sirr commited on
Commit
59be610
·
1 Parent(s): 1b50ae4

Training in progress, step 2

Browse files
Files changed (3) hide show
  1. adapter_config.json +163 -0
  2. adapter_model.bin +3 -0
  3. training_args.bin +3 -0
adapter_config.json ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "gpt2-large",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 8,
11
+ "lora_dropout": 0.1,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "transformer.h.0.attn.c_attn",
18
+ "transformer.h.0.attn.c_proj",
19
+ "transformer.h.1.attn.c_attn",
20
+ "transformer.h.1.attn.c_proj",
21
+ "transformer.h.2.attn.c_attn",
22
+ "transformer.h.2.attn.c_proj",
23
+ "transformer.h.3.attn.c_attn",
24
+ "transformer.h.3.attn.c_proj",
25
+ "transformer.h.4.attn.c_attn",
26
+ "transformer.h.4.attn.c_proj",
27
+ "transformer.h.5.attn.c_attn",
28
+ "transformer.h.5.attn.c_proj",
29
+ "transformer.h.6.attn.c_attn",
30
+ "transformer.h.6.attn.c_proj",
31
+ "transformer.h.7.attn.c_attn",
32
+ "transformer.h.7.attn.c_proj",
33
+ "transformer.h.8.attn.c_attn",
34
+ "transformer.h.8.attn.c_proj",
35
+ "transformer.h.9.attn.c_attn",
36
+ "transformer.h.9.attn.c_proj",
37
+ "transformer.h.10.attn.c_attn",
38
+ "transformer.h.10.attn.c_proj",
39
+ "transformer.h.11.attn.c_attn",
40
+ "transformer.h.11.attn.c_proj",
41
+ "transformer.h.12.attn.c_attn",
42
+ "transformer.h.12.attn.c_proj",
43
+ "transformer.h.13.attn.c_attn",
44
+ "transformer.h.13.attn.c_proj",
45
+ "transformer.h.14.attn.c_attn",
46
+ "transformer.h.14.attn.c_proj",
47
+ "transformer.h.15.attn.c_attn",
48
+ "transformer.h.15.attn.c_proj",
49
+ "transformer.h.16.attn.c_attn",
50
+ "transformer.h.16.attn.c_proj",
51
+ "transformer.h.17.attn.c_attn",
52
+ "transformer.h.17.attn.c_proj",
53
+ "transformer.h.18.attn.c_attn",
54
+ "transformer.h.18.attn.c_proj",
55
+ "transformer.h.19.attn.c_attn",
56
+ "transformer.h.19.attn.c_proj",
57
+ "transformer.h.20.attn.c_attn",
58
+ "transformer.h.20.attn.c_proj",
59
+ "transformer.h.21.attn.c_attn",
60
+ "transformer.h.21.attn.c_proj",
61
+ "transformer.h.22.attn.c_attn",
62
+ "transformer.h.22.attn.c_proj",
63
+ "transformer.h.23.attn.c_attn",
64
+ "transformer.h.23.attn.c_proj",
65
+ "transformer.h.0.mlp.c_fc",
66
+ "transformer.h.0.mlp.c_proj",
67
+ "transformer.h.1.mlp.c_fc",
68
+ "transformer.h.1.mlp.c_proj",
69
+ "transformer.h.2.mlp.c_fc",
70
+ "transformer.h.2.mlp.c_proj",
71
+ "transformer.h.3.mlp.c_fc",
72
+ "transformer.h.3.mlp.c_proj",
73
+ "transformer.h.4.mlp.c_fc",
74
+ "transformer.h.4.mlp.c_proj",
75
+ "transformer.h.5.mlp.c_fc",
76
+ "transformer.h.5.mlp.c_proj",
77
+ "transformer.h.6.mlp.c_fc",
78
+ "transformer.h.6.mlp.c_proj",
79
+ "transformer.h.7.mlp.c_fc",
80
+ "transformer.h.7.mlp.c_proj",
81
+ "transformer.h.8.mlp.c_fc",
82
+ "transformer.h.8.mlp.c_proj",
83
+ "transformer.h.9.mlp.c_fc",
84
+ "transformer.h.9.mlp.c_proj",
85
+ "transformer.h.10.mlp.c_fc",
86
+ "transformer.h.10.mlp.c_proj",
87
+ "transformer.h.11.mlp.c_fc",
88
