AI Briefing#goodfire
The latest AI and developer news about #goodfire, with the original source and a short summary.
The latest AI and developer news about #goodfire, with the original source and a short summary.
Analysis of Qwen3.6-35B-A3B shows that ablating 92.2% of weak neuron activations causes minimal performance loss, proving neurons are tractable units of analysis.
Goodfire identifies interpretability as the key bottleneck for AI alignment and proposes a roadmap involving activation monitoring and intentional design.
The Goodfire research team has released Activation probe technology that utilizes internal model Activation signals to detect Reward Hacking and reduce monitoring costs.
Goodfire has launched a grant program offering up to $1 million in free Silico access to AI interpretability researchers.
Goodfire unveiled the beta of its Silico platform, which automates AI research experiments.
Using Parameter Decomposition technology, they succeeded in precisely removing only the model's German language ability while preserving its abilities in other languages.
This introduces a technique that analyzes data before training to predict changes in model behavior and optimize datasets.
Goodfire has released an early preview of Silico, an interpretability-based AI model design platform.