Open source AI is the way forward and today we're sharing a snapshot of how that's going with the adoption and use of Llama models. Read the full update here https://
go.fb.me/e7odag A few highlights
• Llama is approaching 350M downloads on @HuggingFace
. More than 10x
LLMS
-
Llama Models Approaching 350M Downloads on Hugging Face
By
–
-
Grok generates surreal image: Voldemort Sith Lord judge fusion
By
–
Grok “Generate an image as if Voldemort and a Sith Lord had a baby and he became a judge in Brazil” It’s uncanny!
-
Opinion: RLHF not necessary for stop tokens
By
–
You don't need RLHF for stop tokens; GPT‑3 had that long before RLHF shipped, SFT works fine, and here base Llama 3.1 gets it (presumably) just from Llama 2 prompts floating around online. RLHF does a lot but it's not that deep; I doubt it's adding new concepts.
-
India AI Innovation Centre Launches Indigenous Large Multimodal Models
By
–
The pillar IndiaAI Innovation centre of the INDIAai mission is dedicated to developing and deploying indigenous Large Multimodal Models (LMMs) and domain-specific foundational models in critical sectors.
— IndiaAI (@OfficialINDIAai) 29 août 2024
Visit https://t.co/PTPrQmCzor to know more!@GoI_MeitY @abhish18… pic.twitter.com/qSyREig8e7The pillar IndiaAI Innovation centre of the INDIAai mission is dedicated to developing and deploying indigenous Large Multimodal Models (LMMs) and domain-specific foundational models in critical sectors. Visit https://
indiaai.gov.in to know more! @GoI_MeitY @abhish18 -
OpenRouter sending Llama 3.1 instruct prompts
By
–
It's OpenRouter — AFAIK it's just naively sending the base model the same prompt format used for Llama 3.1 instruct. Other than that, that's the whole dialog; no system prompt.
-
Llama 3.1 base — AI categories and topics
By
–
Also see my previous thread demonstrating Llama 3.1 base:
-
Unabridged Llama 3.1 base outputs show psychosis-like spontaneity
By
–
"Hey what's up." — An unabridged, minimally prompted conversation with Llama 3.1 405B base bf16. Responses exhibit naturalistic, emotional prose and psychosis-like spontaneity, both common in under-prompted base Llama and vividly unlike the output of widely used chat LLMs.
-
Claudette now supports async functionality
By
–
Even more good news on Claudette (
@AnthropicAI Claude's good friend!) — it now supports async too 😀 https://
claudette.answer.ai/#async -
LLM Integration with FastHTML Apps on Replit Platform
By
–
A *lot* of folks are asking how to get LLM help with FastHTML apps. @mattyp has provided a really nice walkthru showing how to set up @Replit to use FastHTML's LLM-ready context docs.
-
ReMamba: Enhancing Mamba Architecture for Long-Sequence Modeling
By
–
ReMamba Equip Mamba with Effective Long-Sequence Modeling discuss: https://
huggingface.co/papers/2408.15
496
… While the Mamba architecture demonstrates superior inference efficiency and competitive performance on short-context natural language processing (NLP) tasks, empirical evidence suggests