farguney commited on
Commit
f744877
·
verified ·
1 Parent(s): 71a414e

Upload g2_delta_s0.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. g2_delta_s0.json +40 -208
g2_delta_s0.json CHANGED
@@ -21,242 +21,74 @@
21
  "lr": 0.0003,
22
  "clip": 1.0,
23
  "warmup": 100,
24
- "eval_events": 24
 
 
 
 
25
  },
26
  "off_law": {},
27
  "device": "cuda",
28
  "runs": [
29
  {
30
- "params": 700802,
31
- "state_elems": 10112,
32
  "history": [
33
  {
34
- "value_acc_d32": 0.0013,
 
35
  "recall_exact_d32": 0.0,
 
36
  "value_acc_d128": 0.0026,
37
  "recall_exact_d128": 0.0,
 
38
  "value_acc_d512": 0.0052,
39
  "recall_exact_d512": 0.0,
 
40
  "value_acc_d1024": 0.0039,
41
  "recall_exact_d1024": 0.0,
 
 
 
 
 
42
  "step": 0,
43
- "train_bpb": 8.0439,
44
- "grad_norm": 1.034613013,
45
- "pairs": 4,
46
- "lr": 0.0003,
47
- "s_per_step": 2.1006
48
- },
49
- {
50
- "value_acc_d32": 0.0,
51
- "recall_exact_d32": 0.0,
52
- "value_acc_d128": 0.0,
53
- "recall_exact_d128": 0.0,
54
- "value_acc_d512": 0.0,
55
- "recall_exact_d512": 0.0,
56
- "value_acc_d1024": 0.0,
57
- "recall_exact_d1024": 0.0,
58
- "step": 500,
59
- "train_bpb": 7.5206,
60
- "grad_norm": 0.6686547995,
61
- "pairs": 4,
62
- "lr": 0.0003,
63
- "s_per_step": 0.4821
64
- },
65
- {
66
- "value_acc_d32": 0.0039,
67
- "recall_exact_d32": 0.0,
68
- "value_acc_d128": 0.0013,
69
- "recall_exact_d128": 0.0,
70
- "value_acc_d512": 0.0013,
71
- "recall_exact_d512": 0.0,
72
- "value_acc_d1024": 0.0026,
73
- "recall_exact_d1024": 0.0,
74
- "step": 1000,
75
- "train_bpb": 7.2557,
76
- "grad_norm": 0.8672067523,
77
- "pairs": 4,
78
- "lr": 0.0003,
79
- "s_per_step": 0.4803
80
- },
81
- {
82
- "value_acc_d32": 0.0039,
83
- "recall_exact_d32": 0.0,
84
- "value_acc_d128": 0.0026,
85
- "recall_exact_d128": 0.0,
86
- "value_acc_d512": 0.0052,
87
- "recall_exact_d512": 0.0,
88
- "value_acc_d1024": 0.0039,
89
- "recall_exact_d1024": 0.0,
90
- "step": 1500,
91
- "train_bpb": 7.2656,
92
- "grad_norm": 4.645389557,
93
- "pairs": 4,
94
- "lr": 0.0003,
95
- "s_per_step": 0.4796
96
- },
97
- {
98
- "value_acc_d32": 0.0026,
99
- "recall_exact_d32": 0.0,
100
- "value_acc_d128": 0.0039,
101
- "recall_exact_d128": 0.0,
102
- "value_acc_d512": 0.0013,
103
- "recall_exact_d512": 0.0,
104
- "value_acc_d1024": 0.0013,
105
- "recall_exact_d1024": 0.0,
106
- "step": 2000,
107
- "train_bpb": 7.2314,
108
- "grad_norm": 0.7635706067,
109
- "pairs": 4,
110
- "lr": 0.0003,
111
- "s_per_step": 0.4791
112
- },
113
- {
114
- "value_acc_d32": 0.0052,
115
- "recall_exact_d32": 0.0,
116
- "value_acc_d128": 0.0,
117
- "recall_exact_d128": 0.0,
118
- "value_acc_d512": 0.0104,
119
- "recall_exact_d512": 0.0,
120
- "value_acc_d1024": 0.0039,
121
- "recall_exact_d1024": 0.0,
122
- "step": 2500,
123
- "train_bpb": 7.2238,
124
- "grad_norm": 0.6498804092,
125
- "pairs": 4,
126
- "lr": 0.0003,
127
- "s_per_step": 0.4787
128
- },
129
- {
