Hugging Face's logo

Datasets:
codeShare
/
lora-training-data

Tags:
Not-For-All-Audiences
Dataset card Files
xet
 Community
1
codeShare commited on
Commit
7eea018
·
verified ·
1 Parent(s): c45aaa9

Upload Inspect_safetensor_weights.ipynb

Browse files
Files changed (1) hide show
  1. Inspect_safetensor_weights.ipynb +581 -0
Inspect_safetensor_weights.ipynb ADDED
@@ -0,0 +1,581 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "nbformat": 4,
3
+ "nbformat_minor": 0,
4
+ "metadata": {
5
+ "colab": {
6
+ "provenance": []
7
+ },
8
+ "kernelspec": {
9
+ "name": "python3",
10
+ "display_name": "Python 3"
11
+ },
12
+ "language_info": {
13
+ "name": "python"
14
+ }
15
+ },
16
+ "cells": [
17
+ {
18
+ "cell_type": "code",
19
+ "metadata": {
20
+ "colab": {
21
+ "base_uri": "https://localhost:8080/"
22
+ },
23
+ "id": "64310d2f",
24
+ "outputId": "6e5e3be2-8d70-44f2-ac23-1078f9f36390"
25
+ },
26
+ "source": [
27
+ "pip install safetensors"
28
+ ],
29
+ "execution_count": 1,
30
+ "outputs": [
31
+ {
32
+ "output_type": "stream",
33
+ "name": "stdout",
34
+ "text": [
35
+ "Requirement already satisfied: safetensors in /usr/local/lib/python3.13/dist-packages (0.8.0)\n"
36
+ ]
37
+ }
38
+ ]
39
+ },
40
+ {
41
+ "cell_type": "code",
42
+ "metadata": {
43
+ "colab": {
44
+ "base_uri": "https://localhost:8080/"
45
+ },
46
+ "id": "a1fcf785",
47
+ "outputId": "943b05c5-978f-4a42-eeea-1b96a07e9146"
48
+ },
49
+ "source": [
50
+ "from safetensors import safe_open\n",
51
+ "\n",
52
+ "file_path = '/content/drive/MyDrive/philosophical-tetsu-e1-moss-image-sft-epoch-01.safetensors'\n",
53
+ "\n",
54
+ "with safe_open(file_path, framework='pt', device='cpu') as f:\n",
55
+ " keys = sorted(f.keys())\n",
56
+ "\n",
57
+ " # Group by the component (e.g. diffusion_model.layers.0.proj.ffn_down)\n",
58
+ " grouped = {}\n",
59
+ " for key in keys:\n",
60
+ " tensor = f.get_tensor(key)\n",
61
+ " shape = list(tensor.shape)\n",
62
+ "\n",
63
+ " # Separate name components\n",
64
+ " parts = key.split('.')\n",
65
+ " # Find where 'lora_A' or 'lora_B' begins to group parent components\n",
66
+ " sub_part_idx = next(i for i, part in enumerate(parts) if 'lora_' in part or part == 'weight')\n",
67
+ " parent_group = '.'.join(parts[:sub_part_idx])\n",
68
+ " sub_group = '.'.join(parts[sub_part_idx:])\n",
69
+ "\n",
70
+ " if parent_group not in grouped:\n",
71
+ " grouped[parent_group] = []\n",
72
+ " grouped[parent_group].append(f\"{sub_group}: {shape}\")\n",
73
+ "\n",
74
+ "# Print each major parent group with its subcategories collapsed onto a single line\n",
75
+ "for parent, dimensions in sorted(grouped.items()):\n",
76
+ " print(f\"{parent} -> {{ {', '.join(dimensions)} }}\")"
77
+ ],
78
+ "execution_count": 3,
79
+ "outputs": [
80
+ {
81
+ "output_type": "stream",
82
+ "name": "stdout",
83
+ "text": [
84
+ "diffusion_model.layers.0.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
85
+ "diffusion_model.layers.0.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
86
+ "diffusion_model.layers.0.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
87
+ "diffusion_model.layers.0.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
88
+ "diffusion_model.layers.0.proj.k.lora_A -> {weight: [32, 3072]}\n",
89
+ "diffusion_model.layers.0.proj.k.lora_B -> {weight: [3072, 32]}\n",
90
+ "diffusion_model.layers.0.proj.o.lora_A -> {weight: [32, 3072]}\n",
91
+ "diffusion_model.layers.0.proj.o.lora_B -> {weight: [3072, 32]}\n",
92
+ "diffusion_model.layers.0.proj.q.lora_A -> {weight: [32, 3072]}\n",
93
+ "diffusion_model.layers.0.proj.q.lora_B -> {weight: [3072, 32]}\n",
94
+ "diffusion_model.layers.0.proj.v.lora_A -> {weight: [32, 3072]}\n",
95
+ "diffusion_model.layers.0.proj.v.lora_B -> {weight: [3072, 32]}\n",
96
+ "diffusion_model.layers.1.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
97
+ "diffusion_model.layers.1.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
98
+ "diffusion_model.layers.1.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
99
+ "diffusion_model.layers.1.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
100
+ "diffusion_model.layers.1.proj.k.lora_A -> {weight: [32, 3072]}\n",
101
+ "diffusion_model.layers.1.proj.k.lora_B -> {weight: [3072, 32]}\n",
102
+ "diffusion_model.layers.1.proj.o.lora_A -> {weight: [32, 3072]}\n",
103
+ "diffusion_model.layers.1.proj.o.lora_B -> {weight: [3072, 32]}\n",
104
+ "diffusion_model.layers.1.proj.q.lora_A -> {weight: [32, 3072]}\n",
105
+ "diffusion_model.layers.1.proj.q.lora_B -> {weight: [3072, 32]}\n",
106
+ "diffusion_model.layers.1.proj.v.lora_A -> {weight: [32, 3072]}\n",
107
+ "diffusion_model.layers.1.proj.v.lora_B -> {weight: [3072, 32]}\n",
108
+ "diffusion_model.layers.10.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
109
+ "diffusion_model.layers.10.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
110
+ "diffusion_model.layers.10.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
111
+ "diffusion_model.layers.10.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
112
+ "diffusion_model.layers.10.proj.k.lora_A -> {weight: [32, 3072]}\n",
113
+ "diffusion_model.layers.10.proj.k.lora_B -> {weight: [3072, 32]}\n",
114
