Sign inGet started
Dataset · Hugging Face

gavinlaw/sglang-omni-fdb-native-runs

SGLang-Omni 原生全双工模型的 Full-Duplex-Bench 运行产物 用 sglang-omni 的全双工评测工具(sgl-project/sglang-omni#2331)对 SGLang-Omni 上的原生全双工模型完整运行 Full-Duplex-Bench(FDB)v1.0 与 v1.5 的录制与评分产物。每次运行一个目录。 20261009-nemotron-voicechat 模型:nvidia/NVIDIA-NemotronLabs-VoiceChat-11B@a4c40ca5b4fe77db13e9840ca4a2b91becf030c8,tokenizer nvidia/NVIDIA-Nemotron-Nano-9B-v2@6533e8de2c68e4536bf7c411d7a3ce5734111476,profile nemotron-voicechat-pr2188。 服务端:sglang-omni main… See the full description on the dataset page: https://huggingface.co/datasets/gavinlaw/sglang-omni-fdb-native-runs.

Author / organizationgavinlaw
License (as declared)cc-by-nc-4.0
Downloads191 (as reported by Hugging Face)
Likes0
Created9 October 2026
Last updated at source16 October 2026
full-duplexspeechbenchmark
Embed this entry
<iframe src="https://silytics.si/embed/item/cfba79af-ef27-405d-9cbd-31d97d32cfb4" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

Popularity, last 2 days

191 now (+10 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.

More from gavinlaw

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.