Promote latest kernel artifacts to main

.gitattributes

@@ -1,35 +1,3 @@
-*.7z filter=lfs diff=lfs merge=lfs -text
-*.arrow filter=lfs diff=lfs merge=lfs -text
-*.bin filter=lfs diff=lfs merge=lfs -text
-*.bz2 filter=lfs diff=lfs merge=lfs -text
-*.ckpt filter=lfs diff=lfs merge=lfs -text
-*.ftz filter=lfs diff=lfs merge=lfs -text
-*.gz filter=lfs diff=lfs merge=lfs -text
-*.h5 filter=lfs diff=lfs merge=lfs -text
-*.joblib filter=lfs diff=lfs merge=lfs -text
-*.lfs.* filter=lfs diff=lfs merge=lfs -text
-*.mlmodel filter=lfs diff=lfs merge=lfs -text
-*.model filter=lfs diff=lfs merge=lfs -text
-*.msgpack filter=lfs diff=lfs merge=lfs -text
-*.npy filter=lfs diff=lfs merge=lfs -text
-*.npz filter=lfs diff=lfs merge=lfs -text
-*.onnx filter=lfs diff=lfs merge=lfs -text
-*.ot filter=lfs diff=lfs merge=lfs -text
-*.parquet filter=lfs diff=lfs merge=lfs -text
-*.pb filter=lfs diff=lfs merge=lfs -text
-*.pickle filter=lfs diff=lfs merge=lfs -text
-*.pkl filter=lfs diff=lfs merge=lfs -text
-*.pt filter=lfs diff=lfs merge=lfs -text
-*.pth filter=lfs diff=lfs merge=lfs -text
-*.rar filter=lfs diff=lfs merge=lfs -text
-*.safetensors filter=lfs diff=lfs merge=lfs -text
-saved_model/**/* filter=lfs diff=lfs merge=lfs -text
-*.tar.* filter=lfs diff=lfs merge=lfs -text
-*.tar filter=lfs diff=lfs merge=lfs -text
-*.tflite filter=lfs diff=lfs merge=lfs -text
-*.tgz filter=lfs diff=lfs merge=lfs -text
-*.wasm filter=lfs diff=lfs merge=lfs -text
-*.xz filter=lfs diff=lfs merge=lfs -text
-*.zip filter=lfs diff=lfs merge=lfs -text
-*.zst filter=lfs diff=lfs merge=lfs -text
-*tfevents* filter=lfs diff=lfs merge=lfs -text
+*.so filter=lfs diff=lfs merge=lfs -text
+*.pyd filter=lfs diff=lfs merge=lfs -text
+*.dylib filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,9 +0,0 @@
-# flashrt/flashrt-rope-train
-
-This repository is a compatibility mirror for older `kernels` clients
-that resolve repositories through the default Hugging Face model repo API.
-
-Canonical Kernel Hub repo: https://huggingface.co/kernels/flashrt/flashrt-rope-train
-
-Do not edit this mirror by hand. It is generated from the Kernel Hub
-`vN` branches and contains the same `build/**` artifacts.

build/torch211-cxx11-cu128-x86_64-linux/__init__.py

@@ -0,0 +1,26 @@
+"""FlashRT RoPE training reference API."""
+from __future__ import annotations
+import torch
+try:
+    from ._ops import ops
+except Exception:  # source-tree tests before kernel-builder creates _ops.py
+    class _SourceOpsFallback:
+        def _flashrt_training_package_marker(self, x):
+            return x
+    ops = _SourceOpsFallback()
+
+def rotate_half(x: torch.Tensor) -> torch.Tensor:
+    half=x.shape[-1]//2; return torch.cat((-x[...,half:], x[...,:half]), dim=-1)
+def _align(freq: torch.Tensor, x: torch.Tensor, unsqueeze_dim: int) -> torch.Tensor:
+    while freq.dim() < x.dim(): freq = freq.unsqueeze(unsqueeze_dim)
+    return freq
+def apply_rope_train(q: torch.Tensor, k: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor, unsqueeze_dim: int = 1):
+    c_q=_align(cos,q,int(unsqueeze_dim)); s_q=_align(sin,q,int(unsqueeze_dim)); c_k=_align(cos,k,int(unsqueeze_dim)); s_k=_align(sin,k,int(unsqueeze_dim))
+    qd = torch.float64 if q.dtype == torch.float64 else torch.float32
+    kd = torch.float64 if k.dtype == torch.float64 else torch.float32
+    return (q.to(qd)*c_q.to(qd)+rotate_half(q.to(qd))*s_q.to(qd)).to(q.dtype), (k.to(kd)*c_k.to(kd)+rotate_half(k.to(kd))*s_k.to(kd)).to(k.dtype)
+def apply_rope_backward_reference(dq: torch.Tensor, dk: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor, unsqueeze_dim: int = 1):
+    return apply_rope_train(dq, dk, cos, -sin, unsqueeze_dim)
+def backend_marker(x: torch.Tensor) -> torch.Tensor:
+    return ops._flashrt_training_package_marker(x)
+__all__=["rotate_half","apply_rope_train","apply_rope_backward_reference","backend_marker"]

