by Meta.

A model’s weights are the values set to each parameter after training the model on a dataset, with the parameters being various factors (such as token size, number of layers) that allow the model to give more complex answers to what’s input by the user.

This led to a flurry of advancements from dedicated open-source community members. Through just the use of their personal hardware, they were able to make leaps and bounds in their quest to place the most powerful AI in the hands of everyday people.

Recommended: A Quick and Dirty Dip Into Cutting-Edge Open-Source LLM Research

OpenAI’s leadership seems to have taken quite a notice of these events, because they seem to be planning to release an open-source LLM, according to a report by Reuters. It’s virtually unanimous that OpenAI’s GPT-4 is the best-performing LLM model out there. So an open-source model from them would be no small event, even if it is weaker than GPT-4.

Finding out exactly how an OpenAI foundation model is built would give the open-source community a wealth of knowledge that they can apply to their other projects.

It would also go to show how seriously OpenAI views open source and the community surrounding it. It would show that they’re fully aware that the only chance for them to maintain their LLM dominance is if they allowed the world to improve and iterate on their designs.

Open-source showing such swift and definite progress toward taking the crown away from OpenAI can hardly be a surprise. The law of the wisdom of the crowd was foretelling of that. The insight and understanding of the relative few can never match the capability of the collective knowledge and experience of the tens of millions.

The Ultimate Open-Source LLM Battle – Who Wins?

In a chatbot arena site managed by LYMSYS, visitors are asked to enter a prompt, and two randomly-selected models will each provide a response.

The model that the user chooses as having given the best response is then raised up on the leaderboard while the other gets lowered.

The following models are the top three highest-performing models in that arena, just behind GPT-4 (ELO rating of 1274), Anthropic’s Claude (rating of 1224), and GPT-3.5-Turbo (rating of 1155).

Vicuna-13B

Trained by LYMSYS, an open research organization based in UC Berkeley, it is the most promising model from the LLaMA leak.

Recommended: 11 Best ChatGPT Alternatives

It reportedly achieves 90% response quality compared to ChatGPT and Google’s Bard, using a casual evaluation method done through GPT-4. They were able to accomplish this with just a training cost of $300. It has a rating of 1083.

Koala-13B

Coming from BAIR, another group within UC Berkeley, this is a dialogue model meant for academic research. It aims to answer the question of whether open-source models can overcome the massive scale advantage of closed models through better curation of training data. It comes in with a rating of 1022.

RWKV-4-Raven-14B

Impressively, this model was developed by a single person known by the username BlinkDL.

Even more impressively, it’s an RNN LLM (Recurrent Neural Network) rather than the ubiquitous Transformer LLM. The advent of Transformers is what led to the power of GPT-4 being achieved.

People like BlinkDL figuring out ways to optimize more archaic architectures could soon lead to a hybrid architecture that overtakes Transformers in both performance and speed. This model’s rating is a respectable 989.

Civilization-Defining Power Through Artificial General Intelligence

Open-source is a term that can bring out patronizing feelings in people because, after all, a lot of the best programs we know today are closed-source and are chosen by billions of people each year. But that is only due to there being no real reason for the wider community to develop superior open-source alternatives preferred by the wider public.

It’s a much different case with AI.

A few companies holding such immense and civilization-defining power for themselves is not a future that anyone who truly understands the capabilities of AI would want.

Artificial general intelligence is just around the corner, and with it, a complete reimagining of society as we know it. It is a tool that every single person should have equal access to. That reality would bring about a golden age that humanity has never before experienced in all its history.

No matter what anyone says, hoarding any AI knowledge for oneself is a complete disservice to the good of humanity.

Rather than being reserved for the privileged few, a world where AI can be developed and iterated upon by any and all is the only way any sort of utopia can be achieved. Through open-source AI, the dreams and optimism of some of our favorite sci-fi stories will finally be brought to life.

Recommended: MiniGPT-4: The Latest Breakthrough in Language Generation Technology

Imagine you have a project that requires you to process tons of text files, and these files are scattered throughout your folder hierarchy. By the time you finish reading this article, you’ll be equipped with the knowledge to efficiently fetch all the .txt files in any folder using Python.

Method 1: The Os Module

The os module can be used to interact effectively with the file system. The method os.listdir() lists all files and directories in your target folder. You’ll use this method along with a for loop and the endswith() method to filter .txt files specifically.

