Add support for loading quantized model
Browse files- configuration_chatglm.py +3 -0
- modeling_chatglm.py +30 -4
- quantization.py +44 -30
configuration_chatglm.py
CHANGED
@@ -70,6 +70,7 @@ class ChatGLMConfig(PretrainedConfig):
|
|
70 |
max_sequence_length=2048,
|
71 |
inner_hidden_size=16384,
|
72 |
position_encoding_2d=True,
|
|
|
73 |
pre_seq_len=None,
|
74 |
prefix_projection=False,
|
75 |
**kwargs
|
@@ -86,8 +87,10 @@ class ChatGLMConfig(PretrainedConfig):
|
|
86 |
self.eos_token_id = eos_token_id
|
87 |
self.pad_token_id = pad_token_id
|
88 |
self.position_encoding_2d = position_encoding_2d
|
|
|
89 |
self.pre_seq_len = pre_seq_len
|
90 |
self.prefix_projection = prefix_projection
|
|
|
91 |
super().__init__(
|
92 |
pad_token_id=pad_token_id,
|
93 |
bos_token_id=bos_token_id,
|
|
|
70 |
max_sequence_length=2048,
|
71 |
inner_hidden_size=16384,
|
72 |
position_encoding_2d=True,
|
73 |
+
quantization_bit=0,
|
74 |
pre_seq_len=None,
|
75 |
prefix_projection=False,
|
76 |
**kwargs
|
|
|
87 |
self.eos_token_id = eos_token_id
|
88 |
self.pad_token_id = pad_token_id
|
89 |
self.position_encoding_2d = position_encoding_2d
|
90 |
+
self.quantization_bit = quantization_bit
|
91 |
self.pre_seq_len = pre_seq_len
|
92 |
self.prefix_projection = prefix_projection
|
93 |
+
|
94 |
super().__init__(
|
95 |
pad_token_id=pad_token_id,
|
96 |
bos_token_id=bos_token_id,
|
modeling_chatglm.py
CHANGED
@@ -139,6 +139,7 @@ class PrefixEncoder(torch.nn.Module):
|
|
139 |
Input shape: (batch-size, prefix-length)
|
140 |
Output shape: (batch-size, prefix-length, 2*layers*hidden)
|
141 |
"""
|
|
|
142 |
def __init__(self, config):
|
143 |
super().__init__()
|
144 |
self.prefix_projection = config.prefix_projection
|
@@ -216,6 +217,13 @@ class RotaryEmbedding(torch.nn.Module):
|
|
216 |
self.cos_cached, self.sin_cached = cos_cached, sin_cached
|
217 |
return self.cos_cached[:seq_len, ...], self.sin_cached[:seq_len, ...]
|
218 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
219 |
|
220 |
def rotate_half(x):
|
221 |
x1, x2 = x[..., :x.shape[-1] // 2], x[..., x.shape[-1] // 2:]
|
@@ -931,7 +939,6 @@ class ChatGLMModel(ChatGLMPreTrainedModel):
|
|
931 |
gmask=use_gmask
|
932 |
)
|
933 |
|
934 |
-
|
935 |
# [seq_len, batch, hidden_size]
|
936 |
hidden_states = inputs_embeds.transpose(0, 1)
|
937 |
|
@@ -999,7 +1006,7 @@ class ChatGLMModel(ChatGLMPreTrainedModel):
|
|
999 |
|
1000 |
|
1001 |
class ChatGLMForConditionalGeneration(ChatGLMPreTrainedModel):
|
1002 |
-
def __init__(self, config):
|
1003 |
super().__init__(config)
|
1004 |
|
1005 |
# self.hidden_size = config.hidden_size
|
@@ -1019,6 +1026,13 @@ class ChatGLMForConditionalGeneration(ChatGLMPreTrainedModel):
|
|
1019 |
dtype=torch.half
|
1020 |
)
|
1021 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1022 |
def get_output_embeddings(self):
|
1023 |
return self.lm_head
|
1024 |
|
@@ -1351,7 +1365,19 @@ class ChatGLMForConditionalGeneration(ChatGLMPreTrainedModel):
|
|
1351 |
break
|
1352 |
yield input_ids
|
1353 |
|
1354 |
-
def quantize(self, bits: int):
|
|
|
|
|
|
|
1355 |
from .quantization import quantize
|
1356 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1357 |
return self
|
|
|
139 |
Input shape: (batch-size, prefix-length)
|
140 |
Output shape: (batch-size, prefix-length, 2*layers*hidden)
|
141 |
"""
|
142 |
+
|
143 |
def __init__(self, config):
|
144 |
super().__init__()
|
145 |
self.prefix_projection = config.prefix_projection
|
|
|
217 |
self.cos_cached, self.sin_cached = cos_cached, sin_cached
|
218 |
return self.cos_cached[:seq_len, ...], self.sin_cached[:seq_len, ...]
