Meriem-DH commited on
Commit
a5ebc53
·
verified ·
1 Parent(s): d86f251

Model save

Browse files
adapter_config.json CHANGED
@@ -28,224 +28,224 @@
28
  "peft_type": "LORA",
29
  "peft_version": "0.18.1",
30
  "qalora_group_size": 16,
31
- "r": 16,
32
  "rank_pattern": {},
33
  "revision": null,
34
  "target_modules": [
35
- "language_model.layers.8.mlp.up_proj",
36
- "18.mlp.gate_proj",
37
- "language_model.layers.13.mlp.gate_proj",
38
- "16.self_attn.o_proj",
39
- "language_model.layers.14.self_attn.q_proj",
40
- "29.self_attn.o_proj",
41
- "language_model.layers.12.mlp.down_proj",
42
- "26.self_attn.q_proj",
43
- "per_layer_model_projection",
44
- "language_model.layers.4.self_attn.o_proj",
45
- "language_model.layers.10.mlp.gate_proj",
46
- "25.mlp.gate_proj",
47
- "33.self_attn.q_proj",
48
- "language_model.layers.14.self_attn.k_proj",
49
  "per_layer_projection",
50
- "17.mlp.up_proj",
51
- "language_model.layers.7.self_attn.o_proj",
52
- "24.mlp.up_proj",
53
- "30.mlp.up_proj",
54
- "language_model.layers.11.mlp.down_proj",
55
- "language_model.layers.6.mlp.up_proj",
56
- "language_model.layers.7.self_attn.q_proj",
57
- "32.mlp.up_proj",
58
- "language_model.layers.6.self_attn.o_proj",
59
- "26.mlp.up_proj",
60
- "19.self_attn.o_proj",
61
- "28.mlp.down_proj",
62
- "30.self_attn.o_proj",
63
- "18.self_attn.q_proj",
64
- "language_model.layers.1.self_attn.v_proj",
65
- "language_model.layers.4.self_attn.v_proj",
 
 
 
 
 
 
 
66
  "23.mlp.gate_proj",
67
- "language_model.layers.8.mlp.down_proj",
68
- "18.self_attn.o_proj",
69
- "language_model.layers.15.self_attn.q_proj",
70
- "26.mlp.gate_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
71
  "language_model.layers.15.mlp.up_proj",
72
- "language_model.layers.15.mlp.down_proj",
73
- "language_model.layers.2.self_attn.k_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
74
  "language_model.layers.12.self_attn.q_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
75
  "33.mlp.gate_proj",
76
- "language_model.layers.9.self_attn.k_proj",
 
 
77
  "language_model.layers.13.mlp.down_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
78
  "25.mlp.up_proj",
79
- "21.self_attn.q_proj",
80
- "language_model.layers.2.mlp.down_proj",
81
- "language_model.layers.11.self_attn.o_proj",
82
- "34.mlp.down_proj",
83
- "language_model.layers.15.mlp.gate_proj",
84
- "language_model.layers.2.mlp.up_proj",
85
- "language_model.layers.7.mlp.gate_proj",
86
- "language_model.layers.10.self_attn.o_proj",
87
- "27.self_attn.q_proj",
88
- "31.self_attn.o_proj",
89
- "language_model.layers.0.self_attn.q_proj",
90
  "language_model.layers.7.mlp.down_proj",
 
 
91
  "27.mlp.gate_proj",
92
- "30.mlp.gate_proj",
93
- "language_model.layers.12.self_attn.o_proj",
94
  "23.self_attn.o_proj",
95
- "language_model.layers.14.mlp.up_proj",
 
 
 
 
 
 
 
 
 
 
96
  "19.mlp.down_proj",
97
- "28.mlp.gate_proj",
98
- "language_model.layers.14.self_attn.o_proj",
99
- "language_model.layers.11.self_attn.k_proj",
100
- "30.self_attn.q_proj",
101
- "17.self_attn.o_proj",
102
- "language_model.layers.2.self_attn.q_proj",
 