build/torch211-cxx11-cu128-x86_64-linux/_ops.py

@@ -0,0 +1,9 @@
+import torch
+from . import _flashrt_rope_train_cuda_eefd4b8
+ops = torch.ops._flashrt_rope_train_cuda_eefd4b8
+
+def add_op_namespace_prefix(op_name: str):
+    """
+    Prefix op by namespace.
+    """
+    return f"_flashrt_rope_train_cuda_eefd4b8::{op_name}"

build/torch211-cxx11-cu128-x86_64-linux/flashrt_rope_train/__init__.py

@@ -0,0 +1,26 @@
+import ctypes
+import importlib.util
+import sys
+from pathlib import Path
+from types import ModuleType
+
+
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
+
+
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))

build/torch211-cxx11-cu128-x86_64-linux/metadata.json

@@ -0,0 +1,23 @@
+{
+  "name": "flashrt-rope-train",
+  "id": "_flashrt_rope_train_cuda_eefd4b8",
+  "version": 1,
+  "license": "Apache-2.0",
+  "python-depends": [],
+  "backend": {
+    "type": "cuda",
+    "archs": [
+      "10.0",
+      "10.1",
+      "12.0+PTX",
+      "7.0",
+      "7.2",
+      "7.5",
+      "8.0",
+      "8.6",
+      "8.7",
+      "8.9",
+      "9.0"
+    ]
+  }
+}

build/torch211-cxx11-cu130-x86_64-linux/__init__.py

@@ -0,0 +1,26 @@
+"""FlashRT RoPE training reference API."""
+from __future__ import annotations
+import torch
+try:
+    from ._ops import ops
+except Exception:  # source-tree tests before kernel-builder creates _ops.py
+    class _SourceOpsFallback:
+        def _flashrt_training_package_marker(self, x):
+            return x
+    ops = _SourceOpsFallback()
+
+def rotate_half(x: torch.Tensor) -> torch.Tensor:
+    half=x.shape[-1]//2; return torch.cat((-x[...,half:], x[...,:half]), dim=-1)
+def _align(freq: torch.Tensor, x: torch.Tensor, unsqueeze_dim: int) -> torch.Tensor:
+    while freq.dim() < x.dim(): freq = freq.unsqueeze(unsqueeze_dim)
+    return freq
+def apply_rope_train(q: torch.Tensor, k: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor, unsqueeze_dim: int = 1):
+    c_q=_align(cos,q,int(unsqueeze_dim)); s_q=_align(sin,q,int(unsqueeze_dim)); c_k=_align(cos,k,int(unsqueeze_dim)); s_k=_align(sin,k,int(unsqueeze_dim))
+    qd = torch.float64 if q.dtype == torch.float64 else torch.float32
+    kd = torch.float64 if k.dtype == torch.float64 else torch.float32
+    return (q.to(qd)*c_q.to(qd)+rotate_half(q.to(qd))*s_q.to(qd)).to(q.dtype), (k.to(kd)*c_k.to(kd)+rotate_half(k.to(kd))*s_k.to(kd)).to(k.dtype)
+def apply_rope_backward_reference(dq: torch.Tensor, dk: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor, unsqueeze_dim: int = 1):
+    return apply_rope_train(dq, dk, cos, -sin, unsqueeze_dim)
+def backend_marker(x: torch.Tensor) -> torch.Tensor:
+    return ops._flashrt_training_package_marker(x)
+__all__=["rotate_half","apply_rope_train","apply_rope_backward_reference","backend_marker"]

