We also noticed a surprising capability in non-coding reasoning, despite the model being trained entirely on code. We benchmarked against models trained for reasoning tasks and replit-code-v1-3b performed incredibly well.
LLMS
-
Replit Open-Sources Complete Code Model 2.7B
By
–
At #ReplitDevDay, we announced we’ve trained and are open-sourcing our first Complete Code model. Introducing replit-code-v1-3b: – 2.7B params
– 20 languages
– 525B tokens
– 40% better than comparable models
– Trained in 10 days Take a look at the benchmarks yourself -
Replit Code Models Outperform Larger Open Source Competitors
By
–
Replit-code-v1-3b & replit-finetuned-v1-3b were trained entirely on code and were meant for single-line code completion. We didn’t expect either to perform so well on HumanEval, but they did. replit-finetuned-v1-3b outperformed all OSS code models, even those 5x its size.
-
Building High-Quality LLMs: Data-Centric AI Approaches
By
–
Building high-quality #LLM is a fairly delicate effort that requires careful planning, systematic benchmarking, and iteration on data, models, & implementation. We are excited to have @togethercompute as a partner at The Future of Data-centric AI conference on Jun 7-8. Join us to
-
AutoTrain Advanced Now Supports LLM Finetuning
By
–
AutoTrain Advanced supports LLM finetuning now! More details come tomorrow! Stay tuned or try without documentation/walkthrough if you dare 😉
-
NeMo Guardrails: Trustworthy LLM Application Development Toolkit
By
–
NeMo Guardrails, a toolkit for developing trustworthy LLM-based conversational applications. It works with all #LLMs and natively supports @langchain
, adding a layer of safety, security, and topical guardrails to existing applications. – @NVIDIAAI https://
nvda.ws/3NfoDrC -
LLMs Boost Productivity of Less Experienced Workers by 35%
By
–
Thanks, Greg! It's great to see the power of LLMs translate into real-world productivity gains. In this case, the less experienced workers got 35% more productive.
-
Master AI-Powered Product Development with Promptimize Test-Driven Engineering
By
–
Master #AI-Powered Product Development with ‘Promptimize’ for Test-Driven #PromptEngineering : https://
preset.io/blog/introduci
ng-promptimize/
… from @preset_data ———
#LLMs #GenerativeAI #DeepLearning #MachineLearning #NLProc #NLU #NLG #ArtificialIntelligence #BigData #DataScience -
Hugging Face and Apache Spark Integration Accelerates LLM Training
By
–
You asked, we delivered It’s now easier than ever to create a #HuggingFace dataset with the cost-effectiveness and performance of #ApacheSpark. Leverage up to 40% faster training and tuning of #LLMs! Learn how to easily get started https://
bit.ly/3NiqFaz