103
  "22.mlp.up_proj",
104
- "language_model.layers.6.self_attn.v_proj",
105
- "language_model.layers.14.mlp.down_proj",
106
- "language_model.layers.4.self_attn.k_proj",
107
- "output_proj",
108
- "32.self_attn.q_proj",
109
- "linear",
110
  "language_model.layers.14.self_attn.v_proj",
111
- "language_model.layers.5.self_attn.q_proj",
112
- "language_model.layers.4.self_attn.q_proj",
113
- "language_model.layers.15.self_attn.o_proj",
114
- "language_model.layers.2.self_attn.o_proj",
115
- "31.mlp.down_proj",
116
- "32.mlp.down_proj",
117
- "29.mlp.up_proj",
118
- "language_model.layers.10.mlp.down_proj",
119
- "language_model.layers.4.mlp.gate_proj",
120
- "language_model.layers.7.self_attn.k_proj",
121
- "language_model.layers.9.mlp.up_proj",
122
- "embedding_projection",
123
- "language_model.layers.3.mlp.gate_proj",
124
- "30.mlp.down_proj",
125
- "29.self_attn.q_proj",
126
- "language_model.layers.0.mlp.up_proj",
127
- "24.mlp.down_proj",
128
  "32.self_attn.o_proj",
129
- "21.mlp.gate_proj",
130
- "language_model.layers.11.mlp.gate_proj",
131
- "28.self_attn.q_proj",
132
- "language_model.layers.8.self_attn.o_proj",
133
- "language_model.layers.13.self_attn.o_proj",
134
- "22.self_attn.o_proj",
135
- "25.mlp.down_proj",
136
- "27.self_attn.o_proj",
137
- "25.self_attn.o_proj",
138
- "language_model.layers.5.mlp.up_proj",
139
- "26.mlp.down_proj",
140
- "34.self_attn.o_proj",
141
- "31.mlp.up_proj",
142
- "18.mlp.up_proj",
143
- "16.self_attn.q_proj",
144
- "23.self_attn.q_proj",
145
- "language_model.layers.10.self_attn.q_proj",
146
- "language_model.layers.5.self_attn.o_proj",
147
- "language_model.layers.12.mlp.up_proj",
148
- "language_model.layers.3.self_attn.k_proj",
149
- "language_model.layers.1.mlp.down_proj",
150
- "20.self_attn.q_proj",
151
- "22.mlp.down_proj",
152
- "language_model.layers.13.self_attn.v_proj",
153
- "34.self_attn.q_proj",
154
- "language_model.layers.3.mlp.down_proj",
155
- "input_proj_linear",
156
- "language_model.layers.5.self_attn.k_proj",
157
- "language_model.layers.0.self_attn.k_proj",
158
- "20.self_attn.o_proj",
159
- "language_model.layers.3.self_attn.o_proj",
160
- "language_model.layers.3.mlp.up_proj",
161
  "language_model.layers.5.self_attn.v_proj",
162
- "relative_k_proj",
163
- "language_model.layers.8.self_attn.q_proj",
 
 
 
