model-trainers of Twitter: why does my model start getting worse over time without the grad norm noticeably increasing — how can this be possible? and how do i fix it? (again, please help 🙂 )
RESEARCH
-
JSONL Dataset Examples with Function Calling Implementation
By
–
That’s the craziest part. I literally copy + pasted 10 JSONL examples (with the JSONL!) from a dataset I had laying around. I made sure that about half of the examples showed some form of function calling.
-
Python ML MLOps CV NLP LLMs Daily Tutorials
By
–
If you are interested in: – Python – Machine Learning – MLOps – CV/NLP – LLMs Find me → @Sumanth_077 Everyday, I share tutorials on above topics! Like/RT the first tweet to help this reach more people!
-
Papers with Code: Over 7000 Public Datasets Available
By
–
5. Papers with Code Papers with Code consist of more than 7000 Public Datasets on different modalities.
-
Awesome Data Github Repository with Public Datasets
By
–
1. Awesome Data Github Repository In this repository, you can find direct links to all the Public datasets, and you can find all sorts of datasets like Images, Text, Audio, and Tabular.
-
Mixtral 8x22B Function Calling Now Officially Working
By
–
Officially have function calling working with base Mixtral 8x22B. This model is insanely powerful, especially with few-shot.
-
Rerank 3 Achieves 2-3x Speed Improvement in Inference Performance
By
–
Rerank 3 is extremely efficient, offering state-of-the-art throughput with a 2-3x improvement in inference speed compared to prior models. We understand that in many business domains, such as customer support, quality results need to be delivered quickly.
-
Rerank 3: Foundation Model for Enterprise Search and RAG
By
–
Introducing Rerank 3: our newest foundation model purpose built to enhance enterprise search and Retrieval Augmented Generation (RAG) systems, enabling accurate retrieval of multi-aspect and semi-structured data in 100+ languages.
-
Meta Kings Cross Event Features AI Leaders Yann LeCun Joelle Pineau
By
–
Amazing day at Meta’s Kings Cross! Got to meet and hear from AI giants like Yann LeCun & Joelle Pineau about the future of tech. The energy, the insights, the people – all top-notch! Thanks, @Meta for a day full of inspiration! #AI #TechInnovation
-
Token Generation Requires Full Model Matrices in Memory
By
–
Not for running models, you need the whole thing in memory because every token that's generated includes calculations run against against the entire collection of matrices