mertcobanov commited on
Commit
32fdc8c
·
verified ·
1 Parent(s): d7629d5

Ollaya package for convaiinnovations/laya

Browse files
README.md CHANGED
@@ -24,8 +24,8 @@ ollaya run laya
24
  ## What is in this repository
25
 
26
  This repository holds only the files Ollaya derives, with no weights. Each graph is an ONNX export of the
27
- original model whose weights **reference the author's own `model.safetensors` by byte offset**,
28
- so `ollaya pull` downloads the weights from the upstream repository, unmodified and pinned to a
29
  commit, and verifies their sha256.
30
 
31
  | Tag | Upstream | Files |
@@ -46,6 +46,10 @@ The checks use typed-decisions plus multilingual and edge cases:
46
  - **fp16:** the same decision on 99.1–99.6% of questions. Nearly all of the differences are
47
  near-ties between the top two options.
48
 
 
 
 
 
49
  ## License
50
 
51
  Same as the upstream model (Apache-2.0). Ollaya itself is Apache-2.0.
 
24
  ## What is in this repository
25
 
26
  This repository holds only the files Ollaya derives, with no weights. Each graph is an ONNX export of the
27
+ original model whose weights **reference the authors' own weight files by byte offset**,
28
+ so `ollaya pull` downloads the weights from the upstream repositories, unmodified and pinned to a
29
  commit, and verifies their sha256.
30
 
31
  | Tag | Upstream | Files |
 
46
  - **fp16:** the same decision on 99.1–99.6% of questions. Nearly all of the differences are
47
  near-ties between the top two options.
48
 
49
+ The fp32 graphs (CPU) are opset 23: attention runs as fused `Attention` nodes, which ONNX Runtime's
50
+ CPU provider runs faster than the decomposed ops, so they need ONNX Runtime 1.23 or newer. The fp16
51
+ graphs (GPU) are opset 20.
52
+
53
  ## License
54
 
55
  Same as the upstream model (Apache-2.0). Ollaya itself is Apache-2.0.
en/decision.json CHANGED
@@ -24,5 +24,5 @@
24
  "act_logits"
25
  ],
26
  "min_markers": 2,
27
- "opset": 20
28
  }
 
24
  "act_logits"
25
  ],
26
  "min_markers": 2,
27
+ "opset": 23
28
  }
en/model-fp32.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c54880e1ab520b5f99ef8417ed2a1f7ef19414e90e87681d1497b39147f8f97a
3
- size 3722668
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9dc0bb17f5f7e389cc4b99cdd14f76a1043a2a60ed91daf2358524b80ab35711
3
+ size 2689236
multilingual/decision.json CHANGED
@@ -24,5 +24,5 @@
24
  "act_logits"
25
  ],
26
  "min_markers": 2,
27
- "opset": 20
28
  }
 
24
  "act_logits"
25
  ],
26
  "min_markers": 2,
27
+ "opset": 23
28
  }
multilingual/model-fp32.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e4437087513397063f5d53a904a2404c4bda3cbf4ab30b3c4813678ea9f49e8e
3
- size 2981576
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ace962240e4cf7c69c0e65ea7224aa791777880003950a81347a206f6bfbed45
3
+ size 2159186
typed-decisions/decision.json CHANGED
@@ -24,5 +24,5 @@
24
  "act_logits"
25
  ],
26
  "min_markers": 2,
27
- "opset": 20
28
  }
 
24
  "act_logits"
25
  ],
26
  "min_markers": 2,
27
+ "opset": 23
28
  }
typed-decisions/model-fp32.onnx CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7d648ef513e8d9f1d66a98ccb1d0a71b6902cb776d5332e90b56cea10c1b6e1e
3
- size 3722669
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f1e62a7eabb6bc5eabf04877348e003b053de9a19f56b9bfd642e74f62a2c6b5
3
+ size 2689237