codeShare commited on
Commit
af3c39a
·
verified ·
1 Parent(s): 1d64249

Upload 95_percent Gemma 4 LoRA adapter to gemma4_lora/

Browse files
gemma4_lora/adapter_config.json CHANGED
@@ -32,220 +32,220 @@
32
  "rank_pattern": {},
33
  "revision": null,
34
  "target_modules": [
35
- "language_model.layers.1.self_attn.k_proj",
36
- "19.self_attn.o_proj",
37
- "22.mlp.down_proj",
38
- "language_model.layers.4.mlp.gate_proj",
39
- "language_model.layers.6.mlp.gate_proj",
40
- "20.self_attn.o_proj",
41
- "16.mlp.up_proj",
42
- "language_model.layers.13.self_attn.k_proj",
43
- "24.self_attn.o_proj",
44
- "language_model.layers.13.mlp.up_proj",
 
45
  "18.self_attn.o_proj",
46
- "language_model.layers.13.self_attn.v_proj",
47
- "language_model.layers.2.mlp.gate_proj",
48
- "34.mlp.up_proj",
49
- "language_model.layers.11.mlp.down_proj",
50
- "19.mlp.up_proj",
51
- "18.mlp.up_proj",
52
- "language_model.layers.7.self_attn.q_proj",
53
- "25.self_attn.o_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
54
  "language_model.layers.8.mlp.up_proj",
55
- "language_model.layers.15.mlp.gate_proj",
56
- "language_model.layers.5.self_attn.v_proj",
57
- "language_model.layers.3.mlp.up_proj",
58
- "22.self_attn.o_proj",
59
- "31.mlp.gate_proj",
60
- "30.mlp.down_proj",
61
- "language_model.layers.4.mlp.up_proj",
62
- "20.mlp.gate_proj",
63
- "21.mlp.gate_proj",
64
- "24.mlp.up_proj",
65
- "language_model.layers.2.mlp.down_proj",
66
- "language_model.layers.15.mlp.down_proj",
67
- "language_model.layers.4.self_attn.q_proj",
68
- "per_layer_input_gate",
69
- "language_model.layers.12.self_attn.o_proj",
70
- "language_model.layers.6.self_attn.v_proj",
71
- "language_model.layers.7.self_attn.k_proj",
72
- "24.self_attn.q_proj",
73
  "language_model.layers.9.self_attn.o_proj",
74
- "20.mlp.down_proj",
75
- "31.self_attn.q_proj",
76
- "30.self_attn.o_proj",
77
- "language_model.layers.14.self_attn.k_proj",
78
- "22.self_attn.q_proj",
79
- "language_model.layers.1.mlp.down_proj",
80
- "18.mlp.down_proj",
81
  "output_proj",
82
- "16.mlp.down_proj",
83
- "32.mlp.down_proj",
84
- "language_model.layers.7.mlp.gate_proj",
85
- "language_model.layers.12.mlp.gate_proj",
86
- "26.self_attn.q_proj",
87
- "language_model.layers.3.self_attn.k_proj",
88
- "33.mlp.gate_proj",
89
- "input_proj_linear",
90
- "language_model.layers.15.mlp.up_proj",
91
- "language_model.layers.6.self_attn.q_proj",
92
- "language_model.layers.5.mlp.up_proj",
93
- "21.mlp.down_proj",
94
- "33.self_attn.q_proj",
95
  "23.mlp.down_proj",
96
- "embedding_projection",
97
- "17.mlp.down_proj",
98
- "26.mlp.up_proj",
99
- "33.self_attn.o_proj",
100
- "language_model.layers.7.self_attn.o_proj",
101
- "31.self_attn.o_proj",
102
- "21.self_attn.o_proj",
103
  "26.mlp.gate_proj",
104
- "language_model.layers.12.mlp.down_proj",
105
- "language_model.layers.5.mlp.gate_proj",
106
- "language_model.layers.0.self_attn.o_proj",
107
- "language_model.layers.1.self_attn.v_proj",
108
- "language_model.layers.3.self_attn.q_proj",
109
- "language_model.layers.1.self_attn.o_proj",
110
- "language_model.layers.9.mlp.up_proj",
111
- "21.self_attn.q_proj",
112
- "29.mlp.up_proj",
113
- "language_model.layers.0.self_attn.v_proj",
 