164
  "19.self_attn.q_proj",
165
- "29.mlp.down_proj",
166
- "22.self_attn.q_proj",
167
- "language_model.layers.8.mlp.gate_proj",
168
- "33.mlp.up_proj",
169
- "language_model.layers.14.mlp.gate_proj",
170
- "language_model.layers.5.mlp.gate_proj",
171
  "language_model.layers.10.self_attn.v_proj",
172
- "language_model.layers.7.self_attn.v_proj",
173
- "language_model.layers.7.mlp.up_proj",
174
  "language_model.layers.6.mlp.down_proj",
175
- "17.self_attn.q_proj",
176
- "language_model.layers.6.self_attn.k_proj",
177
- "language_model.layers.4.mlp.up_proj",
178
- "language_model.layers.0.self_attn.v_proj",
179
- "28.mlp.up_proj",
180
- "language_model.layers.12.self_attn.v_proj",
181
- "language_model.layers.13.self_attn.k_proj",
182
- "language_model.layers.2.mlp.gate_proj",
183
- "29.mlp.gate_proj",
184
- "language_model.layers.0.self_attn.o_proj",
185
- "33.self_attn.o_proj",
186
  "24.self_attn.o_proj",
187
- "24.mlp.gate_proj",
188
- "26.self_attn.o_proj",
189
- "16.mlp.up_proj",
190
- "language_model.layers.3.self_attn.v_proj",
191
- "language_model.layers.4.mlp.down_proj",
192
- "language_model.layers.0.mlp.gate_proj",
193
- "language_model.layers.9.self_attn.q_proj",
194
- "language_model.layers.10.self_attn.k_proj",
195
- "17.mlp.gate_proj",
196
- "34.mlp.up_proj",
197
- "language_model.layers.9.self_attn.o_proj",
198
- "language_model.layers.3.self_attn.q_proj",
199
- "17.mlp.down_proj",
200
- "language_model.layers.9.self_attn.v_proj",
201
- "16.mlp.gate_proj",
202
- "language_model.layers.9.mlp.down_proj",
203
- "33.mlp.down_proj",
204
- "language_model.layers.12.self_attn.k_proj",
205
- "language_model.layers.10.mlp.up_proj",
206
- "25.self_attn.q_proj",
207
- "language_model.layers.1.mlp.up_proj",
208
- "18.mlp.down_proj",
209
- "20.mlp.gate_proj",
210
- "per_layer_input_gate",
211
- "language_model.layers.0.mlp.down_proj",
212
- "34.mlp.gate_proj",
213
  "31.self_attn.q_proj",
214
- "24.self_attn.q_proj",
215
  "23.mlp.up_proj",
216
- "language_model.layers.11.self_attn.v_proj",
217
- "27.mlp.up_proj",
218
- "20.mlp.down_proj",
219
- "language_model.layers.9.mlp.gate_proj",
220
- "16.mlp.down_proj",
221
- "language_model.layers.8.self_attn.k_proj",
222
- "language_model.layers.1.mlp.gate_proj",
223
- "language_model.layers.11.mlp.up_proj",
224
  "19.mlp.up_proj",
225
- "21.mlp.up_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
226
  "language_model.layers.11.self_attn.q_proj",
227
- "20.mlp.up_proj",
228
- "21.self_attn.o_proj",
229
- "language_model.layers.13.self_attn.q_proj",
230
- "language_model.layers.2.self_attn.v_proj",
231
- "19.mlp.gate_proj",
232
- "21.mlp.down_proj",
233
- "language_model.layers.1.self_attn.q_proj",
234
- "language_model.layers.13.mlp.up_proj",
235
- "32.mlp.gate_proj",
236
- "language_model.layers.1.self_attn.o_proj",
237
  "language_model.layers.6.mlp.gate_proj",
238
- "22.mlp.gate_proj",
239
- "language_model.layers.1.self_attn.k_proj",
240
- "language_model.layers.8.self_attn.v_proj",
241
- "language_model.layers.6.self_attn.q_proj",
242
- "input_proj",
243
- "23.mlp.down_proj",
244
- "27.mlp.down_proj",
245
- "31.mlp.gate_proj",
246
- "28.self_attn.o_proj",
247
  "language_model.layers.5.mlp.down_proj",
248
- "language_model.layers.12.mlp.gate_proj"
 
 
 
 
 
 
 
 
 
 
 
 
249
  ],
250
  "target_parameters": null,
251
  "task_type": "CAUSAL_LM",
 
28
  "peft_type": "LORA",
29
  "peft_version": "0.18.1",
30
  "qalora_group_size": 16,
31
+ "r": 8,
32
  "rank_pattern": {},
33
  "revision": null,
34
  "target_modules": [
 
 
 
 
 
 
 
 
 
 
 
 
 
