A great (new) guide and overview on securing LLM systems against prompt injection by @nvidia We did a webinar on prompt injection a few months and the main takeaway was more awareness was needed around this. Great to see posts like this doing that
LLMS
-
Replicate Integration with Vercel AI SDK for Llama 2 Chatbot
By
–
Replicate is now supported by the @vercel AI SDK. `npm ai replicate` Here's how to build your own Llama 2 Chatbot, with streaming: https://
sdk.vercel.ai/docs/guides/pr
oviders/replicate
… -
GPT-4 Performance: Reliability and Future Model Improvements
By
–
with gpt4 its fairly doable. a bit hit and miss with other models tho. i expect this to only improve
-
Multimodal LLMs Enhance Medical Data Integration Capabilities
By
–
Medicine is inherently multimodal, involving a range of data such as medical images, clinical notes, lab tests, and more. Today on the Google Research blog, we discuss a spectrum of approaches for bringing multimodal capabilities to LLMs. Learn more → https://
goo.gle/3Ym4pzZ -
Llama 2: Responsible AI Development Guide for Developers
By
–
With Llama 2 we’re continuing to invest in responsible AI efforts, including a new guide to support devs with best practices and considerations for building products powered by large language models in a responsible manner. Download the full guide https://
bit.ly/3YjrNhC -
Optimized LLM Training Framework Reduces Host Memory Overhead
By
–
If you use an optimized #LLM training framework like https://
pbase.ai/3DHqnE5, you can get the host memory overhead back down to a more reasonable 7 * 4 = 28 GiB of host memory even when training on multiple GPUs. -
Loading Pretrained Checkpoints: Multi-GPU Memory Challenge
By
–
Before you even get to multi-GPU training with model parallel frameworks like #Deepspeed, you need to load the pretrained checkpoint into memory. To make matters worse for machines with multiple GPUs, you need to load the checkpoint into host memory once for each GPU in your job!
-
Llama2 7B Model Training Memory Requirements on Multiple GPUs
By
–
Now training your 7B parameter #Llama2 model in float32 with 8 GPUs requires 7 * 4 * 8 = 224 GiB of host memory just to load it onto the GPUs.
-
Main Problem Teams Face When Fine-tuning LLMs: Out of Memory
By
–
What’s the #1 problem teams encounter when #finetuning an #LLM? The dreaded "Out of memory" error. No, not a CUDA OOM, just a regular host out of memory error.
-
Language Models Guide Robots Performing Everyday Tasks
By
–
“Robot, set up table for pasta”. Check out our work using language models to guide robots performing everyday tasks. 👇 https://t.co/aQN130Wozv
— Fei-Fei Li (@drfeifei) 3 août 2023“Robot, set up table for pasta”. Check out our work using language models to guide robots performing everyday tasks.