+ "diffusion_model.layers.10.proj.o.lora_A -> {weight: [32, 3072]}\n",
115
+ "diffusion_model.layers.10.proj.o.lora_B -> {weight: [3072, 32]}\n",
116
+ "diffusion_model.layers.10.proj.q.lora_A -> {weight: [32, 3072]}\n",
117
+ "diffusion_model.layers.10.proj.q.lora_B -> {weight: [3072, 32]}\n",
118
+ "diffusion_model.layers.10.proj.v.lora_A -> {weight: [32, 3072]}\n",
119
+ "diffusion_model.layers.10.proj.v.lora_B -> {weight: [3072, 32]}\n",
120
+ "diffusion_model.layers.11.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
121
+ "diffusion_model.layers.11.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
122
+ "diffusion_model.layers.11.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
123
+ "diffusion_model.layers.11.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
124
+ "diffusion_model.layers.11.proj.k.lora_A -> {weight: [32, 3072]}\n",
125
+ "diffusion_model.layers.11.proj.k.lora_B -> {weight: [3072, 32]}\n",
126
+ "diffusion_model.layers.11.proj.o.lora_A -> {weight: [32, 3072]}\n",
127
+ "diffusion_model.layers.11.proj.o.lora_B -> {weight: [3072, 32]}\n",
128
+ "diffusion_model.layers.11.proj.q.lora_A -> {weight: [32, 3072]}\n",
129
+ "diffusion_model.layers.11.proj.q.lora_B -> {weight: [3072, 32]}\n",
130
+ "diffusion_model.layers.11.proj.v.lora_A -> {weight: [32, 3072]}\n",
131
+ "diffusion_model.layers.11.proj.v.lora_B -> {weight: [3072, 32]}\n",
132
+ "diffusion_model.layers.12.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
133
+ "diffusion_model.layers.12.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
134
+ "diffusion_model.layers.12.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
135
+ "diffusion_model.layers.12.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
136
+ "diffusion_model.layers.12.proj.k.lora_A -> {weight: [32, 3072]}\n",
137
+ "diffusion_model.layers.12.proj.k.lora_B -> {weight: [3072, 32]}\n",
138
+ "diffusion_model.layers.12.proj.o.lora_A -> {weight: [32, 3072]}\n",
139
+ "diffusion_model.layers.12.proj.o.lora_B -> {weight: [3072, 32]}\n",
140
+ "diffusion_model.layers.12.proj.q.lora_A -> {weight: [32, 3072]}\n",
141
+ "diffusion_model.layers.12.proj.q.lora_B -> {weight: [3072, 32]}\n",
142
+ "diffusion_model.layers.12.proj.v.lora_A -> {weight: [32, 3072]}\n",
143
+ "diffusion_model.layers.12.proj.v.lora_B -> {weight: [3072, 32]}\n",
144
+ "diffusion_model.layers.13.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
145
+ "diffusion_model.layers.13.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
146
+ "diffusion_model.layers.13.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
147
+ "diffusion_model.layers.13.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
148
+ "diffusion_model.layers.13.proj.k.lora_A -> {weight: [32, 3072]}\n",
149
+ "diffusion_model.layers.13.proj.k.lora_B -> {weight: [3072, 32]}\n",
150
+ "diffusion_model.layers.13.proj.o.lora_A -> {weight: [32, 3072]}\n",
151
+ "diffusion_model.layers.13.proj.o.lora_B -> {weight: [3072, 32]}\n",
152
+ "diffusion_model.layers.13.proj.q.lora_A -> {weight: [32, 3072]}\n",
153
+ "diffusion_model.layers.13.proj.q.lora_B -> {weight: [3072, 32]}\n",
154
+ "diffusion_model.layers.13.proj.v.lora_A -> {weight: [32, 3072]}\n",
155
+ "diffusion_model.layers.13.proj.v.lora_B -> {weight: [3072, 32]}\n",
156
+ "diffusion_model.layers.14.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
157
+ "diffusion_model.layers.14.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
158
+ "diffusion_model.layers.14.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
159
+ "diffusion_model.layers.14.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
160
+ "diffusion_model.layers.14.proj.k.lora_A -> {weight: [32, 3072]}\n",
161
+ "diffusion_model.layers.14.proj.k.lora_B -> {weight: [3072, 32]}\n",
162
+ "diffusion_model.layers.14.proj.o.lora_A -> {weight: [32, 3072]}\n",
163
+ "diffusion_model.layers.14.proj.o.lora_B -> {weight: [3072, 32]}\n",
164
+ "diffusion_model.layers.14.proj.q.lora_A -> {weight: [32, 3072]}\n",
165
+ "diffusion_model.layers.14.proj.q.lora_B -> {weight: [3072, 32]}\n",
166
+ "diffusion_model.layers.14.proj.v.lora_A -> {weight: [32, 3072]}\n",
167
+ "diffusion_model.layers.14.proj.v.lora_B -> {weight: [3072, 32]}\n",
168
+ "diffusion_model.layers.15.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
169
+ "diffusion_model.layers.15.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
170
+ "diffusion_model.layers.15.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
171
+ "diffusion_model.layers.15.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
172
+ "diffusion_model.layers.15.proj.k.lora_A -> {weight: [32, 3072]}\n",
173
+ "diffusion_model.layers.15.proj.k.lora_B -> {weight: [3072, 32]}\n",
174
+ "diffusion_model.layers.15.proj.o.lora_A -> {weight: [32, 3072]}\n",
175
+ "diffusion_model.layers.15.proj.o.lora_B -> {weight: [3072, 32]}\n",
176
+ "diffusion_model.layers.15.proj.q.lora_A -> {weight: [32, 3072]}\n",
177
+ "diffusion_model.layers.15.proj.q.lora_B -> {weight: [3072, 32]}\n",
178
+ "diffusion_model.layers.15.proj.v.lora_A -> {weight: [32, 3072]}\n",
179
+ "diffusion_model.layers.15.proj.v.lora_B -> {weight: [3072, 32]}\n",
180
