debgpt 0.5.93

DebGPT -- Chatting LLM with Debian-Specific Knowledge

% DebGPT(1) | Chatting LLM with Debian-Specific Knowledge % Copyright (C) 2024 Mo Zhou [e-mail]; MIT License.

NAME

DebGPT - Chatting LLM with Debian-Specific Knowledge

"AI" = "Artificial Idiot"

SYNOPSIS

debgpt [CLI-behavior-options] [frontend-options] [prompt-composers] [subcommands [subcommand-args]]

DESCRIPTION

This tool is currently experimental.

Large language models (LLMs) are newly emerged tools, which are capable of handling tasks that traditional software could never achieve, such as writing code based on the specification provided by the user. With this tool, we attempt to experiment and explore the possibility of leveraging LLMs to aid Debian development, in any extent.

Essentially, the idea of this tool is to gather some pieces of Debian-specific knowledge, combine them together in a prompt, and then send them all to the LLM. This tool provides convenient functionality for automatically retrieving information from BTS, buildd, Debian Policy, system manual pages, tldr manuals, Debian Developer References, etc. It also provides convenient wrappers for external tools such as git, where debgpt can automatically generate the git commit message and commit the changes for you.

This tool supports multiple frontends, including OpenAI and ZMQ. The ZMQ frontend/backend are provided in this tool to make it self-contained.

QUICK START

First, install DebGPT from PyPI or Git repository:

pip3 install debgpt
pip3 install git+https://salsa.debian.org/deeplearning-team/debgpt.git

The following commands allows you to try DebGPT without going through any configuration immediately after installation.

# The minimum `configuration` required to make debgpt work
export OPENAI_API_KEY="your-api-key"

# start a chat with LLM
debgpt

# try the most important feature of debgpt
debgpt -Hx policy: -a "what is the latest changes in this policy?"

This depends on the commercial OpenAI API service. If you want to switch to a different service provider or self-hosted LLM backend, please read the following sections and configure the software before use.

CONFIGURATION

Upon fresh installation or not configured at all, running debgpt command will simply print the fresh install instructions. Follow the guide to setup.

By default, the configuration file is placed at $HOME/.debgpt/config.toml. Use debgpt genconfig or debgpt config.toml to generate a config template. System-wide configuration file location is not supported.

The minimum configuration needed for debgpt to work only involves one line: openai_api_key = "your-api-key". You can also export the API key in the environment variable OPENAI_API_KEY.

FRONTENDS

Frontend is a client which communicates with an LLM inference backend. The frontend is responsible for sending the user input to the backend, and receive the response from the backend, while maintaining a history.

The tool currently have the following list of frontend implementations. They are specified through the -F | --frontend argument.

• openai: Connects with a OpenAI API-compatible server. For instance, by specifying --openai_base_url, you can switch to a different service provider than the default OpenAI API server.

• anthropic: Connects with Anthropic service. You need to specify --anthropic_api_key or environt variable ANTHROPIC_API_KEY to use this.

• gemini: Connects with Google's Gemini service. You need to specify --gemini_api_key to use this.

• llamafile: Connects with a llamafile (single-file LLM distribution). See https://github.com/Mozilla-Ocho/llamafile for more information. This frontend is implemented in the OpenAI-API compatible way. Setting up --llamafile_base_url to point to the llamafile service you want to use should be enough.

• ollama: Connects with ollama service instance. See https://github.com/ollama/ollama for more information. We currently implement this frontend in the OpenAI-API compatible way. Make sure to specify --ollama_model to the one being served by the ollama service you point to with --ollama_base_url.

• vllm: Connects with a vllm service instance. See https://docs.vllm.ai/en/latest/ for more information. This is a OpenAI-API compatible self-hosted service.

• zmq: Connects with the built-in ZMQ backend. The ZMQ backend is provided for self-hosted LLM inference server. This implementation is very light weight, and not compatible with the OpenAI API. To use this frontend, you may need to set up a corresponding ZMQ backend.

