Goodfire
A research organization specializing in AI interpretability, with research directions including mechanistic interpretability and activation steering. The organization works to understand the internal behavioral mechanisms of neural networks.
Timeline (last 90 days)
Goodfire reported that in an image network, discarding the smallest three-quarters of readings only dropped accuracy from 72% to 71%.
In October 2025, Nick Cammarata published a non-peer-reviewed long article through the interpretability company Goodfire, proposing that the "cat-car neuron" is actually distinguishing "pointy versus round" rather than encoding cats and cars simultaneously.
Goodfire sells interpretability tools, Anthropic invested in Goodfire in 2025, and three Anthropic researchers provided feedback on Cammarata's article.
In a puzzle test on Qwen3-32B, the model gave Santa Claus a 99.7% probability; after turning off 20 neurons auto-labeled "Christmas" or "Santa Claus", the probability only dropped to 76%.
Goodfire's Nick Camerota published the research "Neural Networks Are Already Clean", proposing that neural networks are clean in nature and that the messiness may stem from the observation tools rather than the networks themselves.
Goodfire places more emphasis on tooling and productization, and compared with interpretability research in academia, its researchers' public views inherently carry an industry perspective.
Nick Cammarata is a researcher at Goodfire.
Goodfire is a startup focused on developing AI interpretability tools.
Goodfire researcher Nick Camerata published a long article whose core argument is that the interior of neural networks is not a chaotic superposition but is interpretable.
Base Labs announced an AI safety collaboration with Hugging Face and Goodfire targeting open-source weight models
1 more timeline events
All Facts (11)
In a puzzle test on Qwen3-32B, the model gave Santa Claus a 99.7% probability; after turning off 20 neurons auto-labeled "Christmas" or "Santa Claus", the probability only dropped to 76%.
50%UnverifiedGoodfire reported that in an image network, discarding the smallest three-quarters of readings only dropped accuracy from 72% to 71%.
50%UnverifiedGoodfire sells interpretability tools, Anthropic invested in Goodfire in 2025, and three Anthropic researchers provided feedback on Cammarata's article.
50%UnverifiedGoodfire's Nick Camerota published the research "Neural Networks Are Already Clean", proposing that neural networks are clean in nature and that the messiness may stem from the observation tools rather than the networks themselves.
50%UnverifiedGoodfire places more emphasis on tooling and productization, and compared with interpretability research in academia, its researchers' public views inherently carry an industry perspective.
50%UnverifiedNick Cammarata is a researcher at Goodfire.
50%UnverifiedGoodfire is a startup focused on developing AI interpretability tools.
50%UnverifiedGoodfire researcher Nick Camerata published a long article whose core argument is that the interior of neural networks is not a chaotic superposition but is interpretable.
50%UnverifiedGoodfire focuses on research in the interpretability direction.
50%UnverifiedIn October 2025, Nick Cammarata published a non-peer-reviewed long article through the interpretability company Goodfire, proposing that the "cat-car neuron" is actually distinguishing "pointy versus round" rather than encoding cats and cars simultaneously.
50%UnverifiedBase Labs announced an AI safety collaboration with Hugging Face and Goodfire targeting open-source weight models
60%