←Search results

Search “NVIDIA”

By relevance · Total 62 results Clear search

Tuesday

  1. NVIDIA：Newsroom

    NVIDIA Completes 104-Qubit Fault-Tolerant Quantum Computing Simulation at Jülich Supercomputing Center

    NVIDIA has completed a 104-qubit fault-tolerant quantum computing simulation on the JUPITER supercomputer at the Jülich Supercomputing Center in Germany, calling it the largest complete simulation of a fault-tolerant quantum processor to date. The simulation validated the magic state cultivation method, requiring 90% fewer physical qubits than previous methods, and ran under real noise conditions. The research team first developed the method on the GB200 NVL72 system at NVAQC, then scaled it to 8,192 GPUs on JUPITER, using cuQuantum and CUDA-Q to complete the computation.

    NVIDIA has completed a 104-qubit fault-tolerant quantum computing simulation on the JUPITER supercomputer at the Jülich Supercomputing Center in Germany, calling it the largest complete simulation of a fault-tolerant quantum processor to date. The simulation validated the magic state cultivation method, requiring 90% fewer physical qubits than previous methods, and ran under real noise conditions. The research team first developed the method on the GB200 NVL72 system at NVAQC, then scaled it to

    Matched fieldsTitle、Summary、Body

    #ai-news#Industry news#English text is being prepared.#English text is being prepared.

Thursday

Monday

Tuesday

Wednesday

  1. Hugging Face

    NVIDIA Fine-Tunes the Nemotron Family, Achieving Gold-Medal-Level Results at IOI and IMO

    NVIDIA used the same fine-tuning recipe to turn Nemotron into a specialized model for competitive programming and math olympiad: at IOI 2026, Nemotron-3-Ultra-CC scored 535.4 points (out of 600), and the IMO system scored 30 points (out of 42), exceeding the official gold medal line. The recipe includes a strong base model, domain problems and reasoning trajectory data, SFT and RL post-training, and a generate-evaluate-improve inference loop. The related SFT and RL checkpoints, training datasets, the Nemotron-IMO-Bench benchmark, and the NeMo-Skills inference pipeline have been open-sourced on Hugging Face.

    NVIDIA used the same fine-tuning recipe to turn Nemotron into a specialized model for competitive programming and math olympiad: at IOI 2026, Nemotron-3-Ultra-CC scored 535.4 points (out of 600), and the IMO system scored 30 points (out of 42), exceeding the official gold medal line. The recipe includes a strong base model, domain problems and reasoning trajectory data, SFT and RL post-training, and a generate-evaluate-improve inference loop. The related SFT and RL checkpoints, training datasets

    Matched fieldsTitle、Summary

    #lab#Model releases#English text is being prepared.#Reasoning

Wednesday

Thursday

Friday

Tuesday

Friday

  1. Unite.AI

    Eclypsium demonstrates how compromised hardware can tamper with AI models at inference time

    Eclypsium demonstrated, with NVIDIA's support, that an attacker using a compromised DPU and lax memory isolation can modify runtime model weights during inference, causing the model's answers to change while the model files on disk and hash verification still appear normal. The experiment was conducted in an NVIDIA Grace Hopper environment and depended on conditions such as IOMMU passthrough and lax PCIe isolation, which the researchers explicitly noted are not Linux default settings. The report brings DPUs, GPUs, management controllers and their firmware into the AI security boundary, and notes that model signing, secure boot, runtime isolation and continuous monitoring each answer different questions.

    Eclypsium demonstrated, with NVIDIA's support, that an attacker using a compromised DPU and lax memory isolation can modify runtime model weights during inference, causing the model's answers to change while the model files on disk and hash verification still appear normal. The experiment was conducted in an NVIDIA Grace Hopper environment and depended on conditions such as IOMMU passthrough and lax PCIe isolation, which the researchers explicitly noted are not Linux default settings. The report

    Matched fieldsSummary、Body

    #hot-signal#Safety / alignment#English text is being prepared.#English text is being prepared.

Thursday

Friday

Sunday

Tuesday

Wednesday

Monday

Monday

  1. Story1 reports · 1 facts

    NVIDIA Completes 104-Qubit Fault-Tolerant Quantum Computing Simulation on JUPITER

    NVIDIA announced that it has completed a 104-qubit fault-tolerant quantum computing simulation on the JUPITER supercomputer at the Jülich Supercomputing Centre in Germany, calling it the largest full simulation of a fault-tolerant quantum processor to date. The simulation validated the magic state cultivation method, requiring 90% fewer physical qubits than previous methods and running under realistic noise conditions. The research team first developed the method on the GB200 NVL72 system at NVAQC, then scaled it to 8,192 GPUs on JUPITER, using cuQuantum and CUDA-Q to complete the computation. Currently, the only source for this event is NVIDIA's official announcement, with no subsequent independent developments yet.

    NVIDIA announced that it has completed a 104-qubit fault-tolerant quantum computing simulation on the JUPITER supercomputer at the Jülich Supercomputing Centre in Germany, calling it the largest full simulation of a fault-tolerant quantum processor to date. The simulation validated the magic state cultivation method, requiring 90% fewer physical qubits than previous methods and running under realistic noise conditions. The research team first developed the method on the GB200 NVL72 system at NVA

    Matched fieldsTitle

Monday

  1. FactRelated story

    NVIDIA completes 104-qubit fault-tolerant quantum computing simulation on JUPITER

    NVIDIA announced that it has completed a 104-qubit fault-tolerant quantum computing simulation on the JUPITER supercomputer at the Jülich Supercomputing Centre in Germany, calling it the largest full simulation of a fault-tolerant quantum processor to date. The simulation validated the magic state cultivation method, requiring 90% fewer physical qubits than previous methods and running under realistic noise conditions. The research team first developed the method on the GB200 NVL72 system at NVAQC, then scaled it to 8,192 GPUs on JUPITER, using cuQuantum and CUDA-Q to complete the computation. Currently, the only source for this event is NVIDIA's official announcement, with no subsequent independent developments yet.

    Matched fieldsTitle

    View story:NVIDIA Completes 104-Qubit Fault-Tolerant Quantum Computing Simulation on JUPITER