This repository has been archived by the owner on Dec 2, 2023. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathMultiDevices.py
353 lines (302 loc) · 20.8 KB
/
MultiDevices.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
import torch
from typing import Optional, Tuple
# 来自quantization.py
def quantize(layer, weight_bit_width, empty_init=False):
"""Replace fp16 linear with quantized linear"""
from torch.nn import Linear
from torch.nn.parameter import Parameter
import bz2
import torch
import base64
import ctypes
from transformers.utils import logging
from typing import List
from functools import partial
logger = logging.get_logger(__name__)
try:
from cpm_kernels.kernels.base import LazyKernelCModule, KernelFunction, round_up
class Kernel:
def __init__(self, code: bytes, function_names: List[str]):
self.code = code
self._function_names = function_names
self._cmodule = LazyKernelCModule(self.code)
for name in self._function_names:
setattr(self, name, KernelFunction(self._cmodule, name))
quantization_code = "$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"
kernels = Kernel(
bz2.decompress(base64.b64decode(quantization_code)),
[
"int4WeightCompression",
"int4WeightExtractionFloat",
"int4WeightExtractionHalf",
"int8WeightExtractionFloat",
"int8WeightExtractionHalf",
],
)
except Exception as exception:
kernels = None
logger.warning("Failed to load cpm_kernels:" + str(exception))
class W8A16Linear(torch.autograd.Function):
@staticmethod
def forward(ctx, inp: torch.Tensor, quant_w: torch.Tensor, scale_w: torch.Tensor, weight_bit_width):
ctx.inp_shape = inp.size()
ctx.weight_bit_width = weight_bit_width
out_features = quant_w.size(0)
inp = inp.contiguous().view(-1, inp.size(-1))
weight = extract_weight_to_half(quant_w, scale_w, weight_bit_width)
ctx.weight_shape = weight.size()
output = inp.mm(weight.t())
ctx.save_for_backward(inp, quant_w, scale_w)
return output.view(*(ctx.inp_shape[:-1] + (out_features,)))
@staticmethod
def backward(ctx, grad_output: torch.Tensor):
inp, quant_w, scale_w = ctx.saved_tensors
weight = extract_weight_to_half(quant_w, scale_w, ctx.weight_bit_width)
grad_output = grad_output.contiguous().view(-1, weight.size(0))
grad_input = grad_output.mm(weight)
grad_weight = grad_output.t().mm(inp)
return grad_input.view(ctx.inp_shape), grad_weight.view(ctx.weight_shape), None, None
def compress_int4_weight(weight: torch.Tensor): # (n, m)
with torch.cuda.device(weight.device):
n, m = weight.size(0), weight.size(1)
assert m % 2 == 0
m = m // 2
out = torch.empty(n, m, dtype=torch.int8, device="cuda")
stream = torch.cuda.current_stream()
gridDim = (n, 1, 1)
blockDim = (min(round_up(m, 32), 1024), 1, 1)
kernels.int4WeightCompression(
gridDim,
blockDim,
0,
stream,
[ctypes.c_void_p(weight.data_ptr()), ctypes.c_void_p(out.data_ptr()), ctypes.c_int32(n), ctypes.c_int32(m)],
)
return out
def extract_weight_to_half(weight: torch.Tensor, scale_list: torch.Tensor, source_bit_width: int):
if source_bit_width == 8:
func = kernels.int8WeightExtractionHalf
elif source_bit_width == 4:
func = kernels.int4WeightExtractionHalf
else:
assert False, "Unsupported bit-width"
with torch.cuda.device(weight.device):
n, m = weight.size(0), weight.size(1)
out = torch.empty(n, m * (8 // source_bit_width), dtype=torch.half, device="cuda")
stream = torch.cuda.current_stream()
gridDim = (n, 1, 1)