• dryrun: Fake frontend that does nothing. Instead, we will simply print the generated initial prompt to the screen, so the user can can copy it, and paste into web-based LLMs, including but not limited to ChatGPT (OpenAI), Claude (Anthropic), Bard (google), Gemini (google), HuggingChat (HuggingFace), Perplexity AI, etc. This frontend does not need to connect with any backend.

DISCLAIMER: Unless you connect to a self-hosted LLM Inference backend, we are uncertain how the third-party API servers will handle the data you created. Please refer their corresponding user agreements before adopting one of them. Be aware of such risks, and refrain from sending confidential information such like paid API keys to LLM.

TUTORIAL

The following examples are carefully ordered. You can start from the first example and gradually move to the next one.

1. Basic Usage: Chatting with LLM and CLI Behavior

When no arguments are given, debgpt leads you into a general terminal chatting client with LLM backends. Use debgpt -h to see detailed options.

debgpt

During the interactive chatting mode, you may press / and see a list of available escaped commands that will not be seen as LLM prompt.

• /save <path.txt>: save the last LLM response to the specified file.

• /reset: clear the context. So you can start a new conversation without quiting.

The first prompt can be provided through argument (--ask|-A|-a):

debgpt -A "Who are you? And what can LLM do?"

By specifying the --quit|-Q option, the program will quit after receiving the first response from LLM. For instance, we can let it mimic fortune with temperature 1.0 (--temperature|-T 1.0) for higher randomness:

debgpt -T 1.0 -QA 'Greet with me, and tell me a joke.'

After each session, the chatting history will be saved in ~/.debgpt as a json file in a unique name. The command debgpt replay <file_name> can be used to replay the session in specified file. When <file_name> is not given, debgpt replay will replay the last session.

The program can write the last LLM response to a file through -o <file>, and read question from stdin:

debgpt -Qa 'write a hello world in rakudo for me' -o hello.raku
debgpt -HQ stdin < question.txt | tee result.txt

After gettting familiarized with the fundamental usage and its CLI behavior, we can directly move on to the most important feature of this tool, namely the special prompt composer -- MapReduce.

2. Special MapReduce Prompt Composer for Any Length Context

This MapReduce is a key feature of DebGPT.

Generally, LLMs have a limited context length. If you want to ask a question regarding a very long context, you can split the context into multiple parts, and extract the relevant information from each part. Then, you can ask the LLM to answer the question based on the extracted information.

We have implemented it as a special feature in the debgpt tool. You can use this functionality through the --mapreduce|-x argument. We need the --ask|-A|-a argument to tell LLM what kind of question we want to ask so it can extract the right information. If --ask|-A|-a is not provided, the tool will simply assume that you want to summarize the provided information.

Some usage examples are as follows:

• Load a file and ask a question

debgpt -Hx resume.pdf -A 'Does this person know AI? To what extent?'

• Load a directory and ask a question

debgpt -Hx . -a 'which file implemented mapreduce? how does it work?'
debgpt -Hx . -a 'teach me how to use this software. Is there any hidden functionality that is not written in its readme?'
debgpt -Hx ./debian -A 'how is this package built? how many binary packages will be produced?'

• Load a URL and ask a question

debgpt -Hx 'https://www.debian.org/doc/debian-policy/policy.txt' -A 'what is the purpose of the archive?'

• Load Debian Policy (plain text) and ask a question

debgpt -Hx policy: -A 'what is the changes of the latest version compared to the previous version?'
debgpt -Hx policy: -A 'what package should enter contrib instead of main or non-free?'

• Load Debian Developer Reference (plain text) and ask a question

debgpt -Hx devref: -A 'How can I become a debian developer?'
debgpt -Hx devref: -a 'how does general resolution work?'

• If you don't really bother to read policy: and devref:, or forgot which one is talking about the question in you mind, for instance:

debgpt -H -x policy: -x devref: -a 'which document (and which section) talk about Multi-Arch: ?'

• Load the latest sbuild log file and ask a question

debgpt -Hx sbuild: -A 'why does the build fail? do you have any suggestion?'

