We're happy to be providing the resources for this research, which has resulted in the the best and largest open source CLIP model currently available! ↓
OPEN SOURCE
-
Imagen Outperforms DALL-E2, Stable Diffusion Leads Open Source
By
–
Imagen was released right after DALL-E2, and in their tests with human raters, it outperformed DALL-E2. Stable Diffusion is of course the most useful one as it is open source and people can build on top of it.
-
Flan-T5 Outperforms OPT-IML with Greater Efficiency
By
–
Check out OPT-IML, but a lot of people said they didn't know:
– Flan-T5 checkpoints are publicly available (without requesting access)
– Flan-T5 11B outperforms OPT-IML on MMLU and Big-Bench Hard, despite being 10x more compute efficient Checkpoints: https://
huggingface.co/docs/transform
ers/model_doc/flan-t5
… -
Google and OpenAI: Open Source Data Training Controversy
By
–
Que en concreto aquí se refieren a los datos de entrenamiento usados para los modelos instruct de GPT-3. Pero vamos, que open source no es sólo lo que os interese, colegas de Google
-

Google’s hypocrisy: preaching open source while keeping PaLM closed
By
–
Madre mía, el chiste se cuenta solo… Este párrafo de aquí está recogido en un nuevo paper de Google Research, donde hablan de la importancia del Open Source y poniendo como mal ejemplo a GPT-3. ¡ELLOS! QUE TIENEN A PALM BAJO LLAVE SIENDO SUPERIOR.
-
Documentation Improvements and Bug Fixes Released
By
–
Misc document and Bug Fixes s/o to @ephe_meral @RazRazcle Istora Mandiri, Zach Schillaci, @somejonus
, @eltociear for various documentation improvements and bug fixes! -
PAL Chain Updates: Global Locals and Intermediate Steps
By
–
Some updates to the PAL chain From Nick Furlotte, includes the ability to specify global/locals, and then also return intermediate steps Docs: https://
langchain.readthedocs.io/en/latest/modu
les/chains/examples/pal.html#intermediate-steps
… -

Instruct Embeddings: Task-Tailored Text Embeddings from HuggingFace
By
–
@huggingface Instruct embeddings Instruct embeddings are from an instruction-finetuned embedding model that can generate text embeddings tailored to any task, *just by providing the task description* A joint effort by seanaedmiston and @EnoReyes https://
langchain.readthedocs.io/en/latest/modu
les/utils/combine_docs_examples/embeddings.html#instructembeddings
… -
LangChain 0.0.76 Release: Instruct Embeddings and New Features
By
–
Another big release! 0.0.76 main features: @huggingface Instruct embeddings (seanaedmiston, @EnoReyes
) ngram example selector (
@seanspriggens
) Other features include a new deployment template, easier way to construct LLMChain, and updates to PALChain Lets dive in -

Open Source Vizier: Distributed Experiment Management Platform
By
–
Announcing Open Source Vizier, a standalone Python package designed for managing and optimizing experiments at scale in a reliable distributed manner and for developing and benchmarking algorithms for automated machine learning researchers. Learn more → https://t.co/8kGO9M6jv6 pic.twitter.com/Xvol7qGJ8p
— Google AI (@GoogleAI) 2 février 2023Announcing Open Source Vizier, a standalone Python package designed for managing and optimizing experiments at scale in a reliable distributed manner and for developing and benchmarking algorithms for automated machine learning researchers. Learn more → https://
goo.gle/40CoXoX
