eac123 commited on
Commit
b91dbf2
·
verified ·
1 Parent(s): 6c7042a

Upload results/lr0.0003-fixed4x1_history.json with huggingface_hub

Browse files
results/lr0.0003-fixed4x1_history.json CHANGED
@@ -13,7 +13,7 @@
13
  "loss": 4.2035,
14
  "lr": 3.0299999999999998e-05,
15
  "iters": 1,
16
- "steps_per_s": 21.93,
17
  "flops_cum": 8828127019008.0
18
  },
19
  {
@@ -21,7 +21,7 @@
21
  "loss": 4.1736,
22
  "lr": 6.0299999999999995e-05,
23
  "iters": 1,
24
- "steps_per_s": 21.69,
25
  "flops_cum": 17568846839808.0
26
  },
27
  {
@@ -29,7 +29,7 @@
29
  "loss": 4.1726,
30
  "lr": 9.029999999999999e-05,
31
  "iters": 1,
32
- "steps_per_s": 21.74,
33
  "flops_cum": 26309566660608.0
34
  },
35
  {
@@ -37,7 +37,7 @@
37
  "loss": 4.1751,
38
  "lr": 0.0001203,
39
  "iters": 1,
40
- "steps_per_s": 21.7,
41
  "flops_cum": 35050286481408.0
42
  },
43
  {
@@ -45,7 +45,7 @@
45
  "loss": 4.1623,
46
  "lr": 0.0001503,
47
  "iters": 1,
48
- "steps_per_s": 21.69,
49
  "flops_cum": 43791006302208.0
50
  },
51
  {
@@ -53,7 +53,7 @@
53
  "loss": 4.1639,
54
  "lr": 0.0001803,
55
  "iters": 1,
56
- "steps_per_s": 21.67,
57
  "flops_cum": 52531726123008.0
58
  },
59
  {
@@ -61,7 +61,7 @@
61
  "loss": 4.1456,
62
  "lr": 0.0002103,
63
  "iters": 1,
64
- "steps_per_s": 21.63,
65
  "flops_cum": 61272445943808.0
66
  },
67
  {
@@ -69,7 +69,7 @@
69
  "loss": 4.1389,
70
  "lr": 0.00024029999999999999,
71
  "iters": 1,
72
- "steps_per_s": 21.68,
73
  "flops_cum": 70013165764608.0
74
  },
75
  {
@@ -77,7 +77,7 @@
77
  "loss": 4.112,
78
  "lr": 0.0002703,
79
  "iters": 1,
80
- "steps_per_s": 21.58,
81
  "flops_cum": 78753885585408.0
82
  },
83
  {
@@ -85,7 +85,7 @@
85
  "loss": 4.0971,
86
  "lr": 0.0003,
87
  "iters": 1,
88
- "steps_per_s": 21.58,
89
  "flops_cum": 87494605406208.0
90
  },
91
  {
@@ -93,7 +93,7 @@
93
  "loss": 4.1204,
94
  "lr": 0.00029815325108927063,
95
  "iters": 1,
96
- "steps_per_s": 21.59,
97
  "flops_cum": 96235325227008.0
98
  },
99
  {
@@ -101,7 +101,7 @@
101
  "loss": 4.0513,
102
  "lr": 0.00029265847744427303,
103
  "iters": 1,
104
- "steps_per_s": 21.6,
105
  "flops_cum": 104976045047808.0
106
  },
107
  {
@@ -109,7 +109,7 @@
109
  "loss": 4.0594,
110
  "lr": 0.00028365097862825513,
111
  "iters": 1,
112
- "steps_per_s": 21.57,
113
  "flops_cum": 113716764868608.0
114
  },
115
  {
@@ -117,7 +117,7 @@
117
  "loss": 4.0581,
118
  "lr": 0.0002713525491562421,
119
  "iters": 1,
120
- "steps_per_s": 21.58,
121
  "flops_cum": 122457484689408.0
122
  },
123
  {
@@ -125,7 +125,7 @@
125
  "loss": 4.0422,
126
  "lr": 0.00025606601717798207,
127
  "iters": 1,
128
- "steps_per_s": 21.6,
129
  "flops_cum": 131198204510208.0
130
  },
131
  {
@@ -133,7 +133,7 @@
133
  "loss": 4.0282,
134
  "lr": 0.00023816778784387094,
135
  "iters": 1,
136
- "steps_per_s": 21.58,
137
  "flops_cum": 139938924331008.0
138
  },
139
  {
@@ -141,7 +141,7 @@
141
  "loss": 4.0302,
142
  "lr": 0.00021809857496093199,
143
  "iters": 1,
144
- "steps_per_s": 21.57,
145
  "flops_cum": 148679644151808.0
146
  },
147
  {
@@ -149,7 +149,7 @@
149
  "loss": 4.0033,
150
  "lr": 0.0001963525491562421,
151
  "iters": 1,
152
- "steps_per_s": 21.65,
153
  "flops_cum": 157420363972608.0
154
  },
155
  {
@@ -157,7 +157,7 @@
157
  "loss": 3.9716,
158
  "lr": 0.00017346516975603462,
159
  "iters": 1,
160
- "steps_per_s": 21.62,
161
  "flops_cum": 166161083793408.0
162
  },
163
  {
@@ -165,7 +165,7 @@
165
  "loss": 3.992,
166
  "lr": 0.00015,
167
  "iters": 1,
168
- "steps_per_s": 21.59,
169
  "flops_cum": 174901803614208.0
170
  },
171
  {
@@ -173,7 +173,7 @@
173
  "loss": 3.9775,
174
  "lr": 0.00012653483024396533,
175
  "iters": 1,
176
- "steps_per_s": 21.61,
177
  "flops_cum": 183642523435008.0
178
  },
179
  {
@@ -181,7 +181,7 @@
181
  "loss": 3.9909,
182
  "lr": 0.0001036474508437579,
183
  "iters": 1,
184
- "steps_per_s": 21.62,
185
  "flops_cum": 192383243255808.0
186
  },
187
  {
@@ -189,7 +189,7 @@
189
  "loss": 3.9886,
190
  "lr": 8.190142503906798e-05,
191
  "iters": 1,
192
- "steps_per_s": 21.62,
193
  "flops_cum": 201123963076608.0
194
  },
195
  {
@@ -197,7 +197,7 @@
197
  "loss": 4.0272,
198
  "lr": 6.183221215612904e-05,
199
  "iters": 1,
200
- "steps_per_s": 21.61,
201
  "flops_cum": 209864682897408.0
202
  },
203
  {
@@ -205,7 +205,7 @@
205
  "loss": 4.0201,
206
  "lr": 4.3933982822017876e-05,
207
  "iters": 1,
208
- "steps_per_s": 20.25,
209
  "flops_cum": 218605402718208.0
210
  },
211
  {
@@ -213,7 +213,7 @@
213
  "loss": 3.9474,
214
  "lr": 2.8647450843757897e-05,
215
  "iters": 1,
216
- "steps_per_s": 21.41,
217
  "flops_cum": 227346122539008.0
218
  },
219
  {
@@ -221,7 +221,7 @@
221
  "loss": 3.9986,
222
  "lr": 1.634902137174483e-05,
223
  "iters": 1,
224
- "steps_per_s": 21.61,
225
  "flops_cum": 236086842359808.0
226
  },
227
  {
@@ -229,7 +229,7 @@
229
  "loss": 3.9683,
230
  "lr": 7.34152255572697e-06,
231
  "iters": 1,
232
- "steps_per_s": 21.62,
233
  "flops_cum": 244827562180608.0
234
  },
235
  {
@@ -237,7 +237,7 @@
237
  "loss": 3.9984,
238
  "lr": 1.8467489107293509e-06,
239
  "iters": 1,
240
- "steps_per_s": 21.63,
241
  "flops_cum": 253568282001408.0
242
  },
243
  {
@@ -245,7 +245,7 @@
245
  "loss": 3.9986,
246
  "lr": 1.8505504447152708e-10,
247
  "iters": 1,
248
- "steps_per_s": 21.82,
249
  "flops_cum": 262221594624000.0
250
  }
251
  ],
 
