End of training
Browse files- README.md +2 -2
- all_results.json +15 -0
- eval_results.json +10 -0
- train_results.json +8 -0
- trainer_state.json +832 -0
README.md
CHANGED
@@ -16,8 +16,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
16 |
|
17 |
This model is a fine-tuned version of [models/smolm-mlm/config.json](https://huggingface.co/models/smolm-mlm/config.json) on an unknown dataset.
|
18 |
It achieves the following results on the evaluation set:
|
19 |
-
- Loss: 2.
|
20 |
-
- Accuracy: 0.
|
21 |
|
22 |
## Model description
|
23 |
|
|
|
16 |
|
17 |
This model is a fine-tuned version of [models/smolm-mlm/config.json](https://huggingface.co/models/smolm-mlm/config.json) on an unknown dataset.
|
18 |
It achieves the following results on the evaluation set:
|
19 |
+
- Loss: 2.7042
|
20 |
+
- Accuracy: 0.4475
|
21 |
|
22 |
## Model description
|
23 |
|
all_results.json
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 10.0,
|
3 |
+
"eval_accuracy": 0.4475049050175668,
|
4 |
+
"eval_loss": 2.704235792160034,
|
5 |
+
"eval_runtime": 8.5501,
|
6 |
+
"eval_samples": 70000,
|
7 |
+
"eval_samples_per_second": 8187.065,
|
8 |
+
"eval_steps_per_second": 16.023,
|
9 |
+
"perplexity": 14.942892850119993,
|
10 |
+
"train_loss": 3.079910436302079,
|
11 |
+
"train_runtime": 7252.3341,
|
12 |
+
"train_samples": 763989,
|
13 |
+
"train_samples_per_second": 1053.439,
|
14 |
+
"train_steps_per_second": 16.461
|
15 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 10.0,
|
3 |
+
"eval_accuracy": 0.4475049050175668,
|
4 |
+
"eval_loss": 2.704235792160034,
|
5 |
+
"eval_runtime": 8.5501,
|
6 |
+
"eval_samples": 70000,
|
7 |
+
"eval_samples_per_second": 8187.065,
|
8 |
+
"eval_steps_per_second": 16.023,
|
9 |
+
"perplexity": 14.942892850119993
|
10 |
+
}
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 10.0,
|
3 |
+
"train_loss": 3.079910436302079,
|
4 |
+
"train_runtime": 7252.3341,
|
5 |
+
"train_samples": 763989,
|
6 |
+
"train_samples_per_second": 1053.439,
|
7 |
+
"train_steps_per_second": 16.461
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,832 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 10.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 119380,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.08,
|
13 |
+
"learning_rate": 2.0833333333333333e-05,
|
14 |
+
"loss": 7.7261,
|
15 |
+
"step": 1000
|
16 |
+
},
|
17 |
+
{
|
18 |
+
"epoch": 0.17,
|
19 |
+
"learning_rate": 4.1666666666666665e-05,
|
20 |
+
"loss": 5.5648,
|
21 |
+
"step": 2000
|
22 |
+
},
|
23 |
+
{
|
24 |
+
"epoch": 0.25,
|
25 |
+
"learning_rate": 6.25e-05,
|
26 |
+
"loss": 5.0119,
|
27 |
+
"step": 3000
|
28 |
+
},
|
29 |
+
{
|
30 |
+
"epoch": 0.34,
|
31 |
+
"learning_rate": 8.333333333333333e-05,
|
32 |
+
"loss": 4.6599,
|
33 |
+
"step": 4000
|
34 |
+
},
|
35 |
+
{
|
36 |
+
"epoch": 0.42,
|
37 |
+
"learning_rate": 0.00010416666666666667,
|
38 |
+
"loss": 4.3568,
|
39 |
+
"step": 5000
