Data Machina #177
Latest on diffusion models, improving document AI, BERTopic, Torchscale, unifying VAEs, OpenELM, Universality of neural nets, Causal Discovery
A Smörgåsbord of Diffusion Models. The number of research, projects, and tools in the diffusion models space keeps growing exponentially. It’s hard to keep up. Here are my notes from the last 10 days.
Stable Diffusion v2 is out. Apparently it improves image quality & resolution by 4x. But it seems now the model has been heavily filtered, censored. You can play around with SD v2 in this demo instance from Replicate.
A few days ago, @1littlecoder published a short video on how to use Stable Diffusion v2.0 on Colab with Gradio UI & Hugginface Diffusers.
@qunash open sourced a cool WebUI for Stable Diffusion v2
A new feature of SD v2 is depth2img, which infers the depth of an input image, and then generates new images using both the text and depth information. Quite impressive. Gabor posted about how to use this for full pose control in generative art.
A team @TencentAI just published a paper in which they describe a new approach to generate high-fidelity videos with arbitrary lengths with diffusion models. See paper, demo & code here.
The AI team @NVIDIA published a new way to generate high-quality 3D models from text input. See: Magic3D: High-Resolution Text-to-3D Content Creation (paper, demo.)
Weilun @USTC, pubslished a Pytorch implementation of: SinDiffusion: Learning a Diffusion Model from a Single Natural Image.
At NeurIP2022, a team @Bytedance will show how they efficiently generate video with diffusion models. Check out: Magic Video - paper, demo
Have a nice week.
10 Link-o-Troned
A Pythonista *Experience*
Scripting aRt
Deep & Other Learning Bits
ResearchDocs
El Robótico
data v-i-s-i-o-n-s
DataEng Wranglings
Great [Data] Expectations: Eliminate Your [Data] Pipeline Debt
SQLGlot: A Py-transpiler for DuckDB, Spark, Snowflake & BigQuery
startups -> radar
ML Datasets & Stuff
Postscript, etc
Tips? Suggestions? Feedback? email Carlos
Curated by @ds_ldn in the middle of the night.