+ "diffusion_model.layers.16.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
181
+ "diffusion_model.layers.16.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
182
+ "diffusion_model.layers.16.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
183
+ "diffusion_model.layers.16.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
184
+ "diffusion_model.layers.16.proj.k.lora_A -> {weight: [32, 3072]}\n",
185
+ "diffusion_model.layers.16.proj.k.lora_B -> {weight: [3072, 32]}\n",
186
+ "diffusion_model.layers.16.proj.o.lora_A -> {weight: [32, 3072]}\n",
187
+ "diffusion_model.layers.16.proj.o.lora_B -> {weight: [3072, 32]}\n",
188
+ "diffusion_model.layers.16.proj.q.lora_A -> {weight: [32, 3072]}\n",
189
+ "diffusion_model.layers.16.proj.q.lora_B -> {weight: [3072, 32]}\n",
190
+ "diffusion_model.layers.16.proj.v.lora_A -> {weight: [32, 3072]}\n",
191
+ "diffusion_model.layers.16.proj.v.lora_B -> {weight: [3072, 32]}\n",
192
+ "diffusion_model.layers.17.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
193
+ "diffusion_model.layers.17.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
194
+ "diffusion_model.layers.17.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
195
+ "diffusion_model.layers.17.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
196
+ "diffusion_model.layers.17.proj.k.lora_A -> {weight: [32, 3072]}\n",
197
+ "diffusion_model.layers.17.proj.k.lora_B -> {weight: [3072, 32]}\n",
198
+ "diffusion_model.layers.17.proj.o.lora_A -> {weight: [32, 3072]}\n",
199
+ "diffusion_model.layers.17.proj.o.lora_B -> {weight: [3072, 32]}\n",
200
+ "diffusion_model.layers.17.proj.q.lora_A -> {weight: [32, 3072]}\n",
201
+ "diffusion_model.layers.17.proj.q.lora_B -> {weight: [3072, 32]}\n",
202
+ "diffusion_model.layers.17.proj.v.lora_A -> {weight: [32, 3072]}\n",
203
+ "diffusion_model.layers.17.proj.v.lora_B -> {weight: [3072, 32]}\n",
204
+ "diffusion_model.layers.18.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
205
+ "diffusion_model.layers.18.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
206
+ "diffusion_model.layers.18.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
207
+ "diffusion_model.layers.18.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
208
+ "diffusion_model.layers.18.proj.k.lora_A -> {weight: [32, 3072]}\n",
209
+ "diffusion_model.layers.18.proj.k.lora_B -> {weight: [3072, 32]}\n",
210
+ "diffusion_model.layers.18.proj.o.lora_A -> {weight: [32, 3072]}\n",
211
+ "diffusion_model.layers.18.proj.o.lora_B -> {weight: [3072, 32]}\n",
212
+ "diffusion_model.layers.18.proj.q.lora_A -> {weight: [32, 3072]}\n",
213
+ "diffusion_model.layers.18.proj.q.lora_B -> {weight: [3072, 32]}\n",
214
+ "diffusion_model.layers.18.proj.v.lora_A -> {weight: [32, 3072]}\n",
215
+ "diffusion_model.layers.18.proj.v.lora_B -> {weight: [3072, 32]}\n",
216
+ "diffusion_model.layers.19.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
217
+ "diffusion_model.layers.19.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
218
+ "diffusion_model.layers.19.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
219
+ "diffusion_model.layers.19.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
220
+ "diffusion_model.layers.19.proj.k.lora_A -> {weight: [32, 3072]}\n",
221
+ "diffusion_model.layers.19.proj.k.lora_B -> {weight: [3072, 32]}\n",
222
+ "diffusion_model.layers.19.proj.o.lora_A -> {weight: [32, 3072]}\n",
223
+ "diffusion_model.layers.19.proj.o.lora_B -> {weight: [3072, 32]}\n",
224
+ "diffusion_model.layers.19.proj.q.lora_A -> {weight: [32, 3072]}\n",
225
+ "diffusion_model.layers.19.proj.q.lora_B -> {weight: [3072, 32]}\n",
226
+ "diffusion_model.layers.19.proj.v.lora_A -> {weight: [32, 3072]}\n",
227
+ "diffusion_model.layers.19.proj.v.lora_B -> {weight: [3072, 32]}\n",
228
+ "diffusion_model.layers.2.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
229
+ "diffusion_model.layers.2.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
230
+ "diffusion_model.layers.2.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
231
+ "diffusion_model.layers.2.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
232
+ "diffusion_model.layers.2.proj.k.lora_A -> {weight: [32, 3072]}\n",
233
+ "diffusion_model.layers.2.proj.k.lora_B -> {weight: [3072, 32]}\n",
234
+ "diffusion_model.layers.2.proj.o.lora_A -> {weight: [32, 3072]}\n",
235
+ "diffusion_model.layers.2.proj.o.lora_B -> {weight: [3072, 32]}\n",
236
+ "diffusion_model.layers.2.proj.q.lora_A -> {weight: [32, 3072]}\n",
237
+ "diffusion_model.layers.2.proj.q.lora_B -> {weight: [3072, 32]}\n",
238
+ "diffusion_model.layers.2.proj.v.lora_A -> {weight: [32, 3072]}\n",
239
+ "diffusion_model.layers.2.proj.v.lora_B -> {weight: [3072, 32]}\n",
240
+ "diffusion_model.layers.20.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
241
+ "diffusion_model.layers.20.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
242
+ "diffusion_model.layers.20.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
243
+ "diffusion_model.layers.20.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
244
+ "diffusion_model.layers.20.proj.k.lora_A -> {weight: [32, 3072]}\n",
245
+ "diffusion_model.layers.20.proj.k.lora_B -> {weight: [3072, 32]}\n",
