tweetindex
EN

Daniel Johnson

@_ddjohnson

Member of Technical Staff at @TransluceAI. Building tools to study AI systems and their behaviors. He/him.

2 719Followers
986Following
302Posts total
357KViews on collected posts

Latest posts

@_ddjohnson @GoogleDeepMind Is there something similar for @PyTorch ? 788 views · 4 likes · 0 reposts · 0 replies Open on X →
Want to get started? Penzai's documentation (https://t.co/oHd7FKn62M) includes guided tutorials that show how to visualize, analyze, and fine-tune the Gemma models in Colab. Interpreting attention heads: https://t.co/v77rSQaCgD Low-rank finetuning: https://t.co/AtHK1NkMpX 5.5K views · 41 likes · 4 reposts · 1 replies Open on X →
Penzai's goal is to reduce the barrier of entry for research on understanding pretrained neural networks and steering their behaviors, and to make it easier for researchers to quickly try out new ideas. I'm excited to see what the community can do with it! 4.7K views · 43 likes · 2 reposts · 2 replies Open on X →
Penzai integrates seamlessly with @GoogleColab and the JAX ecosystem. It represents models as legible, editable data structures, to help researchers understand and modify them after they are trained. Built with support from @DougalMaclaurin, @dtarlow2, and @hugo_larochelle! http
Screenshots of interactively exploring the Gemma open-weights language model with Penzai, including a visualization of a parameter tensor for an attention head Linear layer (captioned "Interactively explore model structure!"), and a visualization of a LowRankAdapter layer inserted in place of that Linear layer (captioned "Patch activations or insert new layers!").
7K views · 67 likes · 2 reposts · 1 replies Open on X →
Excited to share Penzai, a JAX research toolkit from @GoogleDeepMind for building, editing, and visualizing neural networks! Penzai makes it easy to see model internals and lets you inject custom logic anywhere. Check it out on GitHub: https://t.co/mas2uiMqj9 https://t.co/HaiYlm
0:25
338.9K views · 2K likes · 386 reposts · 36 replies Open on X →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

338.9K19 Apr
7K
4.7K
5.5K
788

Last 5 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.73%19 Apr
0.99%
1.01%
0.83%
0.51%

Reactions — likes, reposts, replies and quotes — divided by views.

What the audience does

Likes52.8%2 132 in total
Reposts9.8%394 in total
Replies1.0%40 in total
Quotes1.6%64 in total
Bookmarks34.9%1 409 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Similar accounts