Here’s the code snippet:

import os directory = './your_folder/'

txt_files = [] for file in os.listdir(directory): if file.endswith('.txt'): txt_files.append(file) print(txt_files)

This code imports the os module, sets the target directory, and initializes an empty list.

The for loop iterates through all the files and checks for the .txt extension using the endswith() method. Matching files are added to the list, which is printed at the end.

Method 2: The Glob Module (My Fav )

Another solution involves using the glob module, which allows you to find all the file paths in a directory that match a specific pattern. You can use the glob.glob() function to list all .txt files.

Here’s how you can do it:

import glob directory = './your_folder/'

txt_files = glob.glob(f'{directory}*.txt') print(txt_files)

This method imports the glob module, sets the target directory, and retrieves the list of text files using the glob.glob() function that filters file paths based on the given pattern (*.txt). The list of .txt files is then printed.

Method 3: os.listdir() and List Comprehension

The os.listdir() is a simple method to use when listing all files in a directory. You can iterate over all files obtain with this method using a simple list comprehension statement such as [file for file in os.listdir(dir_path) if file.endswith(".txt")].

See this example:

import os dir_path = "your_directory_path"

all_files = os.listdir(dir_path)

txt_files = [file for file in all_files if file.endswith(".txt")] print(txt_files)

This code will list all the text files in the specified directory using os.listdir function.

Method 4: Using os.scandir()

The os.scandir() method can provide more information about each file. Extracting the files from this more information-rich representation is a bit less concise but works just fine in this list comprehension [entry.name for entry in os.scandir(dir_path) if entry.name.endswith(".txt") and entry.is_file()].

For instance, use the following code:

import os dir_path = "your_directory_path"

txt_files = [entry.name for entry in os.scandir(dir_path) if entry.name.endswith(".txt") and entry.is_file()] print(txt_files)

Method 5: Using glob.glob()

For a more concise solution, try the glob.glob() function from the glob library. Here’s the code snippet to list text files:

import glob dir_path = "your_directory_path"

txt_files = glob.glob(f"{dir_path}/*.txt") print(txt_files)

The glob.glob() function returns a list of all text files with the specified pattern (in this case, *.txt).

Method 6: Using pathlib.Path.iterdir()

Finally, the pathlib.Path.iterdir method offers another way to list text files in a directory. To use this method, simply import the pathlib library and write the following code:

from pathlib import Path dir_path = Path("your_directory_path")

txt_files = [file.name for file in dir_path.iterdir() if file.is_file() and file.name.endswith(".txt")] print(txt_files)

In this code, pathlib.Path.iterdir is iterator over the files in the directory and, when combined with list comprehensions, can efficiently list all text files.

Iterating Through Directories

In this section, you’ll learn how to iterate through directories using Python and get all the .txt files in a folder.

We’ll cover three methods: using the for loop method, working with the os.walk() function, and recursively traversing directories with a custom recursive function.

Using the For Loop Method

To get started, we’ll use the os.listdir() function with a for loop. This approach allows you to iterate over all files in a directory and filter by their extension.

This code lists all the .txt files in the specified directory using a simple for loop.

import os directory = 'your_directory_path'

for filename in os.listdir(directory): if filename.endswith('.txt'): print(os.path.join(directory, filename))

Working with the os.walk() Function

The os.walk() function is another powerful tool for iterating over files in directories. It enables you to traverse a directory tree and retrieve all files with a specific extension:

import os root_dir = 'your_directory_path'

for root, dirs, files in os.walk(root_dir): for file in files: if file.endswith('.txt'): print(os.path.join(root, file))

This code explores the entire directory tree, including subdirectories, and prints out the full paths of .txt files.

In fact, we have written a detailed article with a video on the function, feel free to check it out!

Recommended: Python os.walk() – A Simple Illustrated Guide

Recursively Traversing Directories with a Recursive Function

Lastly, you could create a custom recursive function to traverse directories and collect .txt files. This method is particularly useful when working with different operating systems, like Windows and Unix:

from pathlib import Path def find_txt_files(path: Path): txt_files = [] for item in path.iterdir(): if item.is_dir(): txt_files.extend(find_txt_files(item)) elif item.name.endswith('.txt'): txt_files.append(item) return txt_files directory = Path('your_directory_path')

txt_files = find_txt_files(directory)

print(txt_files)

This recursive function explores directories and subdirectories and returns a list of .txt files. This approach is more versatile as it leverages Python 3’s pathlib module.