246
+ "diffusion_model.layers.20.proj.o.lora_A -> {weight: [32, 3072]}\n",
247
+ "diffusion_model.layers.20.proj.o.lora_B -> {weight: [3072, 32]}\n",
248
+ "diffusion_model.layers.20.proj.q.lora_A -> {weight: [32, 3072]}\n",
249
+ "diffusion_model.layers.20.proj.q.lora_B -> {weight: [3072, 32]}\n",
250
+ "diffusion_model.layers.20.proj.v.lora_A -> {weight: [32, 3072]}\n",
251
+ "diffusion_model.layers.20.proj.v.lora_B -> {weight: [3072, 32]}\n",
252
+ "diffusion_model.layers.21.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
253
+ "diffusion_model.layers.21.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
254
+ "diffusion_model.layers.21.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
255
+ "diffusion_model.layers.21.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
256
+ "diffusion_model.layers.21.proj.k.lora_A -> {weight: [32, 3072]}\n",
257
+ "diffusion_model.layers.21.proj.k.lora_B -> {weight: [3072, 32]}\n",
258
+ "diffusion_model.layers.21.proj.o.lora_A -> {weight: [32, 3072]}\n",
259
+ "diffusion_model.layers.21.proj.o.lora_B -> {weight: [3072, 32]}\n",
260
+ "diffusion_model.layers.21.proj.q.lora_A -> {weight: [32, 3072]}\n",
261
+ "diffusion_model.layers.21.proj.q.lora_B -> {weight: [3072, 32]}\n",
262
+ "diffusion_model.layers.21.proj.v.lora_A -> {weight: [32, 3072]}\n",
263
+ "diffusion_model.layers.21.proj.v.lora_B -> {weight: [3072, 32]}\n",
264
+ "diffusion_model.layers.23.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
265
+ "diffusion_model.layers.23.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
266
+ "diffusion_model.layers.23.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
267
+ "diffusion_model.layers.23.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
268
+ "diffusion_model.layers.23.proj.k.lora_A -> {weight: [32, 3072]}\n",
269
+ "diffusion_model.layers.23.proj.k.lora_B -> {weight: [3072, 32]}\n",
270
+ "diffusion_model.layers.23.proj.o.lora_A -> {weight: [32, 3072]}\n",
271
+ "diffusion_model.layers.23.proj.o.lora_B -> {weight: [3072, 32]}\n",
272
+ "diffusion_model.layers.23.proj.q.lora_A -> {weight: [32, 3072]}\n",
273
+ "diffusion_model.layers.23.proj.q.lora_B -> {weight: [3072, 32]}\n",
274
+ "diffusion_model.layers.23.proj.v.lora_A -> {weight: [32, 3072]}\n",
275
+ "diffusion_model.layers.23.proj.v.lora_B -> {weight: [3072, 32]}\n",
276
+ "diffusion_model.layers.24.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
277
+ "diffusion_model.layers.24.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
278
+ "diffusion_model.layers.24.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
279
+ "diffusion_model.layers.24.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
280
+ "diffusion_model.layers.24.proj.k.lora_A -> {weight: [32, 3072]}\n",
281
+ "diffusion_model.layers.24.proj.k.lora_B -> {weight: [3072, 32]}\n",
282
+ "diffusion_model.layers.24.proj.o.lora_A -> {weight: [32, 3072]}\n",
283
+ "diffusion_model.layers.24.proj.o.lora_B -> {weight: [3072, 32]}\n",
284
+ "diffusion_model.layers.24.proj.q.lora_A -> {weight: [32, 3072]}\n",
285
+ "diffusion_model.layers.24.proj.q.lora_B -> {weight: [3072, 32]}\n",
286
+ "diffusion_model.layers.24.proj.v.lora_A -> {weight: [32, 3072]}\n",
287
+ "diffusion_model.layers.24.proj.v.lora_B -> {weight: [3072, 32]}\n",
288
+ "diffusion_model.layers.25.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
289
+ "diffusion_model.layers.25.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
290
+ "diffusion_model.layers.25.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
291
+ "diffusion_model.layers.25.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
292
+ "diffusion_model.layers.25.proj.k.lora_A -> {weight: [32, 3072]}\n",
293
+ "diffusion_model.layers.25.proj.k.lora_B -> {weight: [3072, 32]}\n",
294
+ "diffusion_model.layers.25.proj.o.lora_A -> {weight: [32, 3072]}\n",
295
+ "diffusion_model.layers.25.proj.o.lora_B -> {weight: [3072, 32]}\n",
296
+ "diffusion_model.layers.25.proj.q.lora_A -> {weight: [32, 3072]}\n",
297
+ "diffusion_model.layers.25.proj.q.lora_B -> {weight: [3072, 32]}\n",
298
+ "diffusion_model.layers.25.proj.v.lora_A -> {weight: [32, 3072]}\n",
299
+ "diffusion_model.layers.25.proj.v.lora_B -> {weight: [3072, 32]}\n",
300
+ "diffusion_model.layers.26.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
301
+ "diffusion_model.layers.26.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
302
+ "diffusion_model.layers.26.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
303
+ "diffusion_model.layers.26.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
304
+ "diffusion_model.layers.26.proj.k.lora_A -> {weight: [32, 3072]}\n",
305
+ "diffusion_model.layers.26.proj.k.lora_B -> {weight: [3072, 32]}\n",
306
+ "diffusion_model.layers.26.proj.o.lora_A -> {weight: [32, 3072]}\n",
307
+ "diffusion_model.layers.26.proj.o.lora_B -> {weight: [3072, 32]}\n",
308
+ "diffusion_model.layers.26.proj.q.lora_A -> {weight: [32, 3072]}\n",
309
+ "diffusion_model.layers.26.proj.q.lora_B -> {weight: [3072, 32]}\n",
310
+ "diffusion_model.layers.26.proj.v.lora_A -> {weight: [32, 3072]}\n",
311
+ "diffusion_model.layers.26.proj.v.lora_B -> {weight: [3072, 32]}\n",
312
