End of training
Browse files- README.md +7 -2
- all_results.json +9 -9
- runs/Oct16_20-33-19_workload-ai-workshop/events.out.tfevents.1729112043.workload-ai-workshop.16989.1 +3 -0
- test_results.json +5 -5
- train_results.json +5 -5
- trainer_state.json +316 -316
- validation_results.json +5 -5
README.md
CHANGED
@@ -4,7 +4,7 @@ base_model: BridgeTower/bridgetower-large-itm-mlm-itc
|
|
4 |
tags:
|
5 |
- generated_from_trainer
|
6 |
datasets:
|
7 |
-
- newyorker_caption_contest
|
8 |
model-index:
|
9 |
- name: bridgetower
|
10 |
results: []
|
@@ -15,7 +15,12 @@ should probably proofread and complete it, then remove this comment. -->
|
|
15 |
|
16 |
# bridgetower
|
17 |
|
18 |
-
This model is a fine-tuned version of [BridgeTower/bridgetower-large-itm-mlm-itc](https://huggingface.co/BridgeTower/bridgetower-large-itm-mlm-itc) on the newyorker_caption_contest dataset.
|
|
|
|
|
|
|
|
|
|
|
19 |
|
20 |
## Model description
|
21 |
|
|
|
4 |
tags:
|
5 |
- generated_from_trainer
|
6 |
datasets:
|
7 |
+
- jmhessel/newyorker_caption_contest
|
8 |
model-index:
|
9 |
- name: bridgetower
|
10 |
results: []
|
|
|
15 |
|
16 |
# bridgetower
|
17 |
|
18 |
+
This model is a fine-tuned version of [BridgeTower/bridgetower-large-itm-mlm-itc](https://huggingface.co/BridgeTower/bridgetower-large-itm-mlm-itc) on the jmhessel/newyorker_caption_contest matching dataset.
|
19 |
+
It achieves the following results on the evaluation set:
|
20 |
+
- Loss: 0.1243
|
21 |
+
- Memory Allocated (gb): 51.27
|
22 |
+
- Max Memory Allocated (gb): 60.52
|
23 |
+
- Total Memory Available (gb): 94.62
|
24 |
|
25 |
## Model description
|
26 |
|
all_results.json
CHANGED
@@ -1,15 +1,15 @@
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
-
"eval_loss": 0.
|
4 |
-
"eval_runtime":
|
5 |
-
"eval_samples_per_second":
|
6 |
-
"eval_steps_per_second":
|
7 |
-
"max_memory_allocated (GB)":
|
8 |
"memory_allocated (GB)": 51.27,
|
9 |
"total_flos": 3.0598946525952e+16,
|
10 |
"total_memory_available (GB)": 94.62,
|
11 |
-
"train_loss": 0.
|
12 |
-
"train_runtime":
|
13 |
-
"train_samples_per_second":
|
14 |
-
"train_steps_per_second": 1.
|
15 |
}
|
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
+
"eval_loss": 0.12431247532367706,
|
4 |
+
"eval_runtime": 8.9978,
|
5 |
+
"eval_samples_per_second": 65.301,
|
6 |
+
"eval_steps_per_second": 4.081,
|
7 |
+
"max_memory_allocated (GB)": 60.52,
|
8 |
"memory_allocated (GB)": 51.27,
|
9 |
"total_flos": 3.0598946525952e+16,
|
10 |
"total_memory_available (GB)": 94.62,
|
11 |
+
"train_loss": 0.06082177159737567,
|
12 |
+
"train_runtime": 1079.1005,
|
13 |
+
"train_samples_per_second": 52.128,
|
14 |
+
"train_steps_per_second": 1.304
|
15 |
}
|
runs/Oct16_20-33-19_workload-ai-workshop/events.out.tfevents.1729112043.workload-ai-workshop.16989.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f5e47f19db166d9f51f5eb93aacd1eb69eddbf6ccdd0478eec9ad6f1192e5948
|
3 |
+
size 998
|
test_results.json
CHANGED
@@ -1,10 +1,10 @@
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
-
"eval_loss": 0.
|
4 |
-
"eval_runtime":
|
5 |
-
"eval_samples_per_second":
|
6 |
-
"eval_steps_per_second":
|
7 |
-
"max_memory_allocated (GB)":
|
8 |
"memory_allocated (GB)": 51.27,
|
9 |
"total_memory_available (GB)": 94.62
|
10 |
}
|
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
+
"eval_loss": 0.12431247532367706,
|
4 |
+
"eval_runtime": 8.9978,
|
5 |
+
"eval_samples_per_second": 65.301,
|
6 |
+
"eval_steps_per_second": 4.081,
|
7 |
+
"max_memory_allocated (GB)": 60.52,
|
8 |
"memory_allocated (GB)": 51.27,
|
9 |
"total_memory_available (GB)": 94.62
|
10 |
}
|
train_results.json
CHANGED
@@ -1,11 +1,11 @@
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
-
"max_memory_allocated (GB)":
|
4 |
"memory_allocated (GB)": 50.57,
|
5 |
"total_flos": 3.0598946525952e+16,
|
6 |
"total_memory_available (GB)": 94.62,
|
7 |
-
"train_loss": 0.
|
8 |
-
"train_runtime":
|
9 |
-
"train_samples_per_second":
|
10 |
-
"train_steps_per_second": 1.
|
11 |
}
|
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
+
"max_memory_allocated (GB)": 60.52,
|
4 |
"memory_allocated (GB)": 50.57,
|
5 |
"total_flos": 3.0598946525952e+16,
|
6 |
"total_memory_available (GB)": 94.62,
|
7 |
+
"train_loss": 0.06082177159737567,
|
8 |
+
"train_runtime": 1079.1005,
|
9 |
+
"train_samples_per_second": 52.128,
|
10 |
+
"train_steps_per_second": 1.304
|
11 |
}
|
trainer_state.json
CHANGED
@@ -10,9 +10,9 @@
|
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 0.04081632653061224,
|
13 |
-
"grad_norm":
|
14 |
"learning_rate": 9.918367346938776e-06,
|
15 |
-
"loss": 0.
|
16 |
"max_memory_allocated (GB)": 57.18,
|
17 |
"memory_allocated (GB)": 50.57,
|
18 |
"step": 10,
|
@@ -20,9 +20,9 @@
|
|
20 |
},
|
21 |
{
|
22 |
"epoch": 0.08163265306122448,
|
23 |
-
"grad_norm":
|
24 |
"learning_rate": 9.836734693877552e-06,
|
25 |
-
"loss": 0.
|
26 |
"max_memory_allocated (GB)": 57.18,
|
27 |
"memory_allocated (GB)": 50.57,
|
28 |
"step": 20,
|
@@ -30,9 +30,9 @@
|
|
30 |
},
|
31 |
{
|
32 |
"epoch": 0.12244897959183673,
|
33 |
-
"grad_norm": 5.
|
34 |
"learning_rate": 9.755102040816327e-06,
|
35 |
-
"loss": 0.
|
36 |
"max_memory_allocated (GB)": 57.18,
|
37 |
"memory_allocated (GB)": 50.57,
|
38 |
"step": 30,
|
@@ -40,9 +40,9 @@
|
|
40 |
},
|
41 |
{
|
42 |
"epoch": 0.16326530612244897,
|
43 |
-
"grad_norm": 4.
|
44 |
"learning_rate": 9.673469387755103e-06,
|
45 |
-
"loss": 0.
|
46 |
"max_memory_allocated (GB)": 57.18,
|
47 |
"memory_allocated (GB)": 50.57,
|
48 |
"step": 40,
|
@@ -50,9 +50,9 @@
|
|
50 |
},
|
51 |
{
|
52 |
"epoch": 0.20408163265306123,
|
53 |
-
"grad_norm": 4.
|
54 |
"learning_rate": 9.591836734693878e-06,
|
55 |
-
"loss": 0.
|
56 |
"max_memory_allocated (GB)": 57.18,
|
57 |
"memory_allocated (GB)": 50.57,
|
58 |
"step": 50,
|
@@ -60,9 +60,9 @@
|
|
60 |
},
|
61 |
{
|
62 |
"epoch": 0.24489795918367346,
|
63 |
-
"grad_norm": 2.
|
64 |
"learning_rate": 9.510204081632653e-06,
|
65 |
-
"loss": 0.
