Cortexelus commited on
Commit
62164ab
·
1 Parent(s): 633bd26

Add ONNX files for per-arch TRT compilation

Browse files

- onnx/t5gemma/encoder.onnx
- onnx/same-s/{enc,dec}_dynamic_bf16.onnx
- onnx/same-l/{enc,dec}_dynamic_triton_swa.onnx
- onnx/sa3-{m,sm-music,sm-sfx}/dit.onnx

opset 17, arch-independent. With these, anyone can compile fresh TRT engines
for a new GPU architecture (sm_100, sm_120, ...) without SAT / source ckpts
— just TensorRT + torch + huggingface-hub.

The sa3-m DiT exceeds protobuf's 2GB limit so its weights live in a single
external dit.onnx.data blob alongside the proto (also LFS-tracked).

.gitattributes CHANGED
@@ -36,3 +36,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
36
  Stable_Audio_3.0_Thumbnail_1x1.png filter=lfs diff=lfs merge=lfs -text
37
  *.trt filter=lfs diff=lfs merge=lfs -text
38
  tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
 
 
36
  Stable_Audio_3.0_Thumbnail_1x1.png filter=lfs diff=lfs merge=lfs -text
37
  *.trt filter=lfs diff=lfs merge=lfs -text
38
  tokenizer.json filter=lfs diff=lfs merge=lfs -text
39
+
40
+ # ONNX external-data blobs (for >2GB models)
41
+ *.onnx.data filter=lfs diff=lfs merge=lfs -text
onnx/sa3-m/dit.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6578a9b07dafcd18593c690ecc2b2b20c18f244b617d04133042bb8e720360e2
3
+ size 3878424
onnx/sa3-m/dit.onnx.data ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:985989c387dbe10cfad9a67fb3af9ffcff4cfaf6a86fa98fb6f07957dcfafcc7
3
+ size 5813473856
onnx/sa3-sm-music/dit.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8708f2acc1f2fc5132eccdb930c599d031ad96ad08eed4db1481af165c583d82
3
+ size 1839539410
onnx/sa3-sm-sfx/dit.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e99a37355d36b7b010ede08688806b69b92d2d16730e8d60950226a91cb15029
3
+ size 1839539410
onnx/same-l/dec_dynamic_triton_swa.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf557e8941dda37b70203316232019031ed3ba32a4a994743ef902e55db0780a
3
+ size 1192471398
onnx/same-l/enc_dynamic_triton_swa.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e4c8d0c95351ba842e29b627c463bf44f244301a69ecf5295bb461b7bcf2570e
3
+ size 1192471389
onnx/same-s/dec_dynamic_bf16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:784227d7682c4bf25ccb5d2b67424c7d417f14219118aed2a74e8da525be6464
3
+ size 218676170
onnx/same-s/enc_dynamic_bf16.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87217e0a068f86ec764b0a7257af62457940881c34b2db10785f74c2c1d558da
3
+ size 215532694
onnx/t5gemma/encoder.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:45bb0d030adfb4a16d1d900ce626bafb2d2af6e4d49b7bb584358eabb448be1f
3
+ size 1126948438