+ "diffusion_model.layers.27.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
313
+ "diffusion_model.layers.27.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
314
+ "diffusion_model.layers.27.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
315
+ "diffusion_model.layers.27.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
316
+ "diffusion_model.layers.27.proj.k.lora_A -> {weight: [32, 3072]}\n",
317
+ "diffusion_model.layers.27.proj.k.lora_B -> {weight: [3072, 32]}\n",
318
+ "diffusion_model.layers.27.proj.o.lora_A -> {weight: [32, 3072]}\n",
319
+ "diffusion_model.layers.27.proj.o.lora_B -> {weight: [3072, 32]}\n",
320
+ "diffusion_model.layers.27.proj.q.lora_A -> {weight: [32, 3072]}\n",
321
+ "diffusion_model.layers.27.proj.q.lora_B -> {weight: [3072, 32]}\n",
322
+ "diffusion_model.layers.27.proj.v.lora_A -> {weight: [32, 3072]}\n",
323
+ "diffusion_model.layers.27.proj.v.lora_B -> {weight: [3072, 32]}\n",
324
+ "diffusion_model.layers.28.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
325
+ "diffusion_model.layers.28.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
326
+ "diffusion_model.layers.28.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
327
+ "diffusion_model.layers.28.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
328
+ "diffusion_model.layers.28.proj.k.lora_A -> {weight: [32, 3072]}\n",
329
+ "diffusion_model.layers.28.proj.k.lora_B -> {weight: [3072, 32]}\n",
330
+ "diffusion_model.layers.28.proj.o.lora_A -> {weight: [32, 3072]}\n",
331
+ "diffusion_model.layers.28.proj.o.lora_B -> {weight: [3072, 32]}\n",
332
+ "diffusion_model.layers.28.proj.q.lora_A -> {weight: [32, 3072]}\n",
333
+ "diffusion_model.layers.28.proj.q.lora_B -> {weight: [3072, 32]}\n",
334
+ "diffusion_model.layers.28.proj.v.lora_A -> {weight: [32, 3072]}\n",
335
+ "diffusion_model.layers.28.proj.v.lora_B -> {weight: [3072, 32]}\n",
336
+ "diffusion_model.layers.29.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
337
+ "diffusion_model.layers.29.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
338
+ "diffusion_model.layers.29.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
339
+ "diffusion_model.layers.29.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
340
+ "diffusion_model.layers.29.proj.k.lora_A -> {weight: [32, 3072]}\n",
341
+ "diffusion_model.layers.29.proj.k.lora_B -> {weight: [3072, 32]}\n",
342
+ "diffusion_model.layers.29.proj.o.lora_A -> {weight: [32, 3072]}\n",
343
+ "diffusion_model.layers.29.proj.o.lora_B -> {weight: [3072, 32]}\n",
344
+ "diffusion_model.layers.29.proj.q.lora_A -> {weight: [32, 3072]}\n",
345
+ "diffusion_model.layers.29.proj.q.lora_B -> {weight: [3072, 32]}\n",
346
+ "diffusion_model.layers.29.proj.v.lora_A -> {weight: [32, 3072]}\n",
347
+ "diffusion_model.layers.29.proj.v.lora_B -> {weight: [3072, 32]}\n",
348
+ "diffusion_model.layers.3.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
349
+ "diffusion_model.layers.3.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
350
+ "diffusion_model.layers.3.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
351
+ "diffusion_model.layers.3.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
352
+ "diffusion_model.layers.3.proj.k.lora_A -> {weight: [32, 3072]}\n",
353
+ "diffusion_model.layers.3.proj.k.lora_B -> {weight: [3072, 32]}\n",
354
+ "diffusion_model.layers.3.proj.o.lora_A -> {weight: [32, 3072]}\n",
355
+ "diffusion_model.layers.3.proj.o.lora_B -> {weight: [3072, 32]}\n",
356
+ "diffusion_model.layers.3.proj.q.lora_A -> {weight: [32, 3072]}\n",
357
+ "diffusion_model.layers.3.proj.q.lora_B -> {weight: [3072, 32]}\n",
358
+ "diffusion_model.layers.3.proj.v.lora_A -> {weight: [32, 3072]}\n",
359
+ "diffusion_model.layers.3.proj.v.lora_B -> {weight: [3072, 32]}\n",
360
+ "diffusion_model.layers.30.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
361
+ "diffusion_model.layers.30.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
362
+ "diffusion_model.layers.30.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
363
+ "diffusion_model.layers.30.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
364
+ "diffusion_model.layers.30.proj.k.lora_A -> {weight: [32, 3072]}\n",
365
+ "diffusion_model.layers.30.proj.k.lora_B -> {weight: [3072, 32]}\n",
366
+ "diffusion_model.layers.30.proj.o.lora_A -> {weight: [32, 3072]}\n",
367
+ "diffusion_model.layers.30.proj.o.lora_B -> {weight: [3072, 32]}\n",
368
+ "diffusion_model.layers.30.proj.q.lora_A -> {weight: [32, 3072]}\n",
369
+ "diffusion_model.layers.30.proj.q.lora_B -> {weight: [3072, 32]}\n",
370
+ "diffusion_model.layers.30.proj.v.lora_A -> {weight: [32, 3072]}\n",
371
+ "diffusion_model.layers.30.proj.v.lora_B -> {weight: [3072, 32]}\n",
372
+ "diffusion_model.layers.31.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
373
+ "diffusion_model.layers.31.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
374
+ "diffusion_model.layers.31.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
375
+ "diffusion_model.layers.31.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
376
+ "diffusion_model.layers.31.proj.k.lora_A -> {weight: [32, 3072]}\n",
377
+ "diffusion_model.layers.31.proj.k.lora_B -> {weight: [3072, 32]}\n",