 
114
  "34.self_attn.q_proj",
115
- "language_model.layers.6.self_attn.k_proj",
116
- "28.self_attn.q_proj",
117
- "27.self_attn.q_proj",
118
- "language_model.layers.6.mlp.up_proj",
119
- "19.mlp.down_proj",
120
- "language_model.layers.13.self_attn.o_proj",
121
- "33.mlp.down_proj",
122
- "27.self_attn.o_proj",
123
- "language_model.layers.1.self_attn.q_proj",
124
  "language_model.layers.5.mlp.down_proj",
125
- "language_model.layers.3.mlp.gate_proj",
126
- "32.mlp.up_proj",
127
- "34.self_attn.o_proj",
128
- "20.mlp.up_proj",
129
- "language_model.layers.13.mlp.gate_proj",
130
- "language_model.layers.1.mlp.gate_proj",
 
 
 
 
 
 
 
 
 
 
131
  "17.self_attn.o_proj",
132
- "language_model.layers.4.mlp.down_proj",
133
- "language_model.layers.5.self_attn.k_proj",
134
- "18.mlp.gate_proj",
135
- "language_model.layers.4.self_attn.k_proj",
136
- "language_model.layers.10.self_attn.o_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
137
  "language_model.layers.10.self_attn.v_proj",
138
- "25.self_attn.q_proj",
 
 
 
 
 
 
 
 
 
 
 
 
139
  "27.mlp.gate_proj",
140
- "language_model.layers.2.self_attn.v_proj",
141
- "language_model.layers.2.self_attn.o_proj",
142
- "language_model.layers.13.self_attn.q_proj",
143
- "language_model.layers.0.self_attn.k_proj",
144
- "language_model.layers.10.mlp.up_proj",
145
- "language_model.layers.13.mlp.down_proj",
146
  "26.mlp.down_proj",
147
- "language_model.layers.2.self_attn.k_proj",
148
- "language_model.layers.9.mlp.down_proj",
 