+ "transformer.h.11.mlp.c_proj",
89
+ "transformer.h.12.mlp.c_fc",
90
+ "transformer.h.12.mlp.c_proj",
91
+ "transformer.h.13.mlp.c_fc",
92
+ "transformer.h.13.mlp.c_proj",
93
+ "transformer.h.14.mlp.c_fc",
94
+ "transformer.h.14.mlp.c_proj",
95
+ "transformer.h.15.mlp.c_fc",
96
+ "transformer.h.15.mlp.c_proj",
97
+ "transformer.h.16.mlp.c_fc",
98
+ "transformer.h.16.mlp.c_proj",
99
+ "transformer.h.17.mlp.c_fc",
100
+ "transformer.h.17.mlp.c_proj",
101
+ "transformer.h.18.mlp.c_fc",
102
+ "transformer.h.18.mlp.c_proj",
103
+ "transformer.h.19.mlp.c_fc",
104
+ "transformer.h.19.mlp.c_proj",
105
+ "transformer.h.20.mlp.c_fc",
106
+ "transformer.h.20.mlp.c_proj",
107
+ "transformer.h.21.mlp.c_fc",
108
+ "transformer.h.21.mlp.c_proj",
109
+ "transformer.h.22.mlp.c_fc",
110
+ "transformer.h.22.mlp.c_proj",
111
+ "transformer.h.23.mlp.c_fc",
112
+ "transformer.h.23.mlp.c_proj",
113
+ "transformer.h.24.attn.c_attn",
114
+ "transformer.h.24.attn.c_proj",
115
+ "transformer.h.25.attn.c_attn",
116
+ "transformer.h.25.attn.c_proj",
117
+ "transformer.h.26.attn.c_attn",
118
+ "transformer.h.26.attn.c_proj",
119
+ "transformer.h.27.attn.c_attn",
120
+ "transformer.h.27.attn.c_proj",
121
+ "transformer.h.28.attn.c_attn",
122
+ "transformer.h.28.attn.c_proj",
123
+ "transformer.h.29.attn.c_attn",
124
+ "transformer.h.29.attn.c_proj",
125
+ "transformer.h.30.attn.c_attn",
126
+ "transformer.h.30.attn.c_proj",
127
+ "transformer.h.31.attn.c_attn",
128
+ "transformer.h.31.attn.c_proj",
129
+ "transformer.h.32.attn.c_attn",
130
+ "transformer.h.32.attn.c_proj",
131
+ "transformer.h.33.attn.c_attn",
132
+ "transformer.h.33.attn.c_proj",
133
+ "transformer.h.34.attn.c_attn",
134
+ "transformer.h.34.attn.c_proj",
135
+ "transformer.h.35.attn.c_attn",
136
+ "transformer.h.35.attn.c_proj",
137
+ "transformer.h.24.mlp.c_fc",
138
+ "transformer.h.24.mlp.c_proj",
139
+ "transformer.h.25.mlp.c_fc",
140
+ "transformer.h.25.mlp.c_proj",
141
+ "transformer.h.26.mlp.c_fc",
142
+ "transformer.h.26.mlp.c_proj",
143
+ "transformer.h.27.mlp.c_fc",
144
+ "transformer.h.27.mlp.c_proj",
145
+ "transformer.h.28.mlp.c_fc",
146
+ "transformer.h.28.mlp.c_proj",
147
+ "transformer.h.29.mlp.c_fc",
148
+ "transformer.h.29.mlp.c_proj",
149
+ "transformer.h.30.mlp.c_fc",
150
+ "transformer.h.30.mlp.c_proj",
151
+ "transformer.h.31.mlp.c_fc",
152
+ "transformer.h.31.mlp.c_proj",
153
+ "transformer.h.32.mlp.c_fc",
154
+ "transformer.h.32.mlp.c_proj",
155
+ "transformer.h.33.mlp.c_fc",
156
+ "transformer.h.33.mlp.c_proj",
157
+ "transformer.h.34.mlp.c_fc",
158
+ "transformer.h.34.mlp.c_proj",
159
+ "transformer.h.35.mlp.c_fc",
160
+ "transformer.h.35.mlp.c_proj"
161
+ ],
162
+ "task_type": "CAUSAL_LM"
163
+ }
adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8913fd65d09ef6aa5ea64912c99b326077634c6dc1c99299642706695bc9c1d
3
+ size 23695117
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7882595f14f69f3cd3d75c6d03835a8b0a8b11a3d59094a91169fe18a05c7271
3
+ size 4027