378
+ "diffusion_model.layers.31.proj.o.lora_A -> {weight: [32, 3072]}\n",
379
+ "diffusion_model.layers.31.proj.o.lora_B -> {weight: [3072, 32]}\n",
380
+ "diffusion_model.layers.31.proj.q.lora_A -> {weight: [32, 3072]}\n",
381
+ "diffusion_model.layers.31.proj.q.lora_B -> {weight: [3072, 32]}\n",
382
+ "diffusion_model.layers.31.proj.v.lora_A -> {weight: [32, 3072]}\n",
383
+ "diffusion_model.layers.31.proj.v.lora_B -> {weight: [3072, 32]}\n",
384
+ "diffusion_model.layers.32.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
385
+ "diffusion_model.layers.32.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
386
+ "diffusion_model.layers.32.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
387
+ "diffusion_model.layers.32.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
388
+ "diffusion_model.layers.32.proj.k.lora_A -> {weight: [32, 3072]}\n",
389
+ "diffusion_model.layers.32.proj.k.lora_B -> {weight: [3072, 32]}\n",
390
+ "diffusion_model.layers.32.proj.o.lora_A -> {weight: [32, 3072]}\n",
391
+ "diffusion_model.layers.32.proj.o.lora_B -> {weight: [3072, 32]}\n",
392
+ "diffusion_model.layers.32.proj.q.lora_A -> {weight: [32, 3072]}\n",
393
+ "diffusion_model.layers.32.proj.q.lora_B -> {weight: [3072, 32]}\n",
394
+ "diffusion_model.layers.32.proj.v.lora_A -> {weight: [32, 3072]}\n",
395
+ "diffusion_model.layers.32.proj.v.lora_B -> {weight: [3072, 32]}\n",
396
+ "diffusion_model.layers.33.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
397
+ "diffusion_model.layers.33.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
398
+ "diffusion_model.layers.33.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
399
+ "diffusion_model.layers.33.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
400
+ "diffusion_model.layers.33.proj.k.lora_A -> {weight: [32, 3072]}\n",
401
+ "diffusion_model.layers.33.proj.k.lora_B -> {weight: [3072, 32]}\n",
402
+ "diffusion_model.layers.33.proj.o.lora_A -> {weight: [32, 3072]}\n",
403
+ "diffusion_model.layers.33.proj.o.lora_B -> {weight: [3072, 32]}\n",
404
+ "diffusion_model.layers.33.proj.q.lora_A -> {weight: [32, 3072]}\n",
405
+ "diffusion_model.layers.33.proj.q.lora_B -> {weight: [3072, 32]}\n",
406
+ "diffusion_model.layers.33.proj.v.lora_A -> {weight: [32, 3072]}\n",
407
+ "diffusion_model.layers.33.proj.v.lora_B -> {weight: [3072, 32]}\n",
408
+ "diffusion_model.layers.34.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
409
+ "diffusion_model.layers.34.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
410
+ "diffusion_model.layers.34.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
411
+ "diffusion_model.layers.34.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
412
+ "diffusion_model.layers.34.proj.k.lora_A -> {weight: [32, 3072]}\n",
413
+ "diffusion_model.layers.34.proj.k.lora_B -> {weight: [3072, 32]}\n",
414
+ "diffusion_model.layers.34.proj.o.lora_A -> {weight: [32, 3072]}\n",
415
+ "diffusion_model.layers.34.proj.o.lora_B -> {weight: [3072, 32]}\n",
416
+ "diffusion_model.layers.34.proj.q.lora_A -> {weight: [32, 3072]}\n",
417
+ "diffusion_model.layers.34.proj.q.lora_B -> {weight: [3072, 32]}\n",
418
+ "diffusion_model.layers.34.proj.v.lora_A -> {weight: [32, 3072]}\n",
419
+ "diffusion_model.layers.34.proj.v.lora_B -> {weight: [3072, 32]}\n",
420
+ "diffusion_model.layers.35.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
421
+ "diffusion_model.layers.35.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
422
+ "diffusion_model.layers.35.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
423
+ "diffusion_model.layers.35.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
424
+ "diffusion_model.layers.35.proj.k.lora_A -> {weight: [32, 3072]}\n",
425
+ "diffusion_model.layers.35.proj.k.lora_B -> {weight: [3072, 32]}\n",
426
+ "diffusion_model.layers.35.proj.o.lora_A -> {weight: [32, 3072]}\n",
427
+ "diffusion_model.layers.35.proj.o.lora_B -> {weight: [3072, 32]}\n",
428
+ "diffusion_model.layers.35.proj.q.lora_A -> {weight: [32, 3072]}\n",
429
+ "diffusion_model.layers.35.proj.q.lora_B -> {weight: [3072, 32]}\n",
430
+ "diffusion_model.layers.35.proj.v.lora_A -> {weight: [32, 3072]}\n",
431
+ "diffusion_model.layers.35.proj.v.lora_B -> {weight: [3072, 32]}\n",
432
+ "diffusion_model.layers.36.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
433
+ "diffusion_model.layers.36.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
434
+ "diffusion_model.layers.36.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
435
+ "diffusion_model.layers.36.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
436
+ "diffusion_model.layers.36.proj.k.lora_A -> {weight: [32, 3072]}\n",
437
+ "diffusion_model.layers.36.proj.k.lora_B -> {weight: [3072, 32]}\n",
438
+ "diffusion_model.layers.36.proj.o.lora_A -> {weight: [32, 3072]}\n",
439
+ "diffusion_model.layers.36.proj.o.lora_B -> {weight: [3072, 32]}\n",
440
+ "diffusion_model.layers.36.proj.q.lora_A -> {weight: [32, 3072]}\n",
441
+ "diffusion_model.layers.36.proj.q.lora_B -> {weight: [3072, 32]}\n",
442
+ "diffusion_model.layers.36.proj.v.lora_A -> {weight: [32, 3072]}\n",
443
+ "diffusion_model.layers.36.proj.v.lora_B -> {weight: [3072, 32]}\n",