Filtering Based on File Extension and Size

To get all the .txt files in a folder, you can use the glob module in Python, which provides an easy way to find files matching a specific pattern.

Here’s a simple code snippet to get started:

import glob txt_files = glob.glob('path/to/your/folder/*.txt')

print(txt_files)

This code will provide the absolute paths of all the .txt files within the specified folder.

Now that you have the .txt files, you might want to filter them based on their size. To achieve this, you can use the os module.

Here’s an example of how to filter .txt files by size:

import os

import glob min_size = 1000 # Replace with your desired minimum file size in bytes txt_files = glob.glob('path/to/your/folder/*.txt')

filtered_files = [file for file in txt_files if os.path.getsize(file) >= min_size] print(filtered_files)

In this code, min_size represents the minimum file size in bytes that you wish to retrieve. By using a list comprehension with a condition, you can filter out the files that don’t meet your size requirements.

If you want to find .txt files not only in the target folder but also within its subdirectories, you can use the ** pattern along with the recursive parameter:

txt_files = glob.glob('path/to/your/folder/**/*.txt', recursive=True)

Using this approach, you can easily tailor your search to retrieve specific .txt files based on their size and location. With these tools at hand, you should be able to efficiently filter files in your Python projects.

Operating System Compatibility

Python works well across different operating systems, including Unix and Windows. Thanks to its compatibility , you can consistently use your code on different platforms. For this task, both the os and glob libraries are compatible with Unix and Windows systems, so you don’t have to worry about your text file retrieval code failing on either OS.

To get all the text files in a folder using Python, you can use the os and glob libraries. This works for all operating systems, i.e., Linux, Windows, Ubuntu, macOS.

Here’s a code snippet to achieve this:

import os

import glob os.chdir("your_directory_path")

txt_files = glob.glob('*.txt')

print(txt_files)

Replace “your_directory_path” with the path of your folder containing the txt files.

Recommended Video

This article explores the power of OpenAI’s GPT-4 as a tool for individual creativity and innovation.

I’ll discuss how GPT-4 catalyzes personal growth, boosts productivity, and streamlines complex tasks. Challenging conventional wisdom, I argue for the benefits of independent work with GPT-4 over corporate structures.

By showcasing practical case studies, I aim to inspire individuals to harness AI-powered innovation.

A small quote will be enough to communicate the idea of inspiring freedom that can be unlocked by large language models (LLMs):

“Whenever I have a new idea, I ask GPT-4 to write the most basic version, I provide feedback, it apologizes, and we iterate until we reach the 1.0 version I have in mind.

I use up my GPT-4 quota (25 entries/3 hours) multiple times a day. With the support of GPT-4, I feel unstoppable.

The overnight surge in productivity is intoxicating, not for making money or starting a business, but for the sheer joy of continuously creating ideas from my mind, which feels like happiness.”

Ke Fang

Understanding GPT-4: The Game-Changer in AI

GPT-4, OpenAI’s newest marvel, is not just a text generator. It’s your creative ally, language translator, coding guru, and productivity dynamo!

This self-learning model, built on the internet’s vast text, crafts human-like content with an eye for context and creativity. It outdoes its predecessor, GPT-3, and peers, excelling in translation, summarization, content generation, and coding aid.

Recommended: 10 High-IQ Things GPT-4 Can Do That GPT-3.5 Can’t

But GPT-4 isn’t just about words—it’s a key to your creative vault, productivity boost, and turning dreams into reality. The following sections explore these, offering a guide on making GPT-4 your catalyst for personal growth and innovation.

Personal Growth and Innovation with GPT-4

How can you use it to scale your productivity by an order of magnitude?

GPT-4 can assist in many creative tasks, from content creation to idea generation and problem-solving. With its ability to generate text coherently and contextually accurately, it can help you draft blog posts, create fictional stories, write code, or even compose poetry.

It’s like having a brainstorming partner that never runs out of ideas.

Moreover, GPT-4 can be your personal tutor, assisting you in learning new skills or enhancing your existing ones.