|
219 |
|
220 |
+
def _apply(self, fn):
|
221 |
+
if self.cos_cached is not None:
|
222 |
+
self.cos_cached = fn(self.cos_cached)
|
223 |
+
if self.sin_cached is not None:
|
224 |
+
self.sin_cached = fn(self.sin_cached)
|
225 |
+
return super()._apply(fn)
|
226 |
+
|
227 |
|
228 |
def rotate_half(x):
|
229 |
x1, x2 = x[..., :x.shape[-1] // 2], x[..., x.shape[-1] // 2:]
|
|
|
939 |
gmask=use_gmask
|
940 |
)
|
941 |
|
|
|
942 |
# [seq_len, batch, hidden_size]
|
943 |
hidden_states = inputs_embeds.transpose(0, 1)
|
944 |
|
|
|
1006 |
|
1007 |
|
1008 |
class ChatGLMForConditionalGeneration(ChatGLMPreTrainedModel):
|
1009 |
+
def __init__(self, config: ChatGLMConfig):
|
1010 |
super().__init__(config)
|
1011 |
|
1012 |
# self.hidden_size = config.hidden_size
|
|
|
1026 |
dtype=torch.half
|
1027 |
)
|
1028 |
|
1029 |
+
self.config = config
|
1030 |
+
|
1031 |
+
self.quantized = False
|
1032 |
+
|
1033 |
+
if self.config.quantization_bit:
|
1034 |
+
self.quantize(self.config.quantization_bit, empty_init=True)
|
1035 |
+
|
1036 |
def get_output_embeddings(self):
|
1037 |
return self.lm_head
|
1038 |
|
|
|
1365 |
break
|
1366 |
yield input_ids
|
1367 |
|
1368 |
+
def quantize(self, bits: int, empty_init=False, **kwargs):
|
1369 |
+
if bits == 0:
|
1370 |
+
return
|
1371 |
+
|
1372 |
from .quantization import quantize
|
1373 |
+
|
1374 |
+
if self.quantized:
|
1375 |
+
logger.info("Already quantized.")
|
1376 |
+
return self
|
1377 |
+
|
1378 |
+
self.quantized = True
|
1379 |
+
|
1380 |
+
self.config.quantization_bit = bits
|
1381 |
+
|
1382 |
+
self.transformer = quantize(self.transformer, bits, empty_init=empty_init, **kwargs)
|
1383 |
return self
|
quantization.py
CHANGED
@@ -5,9 +5,40 @@ import bz2
|
|
5 |
import torch
|
6 |
import base64
|
7 |
import ctypes
|
|
|
8 |
|
9 |
from typing import List
|
10 |
-
from
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
|
12 |
|
13 |
class W8A16Linear(torch.autograd.Function):
|
@@ -33,30 +64,6 @@ class W8A16Linear(torch.autograd.Function):
|
|
33 |
return grad_input.view(ctx.inp_shape), grad_weight.view(ctx.weight_shape), None, None
|
34 |
|
35 |
|
36 |
-
class Kernel:
|
37 |
-
def __init__(self, code: bytes, function_names: List[str]):
|
38 |
-
self.code = code
|
39 |
-
self._function_names = function_names
|
40 |
-
self._cmodule = LazyKernelCModule(self.code)
|
41 |
-
|
42 |
-
for name in self._function_names:
|
43 |
-
setattr(self, name, KernelFunction(self._cmodule, name))
|
44 |
-
|
45 |
-
|
46 |
-
quantization_code = "$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"
|
47 |
-
|
48 |
-
kernels = Kernel(
|
49 |
-
bz2.decompress(base64.b64decode(quantization_code)),
|
50 |
-
[
|
51 |
-
"int4WeightCompression",
|
52 |
-
"int4WeightExtractionFloat",
|
53 |
-
"int4WeightExtractionHalf",
|
54 |
-
"int8WeightExtractionFloat",
|
55 |
-
"int8WeightExtractionHalf",
|
56 |
-
],
|
57 |
-
)
|
58 |
-
|
59 |
-
|
60 |
def compress_int4_weight(weight: torch.Tensor): # (n, m)
|
61 |
with torch.cuda.device(weight.device):
|
62 |
n, m = weight.size(0), weight.size(1)
|
@@ -111,18 +118,18 @@ def extract_weight_to_half(weight: torch.Tensor, scale_list: torch.Tensor, sourc
|
|
111 |
|
112 |
|
113 |
class QuantizedLinear(Linear):
|
114 |
-
