Cache What Lasts: Token Retention for Memory-Bounded KV Cache in LLMs
Paper
•
2512.03324
•
Published
TrimKV
A set of models that can run with bounded memory
-
-
ngocbh/TrimKV-Qwen3-4B-Math
Updated • 25 -
ngocbh/TrimKV-Qwen3-1.7B-Math
Updated • 17 -
ngocbh/TrimKV-Qwen3-4B-Instruct-2507
Updated • 9 -
ngocbh/TrimKV-Phi-3-mini-128k-instruct
Updated • 7 -
ngocbh/TrimKV-Qwen3-8B-Math
Updated • 12 -
ngocbh/TrimKV-Qwen3-14B-Math
Updated • 11 -
ngocbh/TrimKV-DeepSeek-R1-Distill-Llama-8B
Updated • 19