Gurkan Ozkan PRO

GurkanOz

gurkanozkan

AI & ML interests

AI, ML, Agent, LLM, GPU

Recent Activity

published an article about 2 months ago

Two Silent Traps in Agentic LLM Evaluation: Vanishing Tool Calls and Disagreeing Judges (Lessons from a DGX Spark Quantization Study)

published an article about 2 months ago

Your Fine-Tune Survives 4-Bit: Trained Agentic Skills Are More Quantization-Robust Than Knowledge

updated a dataset about 2 months ago

GurkanOz/tool-failure-recovery-trajectories

View all activity

Organizations

None yet

Articles 2

Article

Two Silent Traps in Agentic LLM Evaluation: Vanishing Tool Calls and Disagreeing Judges (Lessons from a DGX Spark Quantization Study)

Article

Your Fine-Tune Survives 4-Bit: Trained Agentic Skills Are More Quantization-Robust Than Knowledge

View all Articles

models 3

GurkanOz/qwen3-8b-tool-recovery-FP8-dynamic

Text Generation • 8B • Updated Jul 3 • 6

GurkanOz/qwen3-8b-tool-recovery-NVFP4

Text Generation • 5B • Updated Jul 3 • 9

GurkanOz/qwen3-8b-tool-recovery

Text Generation • 8B • Updated Jul 3 • 8

datasets 1

GurkanOz/tool-failure-recovery-trajectories

Preview • Updated Jul 3 • 171