|
66 |
"max_memory_allocated (GB)": 57.18,
|
67 |
"memory_allocated (GB)": 50.57,
|
68 |
"step": 60,
|
@@ -70,7 +70,7 @@
|
|
70 |
},
|
71 |
{
|
72 |
"epoch": 0.2857142857142857,
|
73 |
-
"grad_norm":
|
74 |
"learning_rate": 9.42857142857143e-06,
|
75 |
"loss": 0.1015,
|
76 |
"max_memory_allocated (GB)": 57.18,
|
@@ -80,9 +80,9 @@
|
|
80 |
},
|
81 |
{
|
82 |
"epoch": 0.32653061224489793,
|
83 |
-
"grad_norm": 1.
|
84 |
"learning_rate": 9.346938775510204e-06,
|
85 |
-
"loss": 0.
|
86 |
"max_memory_allocated (GB)": 57.18,
|
87 |
"memory_allocated (GB)": 50.57,
|
88 |
"step": 80,
|
@@ -90,9 +90,9 @@
|
|
90 |
},
|
91 |
{
|
92 |
"epoch": 0.3673469387755102,
|
93 |
-
"grad_norm":
|
94 |
"learning_rate": 9.26530612244898e-06,
|
95 |
-
"loss": 0.
|
96 |
"max_memory_allocated (GB)": 57.18,
|
97 |
"memory_allocated (GB)": 50.57,
|
98 |
"step": 90,
|
@@ -100,9 +100,9 @@
|
|
100 |
},
|
101 |
{
|
102 |
"epoch": 0.40816326530612246,
|
103 |
-
"grad_norm":
|
104 |
"learning_rate": 9.183673469387756e-06,
|
105 |
-
"loss": 0.
|
106 |
"max_memory_allocated (GB)": 57.18,
|
107 |
"memory_allocated (GB)": 50.57,
|
108 |
"step": 100,
|
@@ -110,9 +110,9 @@
|
|
110 |
},
|
111 |
{
|
112 |
"epoch": 0.4489795918367347,
|
113 |
-
"grad_norm": 3.
|
114 |
"learning_rate": 9.102040816326532e-06,
|
115 |
-
"loss": 0.
|
116 |
"max_memory_allocated (GB)": 57.18,
|
117 |
"memory_allocated (GB)": 50.57,
|
118 |
"step": 110,
|
@@ -120,9 +120,9 @@
|
|
120 |
},
|
121 |
{
|
122 |
"epoch": 0.4897959183673469,
|
123 |
-
"grad_norm":
|
124 |
"learning_rate": 9.020408163265307e-06,
|
125 |
-
"loss": 0.
|
126 |
"max_memory_allocated (GB)": 57.18,
|
127 |
"memory_allocated (GB)": 50.57,
|
128 |
"step": 120,
|
@@ -130,9 +130,9 @@
|
|
130 |
},
|
131 |
{
|
132 |
"epoch": 0.5306122448979592,
|
133 |
-
"grad_norm":
|
134 |
"learning_rate": 8.938775510204082e-06,
|
135 |
-
"loss": 0.
|
136 |
"max_memory_allocated (GB)": 57.18,
|
137 |
"memory_allocated (GB)": 50.57,
|
138 |
"step": 130,
|
@@ -140,9 +140,9 @@
|
|
140 |
},
|
141 |
{
|
142 |
"epoch": 0.5714285714285714,
|
143 |
-
"grad_norm": 1.
|
144 |
"learning_rate": 8.857142857142858e-06,
|
145 |
-
"loss": 0.
|
146 |
"max_memory_allocated (GB)": 57.18,
|
147 |
"memory_allocated (GB)": 50.57,
|
148 |
"step": 140,
|
@@ -150,9 +150,9 @@
|
|
150 |
},
|
151 |
{
|
152 |
"epoch": 0.6122448979591837,
|
153 |
-
"grad_norm": 2.
|
154 |
"learning_rate": 8.775510204081633e-06,
|
155 |
-
"loss": 0.
|
156 |
"max_memory_allocated (GB)": 57.18,
|
157 |
"memory_allocated (GB)": 50.57,
|
158 |
"step": 150,
|
@@ -160,9 +160,9 @@
|
|
160 |
},
|
161 |
{
|
162 |
"epoch": 0.6530612244897959,
|
163 |
-
"grad_norm":
|
164 |
"learning_rate": 8.69387755102041e-06,
|
165 |
-
"loss": 0.
|
166 |
"max_memory_allocated (GB)": 57.18,
|
167 |
"memory_allocated (GB)": 50.57,
|
168 |
"step": 160,
|
@@ -170,9 +170,9 @@
|
|
170 |
},
|
171 |
{
|
172 |
"epoch": 0.6938775510204082,
|
173 |
-
"grad_norm":
|
174 |
"learning_rate": 8.612244897959184e-06,
|
175 |
-
"loss": 0.
|
176 |
"max_memory_allocated (GB)": 57.18,
|
177 |
"memory_allocated (GB)": 50.57,
|
178 |
"step": 170,
|
@@ -180,9 +180,9 @@
|
|
180 |
},
|
181 |
{
|
182 |
"epoch": 0.7346938775510204,
|
183 |
-
"grad_norm": 1.
|
184 |
"learning_rate": 8.530612244897961e-06,
|
185 |
-
"loss": 0.
|
186 |
"max_memory_allocated (GB)": 57.18,
|
187 |
"memory_allocated (GB)": 50.57,
|
188 |
"step": 180,
|
@@ -190,9 +190,9 @@
|
|
190 |
},
|
191 |
{
|
192 |
"epoch": 0.7755102040816326,
|
193 |
-
"grad_norm":
|
194 |
"learning_rate": 8.448979591836736e-06,
|
195 |
-
"loss": 0.
|
196 |
"max_memory_allocated (GB)": 57.18,
|
197 |
"memory_allocated (GB)": 50.57,
|
198 |
"step": 190,
|
@@ -200,9 +200,9 @@
|
|
200 |
},
|
201 |
{
|
202 |
"epoch": 0.8163265306122449,
|
203 |
-
"grad_norm": 0.
|
204 |
"learning_rate": 8.36734693877551e-06,
|
205 |
-
"loss": 0.
|
206 |
"max_memory_allocated (GB)": 57.18,
|
207 |
"memory_allocated (GB)": 50.57,
|
208 |
"step": 200,
|
@@ -210,9 +210,9 @@
|
|
210 |
},
|
211 |
{
|
212 |
"epoch": 0.8571428571428571,
|
213 |
-
"grad_norm": 0.
|
214 |
"learning_rate": 8.285714285714287e-06,
|
215 |
-
"loss": 0.
|
216 |
"max_memory_allocated (GB)": 57.18,
|
217 |
"memory_allocated (GB)": 50.57,
|
218 |
"step": 210,
|
@@ -220,9 +220,9 @@
|
|
220 |
},
|
221 |
{
|
222 |
"epoch": 0.8979591836734694,
|
223 |
-
"grad_norm": 1.
|
224 |
"learning_rate": 8.204081632653062e-06,
|
225 |
-
"loss": 0.
|
226 |
"max_memory_allocated (GB)": 57.18,
|
227 |
"memory_allocated (GB)": 50.57,
|
228 |
"step": 220,
|
@@ -230,9 +230,9 @@
|
|
230 |
},
|
231 |
{
|
232 |
"epoch": 0.9387755102040817,
|
233 |
-
"grad_norm": 0.
|
234 |
"learning_rate": 8.122448979591837e-06,
|
235 |
-
"loss": 0.
|
236 |
"max_memory_allocated (GB)": 57.18,
|
237 |
"memory_allocated (GB)": 50.57,
|
238 |
"step": 230,
|
@@ -240,9 +240,9 @@
|
|
240 |
},
|
241 |
{
|
242 |
"epoch": 0.9795918367346939,
|
243 |
-
"grad_norm": 0.
|
244 |
"learning_rate": 8.040816326530613e-06,
|
245 |
-
"loss": 0.
|
246 |
"max_memory_allocated (GB)": 57.18,
|
247 |
"memory_allocated (GB)": 50.57,
|
248 |
"step": 240,
|
@@ -250,9 +250,9 @@
|
|
250 |
},
|
251 |
{
|
252 |
"epoch": 1.0204081632653061,
|
253 |
-
"grad_norm":
|
254 |
"learning_rate": 7.959183673469388e-06,
|
255 |
-
"loss": 0.
|
256 |
"max_memory_allocated (GB)": 57.18,
|
257 |
"memory_allocated (GB)": 50.57,
|
258 |
"step": 250,
|
@@ -260,9 +260,9 @@
|
|
260 |
},
|
261 |
{
|
262 |
"epoch": 1.0612244897959184,
|
263 |
-
"grad_norm":
|
264 |
"learning_rate": 7.877551020408164e-06,
|
265 |
-
"loss": 0.