blockDim = (min(round_up(m, 32), 1024), 1, 1)
func(
gridDim,
blockDim,
0,
stream,
[
ctypes.c_void_p(weight.data_ptr()),
ctypes.c_void_p(scale_list.data_ptr()),
ctypes.c_void_p(out.data_ptr()),
ctypes.c_int32(n),
ctypes.c_int32(m),
],
)
return out
class QuantizedLinear(Linear):
def __init__(self, weight_bit_width: int, weight_tensor=None, bias_tensor=None, empty_init=False, *args, **kwargs):
super(QuantizedLinear, self).__init__(*args, **kwargs)
self.weight_bit_width = weight_bit_width
shape = self.weight.shape
del self.weight
if weight_tensor is None or empty_init:
self.weight = torch.empty(
shape[0], shape[1] * weight_bit_width // 8, dtype=torch.int8, device=kwargs["device"]
)
self.weight_scale = torch.empty(shape[0], dtype=kwargs["dtype"], device=kwargs["device"])
else:
self.weight_scale = (weight_tensor.abs().max(dim=-1).values / ((2 ** (weight_bit_width - 1)) - 1)).half()
self.weight = torch.round(weight_tensor / self.weight_scale[:, None]).to(torch.int8)
if weight_bit_width == 4:
self.weight = compress_int4_weight(self.weight)
self.weight = Parameter(self.weight.to(kwargs["device"]), requires_grad=False)
self.weight_scale = Parameter(self.weight_scale.to(kwargs["device"]), requires_grad=False)
if bias_tensor is not None:
self.bias = Parameter(bias_tensor.to(kwargs["device"]), requires_grad=False)
else:
self.bias = None
def forward(self, input):
output = W8A16Linear.apply(input, self.weight, self.weight_scale, self.weight_bit_width)
if self.bias is not None:
output = output + self.bias
return output
# 量化开始
layer.attention.query_key_value = QuantizedLinear(
weight_bit_width=weight_bit_width,
weight_tensor=layer.attention.query_key_value.weight.to(torch.cuda.current_device()),
bias_tensor=layer.attention.query_key_value.bias,
in_features=layer.attention.query_key_value.in_features,
out_features=layer.attention.query_key_value.out_features,
bias=True,
dtype=torch.half,
device=layer.attention.query_key_value.weight.device,
empty_init=empty_init
)
layer.attention.dense = QuantizedLinear(
weight_bit_width=weight_bit_width,
weight_tensor=layer.attention.dense.weight.to(torch.cuda.current_device()),
bias_tensor=layer.attention.dense.bias,
in_features=layer.attention.dense.in_features,
out_features=layer.attention.dense.out_features,
bias=True,
dtype=torch.half,
device=layer.attention.dense.weight.device,
empty_init=empty_init
)
layer.mlp.dense_h_to_4h = QuantizedLinear(
weight_bit_width=weight_bit_width,
weight_tensor=layer.mlp.dense_h_to_4h.weight.to(torch.cuda.current_device()),
bias_tensor=layer.mlp.dense_h_to_4h.bias,
in_features=layer.mlp.dense_h_to_4h.in_features,
out_features=layer.mlp.dense_h_to_4h.out_features,
bias=True,
dtype=torch.half,
device=layer.mlp.dense_h_to_4h.weight.device,
empty_init=empty_init
)
layer.mlp.dense_4h_to_h = QuantizedLinear(
weight_bit_width=weight_bit_width,
weight_tensor=layer.mlp.dense_4h_to_h.weight.to(torch.cuda.current_device()),
bias_tensor=layer.mlp.dense_4h_to_h.bias,
in_features=layer.mlp.dense_4h_to_h.in_features,
out_features=layer.mlp.dense_4h_to_h.out_features,
bias=True,
dtype=torch.half,
device=layer.mlp.dense_4h_to_h.weight.device,
empty_init=empty_init
)
return layer
CPU_precision = 'fp32'
GPU_precision = 'fp16'
embeddings = 'cuda:0'
layers = {'cuda:0': '1-28'}
final_layernorm = 'cuda:0'
new_layers = [None for i in range(28)] # ['cuda:1','cuda:0'...]
