Brihi Joshi
@brihi.bsky.social
NLP PhD @ USC
Everything language, humans, felines and music
(and an infrequent shit-poster)
brihijoshi.github.io
Everything language, humans, felines and music
(and an infrequent shit-poster)
brihijoshi.github.io
📖 So fun to collaborate with amazing folks at USC (Xiang Ren, @swabhs.bsky.social ) and Apple (Rik Koncel-Kedziorski, Tim Paek) on this work! 🤝💡📣
Excited to finally share this—let us know what you think! 🙌🏽✨🎉
More details in our draft here: arxiv.org/abs/2504.17993
(8/n)
Excited to finally share this—let us know what you think! 🙌🏽✨🎉
More details in our draft here: arxiv.org/abs/2504.17993
(8/n)
Improving LLM Personas via Rationalization with Psychological Scaffolds
Language models prompted with a user description or persona can predict a user's preferences and opinions, but existing approaches to building personas -- based solely on a user's demographic attribut...
arxiv.org
April 29, 2025 at 1:05 AM
📖 So fun to collaborate with amazing folks at USC (Xiang Ren, @swabhs.bsky.social ) and Apple (Rik Koncel-Kedziorski, Tim Paek) on this work! 🤝💡📣
Excited to finally share this—let us know what you think! 🙌🏽✨🎉
More details in our draft here: arxiv.org/abs/2504.17993
(8/n)
Excited to finally share this—let us know what you think! 🙌🏽✨🎉
More details in our draft here: arxiv.org/abs/2504.17993
(8/n)
🌎 PB&J improvements do not hinge on any particular group—we see gains across education, race, income, and gender. 🌈📊🤝
(7/n)
(7/n)
April 29, 2025 at 1:05 AM
🌎 PB&J improvements do not hinge on any particular group—we see gains across education, race, income, and gender. 🌈📊🤝
(7/n)
(7/n)
💡 It’s not just about having more reasoning tokens—it’s about what those tokens convey. 🧐🔍✨
We found that *human-written rationales* improved personas the most, but LLM-generated ones came close, showing that good scaffolds are very useful for personas. 🧑💻📈🔬
(6/n)
We found that *human-written rationales* improved personas the most, but LLM-generated ones came close, showing that good scaffolds are very useful for personas. 🧑💻📈🔬
(6/n)
April 29, 2025 at 1:05 AM
💡 It’s not just about having more reasoning tokens—it’s about what those tokens convey. 🧐🔍✨
We found that *human-written rationales* improved personas the most, but LLM-generated ones came close, showing that good scaffolds are very useful for personas. 🧑💻📈🔬
(6/n)
We found that *human-written rationales* improved personas the most, but LLM-generated ones came close, showing that good scaffolds are very useful for personas. 🧑💻📈🔬
(6/n)
📊 We tested PB&J on two tasks: predicting opinions and movie ratings. PB&J consistently outperformed methods based on demographics and/or judgments. Using scaffolds boosted performance even more! 🎥🗳️🏆
(5/n)
(5/n)
April 29, 2025 at 1:05 AM
📊 We tested PB&J on two tasks: predicting opinions and movie ratings. PB&J consistently outperformed methods based on demographics and/or judgments. Using scaffolds boosted performance even more! 🎥🗳️🏆
(5/n)
(5/n)
🔧 We propose psychological scaffolds—frameworks based on life experiences, personality traits, or core beliefs—to further guide the structure of LLM rationales. 🛠️🧬🎨
(4/n)
(4/n)
April 29, 2025 at 1:05 AM
🔧 We propose psychological scaffolds—frameworks based on life experiences, personality traits, or core beliefs—to further guide the structure of LLM rationales. 🛠️🧬🎨
(4/n)
(4/n)
🧠 We introduce PB&J (Psychology of Behaviour and Judgments) that draws from folk psychology—the way people naturally explain each other’s actions. PB&J adds LLM-generated rationales to personas, helping models better represent user behavior. 📚🧩💬
(3/n)
(3/n)
April 29, 2025 at 1:05 AM
🧠 We introduce PB&J (Psychology of Behaviour and Judgments) that draws from folk psychology—the way people naturally explain each other’s actions. PB&J adds LLM-generated rationales to personas, helping models better represent user behavior. 📚🧩💬
(3/n)
(3/n)
🤖 LLMs can mimic human behavior with personas, but most current approaches focus only on demographics or past responses as user context, often overlooking the reasons why users think or act a certain way. 🧐💭🔄
(2/n)
(2/n)
April 29, 2025 at 1:05 AM
🤖 LLMs can mimic human behavior with personas, but most current approaches focus only on demographics or past responses as user context, often overlooking the reasons why users think or act a certain way. 🧐💭🔄
(2/n)
(2/n)
omggg it’s amazing 😭 combined together gave me queen vibes!!
November 21, 2024 at 10:30 PM
omggg it’s amazing 😭 combined together gave me queen vibes!!
omg this can be 3D printed at the makerspace!!
November 17, 2024 at 9:10 PM
omg this can be 3D printed at the makerspace!!
November 17, 2024 at 6:57 PM