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 0.5,
|
43 |
+
"learning_rate": 0.000125,
|
44 |
+
"loss": 4.1194,
|
45 |
+
"step": 6000
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 0.59,
|
49 |
+
"learning_rate": 0.00014583333333333335,
|
50 |
+
"loss": 3.8926,
|
51 |
+
"step": 7000
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.67,
|
55 |
+
"learning_rate": 0.00016666666666666666,
|
56 |
+
"loss": 3.778,
|
57 |
+
"step": 8000
|
58 |
+
},
|
59 |
+
{
|
60 |
+
"epoch": 0.75,
|
61 |
+
"learning_rate": 0.0001875,
|
62 |
+
"loss": 3.65,
|
63 |
+
"step": 9000
|
64 |
+
},
|
65 |
+
{
|
66 |
+
"epoch": 0.84,
|
67 |
+
"learning_rate": 0.00020833333333333335,
|
68 |
+
"loss": 3.6165,
|
69 |
+
"step": 10000
|
70 |
+
},
|
71 |
+
{
|
72 |
+
"epoch": 0.92,
|
73 |
+
"learning_rate": 0.00022916666666666666,
|
74 |
+
"loss": 3.5523,
|
75 |
+
"step": 11000
|
76 |
+
},
|
77 |
+
{
|
78 |
+
"epoch": 1.0,
|
79 |
+
"eval_accuracy": 0.3512480392306169,
|
80 |
+
"eval_loss": 3.490408420562744,
|
81 |
+
"eval_runtime": 8.7154,
|
82 |
+
"eval_samples_per_second": 8031.783,
|
83 |
+
"eval_steps_per_second": 15.719,
|
84 |
+
"step": 11938
|
85 |
+
},
|
86 |
+
{
|
87 |
+
"epoch": 1.01,
|
88 |
+
"learning_rate": 0.00025,
|
89 |
+
"loss": 3.4876,
|
90 |
+
"step": 12000
|
91 |
+
},
|
92 |
+
{
|
93 |
+
"epoch": 1.09,
|
94 |
+
"learning_rate": 0.0002708333333333333,
|
95 |
+
"loss": 3.4466,
|
96 |
+
"step": 13000
|
97 |
+
},
|
98 |
+
{
|
99 |
+
"epoch": 1.17,
|
100 |
+
"learning_rate": 0.0002916666666666667,
|
101 |
+
"loss": 3.4306,
|
102 |
+
"step": 14000
|
103 |
+
},
|
104 |
+
{
|
105 |
+
"epoch": 1.26,
|
106 |
+
"learning_rate": 0.0003125,
|
107 |
+
"loss": 3.39,
|
108 |
+
"step": 15000
|
109 |
+
},
|
110 |
+
{
|
111 |
+
"epoch": 1.34,
|
112 |
+
"learning_rate": 0.0003333333333333333,
|
113 |
+
"loss": 3.3891,
|
114 |
+
"step": 16000
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 1.42,
|
118 |
+
"learning_rate": 0.0003541666666666667,
|
119 |
+
"loss": 3.367,
|
120 |
+
"step": 17000
|
121 |
+
},
|
122 |
+
{
|
123 |
+
"epoch": 1.51,
|
124 |
+
"learning_rate": 0.000375,
|
125 |
+
"loss": 3.3546,
|
126 |
+
"step": 18000
|
127 |
+
},
|
128 |
+
{
|
129 |
+
"epoch": 1.59,
|
130 |
+
"learning_rate": 0.0003958333333333333,
|
131 |
+
"loss": 3.3348,
|
132 |
+
"step": 19000
|
133 |
+
},
|
134 |
+
{
|
135 |
+
"epoch": 1.68,
|
136 |
+
"learning_rate": 0.0004166666666666667,
|
137 |
+
"loss": 3.3235,
|
138 |
+
"step": 20000
|
139 |
+
},
|
140 |
+
{
|
141 |
+
"epoch": 1.76,
|
142 |
+
"learning_rate": 0.0004375,
|
143 |
+
"loss": 3.3207,
|
144 |
+
"step": 21000
|
145 |
+
},
|
146 |
+
{
|
147 |
+
"epoch": 1.84,
|
148 |
+
"learning_rate": 0.0004583333333333333,
|
149 |
+
"loss": 3.3028,
|
150 |
+
"step": 22000
|
151 |
+
},
|
152 |
+
{
|
153 |
+
"epoch": 1.93,
|
154 |
+
"learning_rate": 0.0004791666666666667,
|
155 |
+
"loss": 3.3151,
|
156 |
+
"step": 23000
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 2.0,
