Hugging Face's logo

slplab
/
pretrained-asd_w-cot_w-asd_gpt

TensorBoard
Safetensors
Model card Files Metrics Community
slplab commited on
Commit
aaaa1b4
•
1 Parent(s): ac9ab9e

Training in progress, step 5000

Browse files
Files changed (11) hide show
  1. adapter_config.json +195 -0
  2. adapter_model.safetensors +3 -0
  3. runs/Aug18_03-26-27_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723919193.oem-WS-C621E-SAGE-Series.605892.0 +3 -0
  4. runs/Aug18_03-28-59_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723919340.oem-WS-C621E-SAGE-Series.607716.0 +3 -0
  5. runs/Aug18_03-30-30_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723919430.oem-WS-C621E-SAGE-Series.608528.0 +3 -0
  6. runs/Aug18_03-33-46_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723919627.oem-WS-C621E-SAGE-Series.609672.0 +3 -0
  7. runs/Aug18_03-39-22_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723919965.oem-WS-C621E-SAGE-Series.610773.0 +3 -0
  8. runs/Aug18_03-44-57_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723920298.oem-WS-C621E-SAGE-Series.613194.0 +3 -0
  9. runs/Aug18_03-46-29_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723920390.oem-WS-C621E-SAGE-Series.613810.0 +3 -0
  10. runs/Aug18_03-48-10_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723920491.oem-WS-C621E-SAGE-Series.614475.0 +3 -0
  11. training_args.bin +3 -0
adapter_config.json ADDED
@@ -0,0 +1,195 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "kakaobrain/kogpt",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 16,
14
+ "lora_dropout": 0.05,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "transformer.h.21.mlp.fc_out",
24
+ "transformer.h.9.attn.k_proj",
25
+ "transformer.h.6.attn.k_proj",
26
+ "transformer.h.7.attn.q_proj",
27
+ "transformer.h.15.attn.v_proj",
28
+ "transformer.h.18.attn.out_proj",
29
+ "transformer.h.22.attn.k_proj",
30
+ "transformer.h.1.mlp.fc_in",
31
+ "transformer.h.16.attn.v_proj",
32
+ "transformer.h.26.mlp.fc_out",
33
+ "transformer.h.21.attn.v_proj",
34
+ "transformer.h.15.mlp.fc_in",
35
+ "transformer.h.25.attn.out_proj",
36
+ "transformer.h.12.mlp.fc_in",
37
+ "transformer.h.7.mlp.fc_in",
38
+ "transformer.h.26.attn.out_proj",
39
+ "transformer.h.27.attn.out_proj",
40
+ "transformer.h.7.attn.v_proj",
41
+ "transformer.h.23.attn.v_proj",
42
+ "transformer.h.3.mlp.fc_out",
43
+ "transformer.h.5.mlp.fc_in",
44
+ "transformer.h.13.attn.q_proj",
45
+ "transformer.h.6.mlp.fc_out",
46
+ "transformer.h.18.attn.q_proj",
47
+ "transformer.h.11.mlp.fc_out",
48
+ "transformer.h.11.attn.out_proj",
49
+ "transformer.h.1.mlp.fc_out",
50
+ "transformer.h.0.attn.out_proj",
51
+ "transformer.h.4.attn.k_proj",
52
+ "transformer.h.20.mlp.fc_out",
53
+ "transformer.h.17.attn.k_proj",
54
+ "transformer.h.5.attn.v_proj",
55
+ "transformer.h.2.attn.q_proj",
56
+ "transformer.h.1.attn.k_proj",
57
+ "transformer.h.7.mlp.fc_out",
58
+ "transformer.h.9.attn.v_proj",
59
+ "transformer.h.14.attn.v_proj",
60
+ "transformer.h.8.attn.k_proj",
61
+ "transformer.h.16.attn.out_proj",
62
+ "transformer.h.22.mlp.fc_out",
63
+ "transformer.h.11.mlp.fc_in",
64
+ "transformer.h.24.attn.out_proj",
65
+ "transformer.h.10.attn.out_proj",
66
+ "transformer.h.19.attn.q_proj",
67
+ "transformer.h.23.mlp.fc_out",
68
+ "transformer.h.12.mlp.fc_out",
69
+ "transformer.h.19.attn.k_proj",
70
+ "transformer.h.5.attn.q_proj",
71
