The Smol Training Playbook
📚
2.91k
The secrets to building world-class LLMs
plmsmile
plmsmile
AI & ML interests
None yet
Recent Activity
new activity
10 days ago
microsoft/FrogBoss-32B-2510:Will plan to release the FeatAdd 1.2k SFT dataset?
liked
a Space
3 months ago
HuggingFaceTB/smol-training-playbook
upvoted
an
article
8 months ago
GRPO for GUI Grounding Done Right
Organizations
None yet
Will plan to release the FeatAdd 1.2k SFT dataset?
1
#3 opened 10 days ago
by
plmsmile
liked
a
Space
3 months ago
upvoted
an
article
8 months ago
Article
GRPO for GUI Grounding Done Right
•
36
reacted to
VirtualOasis's
post with 🔥
9 months ago
Post
2647
Automatic Multi-Modal Research Agent
I am thinking of building an Automatic Research Agent that can boost creativity!
Input: Topics or data sources
Processing: Automated deep research
Output: multimodal results (such as reports, videos, audio, diagrams) & multi-platform publishing.
There is a three-stage process
In the initial Stage, output for text-based content in markdown format allows for user review before transformation into various other formats, such as PDF or HTML.
The second stage transforms the output into other modalities, like audio, video, diagrams, and translations into different languages.
The final stage focuses on publishing multi-modal content across multiple platforms like X, GitHub, Hugging Face, YouTube, and podcasts, etc.
I am thinking of building an Automatic Research Agent that can boost creativity!
Input: Topics or data sources
Processing: Automated deep research
Output: multimodal results (such as reports, videos, audio, diagrams) & multi-platform publishing.
There is a three-stage process
In the initial Stage, output for text-based content in markdown format allows for user review before transformation into various other formats, such as PDF or HTML.
The second stage transforms the output into other modalities, like audio, video, diagrams, and translations into different languages.
The final stage focuses on publishing multi-modal content across multiple platforms like X, GitHub, Hugging Face, YouTube, and podcasts, etc.
liked
a
model
9 months ago
reacted to
merterbak's
post with 🔥
9 months ago
Post
3813
OpenAI has released BrowseComp an open source benchmark designed to evaluate the web browsing capabilities of AI agents. This dataset comprising 1,266 questions challenges AI models to navigate the web and uncover complex and obscure information. Crafted by human trainers, the questions are intentionally difficult. (unsolvable by another person in under ten minutes and beyond the reach of existing models like ChatGPT with and without browsing and an early version of OpenAI's Deep Research tool.)
Blog Post: https://openai.com/index/browsecomp/
Paper: https://cdn.openai.com/pdf/5e10f4ab-d6f7-442e-9508-59515c65e35d/browsecomp.pdf
Code in simple eval repo: https://github.com/openai/simple-evals
Blog Post: https://openai.com/index/browsecomp/
Paper: https://cdn.openai.com/pdf/5e10f4ab-d6f7-442e-9508-59515c65e35d/browsecomp.pdf
Code in simple eval repo: https://github.com/openai/simple-evals
reacted to
hesamation's
post with ❤️
9 months ago
Post
10519
Google published a 69-page whitepaper on Prompt Engineering and its best practices, a must-read if you are using LLMs in production:
> zero-shot, one-shot, few-shot
> system prompting
> chain-of-thought (CoT)
> ReAct
LINK: https://www.kaggle.com/whitepaper-prompt-engineering
> code prompting
> best practices
> zero-shot, one-shot, few-shot
> system prompting
> chain-of-thought (CoT)
> ReAct
LINK: https://www.kaggle.com/whitepaper-prompt-engineering
> code prompting
> best practices
shame on you, perplexity
❤️
9
3
#17 opened 11 months ago
by
wibrantwu
liked
a
Space
over 1 year ago
updated
3
collections
over 1 year ago