min-dalle VS mesh-transformer-jax

min(DALL·E).

Additionally, it's also open-sourced on GitHub and can be self-hosted, with easy instructions to do so: https://github.com/kuprel/min-dalle

For CPU, even highly-optimized models like mindalle are prohibitively slow.

Maybe you would be interested in this GitHub repo.

These were generated with DALL-E Mega via min-dalle, which is a more advanced version of DALL-E Mini with better visual fidelity (less blurry) but otherwise similar results.

GitHub repo (contains links to Colab notebook and web app at site Replicate[dot]com). The times mentioned in the title don't included setup time.

GPT-J is a LLM case study with two goals: Training a LLM with a data source containing unique material, and using the training frameworkMesh Transformer JAX to achieve a high training efficiency through parallelization. There is no research paper about GPT-J, but on its GitHub pages, the model, different checkpoints, and the complete source code for training is given.

This idea has already been proposed in ViT-22B and GPT-J-6B.

Sure. Here's the repo I used for the fine-tuning: https://github.com/kingoflolz/mesh-transformer-jax. I used 5 epochs, and appart from that I kept the default parameters in the repo.

You can skip to step 4 using something like GPT-J as far as I understand: https://github.com/kingoflolz/mesh-transformer-jax#links

The pretrained model is already available.

The Github repo suggests it's possible you can change the number of checkpoints to make it run on a GPU.

GPT-J, which uses mesh-transformer-jax: https://github.com/kingoflolz/mesh-transformer-jax

They hinted at it in the screenshot, but the goods are linked from the https://6b.eleuther.ai page: https://github.com/kingoflolz/mesh-transformer-jax#gpt-j-6b (Apache 2)