build/torch211-cxx11-cu130-x86_64-linux/_ops.py

@@ -0,0 +1,9 @@
+import torch
+from . import _flashrt_rope_train_cuda_eefd4b8
+ops = torch.ops._flashrt_rope_train_cuda_eefd4b8
+
+def add_op_namespace_prefix(op_name: str):
+    """
+    Prefix op by namespace.
+    """
+    return f"_flashrt_rope_train_cuda_eefd4b8::{op_name}"

build/torch211-cxx11-cu130-x86_64-linux/flashrt_rope_train/__init__.py

@@ -0,0 +1,26 @@
+import ctypes
+import importlib.util
+import sys
+from pathlib import Path
+from types import ModuleType
+
+
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
+
+
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))

build/torch211-cxx11-cu130-x86_64-linux/metadata.json

@@ -0,0 +1,22 @@
+{
+  "name": "flashrt-rope-train",
+  "id": "_flashrt_rope_train_cuda_eefd4b8",
+  "version": 1,
+  "license": "Apache-2.0",
+  "python-depends": [],
+  "backend": {
+    "type": "cuda",
+    "archs": [
+      "10.0",
+      "11.0",
+      "12.0",
+      "12.1+PTX",
+      "7.5",
+      "8.0",
+      "8.6",
+      "8.7",
+      "8.9",
+      "9.0"
+    ]
+  }
+}

build/torch212-cxx11-cu130-x86_64-linux/__init__.py

@@ -0,0 +1,26 @@
+"""FlashRT RoPE training reference API."""
+from __future__ import annotations
+import torch
+try:
+    from ._ops import ops
+except Exception:  # source-tree tests before kernel-builder creates _ops.py
+    class _SourceOpsFallback:
+        def _flashrt_training_package_marker(self, x):
+            return x
+    ops = _SourceOpsFallback()
+
+def rotate_half(x: torch.Tensor) -> torch.Tensor:
+    half=x.shape[-1]//2; return torch.cat((-x[...,half:], x[...,:half]), dim=-1)
+def _align(freq: torch.Tensor, x: torch.Tensor, unsqueeze_dim: int) -> torch.Tensor:
+    while freq.dim() < x.dim(): freq = freq.unsqueeze(unsqueeze_dim)
+    return freq
+def apply_rope_train(q: torch.Tensor, k: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor, unsqueeze_dim: int = 1):
+    c_q=_align(cos,q,int(unsqueeze_dim)); s_q=_align(sin,q,int(unsqueeze_dim)); c_k=_align(cos,k,int(unsqueeze_dim)); s_k=_align(sin,k,int(unsqueeze_dim))
+    qd = torch.float64 if q.dtype == torch.float64 else torch.float32
+    kd = torch.float64 if k.dtype == torch.float64 else torch.float32
+    return (q.to(qd)*c_q.to(qd)+rotate_half(q.to(qd))*s_q.to(qd)).to(q.dtype), (k.to(kd)*c_k.to(kd)+rotate_half(k.to(kd))*s_k.to(kd)).to(k.dtype)
+def apply_rope_backward_reference(dq: torch.Tensor, dk: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor, unsqueeze_dim: int = 1):
+    return apply_rope_train(dq, dk, cos, -sin, unsqueeze_dim)
+def backend_marker(x: torch.Tensor) -> torch.Tensor:
+    return ops._flashrt_training_package_marker(x)
+__all__=["rotate_half","apply_rope_train","apply_rope_backward_reference","backend_marker"]

build/torch212-cxx11-cu130-x86_64-linux/_ops.py

@@ -0,0 +1,9 @@
+import torch
+from . import _flashrt_rope_train_cuda_eefd4b8
+ops = torch.ops._flashrt_rope_train_cuda_eefd4b8
+
+def add_op_namespace_prefix(op_name: str):
+    """
+    Prefix op by namespace.
+    """
+    return f"_flashrt_rope_train_cuda_eefd4b8::{op_name}"