def __init__(self, weight_bit_width: int, weight_tensor=None, bias_tensor=None, *args, **kwargs):
|
115 |
super(QuantizedLinear, self).__init__(*args, **kwargs)
|
116 |
self.weight_bit_width = weight_bit_width
|
117 |
|
118 |
shape = self.weight.shape
|
119 |
del self.weight
|
120 |
|
121 |
-
if weight_tensor is None:
|
122 |
self.weight = torch.empty(
|
123 |
shape[0], shape[1] * weight_bit_width // 8, dtype=torch.int8, device=kwargs["device"]
|
124 |
)
|
125 |
-
self.weight_scale = torch.empty(shape[0], dtype=kwargs["
|
126 |
else:
|
127 |
self.weight_scale = (weight_tensor.abs().max(dim=-1).values / ((2 ** (weight_bit_width - 1)) - 1)).half()
|
128 |
self.weight = torch.round(weight_tensor / self.weight_scale[:, None]).to(torch.int8)
|
@@ -131,7 +138,10 @@ class QuantizedLinear(Linear):
|
|
131 |
|
132 |
self.weight = Parameter(self.weight.to(kwargs["device"]), requires_grad=False)
|
133 |
self.weight_scale = Parameter(self.weight_scale.to(kwargs["device"]), requires_grad=False)
|
134 |
-
|
|
|
|
|
|
|
135 |
|
136 |
def forward(self, input):
|
137 |
output = W8A16Linear.apply(input, self.weight, self.weight_scale, self.weight_bit_width)
|
@@ -140,7 +150,7 @@ class QuantizedLinear(Linear):
|
|
140 |
return output
|
141 |
|
142 |
|
143 |
-
def quantize(model, weight_bit_width):
|
144 |
"""Replace fp16 linear with quantized linear"""
|
145 |
|
146 |
for layer in model.layers:
|
@@ -153,6 +163,7 @@ def quantize(model, weight_bit_width):
|
|
153 |
bias=True,
|
154 |
dtype=torch.half,
|
155 |
device=layer.attention.query_key_value.weight.device,
|
|
|
156 |
)
|
157 |
layer.attention.dense = QuantizedLinear(
|
158 |
weight_bit_width=weight_bit_width,
|
@@ -163,6 +174,7 @@ def quantize(model, weight_bit_width):
|
|
163 |
bias=True,
|
164 |
dtype=torch.half,
|
165 |
device=layer.attention.dense.weight.device,
|
|
|
166 |
)
|
167 |
layer.mlp.dense_h_to_4h = QuantizedLinear(
|
168 |
weight_bit_width=weight_bit_width,
|
@@ -173,6 +185,7 @@ def quantize(model, weight_bit_width):
|
|
173 |
bias=True,
|
174 |
dtype=torch.half,
|
175 |
device=layer.mlp.dense_h_to_4h.weight.device,
|
|
|
176 |
)
|
177 |
layer.mlp.dense_4h_to_h = QuantizedLinear(
|
178 |
weight_bit_width=weight_bit_width,
|
@@ -183,5 +196,6 @@ def quantize(model, weight_bit_width):
|
|
183 |
bias=True,
|
184 |
dtype=torch.half,
|
185 |
device=layer.mlp.dense_4h_to_h.weight.device,
|
|
|
186 |
)
|
187 |
return model
|
|
|
5 |
import torch
|
6 |
import base64
|
7 |
import ctypes
|
8 |
+
from transformers.utils import logging
|
9 |
|
10 |
from typing import List
|
11 |
+
from functools import partial
|
12 |
+
|
13 |
+
logger = logging.get_logger(__name__)
|
14 |
+
|
15 |
+
try:
|
16 |
+
from cpm_kernels.kernels.base import LazyKernelCModule, KernelFunction, round_up
|
17 |
+
|
18 |
+
class Kernel:
|
19 |
+
def __init__(self, code: bytes, function_names: List[str]):
|
20 |
+
self.code = code
|
21 |
+
self._function_names = function_names
|
22 |
+
self._cmodule = LazyKernelCModule(self.code)
|
23 |
+
|
24 |
+
for name in self._function_names:
|
25 |
+
setattr(self, name, KernelFunction(self._cmodule, name))
|
26 |
+
|
27 |
+
quantization_code = "$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"
|
28 |
+
|
29 |
+
kernels = Kernel(
|
30 |
+
bz2.decompress(base64.b64decode(quantization_code)),