|
266 |
"max_memory_allocated (GB)": 57.18,
|
267 |
"memory_allocated (GB)": 50.57,
|
268 |
"step": 260,
|
@@ -270,9 +270,9 @@
|
|
270 |
},
|
271 |
{
|
272 |
"epoch": 1.1020408163265305,
|
273 |
-
"grad_norm": 0.
|
274 |
"learning_rate": 7.79591836734694e-06,
|
275 |
-
"loss": 0.
|
276 |
"max_memory_allocated (GB)": 57.18,
|
277 |
"memory_allocated (GB)": 50.57,
|
278 |
"step": 270,
|
@@ -280,9 +280,9 @@
|
|
280 |
},
|
281 |
{
|
282 |
"epoch": 1.1428571428571428,
|
283 |
-
"grad_norm":
|
284 |
"learning_rate": 7.714285714285716e-06,
|
285 |
-
"loss": 0.
|
286 |
"max_memory_allocated (GB)": 57.18,
|
287 |
"memory_allocated (GB)": 50.57,
|
288 |
"step": 280,
|
@@ -290,9 +290,9 @@
|
|
290 |
},
|
291 |
{
|
292 |
"epoch": 1.183673469387755,
|
293 |
-
"grad_norm":
|
294 |
"learning_rate": 7.63265306122449e-06,
|
295 |
-
"loss": 0.
|
296 |
"max_memory_allocated (GB)": 57.18,
|
297 |
"memory_allocated (GB)": 50.57,
|
298 |
"step": 290,
|
@@ -300,9 +300,9 @@
|
|
300 |
},
|
301 |
{
|
302 |
"epoch": 1.2244897959183674,
|
303 |
-
"grad_norm": 1.
|
304 |
"learning_rate": 7.551020408163265e-06,
|
305 |
-
"loss": 0.
|
306 |
"max_memory_allocated (GB)": 57.18,
|
307 |
"memory_allocated (GB)": 50.57,
|
308 |
"step": 300,
|
@@ -310,7 +310,7 @@
|
|
310 |
},
|
311 |
{
|
312 |
"epoch": 1.2653061224489797,
|
313 |
-
"grad_norm": 1.
|
314 |
"learning_rate": 7.469387755102041e-06,
|
315 |
"loss": 0.0668,
|
316 |
"max_memory_allocated (GB)": 57.18,
|
@@ -320,9 +320,9 @@
|
|
320 |
},
|
321 |
{
|
322 |
"epoch": 1.306122448979592,
|
323 |
-
"grad_norm":
|
324 |
"learning_rate": 7.387755102040817e-06,
|
325 |
-
"loss": 0.
|
326 |
"max_memory_allocated (GB)": 57.18,
|
327 |
"memory_allocated (GB)": 50.57,
|
328 |
"step": 320,
|
@@ -330,9 +330,9 @@
|
|
330 |
},
|
331 |
{
|
332 |
"epoch": 1.346938775510204,
|
333 |
-
"grad_norm": 0.
|
334 |
"learning_rate": 7.306122448979592e-06,
|
335 |
-
"loss": 0.
|
336 |
"max_memory_allocated (GB)": 57.18,
|
337 |
"memory_allocated (GB)": 50.57,
|
338 |
"step": 330,
|
@@ -340,7 +340,7 @@
|
|
340 |
},
|
341 |
{
|
342 |
"epoch": 1.3877551020408163,
|
343 |
-
"grad_norm": 1.
|
344 |
"learning_rate": 7.224489795918368e-06,
|
345 |
"loss": 0.055,
|
346 |
"max_memory_allocated (GB)": 57.18,
|
@@ -350,9 +350,9 @@
|
|
350 |
},
|
351 |
{
|
352 |
"epoch": 1.4285714285714286,
|
353 |
-
"grad_norm": 0.
|
354 |
"learning_rate": 7.1428571428571436e-06,
|
355 |
-
"loss": 0.
|
356 |
"max_memory_allocated (GB)": 57.18,
|
357 |
"memory_allocated (GB)": 50.57,
|
358 |
"step": 350,
|
@@ -360,9 +360,9 @@
|
|
360 |
},
|
361 |
{
|
362 |
"epoch": 1.469387755102041,
|
363 |
-
"grad_norm":
|
364 |
"learning_rate": 7.061224489795919e-06,
|
365 |
-
"loss": 0.
|
366 |
"max_memory_allocated (GB)": 57.18,
|
367 |
"memory_allocated (GB)": 50.57,
|
368 |
"step": 360,
|
@@ -370,9 +370,9 @@
|
|
370 |
},
|
371 |
{
|
372 |
"epoch": 1.510204081632653,
|
373 |
-
"grad_norm":
|
374 |
"learning_rate": 6.979591836734695e-06,
|
375 |
-
"loss": 0.
|
376 |
"max_memory_allocated (GB)": 57.18,
|
377 |
"memory_allocated (GB)": 50.57,
|
378 |
"step": 370,
|
@@ -380,9 +380,9 @@
|
|
380 |
},
|
381 |
{
|
382 |
"epoch": 1.5510204081632653,
|
383 |
-
"grad_norm": 0.
|
384 |
"learning_rate": 6.8979591836734705e-06,
|
385 |
-
"loss": 0.
|
386 |
"max_memory_allocated (GB)": 57.18,
|
387 |
"memory_allocated (GB)": 50.57,
|
388 |
"step": 380,
|
@@ -390,9 +390,9 @@
|
|
390 |
},
|
391 |
{
|
392 |
"epoch": 1.5918367346938775,
|
393 |
-
"grad_norm":
|
394 |
"learning_rate": 6.816326530612245e-06,
|
395 |
-
"loss": 0.
|
396 |
"max_memory_allocated (GB)": 57.18,
|
397 |
"memory_allocated (GB)": 50.57,
|
398 |
"step": 390,
|
@@ -400,9 +400,9 @@
|
|
400 |
},
|
401 |
{
|
402 |
"epoch": 1.6326530612244898,
|
403 |
-
"grad_norm": 0.
|
404 |
"learning_rate": 6.734693877551021e-06,
|
405 |
-
"loss": 0.
|
406 |
"max_memory_allocated (GB)": 57.18,
|
407 |
"memory_allocated (GB)": 50.57,
|
408 |
"step": 400,
|
@@ -410,9 +410,9 @@
|
|
410 |
},
|
411 |
{
|
412 |
"epoch": 1.6734693877551021,
|
413 |
-
"grad_norm": 0.
|
414 |
"learning_rate": 6.653061224489797e-06,
|
415 |
-
"loss": 0.
|
416 |
"max_memory_allocated (GB)": 57.18,
|
417 |
"memory_allocated (GB)": 50.57,
|
418 |
"step": 410,
|
@@ -420,9 +420,9 @@
|
|
420 |
},
|
421 |
{
|
422 |
"epoch": 1.7142857142857144,
|
423 |
-
"grad_norm":
|
424 |
"learning_rate": 6.571428571428572e-06,
|
425 |
-
"loss": 0.
|
426 |
"max_memory_allocated (GB)": 57.18,
|
427 |
"memory_allocated (GB)": 50.57,
|
428 |
"step": 420,
|
@@ -430,9 +430,9 @@
|
|
430 |
},
|
431 |
{
|
432 |
"epoch": 1.7551020408163265,
|
433 |
-
"grad_norm": 0.
|
434 |
"learning_rate": 6.489795918367348e-06,
|
435 |
-
"loss": 0.
|
436 |
"max_memory_allocated (GB)": 57.18,
|
437 |
"memory_allocated (GB)": 50.57,
|
438 |
"step": 430,
|
@@ -440,7 +440,7 @@
|
|
440 |
},
|
441 |
{
|
442 |
"epoch": 1.7959183673469388,
|
443 |
-
"grad_norm":
|
444 |
"learning_rate": 6.408163265306124e-06,
|
445 |
"loss": 0.0606,
|
446 |
"max_memory_allocated (GB)": 57.18,
|
@@ -450,9 +450,9 @@
|
|
450 |
},
|
451 |
{
|
452 |
"epoch": 1.836734693877551,
|
453 |
-
"grad_norm": 0.
|
454 |
"learning_rate": 6.326530612244899e-06,
|
455 |
-
"loss": 0.
|
456 |
"max_memory_allocated (GB)": 57.18,
|
457 |
"memory_allocated (GB)": 50.57,
|
458 |
"step": 450,
|
@@ -460,9 +460,9 @@
|
|
460 |
},
|
461 |
{
|
462 |
"epoch": 1.8775510204081631,
|
463 |
-
"grad_norm": 0.
|
464 |
"learning_rate": 6.244897959183675e-06,
|
465 |
-
"loss": 0.
