Goodfire (@GoodfireAI) / X

Goodfire

713 posts

Goodfire

@GoodfireAI

Using interpretability to understand, learn from, and design AI.

San Francisco

Joined August 2024

Pinned
Goodfire
@GoodfireAI
9h
> replicate J-space on GLM 5.2 > train a reward model and run RL to reduce hallucinations > show me how this model makes cancer predictions Using our platform Silico is like having a team of AI researchers ready to run experiments like these. Private beta is open now. 🧵 (1/6)
00:00
Goodfire reposted
Sauers
@Sauers_
5h
My thoughts after daily driving Silico: It makes research substantially more joyful and exciting, allowing me to accomplish more and explore a more diverse set of methods and ideas. It's sticky; I don’t want to go back to not using it. The agents have more freedom and agency than
00:46
Goodfire
@GoodfireAI
9h
> replicate J-space on GLM 5.2 > train a reward model and run RL to reduce hallucinations > show me how this model makes cancer predictions Using our platform Silico is like having a team of AI researchers ready to run experiments like these. Private beta is open now. 🧵 (1/6)
Goodfire reposted
Bryan Cheong
@bryancsk
7h
The folks at Pangram used theirs and produced a paper from it for a question I posed and it honestly answered the question better than anything I'd have come up with even after several days' work
Ryan Panwar
@RyanPanwar
7h
Silico is a virtual lab for AI research. It designs experiments, manages GPUs, and presents results in a beautiful way. We believe it’s the future of AI research and want to enable everyone to study and understand the nature of intelligence.
Goodfire reposted
Ryan Panwar
@RyanPanwar
7h
Silico is a virtual lab for AI research. It designs experiments, manages GPUs, and presents results in a beautiful way. We believe it’s the future of AI research and want to enable everyone to study and understand the nature of intelligence.
Goodfire
@GoodfireAI
9h
> replicate J-space on GLM 5.2 > train a reward model and run RL to reduce hallucinations > show me how this model makes cancer predictions Using our platform Silico is like having a team of AI researchers ready to run experiments like these. Private beta is open now. 🧵 (1/6)
00:00
Goodfire reposted
Tom McGrath
@banburismus_
8h
silico is really great! if you do ai research you’ll probably find it a big uplift over cc/codex, especially for interpretability
Goodfire
@GoodfireAI
9h
> replicate J-space on GLM 5.2 > train a reward model and run RL to reduce hallucinations > show me how this model makes cancer predictions Using our platform Silico is like having a team of AI researchers ready to run experiments like these. Private beta is open now. 🧵 (1/6)
00:00
Goodfire
@GoodfireAI
9h
Replying to @GoodfireAI
These are just a few examples of what you can do with Silico. Request access to the private beta here:
Contact
From goodfire.ai
Goodfire
@GoodfireAI
9h
Replying to @GoodfireAI
Here, Silico replicated PICASSO on Midnight-12k in one-shot. PICASSO interprets digital pathology models. It breaks what the model sees into readable concepts, shows which drive its cancer predictions, and simulates how changes to the tissue would alter those predictions. (5/6)
GIF
01:51
Chanwoo Kim
@ChanwooKim_
Jun 28
Pathology foundation models can read diagnoses, mutations, and patient outcomes from a routine H&E slide. But what are they actually 𝙨𝙚𝙚𝙞𝙣𝙜? Introducing 𝐏𝐈𝐂𝐀𝐒𝐒𝐎: a framework to make pathology foundation models interpretable and controllable. PICASSO decomposes
Goodfire
@GoodfireAI
9h
Replying to @GoodfireAI
Silico lets us look inside models to see what they’ve learned. Using BSFs on protein language models, it found - without supervision - subspaces in the model whose activations correlate with known protein structures. (4/6)
Goodfire
@GoodfireAI
9h
Replying to @GoodfireAI
Our team spent months developing RLFR, our method which uses probes on a model's internals as reward signals for RL. Silico reproduced it in 2 days, reducing hallucinations in Qwen3-8B by 37% without capability loss. (3/6)
Goodfire
@GoodfireAI
9h
Replying to @GoodfireAI
Silico replicated J-space on GLM-5.2 overnight. It then extended context to ~256k tokens, replicating the key results on multi-hop question answering. (2/6)
Anthropic
@AnthropicAI
Jul 6
New Anthropic research: A global workspace in language models. Of everything happening in your brain right now, only a tiny fraction is consciously accessible—thoughts you can describe, hold in mind, and reason with. We found a strikingly similar divide inside Claude.
00:00
Goodfire reposted
Eric Ho
@eric_ho
Jul 13
just welcomed chris earls, cornell engineering professor, to the team! i'm particularly excited about his work to unlock scientific creativity in frontier AI with interpretability we've hired several professors at @GoodfireAI because we're investing heavily in foundational
Goodfire reposted
Sauers
@Sauers_
Jul 11
Goodfire's Silico decided to show me this result of how representation of days of the week emerges over pretraining
Goodfire
@GoodfireAI
Jul 10
Can LLMs predict the next World Cup champion? Goodfire partnered with @EternisAI to improve how LLM forecasters use available evidence and manage uncertainty. We found models were overconfident in their predictions – but probes significantly improved calibration. (1/6)
Goodfire
@GoodfireAI
Jul 10
Replying to @GoodfireAI
When post-training models for specific tasks, like forecasting, you need clear visibility into model internals & behavior: e.g. what RL is teaching your model, and how to make the most of available data and compute. Interpretability provides the tools for that. (5/6)
Goodfire
@GoodfireAI
Jul 10
Our probes were trained autonomously by Silico, our platform for agentic research. To learn more and request access, visit:
Build AI models the way you write software
From goodfire.ai
Goodfire
@GoodfireAI
Jul 8
Come visit us at booth B102 at ICML to chat about our latest research (and grab a neural geometry sticker)!