+ "transformer.h.4.attn.out_proj",
72
+ "transformer.h.25.attn.k_proj",
73
+ "transformer.h.4.attn.v_proj",
74
+ "transformer.h.1.attn.v_proj",
75
+ "transformer.h.11.attn.k_proj",
76
+ "transformer.h.5.attn.out_proj",
77
+ "transformer.h.16.mlp.fc_in",
78
+ "transformer.h.9.mlp.fc_out",
79
+ "transformer.h.12.attn.q_proj",
80
+ "transformer.h.22.attn.out_proj",
81
+ "transformer.h.10.attn.q_proj",
82
+ "transformer.h.19.attn.v_proj",
83
+ "transformer.h.7.attn.k_proj",
84
+ "transformer.h.22.attn.v_proj",
85
+ "transformer.h.24.attn.k_proj",
86
+ "transformer.h.16.mlp.fc_out",
87
+ "transformer.h.3.attn.v_proj",
88
+ "transformer.h.12.attn.v_proj",
89
+ "transformer.h.13.attn.out_proj",
90
+ "transformer.h.19.attn.out_proj",
91
+ "transformer.h.3.attn.q_proj",
92
+ "transformer.h.20.mlp.fc_in",
93
+ "transformer.h.24.attn.q_proj",
94
+ "transformer.h.14.attn.out_proj",
95
+ "transformer.h.21.attn.out_proj",
96
+ "transformer.h.21.attn.k_proj",
97
+ "transformer.h.16.attn.q_proj",
98
+ "transformer.h.16.attn.k_proj",
99
+ "transformer.h.17.attn.out_proj",
100
+ "transformer.h.14.mlp.fc_out",
101
+ "transformer.h.21.mlp.fc_in",
102
+ "transformer.h.25.mlp.fc_out",
103
+ "transformer.h.15.mlp.fc_out",
104
+ "transformer.h.13.mlp.fc_out",
105
+ "transformer.h.20.attn.k_proj",
106
+ "transformer.h.18.attn.v_proj",
107
+ "transformer.h.26.attn.k_proj",
108
+ "transformer.h.11.attn.q_proj",
109
+ "transformer.h.23.attn.q_proj",
110
+ "transformer.h.13.attn.k_proj",
111
+ "transformer.h.18.mlp.fc_out",
112
+ "transformer.h.10.mlp.fc_out",
113
+ "transformer.h.2.attn.k_proj",
114
+ "transformer.h.10.attn.v_proj",
115
+ "transformer.h.11.attn.v_proj",
116
+ "transformer.h.6.attn.out_proj",
117
+ "transformer.h.19.mlp.fc_in",
118
+ "transformer.h.19.mlp.fc_out",
119
+ "transformer.h.25.mlp.fc_in",
120
+ "transformer.h.26.attn.q_proj",
121
+ "transformer.h.4.mlp.fc_out",
122
+ "transformer.h.0.attn.k_proj",
123
+ "transformer.h.10.mlp.fc_in",
124
+ "transformer.h.14.attn.k_proj",
125
+ "transformer.h.27.attn.k_proj",
126
+ "transformer.h.4.mlp.fc_in",
127
+ "transformer.h.20.attn.out_proj",
128
+ "transformer.h.27.mlp.fc_out",
129
+ "transformer.h.6.mlp.fc_in",
130
+ "transformer.h.13.attn.v_proj",
131
+ "transformer.h.8.attn.out_proj",
132
+ "transformer.h.9.mlp.fc_in",
133
+ "transformer.h.22.mlp.fc_in",
134
+ "transformer.h.3.attn.out_proj",
135
+ "transformer.h.4.attn.q_proj",
136
+ "transformer.h.14.mlp.fc_in",
137
+ "transformer.h.25.attn.q_proj",
138
+ "transformer.h.17.attn.q_proj",
139
+ "transformer.h.7.attn.out_proj",
140
+ "transformer.h.12.attn.k_proj",
141
+ "transformer.h.3.mlp.fc_in",
142
+ "transformer.h.2.attn.out_proj",
143
+ "transformer.h.27.attn.v_proj",
144
+ "transformer.h.8.attn.q_proj",
145
+ "transformer.h.25.attn.v_proj",
146
+ "transformer.h.23.attn.out_proj",
147
+ "transformer.h.24.attn.v_proj",
148
+ "transformer.h.17.attn.v_proj",
149
+ "transformer.h.0.attn.v_proj",
150
+ "transformer.h.23.attn.k_proj",
151
+ "transformer.h.2.attn.v_proj",
152
+ "transformer.h.20.attn.v_proj",
153
+ "transformer.h.26.mlp.fc_in",
154
+ "transformer.h.0.mlp.fc_out",
155
+ "transformer.h.0.mlp.fc_in",
156
+ "transformer.h.3.attn.k_proj",
157
+ "transformer.h.24.mlp.fc_out",
158
+ "transformer.h.2.mlp.fc_in",
159
+ "transformer.h.6.attn.q_proj",
160
+ "transformer.h.9.attn.q_proj",
161
+ "transformer.h.23.mlp.fc_in",
162
+ "transformer.h.2.mlp.fc_out",