 
35
  "per_layer_projection",
36
+ "17.self_attn.q_proj",
37
+ "language_model.layers.10.self_attn.k_proj",
38
+ "language_model.layers.9.mlp.gate_proj",
39
+ "language_model.layers.14.mlp.gate_proj",
40
+ "language_model.layers.1.self_attn.q_proj",
41
+ "26.mlp.down_proj",
42
+ "language_model.layers.14.self_attn.o_proj",
43
+ "language_model.layers.5.mlp.up_proj",
44
+ "input_proj",
45
+ "17.mlp.down_proj",
46
+ "25.self_attn.o_proj",
47
+ "language_model.layers.3.self_attn.v_proj",
48
+ "22.mlp.gate_proj",
49
+ "language_model.layers.9.self_attn.o_proj",
50
+ "33.mlp.down_proj",
51
+ "language_model.layers.1.mlp.gate_proj",
52
+ "language_model.layers.0.mlp.up_proj",
53
+ "language_model.layers.7.self_attn.v_proj",
54
+ "linear",
55
+ "28.self_attn.q_proj",
56
+ "31.mlp.up_proj",
57
+ "language_model.layers.3.mlp.down_proj",
58
+ "24.self_attn.q_proj",
59
  "23.mlp.gate_proj",
60
+ "25.self_attn.q_proj",
61
+ "language_model.layers.13.self_attn.k_proj",
62
+ "language_model.layers.15.self_attn.o_proj",
63
+ "33.self_attn.o_proj",
64
+ "language_model.layers.9.self_attn.q_proj",
65
+ "31.self_attn.o_proj",
66
+ "language_model.layers.8.self_attn.k_proj",
67
+ "language_model.layers.14.mlp.down_proj",
68
+ "19.self_attn.o_proj",
69
+ "21.mlp.gate_proj",
70
+ "27.self_attn.q_proj",
71
+ "language_model.layers.8.mlp.gate_proj",
72
+ "language_model.layers.5.self_attn.o_proj",
73
+ "24.mlp.up_proj",
74
+ "34.self_attn.q_proj",
75
+ "language_model.layers.7.mlp.up_proj",
76
+ "21.self_attn.q_proj",
77
+ "18.mlp.up_proj",
78
+ "16.mlp.down_proj",
79
+ "30.mlp.down_proj",
80
+ "16.mlp.gate_proj",
81
+ "language_model.layers.12.self_attn.v_proj",
82
+ "language_model.layers.0.self_attn.v_proj",
83
+ "language_model.layers.10.mlp.up_proj",
84
+ "language_model.layers.14.mlp.up_proj",
85
  "language_model.layers.15.mlp.up_proj",
86
+ "language_model.layers.0.mlp.down_proj",
87
+ "language_model.layers.10.mlp.down_proj",
88
+ "20.mlp.up_proj",
89
+ "26.mlp.gate_proj",
90
+ "24.mlp.gate_proj",
91
+ "language_model.layers.0.self_attn.k_proj",
92
+ "17.mlp.up_proj",
93
+ "20.self_attn.q_proj",
94
+ "language_model.layers.10.self_attn.o_proj",
95
+ "language_model.layers.11.mlp.gate_proj",
96
+ "16.self_attn.o_proj",
97
+ "embedding_projection",
98
+ "34.mlp.gate_proj",
99
+ "language_model.layers.0.self_attn.o_proj",
100
+ "language_model.layers.3.self_attn.k_proj",
101
+ "output_proj",
102
+ "23.mlp.down_proj",
103
+ "28.mlp.up_proj",
104
+ "language_model.layers.4.self_attn.v_proj",
105
+ "30.self_attn.q_proj",
106
+ "language_model.layers.13.self_attn.o_proj",
107
+ "28.mlp.gate_proj",
108
  "language_model.layers.12.self_attn.q_proj",
109
+ "language_model.layers.3.self_attn.q_proj",
110
+ "language_model.layers.6.self_attn.v_proj",
111
+ "21.mlp.up_proj",
112
+ "17.mlp.gate_proj",
113
+ "29.mlp.down_proj",
114
+ "20.self_attn.o_proj",
115
+ "32.self_attn.q_proj",
116
+ "26.self_attn.q_proj",
117
+ "16.self_attn.q_proj",
118
+ "language_model.layers.1.mlp.down_proj",
119
+ "30.mlp.gate_proj",
120
+ "28.mlp.down_proj",
121
+ "per_layer_input_gate",
122
+ "language_model.layers.2.self_attn.q_proj",
123
+ "language_model.layers.4.self_attn.k_proj",
124
+ "34.mlp.down_proj",
125
  "33.mlp.gate_proj",
126
+ "relative_k_proj",
127
+ "language_model.layers.9.mlp.up_proj",
128
+ "31.mlp.gate_proj",
129
  "language_model.layers.13.mlp.down_proj",
130
+ "language_model.layers.15.self_attn.q_proj",
131
+ "language_model.layers.7.self_attn.q_proj",
132
+ "language_model.layers.4.mlp.down_proj",
133
+ "28.self_attn.o_proj",
134
+ "29.self_attn.q_proj",
135
+ "21.self_attn.o_proj",
136
+ "language_model.layers.14.self_attn.q_proj",
137
+ "language_model.layers.3.mlp.gate_proj",
138
+ "language_model.layers.8.mlp.down_proj",
139
+ "34.self_attn.o_proj",
140
+ "language_model.layers.5.self_attn.k_proj",
141
+ "language_model.layers.12.mlp.gate_proj",
142
+ "language_model.layers.7.self_attn.o_proj",
143
+ "27.mlp.down_proj",
144
+ "18.self_attn.o_proj",
145
+ "language_model.layers.2.self_attn.k_proj",
146
+ "language_model.layers.13.mlp.up_proj",
147
  "25.mlp.up_proj",
148
+ "language_model.layers.2.self_attn.o_proj",
 
