{"id":"arxiv:2311.16452","title":"Can Generalist Foundation Models Outcompete Special-Purpose Tuning? Case Study in Medicine","date":"2023-11-28","section":"Retrieval, embeddings, benchmarks","quality":0.058276,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/","source_url":"https://arxiv.org/abs/2311.16452","text_available":true,"authors":["Harsha Nori","Yin Tat Lee","Sheng Zhang","Dean Carignan"],"outgoing":[{"source":"arxiv:2311.16452","target":"arxiv:2203.11171","evidence":"title","id":"arxiv:2203.11171","title":"Self-Consistency Improves Chain of Thought Reasoning in Language Models","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/02a15b098d96a94d463bb470f59a34723abc34e59196f35c539f9796d8f3c3df/"},{"source":"arxiv:2311.16452","target":"arxiv:2304.15004","evidence":"title","id":"arxiv:2304.15004","title":"Are Emergent Abilities of Large Language Models a Mirage?","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/d7cd8c081ad54f2fdaa2211a4b49dc21f6ff2db7282e287546d210ef90e84323/"},{"source":"arxiv:2311.16452","target":"arxiv:2309.16797","evidence":"id","id":"arxiv:2309.16797","title":"Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/daac3bcfcd7f39084d995905e9ef96ae01c5aab7bfe8e565bd7622b4e44088f7/"}],"incoming":[],"text_parts":[{"start":0,"end":15804,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-1.html"},{"start":15804,"end":31082,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-2.html"},{"start":31082,"end":42015,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-3.html"}],"contents":[{"title":"Can Generalist Foundation Models Outcompete Special-Purpose Tuning? Case Study in Medicine","offset":0,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-1.html#text"},{"title":"Abstract","offset":94,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-1.html#text"},{"title":"1 Introduction","offset":2241,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-1.html#text"},{"title":"2 Background","offset":8112,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-1.html#text"},{"title":"2.1 Foundation Models on Medical Challenge Problems","offset":8129,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-1.html#text"},{"title":"2.2 Prompting Strategies","offset":10137,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-1.html#text"},{"title":"3 Experimental Design","offset":13982,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-1.html#text"},{"title":"3.1 Datasets","offset":14220,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-1.html#text"},{"title":"3.2 Sound Testing Methodology","offset":16591,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-2.html#text"},{"title":"4 Power of Prompting: Exploration and Results","offset":18682,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-2.html#text"},{"title":"4.1 Dynamic Few-shot","offset":19020,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-2.html#text"},{"title":"4.2 Self-Generated Chain of Thought","offset":21181,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-2.html#text"},{"title":"4.3 Choice Shuffling Ensemble","offset":22908,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-2.html#text"},{"title":"4.4 Putting it all together: Medprompt","offset":24282,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-2.html#text"},{"title":"5 Results","offset":27783,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-2.html#text"},{"title":"5.1 Performance on Eyes-Off Data","offset":28666,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-2.html#text"},{"title":"5.2 Insights about Medprompt Components via Ablation Studies","offset":29230,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-2.html#text"},{"title":"5.3 Generalization: Cross-Domain Exploration of Medprompt","offset":32129,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-3.html#text"},{"title":"6 Limitations and Risks","offset":34417,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-3.html#text"},{"title":"7 Summary and Conclusions","offset":38475,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-3.html#text"},{"title":"Acknowledgments","offset":41828,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/part-3.html#text"}],"text_code_points":42015,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/8e8303d0328f412b165c8cdfca91debf742ea45ed7865a0d96ee3905e5bba919/source.md"}