|
466 |
"max_memory_allocated (GB)": 57.18,
|
467 |
"memory_allocated (GB)": 50.57,
|
468 |
"step": 460,
|
@@ -470,7 +470,7 @@
|
|
470 |
},
|
471 |
{
|
472 |
"epoch": 1.9183673469387754,
|
473 |
-
"grad_norm":
|
474 |
"learning_rate": 6.163265306122449e-06,
|
475 |
"loss": 0.0427,
|
476 |
"max_memory_allocated (GB)": 57.18,
|
@@ -480,9 +480,9 @@
|
|
480 |
},
|
481 |
{
|
482 |
"epoch": 1.9591836734693877,
|
483 |
-
"grad_norm": 0.
|
484 |
"learning_rate": 6.0816326530612245e-06,
|
485 |
-
"loss": 0.
|
486 |
"max_memory_allocated (GB)": 57.18,
|
487 |
"memory_allocated (GB)": 50.57,
|
488 |
"step": 480,
|
@@ -490,755 +490,755 @@
|
|
490 |
},
|
491 |
{
|
492 |
"epoch": 2.0,
|
493 |
-
"grad_norm": 0.
|
494 |
"learning_rate": 6e-06,
|
495 |
-
"loss": 0.
|
496 |
-
"max_memory_allocated (GB)":
|
497 |
"memory_allocated (GB)": 50.57,
|
498 |
"step": 490,
|
499 |
"total_memory_available (GB)": 94.62
|
500 |
},
|
501 |
{
|
502 |
"epoch": 2.0408163265306123,
|
503 |
-
"grad_norm": 0.
|
504 |
"learning_rate": 5.918367346938776e-06,
|
505 |
-
"loss": 0.
|
506 |
-
"max_memory_allocated (GB)":
|
507 |
"memory_allocated (GB)": 50.57,
|
508 |
"step": 500,
|
509 |
"total_memory_available (GB)": 94.62
|
510 |
},
|
511 |
{
|
512 |
"epoch": 2.0816326530612246,
|
513 |
-
"grad_norm": 0.
|
514 |
"learning_rate": 5.8367346938775515e-06,
|
515 |
-
"loss": 0.
|
516 |
-
"max_memory_allocated (GB)":
|
517 |
"memory_allocated (GB)": 50.57,
|
518 |
"step": 510,
|
519 |
"total_memory_available (GB)": 94.62
|
520 |
},
|
521 |
{
|
522 |
"epoch": 2.122448979591837,
|
523 |
-
"grad_norm": 1.
|
524 |
"learning_rate": 5.755102040816327e-06,
|
525 |
-
"loss": 0.
|
526 |
-
"max_memory_allocated (GB)":
|
527 |
"memory_allocated (GB)": 50.57,
|
528 |
"step": 520,
|
529 |
"total_memory_available (GB)": 94.62
|
530 |
},
|
531 |
{
|
532 |
"epoch": 2.163265306122449,
|
533 |
-
"grad_norm": 1.
|
534 |
"learning_rate": 5.673469387755103e-06,
|
535 |
-
"loss": 0.
|
536 |
-
"max_memory_allocated (GB)":
|
537 |
"memory_allocated (GB)": 50.57,
|
538 |
"step": 530,
|
539 |
"total_memory_available (GB)": 94.62
|
540 |
},
|
541 |
{
|
542 |
"epoch": 2.204081632653061,
|
543 |
-
"grad_norm": 0.
|
544 |
"learning_rate": 5.591836734693878e-06,
|
545 |
-
"loss": 0.
|
546 |
-
"max_memory_allocated (GB)":
|
547 |
"memory_allocated (GB)": 50.57,
|
548 |
"step": 540,
|
549 |
"total_memory_available (GB)": 94.62
|
550 |
},
|
551 |
{
|
552 |
"epoch": 2.2448979591836733,
|
553 |
-
"grad_norm": 1.
|
554 |
"learning_rate": 5.510204081632653e-06,
|
555 |
-
"loss": 0.
|
556 |
-
"max_memory_allocated (GB)":
|
557 |
"memory_allocated (GB)": 50.57,
|
558 |
"step": 550,
|
559 |
"total_memory_available (GB)": 94.62
|
560 |
},
|
561 |
{
|
562 |
"epoch": 2.2857142857142856,
|
563 |
-
"grad_norm":
|
564 |
"learning_rate": 5.428571428571429e-06,
|
565 |
-
"loss": 0.
|
566 |
-
"max_memory_allocated (GB)":
|
567 |
"memory_allocated (GB)": 50.57,
|
568 |
"step": 560,
|
569 |
"total_memory_available (GB)": 94.62
|
570 |
},
|
571 |
{
|
572 |
"epoch": 2.326530612244898,
|
573 |
-
"grad_norm": 0.
|
574 |
"learning_rate": 5.3469387755102045e-06,
|
575 |
-
"loss": 0.
|
576 |
-
"max_memory_allocated (GB)":
|
577 |
"memory_allocated (GB)": 50.57,
|
578 |
"step": 570,
|
579 |
"total_memory_available (GB)": 94.62
|
580 |
},
|
581 |
{
|
582 |
"epoch": 2.36734693877551,
|
583 |
-
"grad_norm":
|
584 |
"learning_rate": 5.26530612244898e-06,
|
585 |
-
"loss": 0.
|
586 |
-
"max_memory_allocated (GB)":
|
587 |
"memory_allocated (GB)": 50.57,
|
588 |
"step": 580,
|
589 |
"total_memory_available (GB)": 94.62
|
590 |
},
|
591 |
{
|
592 |
"epoch": 2.4081632653061225,
|
593 |
-
"grad_norm": 0.
|
594 |
"learning_rate": 5.183673469387756e-06,
|
595 |
-
"loss": 0.
|
596 |
-
"max_memory_allocated (GB)":
|
597 |
"memory_allocated (GB)": 50.57,
|
598 |
"step": 590,
|
599 |
"total_memory_available (GB)": 94.62
|
600 |
},
|
601 |
{
|
602 |
"epoch": 2.4489795918367347,
|
603 |
-
"grad_norm":
|
604 |
"learning_rate": 5.1020408163265315e-06,
|
605 |
-
"loss": 0.
|
606 |
-
"max_memory_allocated (GB)":
|
607 |
"memory_allocated (GB)": 50.57,
|
608 |
"step": 600,
|
609 |
"total_memory_available (GB)": 94.62
|
610 |
},
|
611 |
{
|
612 |
"epoch": 2.489795918367347,
|
613 |
-
"grad_norm": 0.
|
614 |
"learning_rate": 5.020408163265307e-06,
|
615 |
-
"loss": 0.
|
616 |
-
"max_memory_allocated (GB)":
|
617 |
"memory_allocated (GB)": 50.57,
|
618 |
"step": 610,
|
619 |
"total_memory_available (GB)": 94.62
|
620 |
},
|
621 |
{
|
622 |
"epoch": 2.5306122448979593,
|
623 |
-
"grad_norm":
|
624 |
"learning_rate": 4.938775510204082e-06,
|
625 |
-
"loss": 0.
|
626 |
-
"max_memory_allocated (GB)":
|
627 |
"memory_allocated (GB)": 50.57,
|
628 |
"step": 620,
|
629 |
"total_memory_available (GB)": 94.62
|
630 |
},
|
631 |
{
|
632 |
"epoch": 2.571428571428571,
|
633 |
-
"grad_norm": 0.
|
634 |
"learning_rate": 4.857142857142858e-06,
|
635 |
-
"loss": 0.
|
636 |
-
"max_memory_allocated (GB)":
|
637 |
"memory_allocated (GB)": 50.57,
|
638 |
"step": 630,
|
639 |
"total_memory_available (GB)": 94.62
|
640 |
},
|
641 |
{
|
642 |
"epoch": 2.612244897959184,
|
643 |
-
"grad_norm":
|
644 |
"learning_rate": 4.775510204081633e-06,
|
645 |
-
"loss": 0.
|
646 |
-
"max_memory_allocated (GB)":
|
647 |
"memory_allocated (GB)": 50.57,
|
648 |
"step": 640,
|
649 |
"total_memory_available (GB)": 94.62
|
650 |
},
|
651 |
{
|
652 |
"epoch": 2.6530612244897958,
|
653 |
-
"grad_norm": 0.
|
654 |
"learning_rate": 4.693877551020409e-06,
|
655 |
-
"loss": 0.
