bigdl-core-npu 2.6.0b20241224__cp310-cp310-win_amd64.whl → 2.6.0b20241226__cp310-cp310-win_amd64.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- bigdl-core-npu/npu_llm.dll +0 -0
- {bigdl_core_npu-2.6.0b20241224.dist-info → bigdl_core_npu-2.6.0b20241226.dist-info}/METADATA +1 -1
- {bigdl_core_npu-2.6.0b20241224.dist-info → bigdl_core_npu-2.6.0b20241226.dist-info}/RECORD +9 -9
- intel_npu_acceleration_library/_version.py +1 -1
- intel_npu_acceleration_library/backend/base.py +12 -2
- intel_npu_acceleration_library/backend/bindings.py +2 -1
- intel_npu_acceleration_library/lib/Release/intel_npu_acceleration_library.dll +0 -0
- {bigdl_core_npu-2.6.0b20241224.dist-info → bigdl_core_npu-2.6.0b20241226.dist-info}/WHEEL +0 -0
- {bigdl_core_npu-2.6.0b20241224.dist-info → bigdl_core_npu-2.6.0b20241226.dist-info}/top_level.txt +0 -0
bigdl-core-npu/npu_llm.dll
CHANGED
Binary file
|
@@ -1,10 +1,10 @@
|
|
1
1
|
bigdl-core-npu/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
2
|
-
bigdl-core-npu/npu_llm.dll,sha256=
|
2
|
+
bigdl-core-npu/npu_llm.dll,sha256=E-XoECLDiOffE7l6Od5dC9CWVLCS4vOrO1UY7nigPAQ,3923456
|
3
3
|
bigdl-core-npu/npu_llm.lib,sha256=uG0enTHg0GNdEYMN36DPEKQXuYoioueKYpKXCKR3wyk,44328
|
4
4
|
bigdl-core-npu/include/common.h,sha256=AWpPmCEAz6-oGUtq_ONHlELtOzSMqy4CBTMIppJeOII,2819
|
5
5
|
bigdl-core-npu/include/npu_llm.h,sha256=Ar35OKx1qVbT3Q8TGTx_TQa6IlwIYBpn6oSo9q-t9YM,2509
|
6
6
|
intel_npu_acceleration_library/__init__.py,sha256=ZKTIhGMDjF7P6pF-yX8KWcSXbeHWRk24AO_orsa18f8,536
|
7
|
-
intel_npu_acceleration_library/_version.py,sha256=
|
7
|
+
intel_npu_acceleration_library/_version.py,sha256=uSR3Fqqy5_09oZgqJZWo4fR996ACvzrJvcijDaF_o1Q,112
|
8
8
|
intel_npu_acceleration_library/compiler.py,sha256=3IdgqjamSC8MLexDBJypIeZRiWIcTFnvQSU1LPXUr7Y,6225
|
9
9
|
intel_npu_acceleration_library/device.py,sha256=9bn8eVXJa5cXIqgfLsQAdkMVtVUQABb8z0-mQik5jRg,7424
|
10
10
|
intel_npu_acceleration_library/dtypes.py,sha256=gdd06Wsc9zIZFHlauUEx4xcK9WGTn1Mu6GkuYDJeA-E,4683
|
@@ -12,8 +12,8 @@ intel_npu_acceleration_library/modelling.py,sha256=vSiQOWGJ0l6wGV7zWQtZEkHpnMQIM
|
|
12
12
|
intel_npu_acceleration_library/optimizations.py,sha256=9NY8QoDFbs2LY12jbx6As8g2v0oInX4YzvkjnqViA70,5469
|
13
13
|
intel_npu_acceleration_library/quantization.py,sha256=6N_04h1KX6TNbw-ceANV0Pmk4_lQ2Y9C7Pwn5x-zQzo,5566
|
14
14
|
