wangkevin02/USP
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:8kLicense:mitArchitecture:Transformer0.0K Open Weights Warm

The wangkevin02/USP model is an 8 billion parameter LLaMA-3-base-8B architecture fine-tuned for user simulation. It replicates human-like conversational behavior by emulating diverse user dynamics based on predefined profiles, reconstructing realistic user-LLM dialogues. Utilizing Conditional Supervised Fine-Tuning (SFT) and Reinforcement Learning with Cycle Consistency (RLCC), this model is optimized for generating user utterances in multi-turn dialogues. It supports a maximum context length of 4,096 tokens and is primarily optimized for English language interactions.

Loading preview...