LLaVA-MoD Making LLaVA Tiny via MoE Knowledge Distillation discuss: https://
huggingface.co/papers/2408.15
881
… We introduce LLaVA-MoD, a novel framework designed to enable the efficient training of small-scale Multimodal Language Models (s-MLLM) by distilling knowledge from large-scale MLLM
OPEN SOURCE
-
LLaVA-MoD: Efficient Small Multimodal Models via Knowledge Distillation
By
–
-
Gemini Extensions: dedicated page and OpenStax resources
By
–
There you can notice that Gemini Extensions now have a dedicated page explaining what these extensions can and cannot do "Pull in trustworthy responses based on Rice University’s OpenStax educational resources"
-
Kotaemon: Open-Source Customizable RAG UI for Document Chatting
By
–
kotaemon
— AK (@_akhaliq) 28 août 2024
github: https://t.co/QDZ7nkXohB
An open-source clean & customizable RAG UI for chatting with your documents. Built with both end users and developers in mind. pic.twitter.com/1boG0q4DMkkotaemon github: https://
github.com/Cinnamon/kotae
mon
… An open-source clean & customizable RAG UI for chatting with your documents. Built with both end users and developers in mind. -
Live Broadcast: Training and Using CogVideoX-5B Quickly
By
–
Live broadcast with @_akhaliq , about how to train and quickly use CogVideoX-5B. https://
x.com/i/broadcasts/1
MYGNMennLXKw
… -
Open-source AI: Efficient video training storage for robotics
By
–
One of the essential next step in open-source ai the results of several months of exploration, benchmarking, collaborations – efficient video training/storage for robotics and video models – finally out
-
Paper, repo, and LLM-generated story game Everchanging Quest
By
–
Read their paper (currently exploding the upvote counter):
https://
huggingface.co/papers/2408.14
837
… Their repo: https://
gamengen.github.io In a similar vein, play Everchanging Quest, with LLM-generated stories
https://
huggingface.co/spaces/Jofthom
as/Everchanging-Quest
… -
Fine-Tuning Small Language Models Outperforms General LLMs
By
–
When General Purpose Models Fail, Fine-Tune with Open Source Join us in SF for a panel and Q&A (& plenty of networking!) on how fine-tuning small language models #SLMs can outperform general #LLMs for task-specific use cases. See you there!
-
FastHTML Framework Size Comparison with FastAPI
By
–
I believe that FastHTML is far smaller than FastAPI. It’s definitely not a big framework.
-
Cerebras Inference Achieves Record Throughput for Llama 3.1 Models
By
–
Verified by @ArtificialAnlys, @CerebrasSystems Inference is capable of serving Llama 3.1 70B at 450 tokens/sec and Llama 3.1 8B at 1,850 tokens/sec! https://t.co/hCb9MmSvOo
— AI at Meta (@AIatMeta) 27 août 2024Verified by @ArtificialAnlys
, @cerebras Inference is capable of serving Llama 3.1 70B at 450 tokens/sec and Llama 3.1 8B at 1,850 tokens/sec! -
CogVideoX-5B: Open Weights Text-to-Video AI Model Released
By
–
CogVideoX-5B, Open weights Text to Video AI model is out
— AK (@_akhaliq) 27 août 2024
Bigger size, better quality, lower cost (runs on just 12GB GPU)
model and demo linked pic.twitter.com/6gVU18J4HaCogVideoX-5B, Open weights Text to AI model is out Bigger size, better quality, lower cost (runs on just 12GB GPU) model and demo linked