|
160 |
+
"eval_accuracy": 0.36783764827859605,
|
161 |
+
"eval_loss": 3.319209098815918,
|
162 |
+
"eval_runtime": 8.6438,
|
163 |
+
"eval_samples_per_second": 8098.333,
|
164 |
+
"eval_steps_per_second": 15.85,
|
165 |
+
"step": 23876
|
166 |
+
},
|
167 |
+
{
|
168 |
+
"epoch": 2.01,
|
169 |
+
"learning_rate": 0.0005,
|
170 |
+
"loss": 3.2991,
|
171 |
+
"step": 24000
|
172 |
+
},
|
173 |
+
{
|
174 |
+
"epoch": 2.09,
|
175 |
+
"learning_rate": 0.0004947578108618159,
|
176 |
+
"loss": 3.3032,
|
177 |
+
"step": 25000
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 2.18,
|
181 |
+
"learning_rate": 0.0004895156217236318,
|
182 |
+
"loss": 3.2872,
|
183 |
+
"step": 26000
|
184 |
+
},
|
185 |
+
{
|
186 |
+
"epoch": 2.26,
|
187 |
+
"learning_rate": 0.0004842734325854477,
|
188 |
+
"loss": 3.2814,
|
189 |
+
"step": 27000
|
190 |
+
},
|
191 |
+
{
|
192 |
+
"epoch": 2.35,
|
193 |
+
"learning_rate": 0.0004790312434472636,
|
194 |
+
"loss": 3.2595,
|
195 |
+
"step": 28000
|
196 |
+
},
|
197 |
+
{
|
198 |
+
"epoch": 2.43,
|
199 |
+
"learning_rate": 0.0004737890543090795,
|
200 |
+
"loss": 3.2423,
|
201 |
+
"step": 29000
|
202 |
+
},
|
203 |
+
{
|
204 |
+
"epoch": 2.51,
|
205 |
+
"learning_rate": 0.0004685468651708954,
|
206 |
+
"loss": 3.2336,
|
207 |
+
"step": 30000
|
208 |
+
},
|
209 |
+
{
|
210 |
+
"epoch": 2.6,
|
211 |
+
"learning_rate": 0.00046330467603271125,
|
212 |
+
"loss": 3.2224,
|
213 |
+
"step": 31000
|
214 |
+
},
|
215 |
+
{
|
216 |
+
"epoch": 2.68,
|
217 |
+
"learning_rate": 0.00045806248689452716,
|
218 |
+
"loss": 3.1981,
|
219 |
+
"step": 32000
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 2.76,
|
223 |
+
"learning_rate": 0.000452820297756343,
|
224 |
+
"loss": 3.1808,
|
225 |
+
"step": 33000
|
226 |
+
},
|
227 |
+
{
|
228 |
+
"epoch": 2.85,
|
229 |
+
"learning_rate": 0.00044757810861815897,
|
230 |
+
"loss": 3.1593,
|
231 |
+
"step": 34000
|
232 |
+
},
|
233 |
+
{
|
234 |
+
"epoch": 2.93,
|
235 |
+
"learning_rate": 0.0004423359194799749,
|
236 |
+
"loss": 3.1526,
|
237 |
+
"step": 35000
|
238 |
+
},
|
239 |
+
{
|
240 |
+
"epoch": 3.0,
|
241 |
+
"eval_accuracy": 0.3893979057591623,
|
242 |
+
"eval_loss": 3.154465675354004,
|
243 |
+
"eval_runtime": 8.7169,
|
244 |
+
"eval_samples_per_second": 8030.369,
|
245 |
+
"eval_steps_per_second": 15.717,
|
246 |
+
"step": 35814
|
247 |
+
},
|
248 |
+
{
|
249 |
+
"epoch": 3.02,
|
250 |
+
"learning_rate": 0.00043709373034179073,
|
251 |
+
"loss": 3.1466,
|
252 |
+
"step": 36000
|
253 |
+
},
|
254 |
+
{
|
255 |
+
"epoch": 3.1,
|
256 |
+
"learning_rate": 0.00043185154120360664,
|
257 |
+
"loss": 3.1035,
|
258 |
+
"step": 37000
|
259 |
+
},
|
260 |
+
{
|
261 |
+
"epoch": 3.18,
|
262 |
+
"learning_rate": 0.0004266093520654225,
|
263 |
+
"loss": 3.0893,
|
264 |
+
"step": 38000
|
265 |
+
},
|
266 |
+
{
|
267 |
+
"epoch": 3.27,
|
268 |
+
"learning_rate": 0.0004213671629272384,
|
269 |
+
"loss": 3.1148,
|
270 |
+
"step": 39000