13
  "loss": 4.2035,
14
  "lr": 3.0299999999999998e-05,
15
  "iters": 1,
16
+ "steps_per_s": 21.99,
17
  "flops_cum": 8828127019008.0
18
  },
19
  {
 
21
  "loss": 4.1736,
22
  "lr": 6.0299999999999995e-05,
23
  "iters": 1,
24
+ "steps_per_s": 21.84,
25
  "flops_cum": 17568846839808.0
26
  },
27
  {
 
29
  "loss": 4.1726,
30
  "lr": 9.029999999999999e-05,
31
  "iters": 1,
32
+ "steps_per_s": 21.76,
33
  "flops_cum": 26309566660608.0
34
  },
35
  {
 
37
  "loss": 4.1751,
38
  "lr": 0.0001203,
39
  "iters": 1,
40
+ "steps_per_s": 21.82,
41
  "flops_cum": 35050286481408.0
42
  },
43
  {
 
45
  "loss": 4.1623,
46
  "lr": 0.0001503,
47
  "iters": 1,
48
+ "steps_per_s": 21.76,
49
  "flops_cum": 43791006302208.0
50
  },
51
  {
 
53
  "loss": 4.1639,
54
  "lr": 0.0001803,
55
  "iters": 1,
56
+ "steps_per_s": 21.79,
57
  "flops_cum": 52531726123008.0
58
  },
59
  {
 
61
  "loss": 4.1456,
62
  "lr": 0.0002103,
63
  "iters": 1,
64
+ "steps_per_s": 21.76,
65
  "flops_cum": 61272445943808.0
66
  },
67
  {
 
69
  "loss": 4.1389,
70
  "lr": 0.00024029999999999999,
71
  "iters": 1,
72
+ "steps_per_s": 21.74,
73
  "flops_cum": 70013165764608.0
74
  },
75
  {
 