 
 
 
 
 
 
 
 
 
149
  "language_model.layers.7.mlp.down_proj",
150
+ "language_model.layers.13.self_attn.v_proj",
151
+ "language_model.layers.9.mlp.down_proj",
152
  "27.mlp.gate_proj",
153
+ "language_model.layers.1.self_attn.k_proj",
 
154
  "23.self_attn.o_proj",
155
+ "per_layer_model_projection",
156
+ "language_model.layers.15.mlp.down_proj",
157
+ "language_model.layers.12.self_attn.o_proj",
158
+ "22.self_attn.o_proj",
159
+ "language_model.layers.8.mlp.up_proj",
160
+ "language_model.layers.4.mlp.up_proj",
161
+ "19.mlp.gate_proj",
162
+ "22.self_attn.q_proj",
163
+ "language_model.layers.6.self_attn.q_proj",
164
+ "language_model.layers.1.self_attn.v_proj",
165
+ "language_model.layers.5.mlp.gate_proj",
166
  "19.mlp.down_proj",
167
+ "27.self_attn.o_proj",
168
+ "language_model.layers.8.self_attn.q_proj",
169
+ "26.self_attn.o_proj",
170
+ "34.mlp.up_proj",
171
+ "20.mlp.down_proj",
172
+ "language_model.layers.4.self_attn.q_proj",
173
+ "language_model.layers.6.self_attn.o_proj",
174
  "22.mlp.up_proj",
 
 
 
 
 
 
175
  "language_model.layers.14.self_attn.v_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
176
  "32.self_attn.o_proj",
177
+ "27.mlp.up_proj",
178
+ "language_model.layers.2.mlp.down_proj",
179
+ "language_model.layers.4.mlp.gate_proj",
180
+ "language_model.layers.2.self_attn.v_proj",
181
+ "18.self_attn.q_proj",
182
+ "language_model.layers.10.mlp.gate_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
183
  "language_model.layers.5.self_attn.v_proj",
184
+ "18.mlp.down_proj",
185
+ "20.mlp.gate_proj",
186
+ "language_model.layers.13.self_attn.q_proj",
187
+ "language_model.layers.11.self_attn.k_proj",
188
+ "30.self_attn.o_proj",
189
  "19.self_attn.q_proj",
190
+ "32.mlp.gate_proj",
191
+ "language_model.layers.0.mlp.gate_proj",
192
+ "language_model.layers.1.self_attn.o_proj",
193
+ "language_model.layers.10.self_attn.q_proj",
194
+ "language_model.layers.8.self_attn.o_proj",
 