|
271 |
+
},
|
272 |
+
{
|
273 |
+
"epoch": 3.35,
|
274 |
+
"learning_rate": 0.00041612497378905436,
|
275 |
+
"loss": 3.0801,
|
276 |
+
"step": 40000
|
277 |
+
},
|
278 |
+
{
|
279 |
+
"epoch": 3.43,
|
280 |
+
"learning_rate": 0.0004108827846508702,
|
281 |
+
"loss": 3.0814,
|
282 |
+
"step": 41000
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 3.52,
|
286 |
+
"learning_rate": 0.0004056405955126861,
|
287 |
+
"loss": 3.0665,
|
288 |
+
"step": 42000
|
289 |
+
},
|
290 |
+
{
|
291 |
+
"epoch": 3.6,
|
292 |
+
"learning_rate": 0.00040039840637450197,
|
293 |
+
"loss": 3.0718,
|
294 |
+
"step": 43000
|
295 |
+
},
|
296 |
+
{
|
297 |
+
"epoch": 3.69,
|
298 |
+
"learning_rate": 0.0003951562172363179,
|
299 |
+
"loss": 3.0462,
|
300 |
+
"step": 44000
|
301 |
+
},
|
302 |
+
{
|
303 |
+
"epoch": 3.77,
|
304 |
+
"learning_rate": 0.0003899140280981338,
|
305 |
+
"loss": 3.0328,
|
306 |
+
"step": 45000
|
307 |
+
},
|
308 |
+
{
|
309 |
+
"epoch": 3.85,
|
310 |
+
"learning_rate": 0.0003846718389599497,
|
311 |
+
"loss": 3.0601,
|
312 |
+
"step": 46000
|
313 |
+
},
|
314 |
+
{
|
315 |
+
"epoch": 3.94,
|
316 |
+
"learning_rate": 0.0003794296498217656,
|
317 |
+
"loss": 3.0031,
|
318 |
+
"step": 47000
|
319 |
+
},
|
320 |
+
{
|
321 |
+
"epoch": 4.0,
|
322 |
+
"eval_accuracy": 0.3971048095550437,
|
323 |
+
"eval_loss": 3.0706326961517334,
|
324 |
+
"eval_runtime": 8.6604,
|
325 |
+
"eval_samples_per_second": 8082.758,
|
326 |
+
"eval_steps_per_second": 15.819,
|
327 |
+
"step": 47752
|
328 |
+
},
|
329 |
+
{
|
330 |
+
"epoch": 4.02,
|
331 |
+
"learning_rate": 0.00037418746068358145,
|
332 |
+
"loss": 3.0134,
|
333 |
+
"step": 48000
|
334 |
+
},
|
335 |
+
{
|
336 |
+
"epoch": 4.1,
|
337 |
+
"learning_rate": 0.00036894527154539736,
|
338 |
+
"loss": 3.0064,
|
339 |
+
"step": 49000
|
340 |
+
},
|
341 |
+
{
|
342 |
+
"epoch": 4.19,
|
343 |
+
"learning_rate": 0.00036370308240721327,
|
344 |
+
"loss": 2.9887,
|
345 |
+
"step": 50000
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 4.27,
|
349 |
+
"learning_rate": 0.0003584608932690291,
|
350 |
+
"loss": 2.9757,
|
351 |
+
"step": 51000
|
352 |
+
},
|
353 |
+
{
|
354 |
+
"epoch": 4.36,
|
355 |
+
"learning_rate": 0.0003532187041308451,
|
356 |
+
"loss": 2.9678,
|
357 |
+
"step": 52000
|
358 |
+
},
|
359 |
+
{
|
360 |
+
"epoch": 4.44,
|
361 |
+
"learning_rate": 0.00034797651499266093,
|
362 |
+
"loss": 2.964,
|
363 |
+
"step": 53000
|
364 |
+
},
|
365 |
+
{
|
366 |
+
"epoch": 4.52,
|
367 |
+
"learning_rate": 0.00034273432585447684,
|
368 |
+
"loss": 2.9642,
|
369 |
+
"step": 54000
|
370 |
+
},
|
371 |
+
{
|
372 |
+
"epoch": 4.61,
|
373 |
+
"learning_rate": 0.00033749213671629275,
|
374 |
+
"loss": 2.9797,
|
375 |
+
"step": 55000
|
376 |
+
},
|
377 |
+
{
|
378 |
+
"epoch": 4.69,
|
379 |
+
"learning_rate": 0.0003322499475781086,
|
380 |
+
"loss": 2.9534,
|
381 |
+
"step": 56000
|
382 |
+
},
|
383 |
+
{
|
384 |
+
"epoch": 4.77,
|
385 |
+