• Google search: -x google: will use your prompt as the search query, and answer your question after reading the search results

debgpt -Hx google: -a 'how to start python programming?'

• Google search: -x google:<search_query> gives more control over the search query. Here we let LLM answer the question provided by -a based on the search results of "debian packaging".

debgpt -Hx google:'debian packaging' -a 'how to learn debian packaging?'

The -H argument will skip printing the first prompt generated by debgpt, because it is typically very lengthy, and only useful for debugging and development purpose. To further tweak the mapreduce behavior, you may want to check the --mapreduce_chunksize <int> and --mapreduce_parallelism <int> arguments.

The idea behind this is fairly simple: binary split the gathered information texts until the chunk size is smaller than a pre-defined size, and then pairwise reduce those results using LLM until there is only one chunk left. As a result, this functionality can be very quota-consuming if you are going to deal with long texts. Please keep an eye on your bill when you try this on a paied API service.

3. Standard Prompt Composers for Texts that Fit in Context Window

Prompt Composer is a function that reads the plain text contents from the specified resource, and wrap them as a part of a prompt for the LLM. In the previous section we have seens the special prompt composer MapReduce, which works differently from the standard prompt composers that will be introduced here. Note, the query composers (including special one) can be arbitrarily combined together through command line arguments.

[-f|--file]

The first to introduce is the very general --file|-f query composer, which loads a text file from the specified path.

debgpt -Hf README.md -a 'very briefly teach me how to use this software.'
debgpt -Hf debgpt/policy.py -A 'explain this file'  # --file|-f for small file
debgpt -Hx debgpt/cli.py -a 'explain this file'     # Use --mapreduce|-x if file too large

# Mimicking `licensecheck` command
debgpt -Hf debgpt/frontend.py -A 'Briefly tell me an SPDX identifier of this file.'

The -f|--file argument supports the line range grammar for:

debgpt -Hf pyproject.toml:3-10 -A 'explain it'  # select the [3,10) lines
debgpt -Hf pyproject.toml:-10  -A 'explain it'  # select the [0,10) lines
debgpt -Hf pyproject.toml:3-   -A 'explain it'  # select the [3,end) lines

The rest prompt composers are ordered alphabetically.

[--bts]

Ask LLM to summarize the BTS page for src:pytorch.

debgpt -HQ --bts src:pytorch -A 'Please summarize the above information. Make a table to organize it.'
debgpt -HQ --bts 1056388 -A 'Please summarize the above information.'

[--buildd]

Lookup the build status for package glibc and summarize as a table.

debgpt -HQ --buildd glibc -A 'Please summarize the above information. Make a table to organize it.'

[--cmd]

Being able to pipe the inputs and outputs among different programs is one of the reasons why I love the UNIX philosophy.

For example, we can let debgpt read the command line outputs of apt, and summarize the upgradable packages for us:

debgpt -HQ --cmd 'apt list --upgradable' -A 'Briefly summarize the upgradable packages. You can categorize these packages.' -F openai --openai_model 'gpt-3.5-turbo-16k'

And we can also ask LLM to automatically generate a git commit message for you based on the currently staged changes:

debgpt -HQ --cmd 'git diff --staged' -A 'Briefly describe the change as a git commit message.'

This looks interesting, right? debgpt has a git wrapper that automatically generates the git commit message for the staged contents and commit the message. Just try debgpt git commit --amend to see how it works. This will also be mentioned in the subcommands section.

[--html]

Make the mailing list long story short:

debgpt -H --html 'https://lists.debian.org/debian-project/2023/12/msg00029.html' -A 'Please summarize the above information.',

Explain the differences among voting options:

debgpt -H --html 'https://www.debian.org/vote/2022/vote_003' -A 'Please explain the differences among the above choices.'