444
+ "diffusion_model.layers.37.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
445
+ "diffusion_model.layers.37.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
446
+ "diffusion_model.layers.37.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
447
+ "diffusion_model.layers.37.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
448
+ "diffusion_model.layers.37.proj.k.lora_A -> {weight: [32, 3072]}\n",
449
+ "diffusion_model.layers.37.proj.k.lora_B -> {weight: [3072, 32]}\n",
450
+ "diffusion_model.layers.37.proj.o.lora_A -> {weight: [32, 3072]}\n",
451
+ "diffusion_model.layers.37.proj.o.lora_B -> {weight: [3072, 32]}\n",
452
+ "diffusion_model.layers.37.proj.q.lora_A -> {weight: [32, 3072]}\n",
453
+ "diffusion_model.layers.37.proj.q.lora_B -> {weight: [3072, 32]}\n",
454
+ "diffusion_model.layers.37.proj.v.lora_A -> {weight: [32, 3072]}\n",
455
+ "diffusion_model.layers.37.proj.v.lora_B -> {weight: [3072, 32]}\n",
456
+ "diffusion_model.layers.38.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
457
+ "diffusion_model.layers.38.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
458
+ "diffusion_model.layers.38.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
459
+ "diffusion_model.layers.38.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
460
+ "diffusion_model.layers.38.proj.k.lora_A -> {weight: [32, 3072]}\n",
461
+ "diffusion_model.layers.38.proj.k.lora_B -> {weight: [3072, 32]}\n",
462
+ "diffusion_model.layers.38.proj.o.lora_A -> {weight: [32, 3072]}\n",
463
+ "diffusion_model.layers.38.proj.o.lora_B -> {weight: [3072, 32]}\n",
464
+ "diffusion_model.layers.38.proj.q.lora_A -> {weight: [32, 3072]}\n",
465
+ "diffusion_model.layers.38.proj.q.lora_B -> {weight: [3072, 32]}\n",
466
+ "diffusion_model.layers.38.proj.v.lora_A -> {weight: [32, 3072]}\n",
467
+ "diffusion_model.layers.38.proj.v.lora_B -> {weight: [3072, 32]}\n",
468
+ "diffusion_model.layers.39.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
469
+ "diffusion_model.layers.39.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
470
+ "diffusion_model.layers.39.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
471
+ "diffusion_model.layers.39.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
472
+ "diffusion_model.layers.39.proj.k.lora_A -> {weight: [32, 3072]}\n",
473
+ "diffusion_model.layers.39.proj.k.lora_B -> {weight: [3072, 32]}\n",
474
+ "diffusion_model.layers.39.proj.o.lora_A -> {weight: [32, 3072]}\n",
475
+ "diffusion_model.layers.39.proj.o.lora_B -> {weight: [3072, 32]}\n",
476
+ "diffusion_model.layers.39.proj.q.lora_A -> {weight: [32, 3072]}\n",
477
+ "diffusion_model.layers.39.proj.q.lora_B -> {weight: [3072, 32]}\n",
478
+ "diffusion_model.layers.39.proj.v.lora_A -> {weight: [32, 3072]}\n",
479
+ "diffusion_model.layers.39.proj.v.lora_B -> {weight: [3072, 32]}\n",
480
+ "diffusion_model.layers.4.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
481
+ "diffusion_model.layers.4.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
482
+ "diffusion_model.layers.4.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
483
+ "diffusion_model.layers.4.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
484
+ "diffusion_model.layers.4.proj.k.lora_A -> {weight: [32, 3072]}\n",
485
+ "diffusion_model.layers.4.proj.k.lora_B -> {weight: [3072, 32]}\n",
486
+ "diffusion_model.layers.4.proj.o.lora_A -> {weight: [32, 3072]}\n",
487
+ "diffusion_model.layers.4.proj.o.lora_B -> {weight: [3072, 32]}\n",
488
+ "diffusion_model.layers.4.proj.q.lora_A -> {weight: [32, 3072]}\n",
489
+ "diffusion_model.layers.4.proj.q.lora_B -> {weight: [3072, 32]}\n",
490
+ "diffusion_model.layers.4.proj.v.lora_A -> {weight: [32, 3072]}\n",
491
+ "diffusion_model.layers.4.proj.v.lora_B -> {weight: [3072, 32]}\n",
492
+ "diffusion_model.layers.40.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
493
+ "diffusion_model.layers.40.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
494
+ "diffusion_model.layers.40.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
495
+ "diffusion_model.layers.40.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
496
+ "diffusion_model.layers.40.proj.k.lora_A -> {weight: [32, 3072]}\n",
497
+ "diffusion_model.layers.40.proj.k.lora_B -> {weight: [3072, 32]}\n",
498
+ "diffusion_model.layers.40.proj.o.lora_A -> {weight: [32, 3072]}\n",
499
+ "diffusion_model.layers.40.proj.o.lora_B -> {weight: [3072, 32]}\n",
500
+ "diffusion_model.layers.40.proj.q.lora_A -> {weight: [32, 3072]}\n",
501
+ "diffusion_model.layers.40.proj.q.lora_B -> {weight: [3072, 32]}\n",
502
+ "diffusion_model.layers.40.proj.v.lora_A -> {weight: [32, 3072]}\n",
503
+ "diffusion_model.layers.40.proj.v.lora_B -> {weight: [3072, 32]}\n",
504
+ "diffusion_model.layers.41.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
505
+ "diffusion_model.layers.41.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
506
+ "diffusion_model.layers.41.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
507
+ "diffusion_model.layers.41.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
508
+ "diffusion_model.layers.41.proj.k.lora_A -> {weight: [32, 3072]}\n",
509