"learning_rate": 0.0003270077584399245,
|
386 |
+
"loss": 2.9476,
|
387 |
+
"step": 57000
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 4.86,
|
391 |
+
"learning_rate": 0.0003217655693017404,
|
392 |
+
"loss": 2.9431,
|
393 |
+
"step": 58000
|
394 |
+
},
|
395 |
+
{
|
396 |
+
"epoch": 4.94,
|
397 |
+
"learning_rate": 0.0003165233801635563,
|
398 |
+
"loss": 2.9668,
|
399 |
+
"step": 59000
|
400 |
+
},
|
401 |
+
{
|
402 |
+
"epoch": 5.0,
|
403 |
+
"eval_accuracy": 0.41213198432230425,
|
404 |
+
"eval_loss": 2.9605636596679688,
|
405 |
+
"eval_runtime": 8.7459,
|
406 |
+
"eval_samples_per_second": 8003.788,
|
407 |
+
"eval_steps_per_second": 15.665,
|
408 |
+
"step": 59690
|
409 |
+
},
|
410 |
+
{
|
411 |
+
"epoch": 5.03,
|
412 |
+
"learning_rate": 0.00031128119102537223,
|
413 |
+
"loss": 2.9106,
|
414 |
+
"step": 60000
|
415 |
+
},
|
416 |
+
{
|
417 |
+
"epoch": 5.11,
|
418 |
+
"learning_rate": 0.0003060390018871881,
|
419 |
+
"loss": 2.9053,
|
420 |
+
"step": 61000
|
421 |
+
},
|
422 |
+
{
|
423 |
+
"epoch": 5.19,
|
424 |
+
"learning_rate": 0.000300796812749004,
|
425 |
+
"loss": 2.8983,
|
426 |
+
"step": 62000
|
427 |
+
},
|
428 |
+
{
|
429 |
+
"epoch": 5.28,
|
430 |
+
"learning_rate": 0.00029555462361081984,
|
431 |
+
"loss": 2.8828,
|
432 |
+
"step": 63000
|
433 |
+
},
|
434 |
+
{
|
435 |
+
"epoch": 5.36,
|
436 |
+
"learning_rate": 0.0002903124344726358,
|
437 |
+
"loss": 2.8971,
|
438 |
+
"step": 64000
|
439 |
+
},
|
440 |
+
{
|
441 |
+
"epoch": 5.44,
|
442 |
+
"learning_rate": 0.0002850702453344517,
|
443 |
+
"loss": 2.9074,
|
444 |
+
"step": 65000
|
445 |
+
},
|
446 |
+
{
|
447 |
+
"epoch": 5.53,
|
448 |
+
"learning_rate": 0.00027982805619626756,
|
449 |
+
"loss": 2.8792,
|
450 |
+
"step": 66000
|
451 |
+
},
|
452 |
+
{
|
453 |
+
"epoch": 5.61,
|
454 |
+
"learning_rate": 0.00027458586705808347,
|
455 |
+
"loss": 2.8902,
|
456 |
+
"step": 67000
|
457 |
+
},
|
458 |
+
{
|
459 |
+
"epoch": 5.7,
|
460 |
+
"learning_rate": 0.0002693436779198993,
|
461 |
+
"loss": 2.8666,
|
462 |
+
"step": 68000
|
463 |
+
},
|
464 |
+
{
|
465 |
+
"epoch": 5.78,
|
466 |
+
"learning_rate": 0.00026410148878171523,
|
467 |
+
"loss": 2.8868,
|
468 |
+
"step": 69000
|
469 |
+
},
|
470 |
+
{
|
471 |
+
"epoch": 5.86,
|
472 |
+
"learning_rate": 0.0002588592996435312,
|
473 |
+
"loss": 2.8509,
|
474 |
+
"step": 70000
|
475 |
+
},
|
476 |
+
{
|
477 |
+
"epoch": 5.95,
|
478 |
+
"learning_rate": 0.00025361711050534704,
|
479 |
+
"loss": 2.8548,
|
480 |
+
"step": 71000
|
481 |
+
},
|
482 |
+
{
|
483 |
+
"epoch": 6.0,
|
484 |
+
"eval_accuracy": 0.4243559291065229,
|
485 |
+
"eval_loss": 2.8717477321624756,
|
486 |
+
"eval_runtime": 8.7188,
|
487 |
+
"eval_samples_per_second": 8028.594,
|
488 |
+
"eval_steps_per_second": 15.713,
|
489 |
+
"step": 71628
|
490 |
+
},
|
491 |
+
{
|
492 |
+
"epoch": 6.03,
|
493 |
+
"learning_rate": 0.00024837492136716295,
|
494 |
+
"loss": 2.8484,
|
495 |
+
"step": 72000
|
496 |
