Shanghai AI Laboratory releases 35B agentic model Sage-A1.5
Weights are on Hugging Face for Transformers, vLLM and SGLang; the lab reports SOTA on several agent and science benchmarks.
2
Stories per day · 30 days
Relatedagentic-ai1benchmark1evaluation1llm-agents1mixture-of-experts1open-weights1physics1sage-a1.51
3 stories, newest first · one card per story
Stories tagged #scientific-reasoning
PhysClosureBench finds frontier agents often miss full physics research closure
Claude-Opus-5 scores 6.29/10 on average but reaches full closure in only 32.7% of attempts.
Drift-Constrained Optimization: Only Direction Matters in Fine-Tuning Instruct Models
Drift-Constrained Optimization reformulates fine-tuning as update-direction selection, letting Qwen3 models improve target tasks within a behavioral drift budget.