Today we’re launching SWE-Lancer—a new, more realistic benchmark to evaluate the coding performance of AI models. SWE-Lancer includes over 1,400 freelance software engineering tasks from Upwork, valued at $1 million USD total in real-world payouts.
MACHINE LEARNING
-
Open sourcing ML models and code remains optional but desirable
By
–
No one is obligated to open source their model+code, but it would be refreshing and delightful
-
Request for Recording and Technical Details on Reasoning Model Methodology
By
–
Thanks for the comprehensive summary!! Unfortunately, I wasn't able to tune in to the event but was there a recording or technical blog/paper with info on their reasoning model methodology (re RL, inference-compute scaling strategies etc)?
-
The Future Of Marketing: Smarter Strategies With AI
By
–
The Future Of Marketing: Smarter Strategies With AI
#AI #AIio #BigData #ML #NLU #Futureofwork @lexfridman @sama @kaifulee @ID_AA_Carmack @karpathy @2morrowknight @ylecun http://
ow.ly/P0TT30sK3mS -
Essential Transformers Course and Smolagents Training Resources
By
–
A must-read along with the recent "smolagents" course! Fantastic work by @_lewtun
, @lvwerra and @Thom_Wolf
. It is packed with useful knowledge for working with transformers, even if dating from 2022. Particularly love the focus on data and training everything from the -
Video Generation Models: Diffusion-Based Approach and AI Techniques
By
–
Discover the art of AI video creation in our latest article, "Video Generation Models: A Diffusion-Based Approach."
— Satya Mallick (@LearnOpenCV) 18 février 2025
In the article we have discussed about innovative video generation models like OpenAI's SORA, Google's Veo 2, and Hunyuan Video—exploring how diffusion techniques… pic.twitter.com/1dOCIAVfsDDiscover the art of AI video creation in our latest article, "Video Generation Models: A Diffusion-Based Approach."
In the article we have discussed about innovative video generation models like OpenAI's SORA, Google's Veo 2, and Hunyuan Video—exploring how diffusion techniques -
32B model with 817 examples beats o1-preview on math reasoning
By
–
Less is More for Reasoning (LIMO): a 32B model fine-tuned with 817 examples can beat o1-preview on math reasoning! Do we really need o1's huge RL procedure to see reasoning emerge? It seems not.
Researchers from Shanghai Jiaotong University just demonstrated that carefully -
XAI-Evals Framework for Evaluating Post-Hoc Explanations
By
–
Introducing XAI-Evals: A Framework for Evaluating Post-Hoc Local Explanations!
Our open-source Python package benchmarks methods like SHAP, LIME, Grad-CAM, and more with robust metrics for explainability, faithfulness, and sensitivity. to paper here: https://
aryaxai.com/research-paper
s/xai-evals-a-framework-for-evaluating-post-hoc-local-explanation-methods
… -
Machine Learning for Health 2024 Proceedings Volume 259 Released
By
–
Volume 259 The Proceedings of Machine Learning for Health 2024 https://
proceedings.mlr.press/v259/ Is now available on PMLR. -
2024 Lifelong Learning Conference Proceedings Volume 274 Published
By
–
Volume 274 The Proceedings of the 2024 Conference on Lifelong Learning https://
proceedings.mlr.press/v274/ Is now available on PMLR.