CJ Carr

cortexelus's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

updated a model 13 days ago

stabilityai/stable-audio-3-optimized

Text-to-Audio • Updated 2 days ago • 10.4k • 31

New activity in stabilityai/stable-audio-3-optimized 15 days ago

sm_120 sa3-m: add the fp8 medium-DiT tier (176 fp8 GEMM + 96 bf16 fused MHA, 1.43-1.95x vs fp16mixed)

#7 opened 15 days ago by

cortexelus

sm_120 SAME-L: AOT (graph-capturable) engines + tensorRT/README.md

#6 opened 15 days ago by

cortexelus

New activity in stabilityai/stable-audio-3-optimized 16 days ago

Fix the bf16 medium DiT: bake RoPE's tables into the graph (clipping 3.112% -> 0.014%)

1

#5 opened 16 days ago by

cortexelus

Rebuild medium DiT fp16-mixed engines (sm_90 + new sm_120): attention now fuses, 4.3x faster

2

#4 opened 16 days ago by

cortexelus

Fix medium DiT FP16-mixed ONNX: bound the RoPE island so attention fuses (4.3x faster, cos 1.0000)

#3 opened 16 days ago by

cortexelus

New activity in stabilityai/stable-audio-3-optimized about 1 month ago

Add MLX base-model DiT weights for LoRA training

#2 opened about 1 month ago by

cortexelus

Add bf16 medium DiT TRT engine (sm_90) — FMHA-fused, medium default

#1 opened about 1 month ago by

cortexelus

updated 5 models about 1 month ago

stabilityai/stable-audio-3-small-music-base

Text-to-Audio • 0.6B • Updated Jul 13 • 3.78k • 18

stabilityai/stable-audio-3-small-sfx-base

Text-to-Audio • 0.6B • Updated Jul 13 • 1.52k • 10

stabilityai/stable-audio-3-medium-base

Text-to-Audio • 2B • Updated Jul 13 • 9.3k • 24

stabilityai/stable-audio-3-optimized

Text-to-Audio • Updated 2 days ago • 10.4k • 31

stabilityai/stable-audio-3-optimized

Text-to-Audio • Updated 2 days ago • 10.4k • 31