+
},
|
497 |
+
{
|
498 |
+
"epoch": 6.11,
|
499 |
+
"learning_rate": 0.00024313273222897883,
|
500 |
+
"loss": 2.8266,
|
501 |
+
"step": 73000
|
502 |
+
},
|
503 |
+
{
|
504 |
+
"epoch": 6.2,
|
505 |
+
"learning_rate": 0.0002378905430907947,
|
506 |
+
"loss": 2.8247,
|
507 |
+
"step": 74000
|
508 |
+
},
|
509 |
+
{
|
510 |
+
"epoch": 6.28,
|
511 |
+
"learning_rate": 0.00023264835395261061,
|
512 |
+
"loss": 2.8151,
|
513 |
+
"step": 75000
|
514 |
+
},
|
515 |
+
{
|
516 |
+
"epoch": 6.37,
|
517 |
+
"learning_rate": 0.0002274061648144265,
|
518 |
+
"loss": 2.8078,
|
519 |
+
"step": 76000
|
520 |
+
},
|
521 |
+
{
|
522 |
+
"epoch": 6.45,
|
523 |
+
"learning_rate": 0.0002221639756762424,
|
524 |
+
"loss": 2.8143,
|
525 |
+
"step": 77000
|
526 |
+
},
|
527 |
+
{
|
528 |
+
"epoch": 6.53,
|
529 |
+
"learning_rate": 0.0002169217865380583,
|
530 |
+
"loss": 2.8043,
|
531 |
+
"step": 78000
|
532 |
+
},
|
533 |
+
{
|
534 |
+
"epoch": 6.62,
|
535 |
+
"learning_rate": 0.0002116795973998742,
|
536 |
+
"loss": 2.8159,
|
537 |
+
"step": 79000
|
538 |
+
},
|
539 |
+
{
|
540 |
+
"epoch": 6.7,
|
541 |
+
"learning_rate": 0.00020643740826169007,
|
542 |
+
"loss": 2.7825,
|
543 |
+
"step": 80000
|
544 |
+
},
|
545 |
+
{
|
546 |
+
"epoch": 6.79,
|
547 |
+
"learning_rate": 0.00020119521912350598,
|
548 |
+
"loss": 2.812,
|
549 |
+
"step": 81000
|
550 |
+
},
|
551 |
+
{
|
552 |
+
"epoch": 6.87,
|
553 |
+
"learning_rate": 0.00019595302998532188,
|
554 |
+
"loss": 2.766,
|
555 |
+
"step": 82000
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"epoch": 6.95,
|
559 |
+
"learning_rate": 0.00019071084084713776,
|
560 |
+
"loss": 2.7686,
|
561 |
+
"step": 83000
|
562 |
+
},
|
563 |
+
{
|
564 |
+
"epoch": 7.0,
|
565 |
+
"eval_accuracy": 0.4311035312390553,
|
566 |
+
"eval_loss": 2.8147239685058594,
|
567 |
+
"eval_runtime": 8.6579,
|
568 |
+
"eval_samples_per_second": 8085.129,
|
569 |
+
"eval_steps_per_second": 15.824,
|
570 |
+
"step": 83566
|
571 |
+
},
|
572 |
+
{
|
573 |
+
"epoch": 7.04,
|
574 |
+
"learning_rate": 0.00018546865170895367,
|
575 |
+
"loss": 2.7648,
|
576 |
+
"step": 84000
|
577 |
+
},
|
578 |
+
{
|
579 |
+
"epoch": 7.12,
|
580 |
+
"learning_rate": 0.00018022646257076955,
|
581 |
+
"loss": 2.7609,
|
582 |
+
"step": 85000
|
583 |
+
},
|
584 |
+
{
|
585 |
+
"epoch": 7.2,
|
586 |
+
"learning_rate": 0.00017498427343258543,
|
587 |
+
"loss": 2.7306,
|
588 |
+
"step": 86000
|
589 |
+
},
|
590 |
+
{
|
591 |
+
"epoch": 7.29,
|
592 |
+
"learning_rate": 0.00016974208429440136,
|
593 |
+
"loss": 2.718,
|
594 |
+
"step": 87000
|
595 |
+
},
|
596 |
+
{
|
597 |
+
"epoch": 7.37,
|
598 |
+
"learning_rate": 0.00016449989515621724,
|
599 |
+
"loss": 2.7218,
|
600 |
+
"step": 88000
|
601 |
+
},
|
602 |
+
{
|
603 |
+
"epoch": 7.46,
|
604 |
+
"learning_rate": 0.00015925770601803312,
|
605 |
+
"loss": 2.7624,
|
606 |
+
"step": 89000
|
607 |
+
},
|
608 |
+
{
|
609 |
+
"epoch": 7.54,
|
610 |
+
"learning_rate": 0.00015401551687984903,