[--man, --tldr]

Load the debhelper manpage and ask it to extract a part of it.

debgpt -HQ --man debhelper-compat-upgrade-checklist -A "what's the change between compat 13 and compat 14?"
debgpt -HQ --tldr curl --cmd 'curl -h' -A "download https://localhost/bigfile.iso to /tmp/workspace, in silent mode"

[--pdf]

Load a PDF file and ask a question.

debgpt -H --pdf ./some.pdf -a 'what is this?'

[--policy, --devref]

Load a section of debian policy document, such as section "7.2", and ask a question

debgpt -H --policy 7.2 -A "what is the difference between Depends: and Pre-Depends: ?"
debgpt -H --devref 5.5 -A :summary
    'Please summarize the above information.',

Arbitrary Combination of Prompt Composers

We can add code file and Debian Policy simultaneously. In the following example, we put the debian/control file from the PyTorch package, as well as the Debian Policy section 7.4, and asks the LLM to explain some details:

debgpt -H -f pytorch/debian/control --policy 7.4 -A "Explain what Conflicts+Replaces means in pytorch/debian/control based on the provided policy document"

Similarly, we can also let LLM read the Policy section 4.9.1, and ask it to write some code:

debgpt -H -f pytorch/debian/rules --policy 4.9.1 -A "Implement the support for the 'nocheck' tag based on the example provided in the policy document."

4. External Command Wrapper and Subcommands

Let LLM automatically generate the git commit message, and call git to commit it:

debgpt git commit --amend

If you don't even want to git commit --amend the commited message, just remove --amend from it.

5. Prompt Engineering

An important aspect of using LLMs is prompt engineering. The way you ask a question significantly impacts the quality of the results you will get. Make sure to provide as much information as possible. The following are some references on this topic:

1. OpenAI's Guide https://platform.openai.com/docs/guides/prompt-engineering
2. Chain-of-Thought (CoT): https://arxiv.org/pdf/2205.11916.pdf

6. Frequently Seen Issues

• Context overlength: If the result from query composers is too long, you can switch to the --mapreduce|-x special composer, or switch to a model or backend or service provider that supports longer context.

99. You Name It

The usage of LLM is limited by our imaginations. I am glad to hear from you if you have more good ideas on how we can make LLMs useful for Debian development: https://salsa.debian.org/deeplearning-team/debgpt/-/issues

SELF-CONTAINED BACKEND

Available Backend Implementations

This tool provides one backend implementation: zmq.

• zmq: Only needed when you choose the ZMQ front end for self-hosted LLM inference server.

If you plan to use the openai or dryrun frontends, there is no specific hardware requirement. If you would like to self-host the LLM inference backend (ZMQ backend), powerful hardware is required.

LLM Selections

The concrete hardware requirement depends on the LLM you would like to use. A variety of open-access LLMs can be found here

https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard Generally, when trying to do prompt engineering only, the "instruction-tuned" LLMs and "RL-tuned" (RL is reinforcement learning) LLMs are recommended.

The pretrained (raw) LLMs are not quite useful in this case, as they have not yet gone through instruction tuning, nor reinforcement learning tuning procedure. These pretrained LLMs will more likely generate garbage and not follow your instructions, or simply repeat your instruction. We will only revisit the pretrained LLMs when we plan to start collecting data and fine-tune (e.g., LoRA) a model in the far future.

The following is a list of supported LLMs for self-hosting (this list will be updated when there are new state-of-the-art open-access LLMs available):

• Mistral7B (Mistral-7B-Instruct-v0.2) (default) : This model requires roughly 15GB of disks space to download.

• Mixtral8x7B (Mixtral-8x7B-Instruct-v0.1) : This model is larger yet more powerful than the default LLM. In exchange, it poses even higher hardware requirements. It takes roughly 60~100GB disk space (I forgot this number. Will check later).

Different LLMs will pose different hardware requirements. Please see the "Hardware Requirements" subsection below.

Hardware Requirements

By default, we recommend doing LLM inference in fp16 precision. If the VRAM (such as CUDA memory) is limited, you may also switch to even lower preicisions such as 8bit and 4bit. For pure CPU inference, we only support fp32 precision now.

