Model Detail

HiLS-Attention-7B is a large language model with 7B parameters released by tencent. The model is registered under the text-generation pipeline tag on Hugging Face, distributed under the permissive apache-2.0 license.

HiLS-Attention-7B ships with 7B parameters. Total weight footprint is approximately 7.3 GB, which is the relevant figure when planning local-inference VRAM. The apache-2.0 license is permissive, allowing commercial deployment and derivative work without per-seat fees, though attribution requirements still apply.

HiLS-Attention-7B is best fit for general-purpose chat and instruction-following workloads. Treat this as a starting matrix rather than a benchmark verdict — the right deployment usually depends on the specific evaluation suite that mirrors your workload.