|
611 |
+
"loss": 2.7478,
|
612 |
+
"step": 90000
|
613 |
+
},
|
614 |
+
{
|
615 |
+
"epoch": 7.62,
|
616 |
+
"learning_rate": 0.0001487733277416649,
|
617 |
+
"loss": 2.7372,
|
618 |
+
"step": 91000
|
619 |
+
},
|
620 |
+
{
|
621 |
+
"epoch": 7.71,
|
622 |
+
"learning_rate": 0.00014353113860348082,
|
623 |
+
"loss": 2.7467,
|
624 |
+
"step": 92000
|
625 |
+
},
|
626 |
+
{
|
627 |
+
"epoch": 7.79,
|
628 |
+
"learning_rate": 0.00013828894946529672,
|
629 |
+
"loss": 2.7205,
|
630 |
+
"step": 93000
|
631 |
+
},
|
632 |
+
{
|
633 |
+
"epoch": 7.87,
|
634 |
+
"learning_rate": 0.0001330467603271126,
|
635 |
+
"loss": 2.7269,
|
636 |
+
"step": 94000
|
637 |
+
},
|
638 |
+
{
|
639 |
+
"epoch": 7.96,
|
640 |
+
"learning_rate": 0.00012780457118892848,
|
641 |
+
"loss": 2.7195,
|
642 |
+
"step": 95000
|
643 |
+
},
|
644 |
+
{
|
645 |
+
"epoch": 8.0,
|
646 |
+
"eval_accuracy": 0.4389084267882209,
|
647 |
+
"eval_loss": 2.7656898498535156,
|
648 |
+
"eval_runtime": 8.6162,
|
649 |
+
"eval_samples_per_second": 8124.227,
|
650 |
+
"eval_steps_per_second": 15.9,
|
651 |
+
"step": 95504
|
652 |
+
},
|
653 |
+
{
|
654 |
+
"epoch": 8.04,
|
655 |
+
"learning_rate": 0.0001225623820507444,
|
656 |
+
"loss": 2.7004,
|
657 |
+
"step": 96000
|
658 |
+
},
|
659 |
+
{
|
660 |
+
"epoch": 8.13,
|
661 |
+
"learning_rate": 0.00011732019291256028,
|
662 |
+
"loss": 2.7043,
|
663 |
+
"step": 97000
|
664 |
+
},
|
665 |
+
{
|
666 |
+
"epoch": 8.21,
|
667 |
+
"learning_rate": 0.00011207800377437619,
|
668 |
+
"loss": 2.7322,
|
669 |
+
"step": 98000
|
670 |
+
},
|
671 |
+
{
|
672 |
+
"epoch": 8.29,
|
673 |
+
"learning_rate": 0.00010683581463619207,
|
674 |
+
"loss": 2.6874,
|
675 |
+
"step": 99000
|
676 |
+
},
|
677 |
+
{
|
678 |
+
"epoch": 8.38,
|
679 |
+
"learning_rate": 0.00010159362549800798,
|
680 |
+
"loss": 2.6771,
|
681 |
+
"step": 100000
|
682 |
+
},
|
683 |
+
{
|
684 |
+
"epoch": 8.46,
|
685 |
+
"learning_rate": 9.635143635982387e-05,
|
686 |
+
"loss": 2.6827,
|
687 |
+
"step": 101000
|
688 |
+
},
|
689 |
+
{
|
690 |
+
"epoch": 8.54,
|
691 |
+
"learning_rate": 9.110924722163975e-05,
|
692 |
+
"loss": 2.6756,
|
693 |
+
"step": 102000
|
694 |
+
},
|
695 |
+
{
|
696 |
+
"epoch": 8.63,
|
697 |
+
"learning_rate": 8.586705808345566e-05,
|
698 |
+
"loss": 2.6575,
|
699 |
+
"step": 103000
|
700 |
+
},
|
701 |
+
{
|
702 |
+
"epoch": 8.71,
|
703 |
+
"learning_rate": 8.062486894527155e-05,
|
704 |
+
"loss": 2.669,
|
705 |
+
"step": 104000
|
706 |
+
},
|
707 |
+
{
|
708 |
+
"epoch": 8.8,
|
709 |
+
"learning_rate": 7.538267980708744e-05,
|
710 |
+
"loss": 2.6623,
|
711 |
+
"step": 105000
|
712 |
+
},
|
713 |
+
{
|
714 |
+
"epoch": 8.88,
|
715 |
+
"learning_rate": 7.014049066890334e-05,
|
716 |
+
"loss": 2.6518,
|
717 |
+
"step": 106000
|
718 |
+
},
|
719 |
+
{
|
720 |
+
"epoch": 8.96,
|
721 |
+
"learning_rate": 6.489830153071923e-05,
|
722 |
+
"loss": 2.6659,
|
723 |
+
"step": 107000
|
724 |
+
},
|
