A fine-tuned, INT4-quantized LLaMA-3.2-1B model can run fully offline on Android, and human raters preferred it over a non-fine-tuned quantized baseline for empathy and exploration in mental health conversations.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
An Offline Mobile Conversational Agent for Mental Health Support: Learning from Emotional Dialogues and Psychological Texts with Student-Centered Evaluation
A fine-tuned, INT4-quantized LLaMA-3.2-1B model can run fully offline on Android, and human raters preferred it over a non-fine-tuned quantized baseline for empathy and exploration in mental health conversations.