intel_npu_acceleration_library/backend/__init__.py,sha256=2NP6Ypr1dGUNXmLGW5GD9xrh0U9KJgqxTd_c7su1RUY,857
|
15
|
-
intel_npu_acceleration_library/backend/base.py,sha256=
|
16
|
-
intel_npu_acceleration_library/backend/bindings.py,sha256=
|
15
|
+
intel_npu_acceleration_library/backend/base.py,sha256=KrL4odlFqzodOsRUwQIeq99Fg-R9OxAOltPrv9Jg7w8,10005
|
16
|
+
intel_npu_acceleration_library/backend/bindings.py,sha256=9nL_eoJgWAgnJzDzFsUK-ALn6YSLBR1chTrKV-vQQNg,10446
|
17
17
|
intel_npu_acceleration_library/backend/compression.py,sha256=Avz_zm2s_ELy5peVQ8zFGn8njBfh9nEGR16mflotBic,630
|
18
18
|
intel_npu_acceleration_library/backend/convolution.py,sha256=cN3k78X3Y4Cbf7er-MFq0sJ4OwIvquj8PajpdEDmCo4,2018
|
19
19
|
intel_npu_acceleration_library/backend/factory.py,sha256=RuVxbG-jHRhkkUU2q2kITVov3r-gBb6SGslWS-eL-Sk,41035
|
@@ -198,7 +198,7 @@ intel_npu_acceleration_library/external/openvino/torch/__init__.py,sha256=RXLzsf
|
|
198
198
|
intel_npu_acceleration_library/functional/__init__.py,sha256=WWKwKOh6Sgovv7mKctA872TbLP98Pg5m5-MREvUmlAA,204
|
199
199
|
intel_npu_acceleration_library/functional/scaled_dot_product_attention.py,sha256=yGUcg4tDQOLuUnP1g74cl-ec8TRr2SuAMcNLlN6qLvE,1620
|
200
200
|
intel_npu_acceleration_library/lib/Release/cache.json,sha256=CyrSqZUWo0Ec4_7ydOiuKIC0Gm8AybrGdozUqUuHxBw,8840377
|
201
|
-
intel_npu_acceleration_library/lib/Release/intel_npu_acceleration_library.dll,sha256=
|
201
|
+
intel_npu_acceleration_library/lib/Release/intel_npu_acceleration_library.dll,sha256=WBTGRKJVvmX7Qwyzl1P8PwW3IrtCEgzWcQT3mmgbQJw,341504
|
202
202
|
intel_npu_acceleration_library/lib/Release/openvino.dll,sha256=m7M119p3JBq2YYJJ2zzCaBDz6XivKK3nNykb8L1cvDU,13244768
|
203
203
|
intel_npu_acceleration_library/lib/Release/openvino_auto_batch_plugin.dll,sha256=2v_I9P3Qo0St1bQZMEZscnFOUVvgZQQ0HvQlG3HtTd0,203104
|
204
204
|
intel_npu_acceleration_library/lib/Release/openvino_auto_plugin.dll,sha256=e3Aj9CDRHN30dBEdPSk7OCWe52tWfhI4xeXgyFjuDHg,475488
|
@@ -228,7 +228,7 @@ intel_npu_acceleration_library/nn/functional.py,sha256=UfAKBc0u6RtyaMo14ldH2GpEn
|
|
228
228
|
intel_npu_acceleration_library/nn/linear.py,sha256=Q06SoGQeLaI86nA_ky2GnFC6H2Fw1zyMDILKnpYC2eo,5739
|
229
229
|
intel_npu_acceleration_library/nn/llm.py,sha256=P6dz36Yf6BHtzWcftaghC6QaMI_WeRfQwrCbO7fD6hk,15002
|
230
230
|
intel_npu_acceleration_library/nn/module.py,sha256=EYxoTq6I_YgBDgTF76GPDxHrT8SupOTDGMzQaomBeq8,12667
|
231
|
-
bigdl_core_npu-2.6.