|
656 |
-
"max_memory_allocated (GB)":
|
657 |
"memory_allocated (GB)": 50.57,
|
658 |
"step": 650,
|
659 |
"total_memory_available (GB)": 94.62
|
660 |
},
|
661 |
{
|
662 |
"epoch": 2.693877551020408,
|
663 |
-
"grad_norm":
|
664 |
"learning_rate": 4.612244897959184e-06,
|
665 |
-
"loss": 0.
|
666 |
-
"max_memory_allocated (GB)":
|
667 |
"memory_allocated (GB)": 50.57,
|
668 |
"step": 660,
|
669 |
"total_memory_available (GB)": 94.62
|
670 |
},
|
671 |
{
|
672 |
"epoch": 2.7346938775510203,
|
673 |
-
"grad_norm": 0.
|
674 |
"learning_rate": 4.530612244897959e-06,
|
675 |
-
"loss": 0.
|
676 |
-
"max_memory_allocated (GB)":
|
677 |
"memory_allocated (GB)": 50.57,
|
678 |
"step": 670,
|
679 |
"total_memory_available (GB)": 94.62
|
680 |
},
|
681 |
{
|
682 |
"epoch": 2.7755102040816326,
|
683 |
-
"grad_norm": 1.
|
684 |
"learning_rate": 4.448979591836735e-06,
|
685 |
-
"loss": 0.
|
686 |
-
"max_memory_allocated (GB)":
|
687 |
"memory_allocated (GB)": 50.57,
|
688 |
"step": 680,
|
689 |
"total_memory_available (GB)": 94.62
|
690 |
},
|
691 |
{
|
692 |
"epoch": 2.816326530612245,
|
693 |
-
"grad_norm": 0.
|
694 |
"learning_rate": 4.367346938775511e-06,
|
695 |
-
"loss": 0.
|
696 |
-
"max_memory_allocated (GB)":
|
697 |
"memory_allocated (GB)": 50.57,
|
698 |
"step": 690,
|
699 |
"total_memory_available (GB)": 94.62
|
700 |
},
|
701 |
{
|
702 |
"epoch": 2.857142857142857,
|
703 |
-
"grad_norm": 0.
|
704 |
"learning_rate": 4.2857142857142855e-06,
|
705 |
-
"loss": 0.
|
706 |
-
"max_memory_allocated (GB)":
|
707 |
"memory_allocated (GB)": 50.57,
|
708 |
"step": 700,
|
709 |
"total_memory_available (GB)": 94.62
|
710 |
},
|
711 |
{
|
712 |
"epoch": 2.8979591836734695,
|
713 |
-
"grad_norm":
|
714 |
"learning_rate": 4.204081632653061e-06,
|
715 |
-
"loss": 0.
|
716 |
-
"max_memory_allocated (GB)":
|
717 |
"memory_allocated (GB)": 50.57,
|
718 |
"step": 710,
|
719 |
"total_memory_available (GB)": 94.62
|
720 |
},
|
721 |
{
|
722 |
"epoch": 2.938775510204082,
|
723 |
-
"grad_norm": 1.
|
724 |
"learning_rate": 4.122448979591837e-06,
|
725 |
-
"loss": 0.
|
726 |
-
"max_memory_allocated (GB)":
|
727 |
"memory_allocated (GB)": 50.57,
|
728 |
"step": 720,
|
729 |
"total_memory_available (GB)": 94.62
|
730 |
},
|
731 |
{
|
732 |
"epoch": 2.979591836734694,
|
733 |
-
"grad_norm": 0.
|
734 |
"learning_rate": 4.040816326530612e-06,
|
735 |
-
"loss": 0.
|
736 |
-
"max_memory_allocated (GB)":
|
737 |
"memory_allocated (GB)": 50.57,
|
738 |
"step": 730,
|
739 |
"total_memory_available (GB)": 94.62
|
740 |
},
|
741 |
{
|
742 |
"epoch": 3.020408163265306,
|
743 |
-
"grad_norm": 1.
|
744 |
"learning_rate": 3.959183673469388e-06,
|
745 |
-
"loss": 0.
|
746 |
-
"max_memory_allocated (GB)":
|
747 |
"memory_allocated (GB)": 50.57,
|
748 |
"step": 740,
|
749 |
"total_memory_available (GB)": 94.62
|
750 |
},
|
751 |
{
|
752 |
"epoch": 3.061224489795918,
|
753 |
-
"grad_norm":
|
754 |
"learning_rate": 3.877551020408164e-06,
|
755 |
-
"loss": 0.
|
756 |
-
"max_memory_allocated (GB)":
|
757 |
"memory_allocated (GB)": 50.57,
|
758 |
"step": 750,
|
759 |
"total_memory_available (GB)": 94.62
|
760 |
},
|
761 |
{
|
762 |
"epoch": 3.1020408163265305,
|
763 |
-
"grad_norm": 0.
|
764 |
"learning_rate": 3.795918367346939e-06,
|
765 |
-
"loss": 0.
|
766 |
-
"max_memory_allocated (GB)":
|
767 |
"memory_allocated (GB)": 50.57,
|
768 |
"step": 760,
|
769 |
"total_memory_available (GB)": 94.62
|
770 |
},
|
771 |
{
|
772 |
"epoch": 3.142857142857143,
|
773 |
-
"grad_norm": 0.
|
774 |
"learning_rate": 3.7142857142857146e-06,
|
775 |
-
"loss": 0.
|
776 |
-
"max_memory_allocated (GB)":
|
777 |
"memory_allocated (GB)": 50.57,
|
778 |
"step": 770,
|
779 |
"total_memory_available (GB)": 94.62
|
780 |
},
|
781 |
{
|
782 |
"epoch": 3.183673469387755,
|
783 |
-
"grad_norm": 0.
|
784 |
"learning_rate": 3.6326530612244903e-06,
|
785 |
-
"loss": 0.
|
786 |
-
"max_memory_allocated (GB)":
|
787 |
"memory_allocated (GB)": 50.57,
|
788 |
"step": 780,
|
789 |
"total_memory_available (GB)": 94.62
|
790 |
},
|
791 |
{
|
792 |
"epoch": 3.2244897959183674,
|
793 |
-
"grad_norm": 0.
|
794 |
"learning_rate": 3.5510204081632655e-06,
|
795 |
-
"loss": 0.
|
796 |
-
"max_memory_allocated (GB)":
|
797 |
"memory_allocated (GB)": 50.57,
|
798 |
"step": 790,
|
799 |
"total_memory_available (GB)": 94.62
|
800 |
},
|
801 |
{
|
802 |
"epoch": 3.2653061224489797,
|
803 |
-
"grad_norm": 0.
|
804 |
"learning_rate": 3.469387755102041e-06,
|
805 |
-
"loss": 0.
|
806 |
-
"max_memory_allocated (GB)":
|
807 |
"memory_allocated (GB)": 50.57,
|
808 |
"step": 800,
|
809 |
"total_memory_available (GB)": 94.62
|
810 |
},
|
811 |
{
|
812 |
"epoch": 3.306122448979592,
|
813 |
-
"grad_norm": 0.
|
814 |
"learning_rate": 3.3877551020408168e-06,
|
815 |
-
"loss": 0.
|
816 |
-
"max_memory_allocated (GB)":
|
817 |
"memory_allocated (GB)": 50.57,
|
818 |
"step": 810,
|
819 |
"total_memory_available (GB)": 94.62
|
820 |
},
|
821 |
{
|
822 |
"epoch": 3.3469387755102042,
|
823 |
-
"grad_norm": 0.
|
824 |
"learning_rate": 3.3061224489795924e-06,
|
825 |
-
"loss": 0.
|
826 |
-
"max_memory_allocated (GB)":
|
827 |
"memory_allocated (GB)": 50.57,
|
828 |
"step": 820,
|
829 |
"total_memory_available (GB)": 94.62
|
830 |
},
|
831 |
{
|
832 |
"epoch": 3.387755102040816,
|
833 |
-
"grad_norm": 0.
|
834 |
"learning_rate": 3.2244897959183672e-06,
|
835 |
-
"loss": 0.
|
836 |
-
"max_memory_allocated (GB)":
|
837 |
"memory_allocated (GB)": 50.57,
|
838 |
"step": 830,
|
839 |
"total_memory_available (GB)": 94.62
|
840 |
},
|
841 |
{
|
842 |
"epoch": 3.4285714285714284,
|
843 |
-
"grad_norm": 0.
|
844 |
"learning_rate": 3.142857142857143e-06,
|
845 |
-
"loss": 0.
|
846 |
-
"max_memory_allocated (GB)":
|
847 |
"memory_allocated (GB)": 50.57,
|
848 |
"step": 840,
|
849 |
"total_memory_available (GB)": 94.62
|
850 |
},
|
851 |
{
|
852 |
"epoch": 3.4693877551020407,
|
853 |
-
"grad_norm": 0.
