mooler0410 / llmspracticalguide Goto Github PK

Thank you for your team's efforts, which have given us a better understanding of the development of LLM. In particular, this image
（https://github.com/Mooler0410/LLMsPracticalGuide/blob/main/imgs/tree.png）is drawn so well, can you please tell me how it was drawn?

Is it allowed to use the image of the LLM evolution tree in my thesis (of course with proper citation)?

In Table 1 from the paper and ./imgs/tree.png, GLM belongs to the decoder-only models, while it is catogorized into encoder-decoder models in the Fig. 1 from the paper and ./imgs/survey-gif-test.gif. What makes it different?

What are the "Fine-tuned Models" in this figure, Dose it means small NLP models or just fine-tuned LLM?

Hi there, I'm in the middle of reading the paper
Noticed the really beautiful tree.png of the llm evolution.

Was wondering, have you drawn it manually or is there something that can be used to make it faster? I wish to make a similar tree for a different project, AutoGPT. Ideally I'd love an editor or even a python package where I can input all the data and the image can be generated, this is especially usefull since it can be interactive and or update when new data comes in

Please let me know if there's anything like that

Thanks a lot @Mooler0410 @trx14
All the best and have a good one!

Most Meta models are not really Open Source as per standard definition (https://opensource.org/osd/) as they don’t allow commercial use like the other open source models in the list.

Dear authors,

    Fig 1 in the paper looks fancy, can I know which tool did you use?

Thanks,
Ray

Given the licensing and current performance of the falcon instruct model here https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard it seems like it might merit inclusion.

Hi, thanks for your excellent survey.

We recently proposed an alignment fine-tuning paradigm to enhance the reasoning ability of large language models.

Here are the details of our work:
Title: Making Large Language Models Better Reasoners with Alignment
Link: https://arxiv.org/pdf/2309.02144.pdf

We kindly request that you consider adding our work to this repository and the survey.

Thank you for your time and consideration. 😊

I've seen you wrote that this project is continuesly curate more models
was wondering if you're also updating the tree image in the readme.
hope you do

PS - any word on #22 ? would very appreciate to learn how to render trees like that, how did you do it?

Hello! Our new work titled ‘Uncovering ChatGPT’s Capabilities in Recommender Systems’ has been released and can be found at https://arxiv.org/pdf/2305.02182.pdf. We have also open-sourced our code and detailed results at https://github.com/rainym00d/LLM4RS. We kindly request that you add it to this repository. Thank you!

I am a bit confused the categorization of LLMs into Encoder only, Decoder only, Encoder-Decoder. Finding a bit hard time to understand what these terms actually mean:

Yann Lecun posted that on twitter: https://twitter.com/ylecun/status/1651762787373428736?lang=en

Can you please shed some light?

Thanks

It's so pretty. by manual?

Nice summary. Could we add the discussions of Llama in the blog mentioned. https://jingfengyang.github.io/gpt

Last box says "real-word"; should be "real world"

(Capitalization is also a bit wonky and inconsistent. For example, "Contexts contain enough knowledge" vs. "Difficult Tasks Requiring Emergent abilities (e.g., reasoning)")

Dear authors,

Our NLP application only use natural language understanding to classify user intent. Is it can be viewed as mimicking human?

Thanks.