ayo/ipex-llm
1
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

fix bloom-176b int overflow (#8973)

Browse source
This commit is contained in:
Yishuo Wang 2023-09-14 14:37:57 +08:00 committed by GitHub
parent 7563b26ca9
commit bcf456070c
1 changed files with 8 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
python/llm/src/bigdl/llm/ggml/model/llama/llama_cpp.py
View file

@ -947,7 +947,7 @@ def ggml_quantize_tensor(
src, # type: ctypes.Array[ctypes.c_float] # type: ignore src, # type: ctypes.Array[ctypes.c_float] # type: ignore
dst: ctypes.c_void_p, dst: ctypes.c_void_p,
qtype: ctypes.c_int, qtype: ctypes.c_int,
n: ctypes.c_int, n: ctypes.c_size_t,
k: ctypes.c_int, k: ctypes.c_int,
hist, # type: ctypes.Array[ctypes.c_int64] # type: ignore hist, # type: ctypes.Array[ctypes.c_int64] # type: ignore
) -> int: ) -> int:
@ -958,7 +958,7 @@ _lib.ggml_quantize_tensor.argtypes = [
ctypes.POINTER(ctypes.c_float), ctypes.POINTER(ctypes.c_float),
ctypes.c_void_p, ctypes.c_void_p,
ctypes.c_int, ctypes.c_int,
ctypes.c_int, ctypes.c_size_t,
ctypes.c_int, ctypes.c_int,
ctypes.POINTER(ctypes.c_int64), ctypes.POINTER(ctypes.c_int64),
] ]
@ -986,7 +986,7 @@ _lib.ggml_qk_size.restype = ctypes.c_int
def ggml_dequantize_q4_0( def ggml_dequantize_q4_0(
src: ctypes.c_void_p, src: ctypes.c_void_p,
dst: ctypes.c_void_p, dst: ctypes.c_void_p,
k: ctypes.c_int, k: ctypes.c_size_t,
): ):
_lib.ggml_dequantize_q4_0(src, dst, k) _lib.ggml_dequantize_q4_0(src, dst, k)
@ -994,7 +994,7 @@ def ggml_dequantize_q4_0(
_lib.ggml_dequantize_q4_0.argtypes = [ _lib.ggml_dequantize_q4_0.argtypes = [
ctypes.c_void_p, ctypes.c_void_p,
ctypes.c_void_p, ctypes.c_void_p,
ctypes.c_int, ctypes.c_size_t,
] ]
_lib.ggml_quantize_q4_0.restype = None _lib.ggml_quantize_q4_0.restype = None
@ -1002,7 +1002,7 @@ _lib.ggml_quantize_q4_0.restype = None
def ggml_q_format_convet_cpu2xpu( def ggml_q_format_convet_cpu2xpu(
src: ctypes.c_void_p, src: ctypes.c_void_p,
dst: ctypes.c_void_p, dst: ctypes.c_void_p,
n: ctypes.c_int, n: ctypes.c_size_t,
qtype: ctypes.c_int qtype: ctypes.c_int
): ):
_lib.ggml_q_format_convet_cpu2xpu(src, dst, n, qtype) _lib.ggml_q_format_convet_cpu2xpu(src, dst, n, qtype)
@ -1011,7 +1011,7 @@ def ggml_q_format_convet_cpu2xpu(
_lib.ggml_q_format_convet_cpu2xpu.argtypes = [ _lib.ggml_q_format_convet_cpu2xpu.argtypes = [
ctypes.c_void_p, ctypes.c_void_p,
ctypes.c_void_p, ctypes.c_void_p,
ctypes.c_int, ctypes.c_size_t,
ctypes.c_int, ctypes.c_int,
] ]
_lib.ggml_q_format_convet_cpu2xpu.restype = None _lib.ggml_q_format_convet_cpu2xpu.restype = None
@ -1020,7 +1020,7 @@ _lib.ggml_q_format_convet_cpu2xpu.restype = None
def ggml_q_format_convet_xpu2cpu( def ggml_q_format_convet_xpu2cpu(
src: ctypes.c_void_p, src: ctypes.c_void_p,
dst: ctypes.c_void_p, dst: ctypes.c_void_p,
n: ctypes.c_int, n: ctypes.c_size_t,
qtype: ctypes.c_int qtype: ctypes.c_int
): ):
_lib.ggml_q_format_convet_xpu2cpu(src, dst, n, qtype) _lib.ggml_q_format_convet_xpu2cpu(src, dst, n, qtype)
@ -1029,7 +1029,7 @@ def ggml_q_format_convet_xpu2cpu(
_lib.ggml_q_format_convet_xpu2cpu.argtypes = [ _lib.ggml_q_format_convet_xpu2cpu.argtypes = [
ctypes.c_void_p, ctypes.c_void_p,
ctypes.c_void_p, ctypes.c_void_p,
ctypes.c_int, ctypes.c_size_t,
ctypes.c_int ctypes.c_int
] ]
_lib.ggml_q_format_convet_xpu2cpu.restype = None _lib.ggml_q_format_convet_xpu2cpu.restype = None