NVFP4 multimodal embedding projection
FP4-quantized multimodal embedding projection that merges text embeddings with vision features. Takes text token embeddings and vision token embeddings, applies FP4-quantized linear projections to align dimensions, and merges them based on vision token mask positions. Uses block-wise FP4 quantization (16 elements per block) for the projection weights.
Reported evidence · last observed 2026-08-14. Reported by source; not independently reproduced. No public source. License unknown.
Current records
Workloadsuite of 16 cases · mean latencyvocab_size = 151936 · hidden_size = 8192 · sf_vec_size = 16 · vision_hidden_size = 3584 · num_vision_scale_blocks = 224 · vision_hidden_size_half = 179216 cases
Not measured on H100 for this workload. Challenges →
192.4µs85% SOL1.00×Reported · license unknown · no source2026-08-14
Measured exactly what you asked. Reported by source; not independently reproduced.
296.8µs84% SOL1.05×Reported · license unknown · no source2026-07-31
Measured exactly what you asked. Reported by source; not independently reproduced.
398.0µs84% SOL1.06×Reported · license unknown · no source2026-07-14
Measured exactly what you asked. Reported by source; not independently reproduced.
4103.8µs83% SOL1.12×Reported · license unknown · no source2026-08-10
Measured exactly what you asked. Reported by source; not independently reproduced.
5104.1µs83% SOL1.13×Reported · license unknown · no source2026-08-23
Measured exactly what you asked. Reported by source; not independently reproduced.
6106.4µs82% SOL1.15×Reported · license unknown · no source2026-08-03
Measured exactly what you asked. Reported by source; not independently reproduced.
7106.4µs82% SOL1.15×Reported · license unknown · no source2026-08-02
Measured exactly what you asked. Reported by source; not independently reproduced.
8106.5µs82% SOL1.15×Reported · license unknown · no source2026-07-18
Measured exactly what you asked. Reported by source; not independently reproduced.
9109.9µs82% SOL1.19×Reported · license unknown · no source2026-07-21
Measured exactly what you asked. Reported by source; not independently reproduced.
10114.0µs81% SOL1.23×Reported · license unknown · no source2026-07-31
Measured exactly what you asked. Reported by source; not independently reproduced.
11118.1µs81% SOL1.28×Reported · license unknown · no source2026-07-16
Measured exactly what you asked. Reported by source; not independently reproduced.
12128.5µs79% SOL1.39×Reported · license unknown · no source2026-07-14
Measured exactly what you asked. Reported by source; not independently reproduced.
13196.1µs70% SOL2.12×Reported · license unknown · no source2026-07-13
Measured exactly what you asked. Reported by source; not independently reproduced.
14290.2µs61% SOL3.14×Reported · license unknown · no source2026-07-23
Measured exactly what you asked. Reported by source; not independently reproduced.
15291.8µs61% SOL3.16×Reported · license unknown · no source2026-07-25
Measured exactly what you asked. Reported by source; not independently reproduced.
16439.2µs50% SOL4.75×Reported · license unknown · no source2026-07-22
Measured exactly what you asked. Reported by source; not independently reproduced.