[New Lectures 2025] Stanford CS231N Deep Learning for Computer Vision The new lectures of CS231N are publicly available now. The last public release was 2017, nearly 8 years since then. CS231N covers a range of topics related to deep learning and modern computer vision: – image
@jeande_d
-
CS231N Stanford Course Website and YouTube Lecture Playlist
By
–
CS231N website: https://
cs231n.stanford.edu
Youtube lecture playlist: -
LLMs Learn Well-Formulated Information: Task Formulation Over Technique
By
–
It's incredible that anything we can formulate well, LLMs can learn it. There are thousands of "high-quality information sources that are yet to be llmfied". All along, it seems it has been more about "task formulation & data curation" than specific training techniques. On the
-
Enthusiastic about series, reminder to complete LLM project
By
–
Wow, loving the series!! A reminder to self to finish build llm from scratch 😀
-
CulturalGround Open-Source Multilingual Cultural VQA Dataset Released
By
–
The code used for curating CulturalGround, the largest open-source multilingual cultural VQA dataset, is now publicly available. The proposed pipeline could greatly enhance factual understanding of a wide array of real-world entities in multimodal LLMs, at zero cost. 30M VQA
-
Code Release Coming Soon for CulturalPangea-7B Model
By
–
Thanks for checking out the work, @AbrahamOwos
. We will release the code as soon as possible. In the meantime, you can run model with https://
huggingface.co/neulab/Cultura
lPangea-7B
…. -
CulturalGround: Extracting More Training Data from Cultural Knowledge Bases
By
–
Importantly, CulturalGround is further evidence that we are not out of data yet and we can squeeze out more and more training data from the knowledge bases. By just rethinking where cultural data resides and systematically generating factual questions and answers about the
-
Cross-lingual Model Transfer Analysis and Methodology Performance
By
–
Here we share some interesting analysis we perform on the results. There are many languages and regions than one can train for, so it’s natural to examine where our methodology and model does transfer to cultures and languages we don’t cover in the dataset. We noticed that
-
Overcoming Catastrophic Forgetting in Continual Learning Models
By
–
One critical challenge with continual learning or adding new domain knowledge is catastrophic forgetting, the decline of base model performance with new domain data. Our training methodology overcomes this and not only we progressively improve cultural understanding with more
-
CulturalPangea Excels in Multilingual and English Performance
By
–
CulturalPangea not only achieves good performance on multilingual cultural understanding, but also perform well on English, and overall.