130
- "value_acc_d32": 0.0039,
131
- "recall_exact_d32": 0.0,
132
- "value_acc_d128": 0.0026,
133
- "recall_exact_d128": 0.0,
134
- "value_acc_d512": 0.0013,
135
- "recall_exact_d512": 0.0,
136
- "value_acc_d1024": 0.0065,
137
- "recall_exact_d1024": 0.0,
138
- "step": 3000,
139
- "train_bpb": 7.2264,
140
- "grad_norm": 1.502449512,
141
- "pairs": 4,
142
- "lr": 0.0003,
143
- "s_per_step": 0.4783
144
- },
145
- {
146
- "value_acc_d32": 0.0,
147
- "recall_exact_d32": 0.0,
148
- "value_acc_d128": 0.0013,
149
- "recall_exact_d128": 0.0,
150
- "value_acc_d512": 0.0039,
151
- "recall_exact_d512": 0.0,
152
- "value_acc_d1024": 0.0065,
153
- "recall_exact_d1024": 0.0,
154
- "step": 3500,
155
- "train_bpb": 7.2231,
156
- "grad_norm": 1.429182649,
157
- "pairs": 4,
158
- "lr": 0.0003,
159
- "s_per_step": 0.478
160
- },
161
- {
162
- "value_acc_d32": 0.0026,
163
- "recall_exact_d32": 0.0,
164
- "value_acc_d128": 0.0013,
165
- "recall_exact_d128": 0.0,
166
- "value_acc_d512": 0.0065,
167
- "recall_exact_d512": 0.0,
168
- "value_acc_d1024": 0.0013,
169
- "recall_exact_d1024": 0.0,
170
- "step": 4000,
171
- "train_bpb": 7.2253,
172
- "grad_norm": 1.32773006,
173
- "pairs": 4,
174
- "lr": 0.0003,
175
- "s_per_step": 0.4779
176
- },
177
- {
178
- "value_acc_d32": 0.0039,
179
- "recall_exact_d32": 0.0,
180
- "value_acc_d128": 0.0039,
181
- "recall_exact_d128": 0.0,
182
- "value_acc_d512": 0.0013,
183
- "recall_exact_d512": 0.0,
184
- "value_acc_d1024": 0.0039,
185
- "recall_exact_d1024": 0.0,
186
- "step": 4500,
187
- "train_bpb": 7.2262,
188
- "grad_norm": 3.401660204,
189
- "pairs": 4,
190
- "lr": 0.0003,
191
- "s_per_step": 0.4776
192
- },
193
- {
194
- "value_acc_d32": 0.0065,
195
- "recall_exact_d32": 0.0,
196
- "value_acc_d128": 0.0,
197
- "recall_exact_d128": 0.0,
198
- "value_acc_d512": 0.0039,
199
- "recall_exact_d512": 0.0,
200
- "value_acc_d1024": 0.0052,
201
- "recall_exact_d1024": 0.0,
202
- "step": 5000,
203
- "train_bpb": 7.2229,
204
- "grad_norm": 2.684652328,
205
- "pairs": 4,
206
- "lr": 0.0003,
207
- "s_per_step": 0.4776
208
- },
209
- {
210
- "value_acc_d32": 0.0026,
211
- "recall_exact_d32": 0.0,
212
- "value_acc_d128": 0.0013,
213
- "recall_exact_d128": 0.0,
214
- "value_acc_d512": 0.0026,
215
- "recall_exact_d512": 0.0,
216
- "value_acc_d1024": 0.0065,
217
- "recall_exact_d1024": 0.0,
218
- "step": 5500,
219
- "train_bpb": 7.2148,
220
- "grad_norm": 0.6313759685,
221
- "pairs": 4,
222
- "lr": 0.0003,
223
- "s_per_step": 0.4772
224
- },
225
- {
226
- "value_acc_d32": 0.0065,
227
- "recall_exact_d32": 0.0,
228
- "value_acc_d128": 0.0,
229
- "recall_exact_d128": 0.0,
230
- "value_acc_d512": 0.0039,
231
- "recall_exact_d512": 0.0,
232
- "value_acc_d1024": 0.0026,
233
- "recall_exact_d1024": 0.0,
234
- "step": 5999,
235
- "train_bpb": 7.8635,
236
- "grad_norm": 0.21263735,
237
  "pairs": 4,
238
  "lr": 0.0003,
239
- "s_per_step": 0.4769
 
