Wow, LLaMA-2 is out! Love that it is open source and is open for commercial applications. This version is also tuned with RLHF which suggest it will be good for conversational uses cases. LLaMA-2 was trained on 40% more data than LLaMA-1 and has double context length(4K now).
LLMS
-
Scale AI releases enterprise custom LLM building with Data Engine
By
–
@scale_ai is releasing the easiest way for enterprises to build custom LLMs using Scale's Data Engine. The biggest challenge for fine-tuning is data—diverse prompt-response datasets are Scale's specialty Build your own domain-specific model
-
Open-source and fine-tuning breakthrough excites AI community
By
–
Excited about what this means for the future of open-source and fine-tuning. This is a huge win for everyone in AI!
-
Scale Open-Sources LLM Engine for Self-Hosted Deployment
By
–
We are open-sourcing scale-llm-engine, our library for hosting and fine-tuning open-source LLMs. This can run on your own infra, as well as on Scale's cloud infrastructure. Docs here: https://
scaleapi.github.io/llm-engine/ Github link: -
Scale API releases Llama 2 fine-tuning cookbook in 5 minutes
By
–
We're also releasing a cookbook for fine-tuning Llama 2 in 5 minutes Get started here: https://
scaleapi.github.io/llm-engine/gui
des/fine_tuning/
… Cookbook here: https://
github.com/scaleapi/llm-e
ngine/blob/main/docs/examples/finetuning.ipynb
… -
Scale AI Open-Sources LLM Engine for Llama 2 Fine-Tuning
By
–
With @MetaAI
's the launch of Llama 2—
@scale_ai will also be: open-sourcing scale-llm-engine, our library for hosting and fine-tuning open-source LLMs releasing the fastest way to fine-tune Llama 2 launching Scale Custom LLMs for enterprises Read more in -
Meta’s Llama 2 Statement of Support Released
By
–
Here's the full list: https://
about.fb.com/news/2023/07/l
lama-2-statement-of-support/
… -
Google’s Machine Unlearning: Removing Data from AI Models
By
–
Super interesting work from Google on machine *un*learning. How do you remove data (& its influence) from a model at reasonable cost, without impacting the model's accuracy? It's a growing ask given LLMs' need to remove copyrighted data, private data, etc https://
ai.googleblog.com/2023/06/announ
cing-first-machine-unlearning.html?m=1
… -
Llama Now Available Worldwide for Global Developers
By
–
Llama is now available to people around the world and we can't wait to see what you build.
-
Llama 2 Available on Major Cloud Platforms and HuggingFace
By
–
We're not alone in this belief. We’re collaborating with a broad set of global supporters across tech, academia & policy who believe in our open approach. Llama 2 is available starting today through @Azure & @Windows
, as well as @AWSCloud and @HuggingFace
.