Smaug from Abacus Ai is now available in the Nvidia Playground!
LLMS
-
GPT-4 Humor Capabilities and the Challenge of Original Comedy
By
–
Agree the gpt-4 is the closest to true humor. Still, when doing joke prompts I like to add names of comics or other information (“standup, absurd, etc”) to get better results. Funny original humor is still one of the lowest success rate tasks for an llm imo. Tough challenge
-
Exploring Multiple AI Models: Gemini Ultra, Pro 1.5, and Claude-3
By
–
Gemini ultra is my current main driver. Gemini pro 1.5 (closed beta) for longer context windows. I’m just getting started with Claude-3. Some stuff I like right away, some stuff I don’t (but may be able to prompt my way around it)
-
DataRobot and NVIDIA showcase LLM inference monitoring at GTC
By
–
We’re excited to see you at @NVIDIAGTC this month! Don’t miss a hands-on session with practitioners from DataRobot and @nvidia exploring inference and monitoring capabilities designed for #LLMs to build highly-performant, enterprise-grade AI solutions. Stop by booth #1603 to
-
PyTorch Labs GPT-Fast Repository Link Shared
By
–
Repo: https://
github.com/pytorch-labs/g
pt-fast
… -
PyTorch Releases GPT-Fast for Efficient Text Generation
By
–
Pytorch released GPT-fast!⚡️
— Sumanth (@Sumanth_077) 5 mars 2024
This is a simple and efficient implementation of pytorch-native transformer text generation:
Here are some key features:
– Very low latency
– <1000 lines of python
– No dependencies other than PyTorch and sentencepiece
– int8/int4 quantization
-… pic.twitter.com/chk4ms5nf6Pytorch released GPT-fast! This is a simple and efficient implementation of pytorch-native transformer text generation: Here are some key features: – Very low latency
– <1000 lines of python
– No dependencies other than PyTorch and sentencepiece
– int8/int4 quantization
– -
Abacus AI Launches Affordable LLM Fine-Tunes API Matching GPT-4
By
–
.
@AbacusAI introduces "LLM Fine-Tunes" inference API = Unlocks the power of affordable, Open-Source #LLMs — matches #GPT4 performance and is 20X cheaper! Start here: https://
abacus.ai/llmapi
————
#LLMOps #AI #GenerativeAI #MachineLearning #MLOps #GenAI #DeepLearning #DataScience -
Claude 3 release and top AI tools this week
By
–
Top stories in AI today: -Anthropic releases Claude 3 and new benchmarks
-Madonna adopts Runway for visuals
-Create realistic AI images with legible text
-New AI app tackles ear infection diagnosis
-5 new AI tools & 4 new AI jobs Read more: http://
therundown.ai/p/claude-dethr
ones-gpt4
… -
Open Source AIs Gaming Standard LLM Benchmark Tests
By
–
We really need better benchmarks for LLMs. This paper shows that open source AIs can successfully guess the answer to standard multiple choice tests used to measure AI… even if they aren’t given the question! That suggests these tests aren’t that useful https://
arxiv.org/pdf/2402.12483
.pdf
… -
ChatGPT to introduce GPT Scripts for builders
By
–
ChatGPT is planning to add more tools for GPT builders called GPT Scripts. Those are referring to "GPT Intents" and "GPT State". h/t @btibor91