149
  "20.self_attn.q_proj",
150
- "language_model.layers.2.self_attn.q_proj",
151
- "31.mlp.up_proj",
152
- "17.self_attn.q_proj",
153
- "28.mlp.down_proj",
154
- "language_model.layers.11.self_attn.k_proj",
155
- "language_model.layers.7.mlp.up_proj",
156
- "language_model.layers.0.mlp.gate_proj",
157
- "16.mlp.gate_proj",
158
- "17.mlp.up_proj",
159
- "input_proj",
160
- "language_model.layers.14.self_attn.o_proj",
161
- "30.self_attn.q_proj",
162
- "language_model.layers.3.self_attn.v_proj",
163
- "26.self_attn.o_proj",
164
- "language_model.layers.2.mlp.up_proj",
165
- "27.mlp.down_proj",
166
- "language_model.layers.10.mlp.down_proj",
167
  "language_model.layers.8.mlp.down_proj",
168
  "language_model.layers.7.self_attn.v_proj",
169
- "30.mlp.gate_proj",
170
- "23.mlp.gate_proj",
171
- "22.mlp.up_proj",
172
- "language_model.layers.4.self_attn.o_proj",
173
- "32.self_attn.q_proj",
174
- "22.mlp.gate_proj",
175
- "per_layer_projection",
176
- "17.mlp.gate_proj",
177
- "34.mlp.gate_proj",
178
- "32.self_attn.o_proj",
179
- "relative_k_proj",
180
- "language_model.layers.5.self_attn.o_proj",
181
  "language_model.layers.8.self_attn.k_proj",
182
- "language_model.layers.10.self_attn.k_proj",
183
- "29.self_attn.o_proj",
184
- "language_model.layers.6.mlp.down_proj",
185
- "31.mlp.down_proj",
186
- "language_model.layers.12.mlp.up_proj",
187
- "language_model.layers.9.self_attn.v_proj",
188
- "21.mlp.up_proj",
189
- "language_model.layers.5.self_attn.q_proj",
190
- "29.mlp.gate_proj",
191
- "16.self_attn.o_proj",
192
  "19.self_attn.q_proj",
193
- "language_model.layers.0.mlp.down_proj",
194
- "language_model.layers.12.self_attn.k_proj",
195
- "language_model.layers.9.self_attn.q_proj",
196
- "language_model.layers.7.mlp.down_proj",
197
- "24.mlp.down_proj",
198
- "language_model.layers.14.mlp.gate_proj",
199
- "language_model.layers.14.self_attn.q_proj",
200
- "16.self_attn.q_proj",
201
- "language_model.layers.8.self_attn.o_proj",
202
- "language_model.layers.15.self_attn.q_proj",
203
- "language_model.layers.9.mlp.gate_proj",
204
- "language_model.layers.8.self_attn.v_proj",
205
- "language_model.layers.6.self_attn.o_proj",
206
  "language_model.layers.8.self_attn.q_proj",
207
- "18.self_attn.q_proj",
208
- "language_model.layers.3.self_attn.o_proj",
209
- "34.mlp.down_proj",
210
- "28.self_attn.o_proj",
211
- "language_model.layers.4.self_attn.v_proj",
212
- "25.mlp.gate_proj",
213
- "language_model.layers.14.self_attn.v_proj",
214
- "28.mlp.up_proj",
215
- "language_model.layers.0.mlp.up_proj",
216
- "27.mlp.up_proj",
217
- "25.mlp.down_proj",
218
- "language_model.layers.11.self_attn.v_proj",
219
  "language_model.layers.11.mlp.up_proj",
220
- "language_model.layers.14.mlp.up_proj",
221
- "language_model.layers.1.mlp.up_proj",
222
- "language_model.layers.8.mlp.gate_proj",
223
- "language_model.layers.10.mlp.gate_proj",
224
- "29.mlp.down_proj",
225
- "linear",
226
- "language_model.layers.0.self_attn.q_proj",
227
- "language_model.layers.15.self_attn.o_proj",
228
- "30.mlp.up_proj",
 
 
 
 
 
 
 
 
229
  "language_model.layers.11.self_attn.o_proj",
230
- "language_model.layers.12.self_attn.v_proj",
231
- "33.mlp.up_proj",
232
- "29.self_attn.q_proj",
233
- "24.mlp.gate_proj",
234
- "23.self_attn.q_proj",
235
- "23.mlp.up_proj",
236
- "19.mlp.gate_proj",
 
 
 
 
 
237
  "language_model.layers.9.self_attn.k_proj",
238
- "language_model.layers.10.self_attn.q_proj",
239
- "language_model.layers.12.self_attn.q_proj",
240
- "language_model.layers.11.self_attn.q_proj",
241
- "language_model.layers.3.mlp.down_proj",
 
 
 
 
242
  "23.self_attn.o_proj",
243
- "25.mlp.up_proj",
244
- "28.mlp.gate_proj",
245
- "per_layer_model_projection",
246
- "32.mlp.gate_proj",
247
- "language_model.layers.11.mlp.gate_proj",
248
- "language_model.layers.14.mlp.down_proj"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
249
  ],
250
  "target_parameters": null,
251
  "task_type": "CAUSAL_LM",
 