+ "diffusion_model.layers.41.proj.k.lora_B -> {weight: [3072, 32]}\n",
510
+ "diffusion_model.layers.41.proj.o.lora_A -> {weight: [32, 3072]}\n",
511
+ "diffusion_model.layers.41.proj.o.lora_B -> {weight: [3072, 32]}\n",
512
+ "diffusion_model.layers.41.proj.q.lora_A -> {weight: [32, 3072]}\n",
513
+ "diffusion_model.layers.41.proj.q.lora_B -> {weight: [3072, 32]}\n",
514
+ "diffusion_model.layers.41.proj.v.lora_A -> {weight: [32, 3072]}\n",
515
+ "diffusion_model.layers.41.proj.v.lora_B -> {weight: [3072, 32]}\n",
516
+ "diffusion_model.layers.5.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
517
+ "diffusion_model.layers.5.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
518
+ "diffusion_model.layers.5.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
519
+ "diffusion_model.layers.5.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
520
+ "diffusion_model.layers.5.proj.k.lora_A -> {weight: [32, 3072]}\n",
521
+ "diffusion_model.layers.5.proj.k.lora_B -> {weight: [3072, 32]}\n",
522
+ "diffusion_model.layers.5.proj.o.lora_A -> {weight: [32, 3072]}\n",
523
+ "diffusion_model.layers.5.proj.o.lora_B -> {weight: [3072, 32]}\n",
524
+ "diffusion_model.layers.5.proj.q.lora_A -> {weight: [32, 3072]}\n",
525
+ "diffusion_model.layers.5.proj.q.lora_B -> {weight: [3072, 32]}\n",
526
+ "diffusion_model.layers.5.proj.v.lora_A -> {weight: [32, 3072]}\n",
527
+ "diffusion_model.layers.5.proj.v.lora_B -> {weight: [3072, 32]}\n",
528
+ "diffusion_model.layers.6.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
529
+ "diffusion_model.layers.6.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
530
+ "diffusion_model.layers.6.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
531
+ "diffusion_model.layers.6.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
532
+ "diffusion_model.layers.6.proj.k.lora_A -> {weight: [32, 3072]}\n",
533
+ "diffusion_model.layers.6.proj.k.lora_B -> {weight: [3072, 32]}\n",
534
+ "diffusion_model.layers.6.proj.o.lora_A -> {weight: [32, 3072]}\n",
535
+ "diffusion_model.layers.6.proj.o.lora_B -> {weight: [3072, 32]}\n",
536
+ "diffusion_model.layers.6.proj.q.lora_A -> {weight: [32, 3072]}\n",
537
+ "diffusion_model.layers.6.proj.q.lora_B -> {weight: [3072, 32]}\n",
538
+ "diffusion_model.layers.6.proj.v.lora_A -> {weight: [32, 3072]}\n",
539
+ "diffusion_model.layers.6.proj.v.lora_B -> {weight: [3072, 32]}\n",
540
+ "diffusion_model.layers.7.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
541
+ "diffusion_model.layers.7.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
542
+ "diffusion_model.layers.7.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
543
+ "diffusion_model.layers.7.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
544
+ "diffusion_model.layers.7.proj.k.lora_A -> {weight: [32, 3072]}\n",
545
+ "diffusion_model.layers.7.proj.k.lora_B -> {weight: [3072, 32]}\n",
546
+ "diffusion_model.layers.7.proj.o.lora_A -> {weight: [32, 3072]}\n",
547
+ "diffusion_model.layers.7.proj.o.lora_B -> {weight: [3072, 32]}\n",
548
+ "diffusion_model.layers.7.proj.q.lora_A -> {weight: [32, 3072]}\n",
549
+ "diffusion_model.layers.7.proj.q.lora_B -> {weight: [3072, 32]}\n",
550
+ "diffusion_model.layers.7.proj.v.lora_A -> {weight: [32, 3072]}\n",
551
+ "diffusion_model.layers.7.proj.v.lora_B -> {weight: [3072, 32]}\n",
552
+ "diffusion_model.layers.8.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
553
+ "diffusion_model.layers.8.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
554
+ "diffusion_model.layers.8.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
555
+ "diffusion_model.layers.8.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
556
+ "diffusion_model.layers.8.proj.k.lora_A -> {weight: [32, 3072]}\n",
557
+ "diffusion_model.layers.8.proj.k.lora_B -> {weight: [3072, 32]}\n",
558
+ "diffusion_model.layers.8.proj.o.lora_A -> {weight: [32, 3072]}\n",
559
+ "diffusion_model.layers.8.proj.o.lora_B -> {weight: [3072, 32]}\n",
560
+ "diffusion_model.layers.8.proj.q.lora_A -> {weight: [32, 3072]}\n",
561
+ "diffusion_model.layers.8.proj.q.lora_B -> {weight: [3072, 32]}\n",
562
+ "diffusion_model.layers.8.proj.v.lora_A -> {weight: [32, 3072]}\n",
563
+ "diffusion_model.layers.8.proj.v.lora_B -> {weight: [3072, 32]}\n",
564
+ "diffusion_model.layers.9.proj.ffn_down.lora_A -> {weight: [32, 8192]}\n",
565
+ "diffusion_model.layers.9.proj.ffn_down.lora_B -> {weight: [3072, 32]}\n",
566
+ "diffusion_model.layers.9.proj.ffn_up_gate.lora_A -> {weight: [32, 3072]}\n",
567
+ "diffusion_model.layers.9.proj.ffn_up_gate.lora_B -> {weight: [16384, 32]}\n",
568
+ "diffusion_model.layers.9.proj.k.lora_A -> {weight: [32, 3072]}\n",
569
+ "diffusion_model.layers.9.proj.k.lora_B -> {weight: [3072, 32]}\n",
570
+ "diffusion_model.layers.9.proj.o.lora_A -> {weight: [32, 3072]}\n",
571
+ "diffusion_model.layers.9.proj.o.lora_B -> {weight: [3072, 32]}\n",
572
+ "diffusion_model.layers.9.proj.q.lora_A -> {weight: [32, 3072]}\n",
573
+ "diffusion_model.layers.9.proj.q.lora_B -> {weight: [3072, 32]}\n",
574
+ "diffusion_model.layers.9.proj.v.lora_A -> {weight: [32, 3072]}\n",
575
+ "diffusion_model.layers.9.proj.v.lora_B -> {weight: [3072, 32]}\n"
576
+ ]
577
+ }
578
+ ]
579
+ }
580
+ ]
581
+ }