build/torch212-cxx11-cu130-x86_64-linux/flashrt_rope_train/__init__.py

@@ -0,0 +1,26 @@
+import ctypes
+import importlib.util
+import sys
+from pathlib import Path
+from types import ModuleType
+
+
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
+
+
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))

build/torch212-cxx11-cu130-x86_64-linux/metadata.json

@@ -0,0 +1,22 @@
+{
+  "name": "flashrt-rope-train",
+  "id": "_flashrt_rope_train_cuda_eefd4b8",
+  "version": 1,
+  "license": "Apache-2.0",
+  "python-depends": [],
+  "backend": {
+    "type": "cuda",
+    "archs": [
+      "10.0",
+      "11.0",
+      "12.0",
+      "12.1+PTX",
+      "7.5",
+      "8.0",
+      "8.6",
+      "8.7",
+      "8.9",
+      "9.0"
+    ]
+  }
+}

build/torch212-cxx11-cu132-x86_64-linux/__init__.py

@@ -0,0 +1,26 @@
+"""FlashRT RoPE training reference API."""
+from __future__ import annotations
+import torch
+try:
+    from ._ops import ops
+except Exception:  # source-tree tests before kernel-builder creates _ops.py
+    class _SourceOpsFallback:
+        def _flashrt_training_package_marker(self, x):
+            return x
+    ops = _SourceOpsFallback()
+
+def rotate_half(x: torch.Tensor) -> torch.Tensor:
+    half=x.shape[-1]//2; return torch.cat((-x[...,half:], x[...,:half]), dim=-1)
+def _align(freq: torch.Tensor, x: torch.Tensor, unsqueeze_dim: int) -> torch.Tensor:
+    while freq.dim() < x.dim(): freq = freq.unsqueeze(unsqueeze_dim)
+    return freq
+def apply_rope_train(q: torch.Tensor, k: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor, unsqueeze_dim: int = 1):
+    c_q=_align(cos,q,int(unsqueeze_dim)); s_q=_align(sin,q,int(unsqueeze_dim)); c_k=_align(cos,k,int(unsqueeze_dim)); s_k=_align(sin,k,int(unsqueeze_dim))
+    qd = torch.float64 if q.dtype == torch.float64 else torch.float32
+    kd = torch.float64 if k.dtype == torch.float64 else torch.float32
+    return (q.to(qd)*c_q.to(qd)+rotate_half(q.to(qd))*s_q.to(qd)).to(q.dtype), (k.to(kd)*c_k.to(kd)+rotate_half(k.to(kd))*s_k.to(kd)).to(k.dtype)
+def apply_rope_backward_reference(dq: torch.Tensor, dk: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor, unsqueeze_dim: int = 1):
+    return apply_rope_train(dq, dk, cos, -sin, unsqueeze_dim)
+def backend_marker(x: torch.Tensor) -> torch.Tensor:
+    return ops._flashrt_training_package_marker(x)
+__all__=["rotate_half","apply_rope_train","apply_rope_backward_reference","backend_marker"]

build/torch212-cxx11-cu132-x86_64-linux/_ops.py

@@ -0,0 +1,9 @@
+import torch
+from . import _flashrt_rope_train_cuda_eefd4b8
+ops = torch.ops._flashrt_rope_train_cuda_eefd4b8
+
+def add_op_namespace_prefix(op_name: str):
+    """
+    Prefix op by namespace.
+    """
+    return f"_flashrt_rope_train_cuda_eefd4b8::{op_name}"

build/torch212-cxx11-cu132-x86_64-linux/flashrt_rope_train/__init__.py

@@ -0,0 +1,26 @@
+import ctypes
+import importlib.util
+import sys
+from pathlib import Path
+from types import ModuleType
+
+
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
+
+
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))

build/torch212-cxx11-cu132-x86_64-linux/metadata.json

@@ -0,0 +1,22 @@
+{
+  "name": "flashrt-rope-train",
+  "id": "_flashrt_rope_train_cuda_eefd4b8",
+  "version": 1,
+  "license": "Apache-2.0",
+  "python-depends": [],
+  "backend": {
+    "type": "cuda",
+    "archs": [
+      "10.0",
+      "11.0",
+      "12.0",
+      "12.1+PTX",
+      "7.5",
+      "8.0",
+      "8.6",
+      "8.7",
+      "8.9",
+      "9.0"
+    ]
+  }
+}