Note, Multi-GPU inference is supported by the underlying transformers library. If you have multiple GPUs, this memory requirement is roughly divided by your number of GPUs.

Hardware requirements for the Mistral7B LLM:

• Mistral7B + fp16 (cuda): 24GB+ VRAM preferred, but needs a 48GB GPU to run all the demos (some of them have a context as long as 8k). Example: Nvidia RTX A5000, Nvidia RTX 4090.
• Mistral7B + 8bit (cuda): 12GB+ VRAM at minimum, but 24GB+ preferred so you can run all demos.
• Mistral7B + 4bit (cuda): 6GB+ VRAM at minimum but 12GB+ preferred so you can run all demos. Example: Nvidia RTX 4070 (mobile) 8GB.
• Mistral7B + fp32 (cpu): Requires 64GB+ of RAM, but a CPU is 100~400 times slower than a GPU for this workload and thus not recommended.

Hardware requirement for the Mixtral8x7B LLM:

• Mixtral8x7B + fp16 (cuda): 90GB+ VRAM.
• Mixtral8x7B + 8bit (cuda): 45GB+ VRAM.
• Mixtral8x7B + 4bit (cuda): 23GB+ VRAM, but in order to make it work with long context such as 8k tokens, you still need 2x 48GB GPUs in 4bit precision.

See https://huggingface.co/blog/mixtral for more.

Usage of the ZMQ Backend

If you want to run the default LLM with different precisions:

debgpt backend --max_new_tokens=1024 --device cuda --precision fp16
debgpt backend --max_new_tokens=1024 --device cuda --precision bf16
debgpt backend --max_new_tokens=1024 --device cuda --precision 8bit
debgpt backend --max_new_tokens=1024 --device cuda --precision 4bit

The only supported precision on CPU is fp32 (for now). If you want to fall back to CPU computation (very slow):

debgpt backend --max_new_tokens=1024 --device cpu --precision fp32

If you want to run a different LLM, such as Mixtral8x7B than the default Mistral7B:

debgpt backend --max_new_tokens=1024 --device cuda --precision 4bit --llm Mixtral8x7B

The argument --max_new_tokens does not matter much and you can adjust it (it is the maximum length of each llm reply). You can adjust it as wish.

TODO

The following is the current TODO List.Some ideas might be a little bit far away.

1. https://github.com/openai/chatgpt-retrieval-plugin
2. implement --archwiki --gentoowiki --debianwiki --fedorawiki --wikipedia (although the LLM have already read the wikipedia dump many times)
3. analyze udd, ddpo, contributors, nm
4. How can LLM help CPython transition? failing tests, API changes, etc.
5. What else can we do about the Debian patching workflow? adding patch description?
6. find upstream bug that matches debian bug (bug triage)
7. connect with debian codesearch API https://codesearch.debian.net/faq
8. Let LLM imitate Janitor, and possibly do some more complicated things
9. Extend Lintian with LLM for complicated checks?
10. Let LLM do mentoring (lists.debian.org/debian-mentors) e.g., reviewing a .dsc package. This is very difficult given limited context length. Maybe LLMs are not yet smart enough to do this.
11. Apart from the str type, the frontend supports other return types like List or Dict (for advanced usage such as in-context learning) are possible (see debgpt/frontend.py :: ZMQFrontend.query, but those are not explored yet.
12. The current implementation stays at prompt-engineering an existing Chatting LLM with debian-specific documents, like debian-policy, debian developer references, and some man pages. In the future, we may want to explore how we can use larger datasets like Salsa dump, Debian mailing list dump, etc. LoRA or RAG or any new methods are to be investegated with the datasets. Also see follow-ups at https://lists.debian.org/debian-project/2023/12/msg00028.html
13. Should we really train or fine-tune a model? How do we organize the data?

REFERENCES

[1] Access large language models from the command-line : https://github.com/simonw/llm

[2] Turn your task descriptions into precise shell commands : https://github.com/sderev/shellgenius

LICENSE

Copyright (C) 2024 Mo Zhou [e-mail]; MIT/Expat License