Want to learn a new programming language? GPT-4 can provide examples, answer your queries, and guide you through the process.

Want to learn a new human language? GPT-4 can be your friend you can talk to learn and gain proficiency. You can ask it any words if you struggle in the beginning.

Want to understand complex scientific concepts? GPT-4 is proficient enough to be able to teach you the basics of almost any field in philosophy, history, economics, and many more.

More than just fostering creativity and knowledge, GPT-4 encourages innovation by enabling you to take on projects beyond your current abilities. It’s like a co-pilot, guiding you through tasks you’ve never done before—be it developing a web application, creating a podcast, or launching a startup.

With GPT-4 by your side, the potential for personal growth and innovation is boundless.

10X Your Productivity with GPT-4

One key advantage of GPT-4 is its ability to handle tasks traditionally requiring specialized knowledge, thus freeing you to focus on other aspects of your work.

For instance, GPT-4 can write interfaces based on documentation and APIs, enabling even non-programmers to create software.

ChatGPT is the ultimate “No Code Tool” – a recent hype term on Twitter.

Similarly, it can generate accurate translations, aiding in quick and efficient product localization.

GPT-4 also shines in automation. With the ability to generate responses, GPT-4 can serve as a personal assistant, automating your emails, calendar appointments, and even customer service responses. That is if you know how to use ChatGPT Plugins.

Moreover, GPT-4 can quickly sift through large volumes of data and summarize key information, aiding in research tasks, market analysis, and decision-making.

By leveraging GPT-4, you can significantly increase your productivity, freeing up time for strategic thinking, creative ideation, and skill enhancement. GPT-4 is not just a tool, but a virtual collaborator driving efficiency and productivity.

Being Alone with GPT-4 Beats a Corporate Team

The conventional wisdom of a structured, corporate environment as the best place for career growth is being challenged by the potential of independent work amplified by AI tools like GPT-4.

Independence with GPT-4 can often prove more fruitful than being part of a large corporation.

The power of GPT-4 lies in its ability to augment individual abilities, acting as a multiplier of your skills and creativity. In a corporate setting, the focus is often on specialized roles, and the potential of a tool like GPT-4 can be diluted by bureaucracy, poor code, or lack of agility.

On the other hand, as an independent worker, the capabilities of GPT-4 can be fully harnessed. It can assist you in diverse tasks, enable rapid prototyping, and facilitate quick iterations of your ideas.

In effect, it can act as your personal team of specialists, available round the clock, without the constraints of a traditional office environment.

Moreover, working independently with GPT-4 promotes lifelong learning, forcing you to continually adapt and grow. It provides the opportunity to step out of your comfort zone, try new things, and even take risks that might be impossible within a corporate structure.

Don’t underestimate the power of being forced to learn and adapt in our rapidly changing times. The last thing you want to do is sleep through the tech developments during the next decade, which will be the most disruptive decade in the history of humanity!

Feel free to join the Finxter email newsletter to stay updated on the most profound trends and tools in AIs and LLMs.

Embracing independence with GPT-4 may not be the path of least resistance, but for those ready to navigate the uncharted waters of AI-driven innovation, it promises exciting opportunities and immense personal growth.

Real-World Examples: Harnessing GPT-4 for Individual Success

It’s one thing to talk about the potential of GPT-4 in abstract terms, but real-world examples truly illustrate its power.

Here are a few instances where individuals have harnessed GPT-4 to achieve remarkable success.

Case Study 1 – Building a Podcast Search Website: One individual with no prior experience in front-end coding was able to build a podcast search website after just 48 hours of conversation with GPT-4. This case study highlights GPT-4’s ability to provide step-by-step instructions, breaking down complex tasks into manageable actions, and facilitating the creation of a fully functional web application.

Case Study 2 – Developing a Chrome Extension: Another example is of an individual who developed a Chrome extension to skip certain timestamps when watching videos on web pages. GPT-4 guided the creation process, and within 15 minutes, a new tool was born. This showcases GPT-4’s proficiency in coding assistance and its potential to speed up development tasks significantly.

Case Study 3 – Launching Multiple Apps: Perhaps one of the most impressive examples is how this single developer, within five months, single-handedly launched five iOS apps. This underlines GPT-4’s capacity as a versatile partner capable of assisting in complex, multifaceted projects.