32
  "rank_pattern": {},
33
  "revision": null,
34
  "target_modules": [
35
+ "language_model.layers.14.self_attn.v_proj",
36
+ "language_model.layers.9.self_attn.v_proj",
37
+ "language_model.layers.13.self_attn.o_proj",
38
+ "32.mlp.gate_proj",
39
+ "32.mlp.up_proj",
40
+ "26.self_attn.q_proj",
41
+ "19.mlp.gate_proj",
42
+ "language_model.layers.4.self_attn.o_proj",
43
+ "18.mlp.gate_proj",
44
+ "28.mlp.down_proj",
45
+ "30.self_attn.o_proj",
46
  "18.self_attn.o_proj",
47
+ "16.mlp.up_proj",
48
+ "33.self_attn.q_proj",
49
+ "17.mlp.up_proj",
50
+ "language_model.layers.11.self_attn.k_proj",
51
+ "language_model.layers.9.mlp.down_proj",
52
+ "language_model.layers.4.self_attn.k_proj",
53
+ "23.self_attn.q_proj",
54
+ "28.self_attn.q_proj",
55
+ "22.mlp.up_proj",
56
+ "31.mlp.up_proj",
57
+ "language_model.layers.7.mlp.down_proj",
58
+ "language_model.layers.8.self_attn.o_proj",
59
+ "language_model.layers.1.self_attn.v_proj",
60
+ "language_model.layers.10.self_attn.k_proj",
61
+ "language_model.layers.10.self_attn.q_proj",
62
+ "31.self_attn.o_proj",
63
+ "language_model.layers.5.self_attn.q_proj",
64
+ "language_model.layers.12.self_attn.k_proj",
65
+ "language_model.layers.0.self_attn.v_proj",
66
+ "20.mlp.up_proj",
67
+ "language_model.layers.6.self_attn.o_proj",
68
+ "language_model.layers.0.mlp.gate_proj",
69
+ "20.self_attn.o_proj",
70
+ "language_model.layers.14.self_attn.o_proj",
71
  "language_model.layers.8.mlp.up_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
72
  "language_model.layers.9.self_attn.o_proj",
73
+ "language_model.layers.8.self_attn.v_proj",
 
 
 
 
 
 
74
  "output_proj",
75
+ "21.mlp.up_proj",
76
+ "language_model.layers.0.mlp.down_proj",
77
+ "34.self_attn.o_proj",
78
+ "language_model.layers.13.mlp.gate_proj",
79
+ "language_model.layers.10.self_attn.o_proj",
80
+ "28.mlp.up_proj",
81
+ "30.mlp.up_proj",
82
+ "31.mlp.gate_proj",
 
 
 
 
 
83
  "23.mlp.down_proj",
 
 
 
 
 
 
 
84
  "26.mlp.gate_proj",
85
+ "21.mlp.down_proj",
86
+ "23.mlp.gate_proj",
87
+ "language_model.layers.0.mlp.up_proj",
88
+ "language_model.layers.4.mlp.gate_proj",
89
+ "language_model.layers.15.mlp.down_proj",
90
+ "language_model.layers.3.self_attn.o_proj",
91
+ "24.self_attn.q_proj",
92
+ "25.mlp.gate_proj",
93
+ "language_model.layers.3.mlp.gate_proj",
94
+ "language_model.layers.3.mlp.down_proj",
95
+ "19.mlp.up_proj",
96
+ "language_model.layers.14.mlp.up_proj",
97
  "34.self_attn.q_proj",
98
+ "language_model.layers.11.mlp.down_proj",
99
+ "language_model.layers.2.self_attn.q_proj",
100
+ "34.mlp.up_proj",
 
 
 
 
 