195
  "language_model.layers.10.self_attn.v_proj",
 
 
196
  "language_model.layers.6.mlp.down_proj",
 
 
 
 
 
 
 
 
 
 
 
197
  "24.self_attn.o_proj",
198
+ "33.self_attn.q_proj",
199
+ "26.mlp.up_proj",
200
+ "language_model.layers.12.mlp.down_proj",
201
+ "language_model.layers.9.self_attn.k_proj",
202
+ "30.mlp.up_proj",
203
+ "language_model.layers.11.mlp.up_proj",
204
+ "23.self_attn.q_proj",
205
+ "17.self_attn.o_proj",
206
+ "language_model.layers.0.self_attn.q_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
207
  "31.self_attn.q_proj",
208
+ "25.mlp.gate_proj",
209
  "23.mlp.up_proj",
210
+ "31.mlp.down_proj",
211
+ "language_model.layers.9.self_attn.v_proj",
212
+ "language_model.layers.1.mlp.up_proj",
213
+ "language_model.layers.12.self_attn.k_proj",
214
+ "22.mlp.down_proj",
215
+ "24.mlp.down_proj",
 
 
216
  "19.mlp.up_proj",
217
+ "language_model.layers.14.self_attn.k_proj",
218
+ "29.mlp.gate_proj",
219
+ "29.self_attn.o_proj",
220
+ "18.mlp.gate_proj",
221
+ "language_model.layers.3.self_attn.o_proj",
222
+ "language_model.layers.4.self_attn.o_proj",
223
+ "language_model.layers.6.self_attn.k_proj",
224
+ "language_model.layers.15.mlp.gate_proj",
225
+ "32.mlp.up_proj",
226
+ "32.mlp.down_proj",
227
+ "language_model.layers.2.mlp.gate_proj",
228
+ "29.mlp.up_proj",
229
+ "language_model.layers.7.self_attn.k_proj",
230
+ "language_model.layers.12.mlp.up_proj",
231
  "language_model.layers.11.self_attn.q_proj",
232
+ "language_model.layers.11.self_attn.v_proj",
 
 
 
 
 
 
 
 
 
233
  "language_model.layers.6.mlp.gate_proj",
234
+ "language_model.layers.3.mlp.up_proj",
 
 
 
 
 
 
 
 
235
  "language_model.layers.5.mlp.down_proj",
236
+ "language_model.layers.13.mlp.gate_proj",
237
+ "16.mlp.up_proj",
238
+ "33.mlp.up_proj",
239
+ "language_model.layers.7.mlp.gate_proj",
240
+ "language_model.layers.5.self_attn.q_proj",
241
+ "21.mlp.down_proj",
242
+ "25.mlp.down_proj",
243
+ "language_model.layers.6.mlp.up_proj",
244
+ "language_model.layers.8.self_attn.v_proj",
245
+ "input_proj_linear",
246
+ "language_model.layers.11.self_attn.o_proj",
247
+ "language_model.layers.2.mlp.up_proj",
248
+ "language_model.layers.11.mlp.down_proj"
249
  ],
250
  "target_parameters": null,
251
  "task_type": "CAUSAL_LM",
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:249ad1e7bbab3244aa1c58d024eb40cb9fb3b74ac0b71195a85ece752db6142b
3
- size 1686620368
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40cd7413b7ae7b560eb7110fb4894be8b019eafe55242d3661e5857f2077e88f
3
+ size 1648698552
runs/Apr09_22-03-00_14d6a481b527/events.out.tfevents.1775772180.14d6a481b527.3695.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d56a46056b6bb87f069084cfd8966156784e13dd22940ca94117a7b11d7e947d
3
+ size 22278
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:426e908b596497a7a983fe4f185f4f8d689e703c3a552be4ee16d8030e55bc0e
3
  size 5713
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:49a44881f2e2c8906480ffe626f2e9d8568f3048d66fa093da311cb648a3a4b6
3
  size 5713