TIGER-Lab/AceCodeRM-7B

Updated Apr 9 • 137 • 4

Note The state-of-the-art 7B reward model for code generation

TIGER-Lab/AceCodeRM-32B

Updated Apr 9 • 14 • 7

Note The state-of-the-art 32B reward model for code generation

Note The first large-scale coding dataset with an average of 16 test cases per prompt, synthesized by GPT-4o-mini

TIGER-Lab/AceCoder-Qwen2.5-Coder-7B-Base-Rule

Updated Feb 8 • 53 • 1

Note DeepSeek-R1 style RL-tuned model with binary pass rate as the verifiable reward