|
232
|
-
bigdl_core_npu-2.6.
|
233
|
-
bigdl_core_npu-2.6.
|
234
|
-
bigdl_core_npu-2.6.
|
231
|
+
bigdl_core_npu-2.6.0b20241226.dist-info/METADATA,sha256=s4raV1hQE24vgYfahkhLmod_YL2uCBORa-JOeCe-VV8,1541
|
232
|
+
bigdl_core_npu-2.6.0b20241226.dist-info/WHEEL,sha256=tcd-HDpskugT8GYYKyyid0lOlzoZtZdWwcrj5ormtfo,101
|
233
|
+
bigdl_core_npu-2.6.0b20241226.dist-info/top_level.txt,sha256=iMQZlTsFPJjlD-Y0MqZEP_9ifI0LlbNCJIOTaMoGMjk,46
|
234
|
+
bigdl_core_npu-2.6.0b20241226.dist-info/RECORD,,
|
@@ -53,13 +53,14 @@ class BaseNPUBackend:
|
|
53
53
|
):
|
54
54
|
backend_lib.destroyNNFactory(self._mm)
|
55
55
|
|
56
|
-
def save(self, path: str):
|
56
|
+
def save(self, path: str, compress_to_fp16: bool = True):
|
57
57
|
"""Save the Openvino model.
|
58
58
|
|
59
59
|
Args:
|
60
60
|
path (str): the model save path
|
61
|
+
compress_to_fp16 (bool): whether to compress floating point weights to FP16 (default: True).
|
61
62
|
"""
|
62
|
-
backend_lib.saveModel(self._mm, ctypes.c_char_p(path.encode()))
|
63
|
+
backend_lib.saveModel(self._mm, ctypes.c_char_p(path.encode()), compress_to_fp16)
|
63
64
|
|
64
65
|
def saveCompiledModel(self, path: str):
|
65
66
|
"""Save the compiled model.
|
@@ -68,6 +69,15 @@ class BaseNPUBackend:
|
|
68
69
|
path (str): the compiled model save path
|
69
70
|
"""
|
70
71
|
backend_lib.saveCompiledModel(self._mm, ctypes.c_char_p(path.encode()))
|
72
|
+
|
73
|
+
def serialize(self, xml_path: str, bin_path: str):
|
74
|
+
"""Serialize the Openvino model.
|
75
|
+
|
76
|
+
Args:
|
77
|
+
xml_path (str): the model save xml path
|
78
|
+
bin_path (str): the model save bin path
|
79
|
+
"""
|
80
|
+
backend_lib.serializeModel(self._mm, ctypes.c_char_p(xml_path.encode()), ctypes.c_char_p(bin_path.encode()))
|
71
81
|
|
72
82
|
|
73
83
|
class BaseNPUBackendWithPrefetch(BaseNPUBackend):
|
@@ -67,8 +67,9 @@ def init_common(lib: ctypes.CDLL):
|
|
67
67
|
Args:
|
68
68
|
lib (ctypes.CDLL): Intel® NPU Acceleration Library runtime library
|
69
69
|
"""
|
70
|
-
lib.saveModel.argtypes = [handler, ctypes.c_char_p]
|
70
|
+
lib.saveModel.argtypes = [handler, ctypes.c_char_p, ctypes.c_bool]
|
71
71
|
lib.saveCompiledModel.argtypes = [handler, ctypes.c_char_p]
|
72
|
+
lib.serializeModel.argtypes = [handler, ctypes.c_char_p, ctypes.c_char_p]
|
72
73
|
|
73
74
|
# Set input activations
|
74
75
|
lib.set_activation.argtypes = [handler, ctypes.c_void_p, ctypes.c_int]
|
Binary file
|
File without changes
|
{bigdl_core_npu-2.6.0b20241224.dist-info → bigdl_core_npu-2.6.0b20241226.dist-info}/top_level.txt
RENAMED
File without changes
|