|
854 |
"learning_rate": 3.0612244897959185e-06,
|
855 |
-
"loss": 0.
|
856 |
-
"max_memory_allocated (GB)":
|
857 |
"memory_allocated (GB)": 50.57,
|
858 |
"step": 850,
|
859 |
"total_memory_available (GB)": 94.62
|
860 |
},
|
861 |
{
|
862 |
"epoch": 3.510204081632653,
|
863 |
-
"grad_norm":
|
864 |
"learning_rate": 2.979591836734694e-06,
|
865 |
-
"loss": 0.
|
866 |
-
"max_memory_allocated (GB)":
|
867 |
"memory_allocated (GB)": 50.57,
|
868 |
"step": 860,
|
869 |
"total_memory_available (GB)": 94.62
|
870 |
},
|
871 |
{
|
872 |
"epoch": 3.5510204081632653,
|
873 |
-
"grad_norm": 0.
|
874 |
"learning_rate": 2.8979591836734694e-06,
|
875 |
-
"loss": 0.
|
876 |
-
"max_memory_allocated (GB)":
|
877 |
"memory_allocated (GB)": 50.57,
|
878 |
"step": 870,
|
879 |
"total_memory_available (GB)": 94.62
|
880 |
},
|
881 |
{
|
882 |
"epoch": 3.5918367346938775,
|
883 |
-
"grad_norm":
|
884 |
"learning_rate": 2.816326530612245e-06,
|
885 |
-
"loss": 0.
|
886 |
-
"max_memory_allocated (GB)":
|
887 |
"memory_allocated (GB)": 50.57,
|
888 |
"step": 880,
|
889 |
"total_memory_available (GB)": 94.62
|
890 |
},
|
891 |
{
|
892 |
"epoch": 3.63265306122449,
|
893 |
-
"grad_norm": 0.
|
894 |
"learning_rate": 2.7346938775510207e-06,
|
895 |
-
"loss": 0.
|
896 |
-
"max_memory_allocated (GB)":
|
897 |
"memory_allocated (GB)": 50.57,
|
898 |
"step": 890,
|
899 |
"total_memory_available (GB)": 94.62
|
900 |
},
|
901 |
{
|
902 |
"epoch": 3.673469387755102,
|
903 |
-
"grad_norm": 0.
|
904 |
"learning_rate": 2.6530612244897964e-06,
|
905 |
-
"loss": 0.
|
906 |
-
"max_memory_allocated (GB)":
|
907 |
"memory_allocated (GB)": 50.57,
|
908 |
"step": 900,
|
909 |
"total_memory_available (GB)": 94.62
|
910 |
},
|
911 |
{
|
912 |
"epoch": 3.7142857142857144,
|
913 |
-
"grad_norm": 0.
|
914 |
"learning_rate": 2.571428571428571e-06,
|
915 |
-
"loss": 0.
|
916 |
-
"max_memory_allocated (GB)":
|
917 |
"memory_allocated (GB)": 50.57,
|
918 |
"step": 910,
|
919 |
"total_memory_available (GB)": 94.62
|
920 |
},
|
921 |
{
|
922 |
"epoch": 3.7551020408163263,
|
923 |
-
"grad_norm": 0.
|
924 |
"learning_rate": 2.489795918367347e-06,
|
925 |
-
"loss": 0.
|
926 |
-
"max_memory_allocated (GB)":
|
927 |
"memory_allocated (GB)": 50.57,
|
928 |
"step": 920,
|
929 |
"total_memory_available (GB)": 94.62
|
930 |
},
|
931 |
{
|
932 |
"epoch": 3.795918367346939,
|
933 |
-
"grad_norm": 0.
|
934 |
"learning_rate": 2.4081632653061225e-06,
|
935 |
-
"loss": 0.
|
936 |
-
"max_memory_allocated (GB)":
|
937 |
"memory_allocated (GB)": 50.57,
|
938 |
"step": 930,
|
939 |
"total_memory_available (GB)": 94.62
|
940 |
},
|
941 |
{
|
942 |
"epoch": 3.836734693877551,
|
943 |
-
"grad_norm":
|
944 |
"learning_rate": 2.326530612244898e-06,
|
945 |
"loss": 0.0489,
|
946 |
-
"max_memory_allocated (GB)":
|
947 |
"memory_allocated (GB)": 50.57,
|
948 |
"step": 940,
|
949 |
"total_memory_available (GB)": 94.62
|
950 |
},
|
951 |
{
|
952 |
"epoch": 3.877551020408163,
|
953 |
-
"grad_norm": 0.
|
954 |
"learning_rate": 2.244897959183674e-06,
|
955 |
-
"loss": 0.
|
956 |
-
"max_memory_allocated (GB)":
|
957 |
"memory_allocated (GB)": 50.57,
|
958 |
"step": 950,
|
959 |
"total_memory_available (GB)": 94.62
|
960 |
},
|
961 |
{
|
962 |
"epoch": 3.9183673469387754,
|
963 |
-
"grad_norm": 0.
|
964 |
"learning_rate": 2.1632653061224495e-06,
|
965 |
-
"loss": 0.
|
966 |
-
"max_memory_allocated (GB)":
|
967 |
"memory_allocated (GB)": 50.57,
|
968 |
"step": 960,
|
969 |
"total_memory_available (GB)": 94.62
|
970 |
},
|
971 |
{
|
972 |
"epoch": 3.9591836734693877,
|
973 |
-
"grad_norm": 1.
|
974 |
"learning_rate": 2.0816326530612247e-06,
|
975 |
-
"loss": 0.
|
976 |
-
"max_memory_allocated (GB)":
|
977 |
"memory_allocated (GB)": 50.57,
|
978 |
"step": 970,
|
979 |
"total_memory_available (GB)": 94.62
|
980 |
},
|
981 |
{
|
982 |
"epoch": 4.0,
|
983 |
-
"grad_norm": 0.
|
984 |
"learning_rate": 2.0000000000000003e-06,
|
985 |
-
"loss": 0.
|
986 |
-
"max_memory_allocated (GB)":
|
987 |
"memory_allocated (GB)": 50.57,
|
988 |
"step": 980,
|
989 |
"total_memory_available (GB)": 94.62
|
990 |
},
|
991 |
{
|
992 |
"epoch": 4.040816326530612,
|
993 |
-
"grad_norm": 0.
|
994 |
"learning_rate": 1.9183673469387756e-06,
|
995 |
"loss": 0.0483,
|
996 |
-
"max_memory_allocated (GB)":
|
997 |
"memory_allocated (GB)": 50.57,
|
998 |
"step": 990,
|
999 |
"total_memory_available (GB)": 94.62
|
1000 |
},
|
1001 |
{
|
1002 |
"epoch": 4.081632653061225,
|
1003 |
-
"grad_norm":
|
1004 |
"learning_rate": 1.8367346938775512e-06,
|
1005 |
-
"loss": 0.
|
1006 |
-
"max_memory_allocated (GB)":
|
1007 |
"memory_allocated (GB)": 50.57,
|
1008 |
"step": 1000,
|
1009 |
"total_memory_available (GB)": 94.62
|
1010 |
},
|
1011 |
{
|
1012 |
"epoch": 4.122448979591836,
|
1013 |
-
"grad_norm": 0.
|
1014 |
"learning_rate": 1.7551020408163267e-06,
|
1015 |
-
"loss": 0.
|
1016 |
-
"max_memory_allocated (GB)":
|
1017 |
"memory_allocated (GB)": 50.57,
|
1018 |
"step": 1010,
|
1019 |
"total_memory_available (GB)": 94.62
|
1020 |
},
|
1021 |
{
|
1022 |
"epoch": 4.163265306122449,
|
1023 |
-
"grad_norm": 0.
|
1024 |
"learning_rate": 1.6734693877551023e-06,
|
1025 |
-
"loss": 0.
|
1026 |
-
"max_memory_allocated (GB)":
|
1027 |
"memory_allocated (GB)": 50.57,
|
1028 |
"step": 1020,
|
1029 |
"total_memory_available (GB)": 94.62
|
1030 |
},
|
1031 |
{
|
1032 |
"epoch": 4.204081632653061,
|
1033 |
-
"grad_norm": 0.
|
1034 |
"learning_rate": 1.5918367346938775e-06,
|
1035 |
-
"loss": 0.
|
1036 |
-
"max_memory_allocated (GB)":
|
1037 |
"memory_allocated (GB)": 50.57,
|
1038 |
"step": 1030,
|
1039 |
"total_memory_available (GB)": 94.62
|
1040 |
},
|
1041 |
{
|
1042 |
"epoch": 4.244897959183674,
|
1043 |
-
"grad_norm": 0.