77
  "loss": 4.112,
78
  "lr": 0.0002703,
79
  "iters": 1,
80
+ "steps_per_s": 21.75,
81
  "flops_cum": 78753885585408.0
82
  },
83
  {
 
85
  "loss": 4.0971,
86
  "lr": 0.0003,
87
  "iters": 1,
88
+ "steps_per_s": 21.75,
89
  "flops_cum": 87494605406208.0
90
  },
91
  {
 
93
  "loss": 4.1204,
94
  "lr": 0.00029815325108927063,
95
  "iters": 1,
96
+ "steps_per_s": 21.74,
97
  "flops_cum": 96235325227008.0
98
  },
99
  {
 
101
  "loss": 4.0513,
102
  "lr": 0.00029265847744427303,
103
  "iters": 1,
104
+ "steps_per_s": 21.7,
105
  "flops_cum": 104976045047808.0
106
  },
107
  {
 
109
  "loss": 4.0594,
110
  "lr": 0.00028365097862825513,
111
  "iters": 1,
112
+ "steps_per_s": 21.73,
113
  "flops_cum": 113716764868608.0
114
  },
115
  {
 
117
  "loss": 4.0581,
118
  "lr": 0.0002713525491562421,
119
  "iters": 1,
120
+ "steps_per_s": 21.72,
121
  "flops_cum": 122457484689408.0
122
  },
123
  {
 
125
  "loss": 4.0422,
126
  "lr": 0.00025606601717798207,
127
  "iters": 1,
128
+ "steps_per_s": 21.66,
129
  "flops_cum": 131198204510208.0
130
  },
131
  {
 
133
  "loss": 4.0282,
134
  "lr": 0.00023816778784387094,
135
  "iters": 1,
136
+ "steps_per_s": 21.67,
137
  "flops_cum": 139938924331008.0
138
  },
139
  {
 
141
  "loss": 4.0302,
142
  "lr": 0.00021809857496093199,
143
  "iters": 1,
144
+ "steps_per_s": 21.7,
145
  "flops_cum": 148679644151808.0
146
  },
147
  {
 
149
  "loss": 4.0033,
150
  "lr": 0.0001963525491562421,
151
  "iters": 1,
152
+ "steps_per_s": 21.76,
153
  "flops_cum": 157420363972608.0
154
  },
155
  {
 
157
  "loss": 3.9716,
158
  "lr": 0.00017346516975603462,
159
  "iters": 1,
160
+ "steps_per_s": 21.74,
161
  "flops_cum": 166161083793408.0
162
  },
163
  {
 
165
  "loss": 3.992,
166
  "lr": 0.00015,
167
  "iters": 1,
168
+ "steps_per_s": 21.72,
169
  "flops_cum": 174901803614208.0
170
  },
171
  {
 
173
  "loss": 3.9775,
174
  "lr": 0.00012653483024396533,
175
  "iters": 1,
176
+ "steps_per_s": 21.71,
177
  "flops_cum": 183642523435008.0
178
  },
179
  {
 
181
  "loss": 3.9909,
182
  "lr": 0.0001036474508437579,
183
  "iters": 1,
184
+ "steps_per_s": 21.69,
185
  "flops_cum": 192383243255808.0
186
  },
187
  {
 
189
  "loss": 3.9886,
190
  "lr": 8.190142503906798e-05,
191
  "iters": 1,
192
+ "steps_per_s": 21.73,
193
  "flops_cum": 201123963076608.0
194
  },
195
  {
 
197
  "loss": 4.0272,
198
  "lr": 6.183221215612904e-05,
199
  "iters": 1,
200
+ "steps_per_s": 21.71,
201
  "flops_cum": 209864682897408.0
202
  },
203
  {
 
205
  "loss": 4.0201,
206
  "lr": 4.3933982822017876e-05,
207
  "iters": 1,
208
+ "steps_per_s": 20.3,
209
  "flops_cum": 218605402718208.0
210
  },
211
  {
 
213
  "loss": 3.9474,
214
  "lr": 2.8647450843757897e-05,
215
  "iters": 1,
216
+ "steps_per_s": 21.66,
217
  "flops_cum": 227346122539008.0
218
  },
219
  {
 
221
  "loss": 3.9986,
222
  "lr": 1.634902137174483e-05,
223
  "iters": 1,
224
+ "steps_per_s": 21.65,
225
  "flops_cum": 236086842359808.0
226
  },
227
  {
 
229
  "loss": 3.9683,
230
  "lr": 7.34152255572697e-06,
231
  "iters": 1,
232
+ "steps_per_s": 21.71,
233
  "flops_cum": 244827562180608.0
234
  },
235
  {
 
237
  "loss": 3.9984,
238
  "lr": 1.8467489107293509e-06,
239
  "iters": 1,
240
+ "steps_per_s": 21.71,
241
  "flops_cum": 253568282001408.0
242
  },
243
  {
 
245
  "loss": 3.9986,
246
  "lr": 1.8505504447152708e-10,
247
  "iters": 1,
248
+ "steps_per_s": 21.93,
249
  "flops_cum": 262221594624000.0
250
  }
251
  ],