 
101
  "language_model.layers.5.mlp.down_proj",
102
+ "per_layer_model_projection",
103
+ "language_model.layers.10.mlp.gate_proj",
104
+ "33.self_attn.o_proj",
105
+ "16.self_attn.q_proj",
106
+ "34.mlp.gate_proj",
107
+ "25.mlp.down_proj",
108
+ "32.self_attn.q_proj",
109
+ "24.self_attn.o_proj",
110
+ "33.mlp.up_proj",
111
+ "language_model.layers.10.mlp.up_proj",
112
+ "language_model.layers.0.self_attn.o_proj",
113
+ "34.mlp.down_proj",
114
+ "16.mlp.down_proj",
115
+ "29.mlp.up_proj",
116
+ "language_model.layers.15.self_attn.o_proj",
117
+ "30.self_attn.q_proj",
118
  "17.self_attn.o_proj",
119
+ "language_model.layers.3.self_attn.k_proj",
120
+ "language_model.layers.1.self_attn.k_proj",
121
+ "per_layer_projection",
122
+ "language_model.layers.14.mlp.gate_proj",
123
+ "language_model.layers.15.mlp.up_proj",
124
+ "language_model.layers.7.self_attn.o_proj",
125
+ "16.mlp.gate_proj",
126
+ "language_model.layers.2.self_attn.o_proj",
127
+ "17.self_attn.q_proj",
128
+ "24.mlp.gate_proj",
129
+ "language_model.layers.7.self_attn.k_proj",
130
+ "28.self_attn.o_proj",
131
+ "27.mlp.up_proj",
132
+ "language_model.layers.6.self_attn.v_proj",
133
+ "language_model.layers.12.mlp.up_proj",
134
+ "29.self_attn.o_proj",
135
+ "language_model.layers.9.mlp.gate_proj",
136
+ "language_model.layers.11.mlp.gate_proj",
137
+ "language_model.layers.13.self_attn.v_proj",
138
  "language_model.layers.10.self_attn.v_proj",
139
+ "language_model.layers.15.mlp.gate_proj",
140
+ "language_model.layers.1.mlp.down_proj",
141
+ "language_model.layers.7.self_attn.q_proj",
142
+ "21.mlp.gate_proj",
143
+ "language_model.layers.7.mlp.gate_proj",
144
+ "29.mlp.down_proj",
145
+ "language_model.layers.3.mlp.up_proj",
146
+ "23.mlp.up_proj",
147
+ "language_model.layers.13.mlp.up_proj",
148
+ "30.mlp.down_proj",
149
+ "input_proj",
150
+ "24.mlp.down_proj",
151
+ "language_model.layers.8.mlp.gate_proj",
152
  "27.mlp.gate_proj",
153
+ "language_model.layers.1.mlp.gate_proj",
 
 
 
 
 
154
  "26.mlp.down_proj",
155
+ "language_model.layers.11.self_attn.q_proj",
156
+ "per_layer_input_gate",
157
+ "language_model.layers.15.self_attn.q_proj",
158
  "20.self_attn.q_proj",
159
+ "language_model.layers.5.self_attn.o_proj",
160
+ "language_model.layers.4.mlp.up_proj",
161
+ "31.self_attn.q_proj",
162
+ "language_model.layers.2.mlp.gate_proj",
163
+ "20.mlp.gate_proj",
164
+ "21.self_attn.q_proj",
165
+ "language_model.layers.11.self_attn.v_proj",
166
+ "language_model.layers.6.mlp.gate_proj",
167
+ "language_model.layers.3.self_attn.q_proj",
168
+ "22.self_attn.q_proj",
169
+ "28.mlp.gate_proj",
170
+ "29.mlp.gate_proj",
171
+ "language_model.layers.6.self_attn.k_proj",
172
+ "17.mlp.gate_proj",
173
+ "language_model.layers.12.mlp.down_proj",
 
 
174
  "language_model.layers.8.mlp.down_proj",
175
  "language_model.layers.7.self_attn.v_proj",
176
+ "language_model.layers.9.self_attn.q_proj",
177
+ "input_proj_linear",
178
+ "language_model.layers.12.self_attn.v_proj",
 
 
 
 
 
 
 
 
 
179
  "language_model.layers.8.self_attn.k_proj",
180
+ "language_model.layers.1.mlp.up_proj",
 
 
 
 
 
 
 
 
 