|
1044 |
"learning_rate": 1.5102040816326532e-06,
|
1045 |
-
"loss": 0.
|
1046 |
-
"max_memory_allocated (GB)":
|
1047 |
"memory_allocated (GB)": 50.57,
|
1048 |
"step": 1040,
|
1049 |
"total_memory_available (GB)": 94.62
|
1050 |
},
|
1051 |
{
|
1052 |
"epoch": 4.285714285714286,
|
1053 |
-
"grad_norm": 0.
|
1054 |
"learning_rate": 1.4285714285714286e-06,
|
1055 |
-
"loss": 0.
|
1056 |
-
"max_memory_allocated (GB)":
|
1057 |
"memory_allocated (GB)": 50.57,
|
1058 |
"step": 1050,
|
1059 |
"total_memory_available (GB)": 94.62
|
1060 |
},
|
1061 |
{
|
1062 |
"epoch": 4.326530612244898,
|
1063 |
-
"grad_norm": 0.
|
1064 |
"learning_rate": 1.3469387755102043e-06,
|
1065 |
-
"loss": 0.
|
1066 |
-
"max_memory_allocated (GB)":
|
1067 |
"memory_allocated (GB)": 50.57,
|
1068 |
"step": 1060,
|
1069 |
"total_memory_available (GB)": 94.62
|
1070 |
},
|
1071 |
{
|
1072 |
"epoch": 4.36734693877551,
|
1073 |
-
"grad_norm": 0.
|
1074 |
"learning_rate": 1.2653061224489795e-06,
|
1075 |
-
"loss": 0.
|
1076 |
-
"max_memory_allocated (GB)":
|
1077 |
"memory_allocated (GB)": 50.57,
|
1078 |
"step": 1070,
|
1079 |
"total_memory_available (GB)": 94.62
|
1080 |
},
|
1081 |
{
|
1082 |
"epoch": 4.408163265306122,
|
1083 |
-
"grad_norm": 0.
|
1084 |
"learning_rate": 1.1836734693877552e-06,
|
1085 |
-
"loss": 0.
|
1086 |
-
"max_memory_allocated (GB)":
|
1087 |
"memory_allocated (GB)": 50.57,
|
1088 |
"step": 1080,
|
1089 |
"total_memory_available (GB)": 94.62
|
1090 |
},
|
1091 |
{
|
1092 |
"epoch": 4.448979591836735,
|
1093 |
-
"grad_norm": 0.
|
1094 |
"learning_rate": 1.1020408163265308e-06,
|
1095 |
-
"loss": 0.
|
1096 |
-
"max_memory_allocated (GB)":
|
1097 |
"memory_allocated (GB)": 50.57,
|
1098 |
"step": 1090,
|
1099 |
"total_memory_available (GB)": 94.62
|
1100 |
},
|
1101 |
{
|
1102 |
"epoch": 4.489795918367347,
|
1103 |
-
"grad_norm": 0.
|
1104 |
"learning_rate": 1.0204081632653063e-06,
|
1105 |
-
"loss": 0.
|
1106 |
-
"max_memory_allocated (GB)":
|
1107 |
"memory_allocated (GB)": 50.57,
|
1108 |
"step": 1100,
|
1109 |
"total_memory_available (GB)": 94.62
|
1110 |
},
|
1111 |
{
|
1112 |
"epoch": 4.530612244897959,
|
1113 |
-
"grad_norm": 0.
|
1114 |
"learning_rate": 9.387755102040817e-07,
|
1115 |
-
"loss": 0.
|
1116 |
-
"max_memory_allocated (GB)":
|
1117 |
"memory_allocated (GB)": 50.57,
|
1118 |
"step": 1110,
|
1119 |
"total_memory_available (GB)": 94.62
|
1120 |
},
|
1121 |
{
|
1122 |
"epoch": 4.571428571428571,
|
1123 |
-
"grad_norm": 0.
|
1124 |
"learning_rate": 8.571428571428572e-07,
|
1125 |
-
"loss": 0.
|
1126 |
-
"max_memory_allocated (GB)":
|
1127 |
"memory_allocated (GB)": 50.57,
|
1128 |
"step": 1120,
|
1129 |
"total_memory_available (GB)": 94.62
|
1130 |
},
|
1131 |
{
|
1132 |
"epoch": 4.612244897959184,
|
1133 |
-
"grad_norm": 0.
|
1134 |
"learning_rate": 7.755102040816327e-07,
|
1135 |
-
"loss": 0.
|
1136 |
-
"max_memory_allocated (GB)":
|
1137 |
"memory_allocated (GB)": 50.57,
|
1138 |
"step": 1130,
|
1139 |
"total_memory_available (GB)": 94.62
|
1140 |
},
|
1141 |
{
|
1142 |
"epoch": 4.653061224489796,
|
1143 |
-
"grad_norm":
|
1144 |
"learning_rate": 6.938775510204082e-07,
|
1145 |
-
"loss": 0.
|
1146 |
-
"max_memory_allocated (GB)":
|
1147 |
"memory_allocated (GB)": 50.57,
|
1148 |
"step": 1140,
|
1149 |
"total_memory_available (GB)": 94.62
|
1150 |
},
|
1151 |
{
|
1152 |
"epoch": 4.6938775510204085,
|
1153 |
-
"grad_norm": 0.
|
1154 |
"learning_rate": 6.122448979591837e-07,
|
1155 |
-
"loss": 0.
|
1156 |
-
"max_memory_allocated (GB)":
|
1157 |
"memory_allocated (GB)": 50.57,
|
1158 |
"step": 1150,
|
1159 |
"total_memory_available (GB)": 94.62
|
1160 |
},
|
1161 |
{
|
1162 |
"epoch": 4.73469387755102,
|
1163 |
-
"grad_norm":
|
1164 |
"learning_rate": 5.306122448979592e-07,
|
1165 |
-
"loss": 0.
|
1166 |
-
"max_memory_allocated (GB)":
|
1167 |
"memory_allocated (GB)": 50.57,
|
1168 |
"step": 1160,
|
1169 |
"total_memory_available (GB)": 94.62
|
1170 |
},
|
1171 |
{
|
1172 |
"epoch": 4.775510204081632,
|
1173 |
-
"grad_norm": 0.
|
1174 |
"learning_rate": 4.489795918367347e-07,
|
1175 |
-
"loss": 0.
|
1176 |
-
"max_memory_allocated (GB)":
|
1177 |
"memory_allocated (GB)": 50.57,
|
1178 |
"step": 1170,
|
1179 |
"total_memory_available (GB)": 94.62
|
1180 |
},
|
1181 |
{
|
1182 |
"epoch": 4.816326530612245,
|
1183 |
-
"grad_norm": 0.
|
1184 |
"learning_rate": 3.6734693877551025e-07,
|
1185 |
-
"loss": 0.
|
1186 |
-
"max_memory_allocated (GB)":
|
1187 |
"memory_allocated (GB)": 50.57,
|
1188 |
"step": 1180,
|
1189 |
"total_memory_available (GB)": 94.62
|
1190 |
},
|
1191 |
{
|
1192 |
"epoch": 4.857142857142857,
|
1193 |
-
"grad_norm":
|
1194 |
"learning_rate": 2.8571428571428575e-07,
|
1195 |
-
"loss": 0.
|
1196 |
-
"max_memory_allocated (GB)":
|
1197 |
"memory_allocated (GB)": 50.57,
|
1198 |
"step": 1190,
|
1199 |
"total_memory_available (GB)": 94.62
|
1200 |
},
|
1201 |
{
|
1202 |
"epoch": 4.8979591836734695,
|
1203 |
-
"grad_norm": 0.
|
1204 |
"learning_rate": 2.0408163265306121e-07,
|
1205 |
-
"loss": 0.
|
1206 |
-
"max_memory_allocated (GB)":
|
1207 |
"memory_allocated (GB)": 50.57,
|
1208 |
"step": 1200,
|
1209 |
"total_memory_available (GB)": 94.62
|
1210 |
},
|
1211 |
{
|
1212 |
"epoch": 4.938775510204081,
|
1213 |
-
"grad_norm":
|
1214 |
"learning_rate": 1.2244897959183673e-07,
|
1215 |
-
"loss": 0.
|
1216 |
-
"max_memory_allocated (GB)":
|
1217 |
"memory_allocated (GB)": 50.57,
|
1218 |
"step": 1210,
|
1219 |
"total_memory_available (GB)": 94.62
|
1220 |
},
|
1221 |
{
|
1222 |
"epoch": 4.979591836734694,
|
1223 |
-
"grad_norm":
|
1224 |
"learning_rate": 4.0816326530612253e-08,
|
1225 |
-
"loss": 0.