725 |
+
{
|
726 |
+
"epoch": 9.0,
|
727 |
+
"eval_accuracy": 0.4483279917760174,
|
728 |
+
"eval_loss": 2.710298538208008,
|
729 |
+
"eval_runtime": 8.5942,
|
730 |
+
"eval_samples_per_second": 8145.024,
|
731 |
+
"eval_steps_per_second": 15.941,
|
732 |
+
"step": 107442
|
733 |
+
},
|
734 |
+
{
|
735 |
+
"epoch": 9.05,
|
736 |
+
"learning_rate": 5.965611239253512e-05,
|
737 |
+
"loss": 2.6505,
|
738 |
+
"step": 108000
|
739 |
+
},
|
740 |
+
{
|
741 |
+
"epoch": 9.13,
|
742 |
+
"learning_rate": 5.441392325435102e-05,
|
743 |
+
"loss": 2.6367,
|
744 |
+
"step": 109000
|
745 |
+
},
|
746 |
+
{
|
747 |
+
"epoch": 9.21,
|
748 |
+
"learning_rate": 4.917173411616691e-05,
|
749 |
+
"loss": 2.6232,
|
750 |
+
"step": 110000
|
751 |
+
},
|
752 |
+
{
|
753 |
+
"epoch": 9.3,
|
754 |
+
"learning_rate": 4.3929544977982805e-05,
|
755 |
+
"loss": 2.6453,
|
756 |
+
"step": 111000
|
757 |
+
},
|
758 |
+
{
|
759 |
+
"epoch": 9.38,
|
760 |
+
"learning_rate": 3.8687355839798705e-05,
|
761 |
+
"loss": 2.6565,
|
762 |
+
"step": 112000
|
763 |
+
},
|
764 |
+
{
|
765 |
+
"epoch": 9.47,
|
766 |
+
"learning_rate": 3.344516670161459e-05,
|
767 |
+
"loss": 2.6244,
|
768 |
+
"step": 113000
|
769 |
+
},
|
770 |
+
{
|
771 |
+
"epoch": 9.55,
|
772 |
+
"learning_rate": 2.820297756343049e-05,
|
773 |
+
"loss": 2.6429,
|
774 |
+
"step": 114000
|
775 |
+
},
|
776 |
+
{
|
777 |
+
"epoch": 9.63,
|
778 |
+
"learning_rate": 2.2960788425246382e-05,
|
779 |
+
"loss": 2.6117,
|
780 |
+
"step": 115000
|
781 |
+
},
|
782 |
+
{
|
783 |
+
"epoch": 9.72,
|
784 |
+
"learning_rate": 1.771859928706228e-05,
|
785 |
+
"loss": 2.6289,
|
786 |
+
"step": 116000
|
787 |
+
},
|
788 |
+
{
|
789 |
+
"epoch": 9.8,
|
790 |
+
"learning_rate": 1.2476410148878172e-05,
|
791 |
+
"loss": 2.6174,
|
792 |
+
"step": 117000
|
793 |
+
},
|
794 |
+
{
|
795 |
+
"epoch": 9.88,
|
796 |
+
"learning_rate": 7.234221010694066e-06,
|
797 |
+
"loss": 2.6142,
|
798 |
+
"step": 118000
|
799 |
+
},
|
800 |
+
{
|
801 |
+
"epoch": 9.97,
|
802 |
+
"learning_rate": 1.99203187250996e-06,
|
803 |
+
"loss": 2.6254,
|
804 |
+
"step": 119000
|
805 |
+
},
|
806 |
+
{
|
807 |
+
"epoch": 10.0,
|
808 |
+
"eval_accuracy": 0.4510256799878438,
|
809 |
+
"eval_loss": 2.699005126953125,
|
810 |
+
"eval_runtime": 8.5626,
|
811 |
+
"eval_samples_per_second": 8175.05,
|
812 |
+
"eval_steps_per_second": 16.0,
|
813 |
+
"step": 119380
|
814 |
+
},
|
815 |
+
{
|
816 |
+
"epoch": 10.0,
|
817 |
+
"step": 119380,
|
818 |
+
"total_flos": 6556448698096128.0,
|
819 |
+
"train_loss": 3.079910436302079,
|
820 |
+
"train_runtime": 7252.3341,
|
821 |
+
"train_samples_per_second": 1053.439,
|
822 |
+
"train_steps_per_second": 16.461
|
823 |
+
}
|
824 |
+
],
|
825 |
+
"logging_steps": 1000,
|
826 |
+
"max_steps": 119380,
|
827 |
+
"num_train_epochs": 10,
|
828 |
+
"save_steps": 2000,
|
829 |
+
"total_flos": 6556448698096128.0,
|
830 |
+
"trial_name": null,
|
831 |
+
"trial_params": null
|
832 |
+
}
|