181
  "19.self_attn.q_proj",
182
+ "25.mlp.up_proj",
183
+ "language_model.layers.2.self_attn.v_proj",
184
+ "language_model.layers.1.self_attn.q_proj",
185
+ "language_model.layers.4.mlp.down_proj",
186
+ "language_model.layers.5.mlp.up_proj",
 
 
 
 
 
 
 
 
187
  "language_model.layers.8.self_attn.q_proj",
 
 
 
 
 
 
 
 
 
 
 
 
188
  "language_model.layers.11.mlp.up_proj",
189
+ "language_model.layers.13.self_attn.k_proj",
190
+ "33.mlp.down_proj",
191
+ "18.mlp.down_proj",
192
+ "language_model.layers.2.mlp.up_proj",
193
+ "language_model.layers.5.mlp.gate_proj",
194
+ "21.self_attn.o_proj",
195
+ "language_model.layers.14.mlp.down_proj",
196
+ "22.self_attn.o_proj",
197
+ "32.mlp.down_proj",
198
+ "language_model.layers.10.mlp.down_proj",
199
+ "embedding_projection",
200
+ "25.self_attn.o_proj",
201
+ "24.mlp.up_proj",
202
+ "19.mlp.down_proj",
203
+ "20.mlp.down_proj",
204
+ "language_model.layers.6.mlp.down_proj",
205
+ "language_model.layers.4.self_attn.v_proj",
206
  "language_model.layers.11.self_attn.o_proj",
207
+ "language_model.layers.12.self_attn.o_proj",
208
+ "27.mlp.down_proj",
209
+ "language_model.layers.0.self_attn.q_proj",
210
+ "27.self_attn.o_proj",
211
+ "32.self_attn.o_proj",
212
+ "language_model.layers.5.self_attn.v_proj",
213
+ "33.mlp.gate_proj",
214
+ "language_model.layers.2.mlp.down_proj",
215
+ "26.self_attn.o_proj",
216
+ "19.self_attn.o_proj",
217
+ "language_model.layers.7.mlp.up_proj",
218
+ "language_model.layers.0.self_attn.k_proj",
219
  "language_model.layers.9.self_attn.k_proj",
220
+ "27.self_attn.q_proj",
221
+ "language_model.layers.4.self_attn.q_proj",
222
+ "language_model.layers.13.mlp.down_proj",
223
+ "18.mlp.up_proj",
224
+ "relative_k_proj",
225
+ "31.mlp.down_proj",
226
+ "linear",
227
+ "language_model.layers.9.mlp.up_proj",
228
  "23.self_attn.o_proj",
229
+ "language_model.layers.5.self_attn.k_proj",
230
+ "17.mlp.down_proj",
231
+ "26.mlp.up_proj",
232
+ "language_model.layers.6.mlp.up_proj",
233
+ "language_model.layers.13.self_attn.q_proj",
234
+ "language_model.layers.2.self_attn.k_proj",
235
+ "language_model.layers.12.mlp.gate_proj",
236
+ "25.self_attn.q_proj",
237
+ "language_model.layers.3.self_attn.v_proj",
238
+ "22.mlp.gate_proj",
239
+ "language_model.layers.14.self_attn.q_proj",
240
+ "29.self_attn.q_proj",
241
+ "16.self_attn.o_proj",
242
+ "22.mlp.down_proj",
243
+ "language_model.layers.1.self_attn.o_proj",
244
+ "language_model.layers.14.self_attn.k_proj",
245
+ "18.self_attn.q_proj",
246
+ "language_model.layers.6.self_attn.q_proj",
247
+ "language_model.layers.12.self_attn.q_proj",
248
+ "30.mlp.gate_proj"
249
  ],
250
  "target_parameters": null,
251
  "task_type": "CAUSAL_LM",
gemma4_lora/adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:35d8dc30124a1bdf360931a39927e4c6b9ea8b6b3a3a5f147b7a8daba05f9c43
3
  size 151844848
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e9473003ac19644fdfff066ec4d633e843127aada38dc4fbc5f399dcbf6be7c9
3
  size 151844848