240
  }
241
  ],
242
  "final": {
243
- "value_acc_d32": 0.0065,
 
244
  "recall_exact_d32": 0.0,
245
- "value_acc_d128": 0.0,
 
246
  "recall_exact_d128": 0.0,
247
- "value_acc_d512": 0.0039,
 
248
  "recall_exact_d512": 0.0,
249
- "value_acc_d1024": 0.0026,
 
250
  "recall_exact_d1024": 0.0,
251
- "step": 5999,
252
- "train_bpb": 7.8635,
253
- "grad_norm": 0.21263735,
 
 
 
 
 
254
  "pairs": 4,
255
  "lr": 0.0003,
256
- "s_per_step": 0.4769
 
257
  }
258
  }
259
  ],
260
- "partial": false,
261
  "result_file": "g2_delta_s0.json"
262
  }
 
21
  "lr": 0.0003,
22
  "clip": 1.0,
23
  "warmup": 100,
24
+ "eval_events": 24,
25
+ "law_rev": "candm-v1",
26
+ "kappa": 8.0,
27
+ "echo_phase": 1500,
28
+ "echo_delay": 32
29
  },
30
  "off_law": {},
31
  "device": "cuda",
32
  "runs": [
33
  {
34
+ "params": 536835,
35
+ "state_elems": 10176,
36
  "history": [
37
  {
38
+ "law_rev": "candm-v1",
39
+ "value_acc_d32": 0.0026,
40
  "recall_exact_d32": 0.0,
41
+ "key_acc_d32": 0.0104,
42
  "value_acc_d128": 0.0026,
43
  "recall_exact_d128": 0.0,
44
+ "key_acc_d128": 0.0052,
45
  "value_acc_d512": 0.0052,
46
  "recall_exact_d512": 0.0,
47
+ "key_acc_d512": 0.0,
48
  "value_acc_d1024": 0.0039,
49
  "recall_exact_d1024": 0.0,
50
+ "key_acc_d1024": 0.0026,
51
+ "alpha_mean": 0.999,
52
+ "beta_mean": 0.0502,
53
+ "m_fro": 3.695,
54
+ "p_fro": 4.899,
55
  "step": 0,
56
+ "train_bpb": 15.5337,
57
+ "grad_norm": 16.24673271,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
58
  "pairs": 4,
59
  "lr": 0.0003,
60
+ "eta": 1.0,
61
+ "s_per_step": 3.7504
62
  }
63
  ],
64
  "final": {
65
+ "law_rev": "candm-v1",
66
+ "value_acc_d32": 0.0026,
67
  "recall_exact_d32": 0.0,
68
+ "key_acc_d32": 0.0104,
69
+ "value_acc_d128": 0.0026,
70
  "recall_exact_d128": 0.0,
71
+ "key_acc_d128": 0.0052,
72
+ "value_acc_d512": 0.0052,
73
  "recall_exact_d512": 0.0,
74
+ "key_acc_d512": 0.0,
75
+ "value_acc_d1024": 0.0039,
76
  "recall_exact_d1024": 0.0,
77
+ "key_acc_d1024": 0.0026,
78
+ "alpha_mean": 0.999,
79
+ "beta_mean": 0.0502,
80
+ "m_fro": 3.695,
81
+ "p_fro": 4.899,
82
+ "step": 0,
83
+ "train_bpb": 15.5337,
84
+ "grad_norm": 16.24673271,
85
  "pairs": 4,
86
  "lr": 0.0003,
87
+ "eta": 1.0,
88
+ "s_per_step": 3.7504
89
  }
90
  }
91
  ],
92
+ "partial": true,
93
  "result_file": "g2_delta_s0.json"
94
  }