163
+ "transformer.h.5.attn.k_proj",
164
+ "transformer.h.5.mlp.fc_out",
165
+ "transformer.h.15.attn.out_proj",
166
+ "transformer.h.10.attn.k_proj",
167
+ "transformer.h.6.attn.v_proj",
168
+ "transformer.h.20.attn.q_proj",
169
+ "transformer.h.17.mlp.fc_in",
170
+ "transformer.h.8.mlp.fc_in",
171
+ "transformer.h.26.attn.v_proj",
172
+ "transformer.h.13.mlp.fc_in",
173
+ "transformer.h.12.attn.out_proj",
174
+ "transformer.h.14.attn.q_proj",
175
+ "transformer.h.21.attn.q_proj",
176
+ "transformer.h.22.attn.q_proj",
177
+ "transformer.h.1.attn.out_proj",
178
+ "transformer.h.27.mlp.fc_in",
179
+ "transformer.h.0.attn.q_proj",
180
+ "transformer.h.15.attn.q_proj",
181
+ "transformer.h.15.attn.k_proj",
182
+ "transformer.h.24.mlp.fc_in",
183
+ "transformer.h.27.attn.q_proj",
184
+ "transformer.h.17.mlp.fc_out",
185
+ "transformer.h.18.mlp.fc_in",
186
+ "transformer.h.8.mlp.fc_out",
187
+ "transformer.h.18.attn.k_proj",
188
+ "transformer.h.1.attn.q_proj",
189
+ "transformer.h.9.attn.out_proj",
190
+ "transformer.h.8.attn.v_proj"
191
+ ],
192
+ "task_type": "CAUSAL_LM",
193
+ "use_dora": false,
194
+ "use_rslora": false
195
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae919e991fcdbd8723ab38cfad4a2f87224da44a510517049961d616dc1e8217
3
+ size 66103768
runs/Aug18_03-26-27_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723919193.oem-WS-C621E-SAGE-Series.605892.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e4d95372972b86390b060b1245f3ddd8ba5222ba1a859fed0f6749d1e61abd89
3
+ size 10618
runs/Aug18_03-28-59_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723919340.oem-WS-C621E-SAGE-Series.607716.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:442ecff0a8cda33ca573c2ad30a023d19f6a4f59d3729b9ca2d7269dbf818627
3
+ size 5353
runs/Aug18_03-30-30_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723919430.oem-WS-C621E-SAGE-Series.608528.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:22d32ed20fca14870fdbadf6b8eb4c842a89c4f3d0de794c3056a8e3168aa277
3
+ size 5353
runs/Aug18_03-33-46_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723919627.oem-WS-C621E-SAGE-Series.609672.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c28b7c6b9e6e76d798d1f52d228ff3d99b2885e9415d23bbe29ef4e4ae5585d0
3
+ size 5353
runs/Aug18_03-39-22_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723919965.oem-WS-C621E-SAGE-Series.610773.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11cdc0b40107500fb492e756bbc317063fdfd54d39139632fe886652db620693
3
+ size 5834
runs/Aug18_03-44-57_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723920298.oem-WS-C621E-SAGE-Series.613194.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b441e993067832e8e3d88575b8acb119f09e34ad96701e760bb5f0daaa36f7ed
3
+ size 5834
runs/Aug18_03-46-29_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723920390.oem-WS-C621E-SAGE-Series.613810.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b148982a5ada2435fe245e9061ba9ba690726415f95c29a6171f9771efb73912
3
+ size 5835
runs/Aug18_03-48-10_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723920491.oem-WS-C621E-SAGE-Series.614475.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61723f2c029404edf8ea3763c313fe4f230d5449f18874334e924dc638e22b87
3
+ size 8244
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cbb7025434e4869a894ec7906df7d2e533e50d0f606ad204527115110c0e50a1
3
+ size 5304