|
31 |
+
[
|
32 |
+
"int4WeightCompression",
|
33 |
+
"int4WeightExtractionFloat",
|
34 |
+
"int4WeightExtractionHalf",
|
35 |
+
"int8WeightExtractionFloat",
|
36 |
+
"int8WeightExtractionHalf",
|
37 |
+
],
|
38 |
+
)
|
39 |
+
except Exception as exception:
|
40 |
+
kernels = None
|
41 |
+
logger.warning("Failed to load cpm_kernels:" + str(exception))
|
42 |
|
43 |
|
44 |
class W8A16Linear(torch.autograd.Function):
|
|
|
64 |
return grad_input.view(ctx.inp_shape), grad_weight.view(ctx.weight_shape), None, None
|
65 |
|
66 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
67 |
def compress_int4_weight(weight: torch.Tensor): # (n, m)
|
68 |
with torch.cuda.device(weight.device):
|
69 |
n, m = weight.size(0), weight.size(1)
|
|
|
118 |
|
119 |
|
120 |
class QuantizedLinear(Linear):
|
121 |
+
def __init__(self, weight_bit_width: int, weight_tensor=None, bias_tensor=None, empty_init=False, *args, **kwargs):
|
122 |
super(QuantizedLinear, self).__init__(*args, **kwargs)
|
123 |
self.weight_bit_width = weight_bit_width
|
124 |
|
125 |
shape = self.weight.shape
|
126 |
del self.weight
|
127 |
|
128 |
+
if weight_tensor is None or empty_init:
|
129 |
self.weight = torch.empty(
|
130 |
shape[0], shape[1] * weight_bit_width // 8, dtype=torch.int8, device=kwargs["device"]
|
131 |
)
|
132 |
+
self.weight_scale = torch.empty(shape[0], dtype=kwargs["dtype"], device=kwargs["device"])
|
133 |
else:
|
134 |
self.weight_scale = (weight_tensor.abs().max(dim=-1).values / ((2 ** (weight_bit_width - 1)) - 1)).half()
|
135 |
self.weight = torch.round(weight_tensor / self.weight_scale[:, None]).to(torch.int8)
|
|
|
138 |
|
139 |
self.weight = Parameter(self.weight.to(kwargs["device"]), requires_grad=False)
|
140 |
self.weight_scale = Parameter(self.weight_scale.to(kwargs["device"]), requires_grad=False)
|
141 |
+
if bias_tensor is not None:
|
142 |
+
self.bias = Parameter(bias_tensor.to(kwargs["device"]), requires_grad=False)
|
143 |
+
else:
|
144 |
+
self.bias = None
|
145 |
|
146 |
def forward(self, input):
|
147 |
output = W8A16Linear.apply(input, self.weight, self.weight_scale, self.weight_bit_width)
|
|
|
150 |
return output
|
151 |
|
152 |
|
153 |
+
def quantize(model, weight_bit_width, empty_init=False, **kwargs):
|
154 |
"""Replace fp16 linear with quantized linear"""
|
155 |
|
156 |
for layer in model.layers:
|
|
|
163 |
bias=True,
|
164 |
dtype=torch.half,
|
165 |
device=layer.attention.query_key_value.weight.device,
|
166 |
+
empty_init=empty_init
|
167 |
)
|
168 |
layer.attention.dense = QuantizedLinear(
|
169 |
weight_bit_width=weight_bit_width,
|
|
|
174 |
bias=True,
|
175 |
dtype=torch.half,
|
176 |
device=layer.attention.dense.weight.device,
|
177 |
+
empty_init=empty_init
|
178 |
)
|
179 |
layer.mlp.dense_h_to_4h = QuantizedLinear(
|
180 |
weight_bit_width=weight_bit_width,
|
|
|
185 |
bias=True,
|
186 |
dtype=torch.half,
|
187 |
device=layer.mlp.dense_h_to_4h.weight.device,
|
188 |
+
empty_init=empty_init
|
189 |
)
|
190 |
layer.mlp.dense_4h_to_h = QuantizedLinear(
|
191 |
weight_bit_width=weight_bit_width,
|
|
|
196 |
bias=True,
|
197 |
dtype=torch.half,
|
198 |
device=layer.mlp.dense_4h_to_h.weight.device,
|
199 |
+
empty_init=empty_init
|
200 |
)
|
201 |
return model
|