def CPU_weight_type(_nn):
global CPU_precision
if(CPU_precision == 'fp32'):
return _nn.float()
elif(CPU_precision == "bf16"):
return _nn.bfloat16()
def quantize_func(_nn):
global GPU_precision
if(GPU_precision == 'fp16'):
return _nn
elif(GPU_precision == 'int8'):
print('int8', end=' -> ')
return quantize(_nn, 8)
elif(GPU_precision == 'int4'):
# print('建议使用已量化的模型')
print('int4', end=' -> ')
return quantize(_nn, 4)
class layers_data_temp():
def __init__(self) -> None:
self.position_ids = None
self.attention_mask = None
class hook_layer(torch.nn.Module):
def __init__(self, layer, device, data_temp, tag) -> None:
super().__init__()
self.layer = CPU_weight_type(layer).to(device) if device == 'cpu' else quantize_func(layer).to(device)
print(device)
self.device = device
self.device_index = None if device == 'cpu' else int(device.split(':')[1])
self.data_temp = data_temp
self.tag = tag
def ToDevice_hidden_states(self, _nn):
# print(self.tag, _nn.device, '->', self.device)
if(self.device == 'cpu'):
return CPU_weight_type(_nn).to(self.device)
else:
return _nn.half().to(self.device)
def ToDevice(self, _nn):
# print(self.tag, _nn.device, '->', self.device)
return _nn.to(self.device)
def forward(self,
hidden_states: torch.Tensor,
position_ids,
attention_mask: torch.Tensor,
layer_id,
layer_past: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,
use_cache: bool = False,
output_attentions: bool = False,):
if(layer_id == 0 or hidden_states.device.index != self.device_index):
hidden_states = self.ToDevice_hidden_states(hidden_states)
self.data_temp.position_ids = self.ToDevice(position_ids)
self.data_temp.attention_mask = self.ToDevice(attention_mask)
output = self.layer(hidden_states,
self.data_temp.position_ids,
self.data_temp.attention_mask,
layer_id,
layer_past,
use_cache,
output_attentions)
return output
class hook_easy(torch.nn.Module):
def __init__(self, nn, device, tag) -> None:
super().__init__()
self.nn = CPU_weight_type(nn).to(device) if device == 'cpu' else nn.to(device)
print(device)
self.device = device
self.device_index = None if device == 'cpu' else int(device.split(':')[1])
self.tag = tag
def ToDevice(self, _nn):
# print(self.tag, _nn.device, '->', self.device)
if(self.tag == 'final_layernorm'):
if(self.device == 'cpu'):
return CPU_weight_type(_nn).to(self.device)
else:
return _nn.half().to(self.device)
else:
if(self.device == 'cpu'):
return CPU_weight_type(_nn).to(self.device)
else:
return _nn.to(self.device)
def forward(self, input):
if(input.device.index != self.device_index):
input = self.ToDevice(input)
output = self.nn(input)
return output
def hook(model):
global embeddings,new_layers,final_layernorm
print('word_embeddings', end=' -> ')
model.transformer.word_embeddings = hook_easy(model.transformer.word_embeddings, embeddings, 'word_embeddings')
data_temp = layers_data_temp()# 创建layers临时数据实例
for index, _ in enumerate(model.transformer.layers):
print('layer', index, end=' -> ')
model.transformer.layers[index] = hook_layer(model.transformer.layers[index], new_layers[index], data_temp, 'layer:' + str(index))
print('final_layernorm', end=' -> ')
model.transformer.final_layernorm = hook_easy(model.transformer.final_layernorm, final_layernorm, 'final_layernorm')
print('lm_head', end=' -> ')
model.lm_head = hook_easy(model.lm_head, embeddings, 'lm_head')
print('hooked.')
return model
def PickupLayersParameter():
global layers
# 处理layers参数
if(layers is None or len(layers) < 1):
raise 'bad layer parameter'
check_id = set(range(1, 28 + 1))
layers_num = 0
for i in layers:
layer_id = layers[i].split('-')
layers[i] = set(range(int(layer_id[0]), int(layer_id[1]) + 1))
if(not(layers[i] <= check_id)):
raise 'found bad layer id.'
layers_num += len(layers[i])
if(layers_num != 28):
raise 'the layer num is not 28.'
for i in layers:
for ii in layers[i]:
new_layers[ii - 1] = i
def ConfigMultiDevices(model):
global embeddings,layers,final_layernorm
PickupLayersParameter()
model = hook(model)
return model