{"generator":"Jekyll","link":[{"@attributes":{"href":"https:\/\/adamkarvonen.github.io\/feed.xml","rel":"self","type":"application\/atom+xml"}},{"@attributes":{"href":"https:\/\/adamkarvonen.github.io\/","rel":"alternate","type":"text\/html"}}],"updated":"2026-06-17T17:58:54+00:00","id":"https:\/\/adamkarvonen.github.io\/feed.xml","title":"Adam Karvonen","subtitle":"Adam Karvonen's personal website","entry":[{"title":"LLM inference is nearly deterministic. We use this to audit providers","link":{"@attributes":{"href":"https:\/\/adamkarvonen.github.io\/machine_learning\/2025\/11\/28\/difr.html","rel":"alternate","type":"text\/html","title":"LLM inference is nearly deterministic. We use this to audit providers"}},"published":"2025-11-28T19:46:58+00:00","updated":"2025-11-28T19:46:58+00:00","id":"https:\/\/adamkarvonen.github.io\/machine_learning\/2025\/11\/28\/difr","content":{"@attributes":{"type":"html"}},"author":{"name":{}},"category":{"@attributes":{"term":"machine_learning"}},"summary":{"@attributes":{"type":"html"}}},{"title":"Frontier AI Models Still Fail at Basic Physical Tasks: A Manufacturing Case Study","link":{"@attributes":{"href":"https:\/\/adamkarvonen.github.io\/machine_learning\/2025\/04\/13\/llm-manufacturing-eval.html","rel":"alternate","type":"text\/html","title":"Frontier AI Models Still Fail at Basic Physical Tasks: A Manufacturing Case Study"}},"published":"2025-04-13T19:46:58+00:00","updated":"2025-04-13T19:46:58+00:00","id":"https:\/\/adamkarvonen.github.io\/machine_learning\/2025\/04\/13\/llm-manufacturing-eval","content":{"@attributes":{"type":"html"}},"author":{"name":{}},"category":{"@attributes":{"term":"machine_learning"}},"summary":{"@attributes":{"type":"html"}}},{"title":"Using an LLM perplexity filter to detect weight exfiltration","link":{"@attributes":{"href":"https:\/\/adamkarvonen.github.io\/machine_learning\/2024\/07\/21\/weight-exfiltration.html","rel":"alternate","type":"text\/html","title":"Using an LLM perplexity filter to detect weight exfiltration"}},"published":"2024-07-21T19:46:58+00:00","updated":"2024-07-21T19:46:58+00:00","id":"https:\/\/adamkarvonen.github.io\/machine_learning\/2024\/07\/21\/weight-exfiltration","content":{"@attributes":{"type":"html"}},"author":{"name":{}},"category":{"@attributes":{"term":"machine_learning"}},"summary":{"@attributes":{"type":"html"}}},{"title":"Evaluating Sparse Autoencoders with Board Games","link":{"@attributes":{"href":"https:\/\/adamkarvonen.github.io\/machine_learning\/2024\/06\/12\/sae-board-game-eval.html","rel":"alternate","type":"text\/html","title":"Evaluating Sparse Autoencoders with Board Games"}},"published":"2024-06-12T19:46:58+00:00","updated":"2024-06-12T19:46:58+00:00","id":"https:\/\/adamkarvonen.github.io\/machine_learning\/2024\/06\/12\/sae-board-game-eval","content":{"@attributes":{"type":"html"}},"author":{"name":{}},"category":{"@attributes":{"term":"machine_learning"}},"summary":{"@attributes":{"type":"html"}}},{"title":"An Intuitive Explanation of Sparse Autoencoders for LLM Interpretability","link":{"@attributes":{"href":"https:\/\/adamkarvonen.github.io\/machine_learning\/2024\/06\/11\/sae-intuitions.html","rel":"alternate","type":"text\/html","title":"An Intuitive Explanation of Sparse Autoencoders for LLM Interpretability"}},"published":"2024-06-11T19:46:58+00:00","updated":"2024-06-11T19:46:58+00:00","id":"https:\/\/adamkarvonen.github.io\/machine_learning\/2024\/06\/11\/sae-intuitions","content":{"@attributes":{"type":"html"}},"author":{"name":{}},"category":{"@attributes":{"term":"machine_learning"}},"summary":{"@attributes":{"type":"html"}}},{"title":"Manipulating Chess-GPT\u2019s World Model","link":{"@attributes":{"href":"https:\/\/adamkarvonen.github.io\/machine_learning\/2024\/03\/20\/chess-gpt-interventions.html","rel":"alternate","type":"text\/html","title":"Manipulating Chess-GPT\u2019s World Model"}},"published":"2024-03-20T19:46:58+00:00","updated":"2024-03-20T19:46:58+00:00","id":"https:\/\/adamkarvonen.github.io\/machine_learning\/2024\/03\/20\/chess-gpt-interventions","content":{"@attributes":{"type":"html"}},"author":{"name":{}},"category":{"@attributes":{"term":"machine_learning"}},"summary":{"@attributes":{"type":"html"}}},{"title":"Chess-GPT\u2019s Internal World Model","link":{"@attributes":{"href":"https:\/\/adamkarvonen.github.io\/machine_learning\/2024\/01\/03\/chess-world-models.html","rel":"alternate","type":"text\/html","title":"Chess-GPT\u2019s Internal World Model"}},"published":"2024-01-03T22:46:58+00:00","updated":"2024-01-03T22:46:58+00:00","id":"https:\/\/adamkarvonen.github.io\/machine_learning\/2024\/01\/03\/chess-world-models","content":{"@attributes":{"type":"html"}},"author":{"name":{}},"category":{"@attributes":{"term":"machine_learning"}},"summary":{"@attributes":{"type":"html"}}}]}