Williams07's picture
Add benchmark evaluation results
8a8ae50 verified
Raw History Blame Contribute Delete
960 Bytes
- dataset:
id: ScaleAI/SWE-bench_Pro
task_id: SWE_Bench_Pro
revision: 2d52cb3df914a3fcf80c7f66738b3a88ae37fc50
value: 77.00
source:
url: https://huggingface.co/Logics-MLLM/Logics-SWE-Qwen3.6-35B-A3B
name: Logics-SWE 35B-A3B model card
notes: "SWE-bench Pro V2, 642 tasks; mean of three rounds; mini-SWE-agent 2.4.6 with patch replay."
- dataset:
id: ScaleAI/SWE-bench_Pro
task_id: SWE_Bench_Pro_hard
value: 33.99
source:
url: https://huggingface.co/Logics-MLLM/Logics-SWE-Qwen3.6-35B-A3B
name: Logics-SWE 35B-A3B model card
notes: "Three-round mean; 51-task Pro V2 subset; mini-SWE-agent 2.4.6 with patch replay."
- dataset:
id: SWE-bench/SWE-bench_Multilingual
task_id: "swe_bench_multilingual_%_resolved"
value: 66.44
source:
url: https://huggingface.co/Logics-MLLM/Logics-SWE-Qwen3.6-35B-A3B
name: Logics-SWE 35B-A3B model card
notes: "Three-round mean; 300 tasks per round; SWE-agent."