From @bindureddy at @abacusai >> New way to fine tune #LLMs for better performance on downstream tasks (e.g, reasoning & summarization) Research paper: https://
arxiv.org/abs/2402.13228 The model: https://
huggingface.co/abacusai/Smaug
-Mixtral-v0.1
… #GenerativeAI #AI #GenAI #ML #DataScience
LLMS
-
New LLM Fine-Tuning Method for Enhanced Downstream Task Performance
By
–
-
OLMo 7B: New Open Source LLM from Allen AI
By
–
A new open source, state-of-the-art #LLM to empower academics is here Databricks Chief Scientist Jonathan Frankle shares how @allen_ai trained OLMo 7B on Databricks #MosaicAI to help advance #AI through open research. Learn more
-
Math Skills Gap: LLMs’ Key Limitation Toward AGI
By
–
Math is one of the only skills LLMs significantly lack compared to humans It’s a loose term, but when AI understands the world as well as any human, most people believe that’s AGI
-
LLM-based SQL Agent for SAP HANA Database Queries
By
–
Discover the power of LLM based SQL Agent for SAP HANA queries Lots of enterprise data is contained in SAP HANA databases. This great guide by Purankhoeval goes over how to connect a LangChain SQL Agent to these types of databases Blog: https://
medium.com/@purankhoeval/
discover-the-power-of-llm-based-sql-agent-for-sap-hana-queries-4c97343f69e2
… -
Q* Model Solves Math Problems as Step Toward AGI
By
–
So what is Q*?
— Rowan Cheung (@rowancheung) 3 mars 2024
According to leaks, all we know is the new model is “able to solve math problems at grade-school level"
Math might not seem impressive to most, but it's a huge step toward creating AGI. pic.twitter.com/Va2cX9agBuSo what is Q*? According to leaks, all we know is the new model is “able to solve math problems at grade-school level" Math might not seem impressive to most, but it's a huge step toward creating AGI.
-
LLMs for Tabular Data: Techniques, Models and Research Directions
By
–
9/ LLMs on Tabular Data – an overview of LLMs for tabular data tasks including key techniques, metrics, datasets, models, and optimization approaches; it covers limitations and unexplored ideas with insights for future research directions.
-
StarCoder 2: Open LLM Family for Code Generation
By
–
8/ StarCoder 2 – a family of open LLMs for code with three different sizes (3B, 7B, and 15B); the 15B model was trained on 14 trillion tokens and 600+ programming languages with a context window of 16K token and employing a fill-in-the-middle objective.
-
Comprehensive Overview and Analysis of LLM Datasets
By
–
4/ Dataset for LLMs – a comprehensive overview (180+ pages) and analysis of LLM datasets.
-
LearnAct: Open-Action Learning Strategy for Language Agents
By
–
5/ LearnAct – explores open-action learning for language agents through an iterative learning strategy that creates and improves actions using Python functions.
-
Mistral Large: Multilingual LLM with Advanced Reasoning and Code
By
–
2/ Mistral Large – a new LLM with strong multilingual, reasoning, maths, and code generation capabilities.