These case studies demonstrate GPT-4’s potential as an extraordinary tool for individuals. They serve as tangible proof that with the power of GPT-4, individuals can achieve their ideas and dream bigger than ever before.

“Many people may not be aware of the leverage GPT-4 offers to individuals and may not have considered the endless possibilities it enables. What I am doing is simply informing them of these aspects. I hope this is useful to the 0.01% of readers.” — Ke Fang

Where to Go From Here: The Journey Towards Mastery with Finxter Academy

You’ve seen the transformative power of GPT-4 and the immense opportunities it presents. But the question is – how do you acquire the skills to leverage this extraordinary tool effectively?

This is where Finxter Academy comes in:

At Finxter Academy, we’ve tailored our courses to help you harness exponential technologies like Python, machine learning, ChatGPT, and blockchain development. We aim to provide you with the knowledge and practical skills to master these technologies and create your path to success.

Python Courses: Python is the cornerstone of many AI-related projects. It is the language behind GPT-4 and other large language models. Mastering Python can enable you to interact more efficiently with GPT-4, customize its functionalities, and even build AI-powered applications.

Machine Learning Courses: Our courses in machine learning are designed to demystify the principles behind AI models like GPT-4. You’ll gain insights into how these models learn, their underlying algorithms, and how you can use machine learning for your innovative projects.

ChatGPT Courses: Given the central role of GPT-4 in the future of personal growth and innovation, we’ve developed dedicated courses on ChatGPT. These will provide in-depth knowledge on using GPT-4 effectively, enabling you to unlock its full potential.

Blockchain Development Courses: The future is not just AI; it’s a blend of multiple exponential technologies. Our blockchain development courses will empower you to delve into this revolutionary technology, opening up new avenues for innovation.

The potential of GPT-4 for individual growth and innovation is undeniable. By equipping yourself with the right skills at Finxter Academy, you can ensure you are fully prepared to harness this potential. We invite you to embark on this journey of lifelong learning and relentless innovation. Let’s set sail into the vast ocean of AI-powered possibilities together.

Join Finxter Academy today.

MPT-7B is a large language model (LLM) standard developed by MosaicML, for open-source, commercially usable LLMs and a groundbreaking innovation in natural language processing technology.

With nearly 7 billion parameters, MPT-7B offers impressive performance and has been trained on a diverse dataset of 1 trillion tokens, including text and code. As a part of the MosaicPretrainedTransformer (MPT) family, it utilizes a modified transformer architecture, optimized for efficient training and inference, setting a new standard for open-source, commercially usable language models.

MosaicML achieved an impressive feat by training MPT-7B on their platform in just 9.5 days, with zero human intervention, at a cost of around $200,000. This model not only offers unparalleled quality but also mirrors the performance of Meta’s LLaMA-7B while maintaining an open-source status, making it ideal for commercial use.

MPT-7B’s lineup includes various specialized models like MPT-7B-Instruct, MPT-7B-Chat, and MPT-7B-StoryWriter-65k+, each catering to different use cases. By offering powerful performance and extensive functionality, MPT-7B emerges as a leading contender in the global LLM landscape.

MPT-7B Huggingface

MPT-7B is a large language model developed by MosaicML and available on Hugging Face for easy usage. It is designed for efficient training and inference, suitable for commercial use and outperforms other models in various benchmarks.

LLM

As a large language model (LLM), MPT-7B is trained from scratch on 1T tokens of text and code. It utilizes a modified transformer architecture for better efficiency and matches the quality of other LLMs while being open-source.

Comparison to Other LLMs

The MPT-7B is an impressive language learning model (LLM) that demonstrates performance comparable to the LLaMA-7B model, and even outpaces other open-source models ranging from 7B to 20B parameters in terms of standard academic tasks. (source)

Quality evaluations involving a compilation of 11 open-source benchmarks commonly used for in-context learning (ICL), in addition to a self-curated Jeopardy benchmark to test factual accuracy in responses, demonstrate the robust performance of MPT-7B.

Remarkably, zero-shot accuracy comparisons between MPT-7B, LLaMA-7B, and other open-source models revealed that MPT-7B and LLaMA-7B share a similar level of quality across all tasks, with each model earning the highest scores on 6 out of the 12 tasks.

