Genghan
/

sft-deepseek-coder-33b-instruct_GRPO_nki_pure_0921_cluster4

Model card Files Files and versions

RL+SFT with kernels generated by AccelOpt+NKIBench. Paper

Downloads last month: 13

Safetensors

Model size

33B params

Tensor type

BF16

·

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Dataset used to train Genghan/sft-deepseek-coder-33b-instruct_GRPO_nki_pure_0921_cluster4

Paper for Genghan/sft-deepseek-coder-33b-instruct_GRPO_nki_pure_0921_cluster4

AccelOpt: A Self-Improving LLM Agentic System for AI Accelerator Kernel Optimization

Paper • 2511.15915 • Published 7 days ago • 2