Despite their comparable performance, MPT-7B and LLaMA-7B noticeably surpass other open-source language models, including those with substantially larger parameter counts.

These results, made possible through the MosaicML LLM Foundry’s ICL evaluation framework, are of particular importance as they were achieved under fair and consistent conditions without the use of prompt strings or prompt tuning.

Furthermore, this evaluation suite brings with it an invitation to the community to engage in model evaluations and contribute additional datasets and ICL task types for continued advancements in the evaluation process.

I also find a nice video on the model, check it out right here:

Commercial Use and Licences

MPT-7B is released under the Apache 2.0, CC-By-SA-3.0, and CC-By-SA-4.0 licenses on Huggingface, not GitHub, to my knowledge, making it usable for commercial applications without any restrictions.

1. Apache 2.0: It is an open-source software license that permits users to freely use, modify, and distribute the licensed work, while also providing explicit grant of patent rights from contributors to users.
2. CC-BY-SA-3.0: Creative Commons Attribution-ShareAlike 3.0 is a license that allows for free distribution, remixing, tweaking, and building upon a work, even commercially, as long as the new creation is credited and licensed under the identical terms.
3. CC-BY-SA-4.0: This is an updated version of the Creative Commons Attribution-ShareAlike license that similarly allows anyone to remix, adapt, and build upon a work, even for commercial purposes, provided that they credit the original creation and license their new creations under the identical terms, but with a few enhancements in terms of internationalization and adaptability to new technologies compared to its predecessor.

Chat

The MPT-7B model has a specific version called MPT-7B-Chat that is designed for conversational use cases, making it a great option for building chatbots and virtual assistants.

Here’s another sample chat from the original website:

Storywriter 65K

I was always frustrated with ChatGPTs length limitations. Storywriter 65k is a nice open-source solution to it!

MPT-7B has a StoryWriter variant that focuses on generating coherent and engaging stories. This StoryWriter version is an excellent choice for content generation tasks. The MPT-7B-StoryWriter-65k+ version is designed to handle even longer stories, suitable for applications requiring extended narrative output.

Here’s an example prompt (source):

MPT-7B-Instruct

The Instruct version of MPT-7B is optimized for providing detailed instructions and guidance based on user input, making it a perfect fit for instructional applications and virtual learning.

Context Length

MPT-7B large language models are designed to handle varying context lengths depending on the use case. Longer context lengths allow for better understanding and more accurate responses in conversational scenarios.

Tokens, Meta, and Datasets

MPT-7B utilizes 1T tokens in various data sources such as the Books3 dataset created by EleutherAI and the Evol-Instruct dataset.

Meta-information about MPT-7B, such as its architecture and training methodology, can be found in the documentation.

Datasets used for training MPT-7B include Books3, Alpaca, and Evol-Instruct, which cover different types of text content to create a diverse language model.

(source)

You can check out their great GitHub repository MosaicML Streaming to train your LLMs easily from cloud storage (multi-node, distributed training for large models)!

Access

MPT-7B is easy to access through its Hugging Face implementation, making it straightforward to deploy and integrate into various projects and applications.

Benchmarks

MPT-7B has been benchmarked against several other large language models and matches the performance of LLaMA, as shown above, while being open-source and commercially friendly.

Unfortunately, I didn’t find an independently-researched benchmark that was not provided by their creators MosaicML. More research is definitely needed! If you’re an ML researcher, why not fill this research gap?

Databricks Dolly-15K, Sharegpt-Vicuna, HC3, Anthropic Helpful and Harmless Datasets

MPT-7B is designed to work effectively with various language models and datasets such as Databricks Dolly-15K, Sharegpt-Vicuna, HC3, and Anthropic’s Helpful and Harmless datasets.

Pricing

While there is no direct pricing associated with MPT-7B, users may experience costs associated with infrastructure, compute resources, and deployment depending on their requirements.

Thanks for reading the article! Feel free to join 100,000 coders in my free email newsletter on AI and exponential technologies such as blockchain development and Python!

Also, you can download a fun cheat sheet here:

OpenAI Glossary Cheat Sheet (100% Free PDF Download)

Finally, check out our free cheat sheet on OpenAI terminology, many Finxters have told me they love it!

Recommended: OpenAI Terminology Cheat Sheet (Free Download PDF)