|
1226 |
-
"max_memory_allocated (GB)":
|
1227 |
"memory_allocated (GB)": 50.57,
|
1228 |
"step": 1220,
|
1229 |
"total_memory_available (GB)": 94.62
|
1230 |
},
|
1231 |
{
|
1232 |
"epoch": 5.0,
|
1233 |
-
"max_memory_allocated (GB)":
|
1234 |
"memory_allocated (GB)": 50.57,
|
1235 |
"step": 1225,
|
1236 |
"total_flos": 3.0598946525952e+16,
|
1237 |
"total_memory_available (GB)": 94.62,
|
1238 |
-
"train_loss": 0.
|
1239 |
-
"train_runtime":
|
1240 |
-
"train_samples_per_second":
|
1241 |
-
"train_steps_per_second": 1.
|
1242 |
}
|
1243 |
],
|
1244 |
"logging_steps": 10,
|
|
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 0.04081632653061224,
|
13 |
+
"grad_norm": 16.731555938720703,
|
14 |
"learning_rate": 9.918367346938776e-06,
|
15 |
+
"loss": 0.2616,
|
16 |
"max_memory_allocated (GB)": 57.18,
|
17 |
"memory_allocated (GB)": 50.57,
|
18 |
"step": 10,
|
|
|
20 |
},
|
21 |
{
|
22 |
"epoch": 0.08163265306122448,
|
23 |
+
"grad_norm": 7.408310890197754,
|
24 |
"learning_rate": 9.836734693877552e-06,
|
25 |
+
"loss": 0.1553,
|
26 |
"max_memory_allocated (GB)": 57.18,
|
27 |
"memory_allocated (GB)": 50.57,
|
28 |
"step": 20,
|
|
|
30 |
},
|
31 |
{
|
32 |
"epoch": 0.12244897959183673,
|
33 |
+
"grad_norm": 5.734114170074463,
|
34 |
"learning_rate": 9.755102040816327e-06,
|
35 |
+
"loss": 0.1248,
|
36 |
"max_memory_allocated (GB)": 57.18,
|
37 |
"memory_allocated (GB)": 50.57,
|
38 |
"step": 30,
|
|
|
40 |
},
|
41 |
{
|
42 |
"epoch": 0.16326530612244897,
|
43 |
+
"grad_norm": 4.232840061187744,
|
44 |
"learning_rate": 9.673469387755103e-06,
|
45 |
+
"loss": 0.1106,
|
46 |
"max_memory_allocated (GB)": 57.18,
|
47 |
"memory_allocated (GB)": 50.57,
|
48 |
"step": 40,
|
|
|
50 |
},
|
51 |
{
|
52 |
"epoch": 0.20408163265306123,
|
53 |
+
"grad_norm": 4.877067565917969,
|
54 |
"learning_rate": 9.591836734693878e-06,
|
55 |
+
"loss": 0.1233,
|
56 |
"max_memory_allocated (GB)": 57.18,
|
57 |
"memory_allocated (GB)": 50.57,
|
58 |
"step": 50,
|
|
|
60 |
},
|
61 |
{
|
62 |
"epoch": 0.24489795918367346,
|
63 |
+
"grad_norm": 2.0087852478027344,
|
64 |
"learning_rate": 9.510204081632653e-06,
|
65 |
+
"loss": 0.0799,
|
66 |
"max_memory_allocated (GB)": 57.18,
|
67 |
"memory_allocated (GB)": 50.57,
|
68 |
"step": 60,
|
|
|
70 |
},
|
71 |
{
|
72 |
"epoch": 0.2857142857142857,
|
73 |
+
"grad_norm": 4.425936698913574,
|
74 |
"learning_rate": 9.42857142857143e-06,
|
75 |
"loss": 0.1015,
|
76 |
"max_memory_allocated (GB)": 57.18,
|
|
|
80 |
},
|
81 |
{
|
82 |
"epoch": 0.32653061224489793,
|
83 |
+
"grad_norm": 1.8348758220672607,
|
84 |
"learning_rate": 9.346938775510204e-06,
|
85 |
+
"loss": 0.1094,
|
86 |
"max_memory_allocated (GB)": 57.18,
|
87 |
"memory_allocated (GB)": 50.57,
|
88 |
"step": 80,
|
|
|
90 |
},
|
91 |
{
|
92 |
"epoch": 0.3673469387755102,
|
93 |
+
"grad_norm": 5.009132385253906,
|
94 |
"learning_rate": 9.26530612244898e-06,
|
95 |
+
"loss": 0.0837,
|
96 |
"max_memory_allocated (GB)": 57.18,
|
97 |
"memory_allocated (GB)": 50.57,
|
98 |
"step": 90,
|
|
|
100 |
},
|
101 |
{
|
102 |
"epoch": 0.40816326530612246,
|
103 |
+
"grad_norm": 6.888875484466553,
|
104 |
"learning_rate": 9.183673469387756e-06,
|
105 |
+
"loss": 0.075,
|
106 |
"max_memory_allocated (GB)": 57.18,
|
107 |
"memory_allocated (GB)": 50.57,
|
108 |
"step": 100,
|
|
|
110 |
},
|
111 |
{
|
112 |
"epoch": 0.4489795918367347,
|
113 |
+
"grad_norm": 3.038123369216919,
|
114 |
"learning_rate": 9.102040816326532e-06,
|
115 |
+
"loss": 0.0709,
|
116 |
"max_memory_allocated (GB)": 57.18,
|
117 |
"memory_allocated (GB)": 50.57,
|
118 |
"step": 110,
|
|
|
120 |
},
|
121 |
{
|
122 |
"epoch": 0.4897959183673469,
|
123 |
+
"grad_norm": 2.6441116333007812,
|
124 |
"learning_rate": 9.020408163265307e-06,
|
125 |
+
"loss": 0.0706,
|
126 |
"max_memory_allocated (GB)": 57.18,
|
127 |
"memory_allocated (GB)": 50.57,
|
128 |
"step": 120,
|
|
|
130 |
},
|
131 |
{
|
132 |
"epoch": 0.5306122448979592,
|
133 |
+
"grad_norm": 3.5273971557617188,
|
134 |
"learning_rate": 8.938775510204082e-06,
|
135 |
+
"loss": 0.0597,
|
136 |
"max_memory_allocated (GB)": 57.18,
|
137 |
"memory_allocated (GB)": 50.57,
|
138 |
"step": 130,
|
|
|
140 |
},
|
141 |
{
|
142 |
"epoch": 0.5714285714285714,
|
143 |
+
"grad_norm": 1.2711573839187622,
|
144 |
"learning_rate": 8.857142857142858e-06,
|
145 |
+
"loss": 0.0499,
|
146 |
"max_memory_allocated (GB)": 57.18,
|
147 |
"memory_allocated (GB)": 50.57,
|
148 |
"step": 140,
|
|
|
150 |
},
|
151 |
{
|
152 |
"epoch": 0.6122448979591837,
|
153 |
+
"grad_norm": 2.2461705207824707,
|
154 |
"learning_rate": 8.775510204081633e-06,
|
155 |
+
"loss": 0.0636,
|
156 |
"max_memory_allocated (GB)": 57.18,
|
157 |
"memory_allocated (GB)": 50.57,
|
158 |
"step": 150,
|
|
|
160 |
},
|
161 |
{
|
162 |
"epoch": 0.6530612244897959,
|
163 |
+
"grad_norm": 2.0061120986938477,
|
164 |
"learning_rate": 8.69387755102041e-06,
|
165 |
+
"loss": 0.0708,
|
166 |
"max_memory_allocated (GB)": 57.18,
|
167 |
"memory_allocated (GB)": 50.57,
|
168 |
"step": 160,
|
|
|
170 |
},
|
171 |
{
|
172 |
"epoch": 0.6938775510204082,
|
173 |
+
"grad_norm": 1.7886881828308105,
|
174 |
"learning_rate": 8.612244897959184e-06,
|
175 |
+
"loss": 0.0753,
|
176 |
"max_memory_allocated (GB)": 57.18,
|
177 |
"memory_allocated (GB)": 50.57,
|
178 |
"step": 170,
|
|
|
180 |
},
|
181 |
{
|
182 |
"epoch": 0.7346938775510204,
|
183 |
+
"grad_norm": 1.8707109689712524,
|
184 |
"learning_rate": 8.530612244897961e-06,
|
185 |
+
"loss": 0.0612,
|
186 |
"max_memory_allocated (GB)": 57.18,
|
187 |
"memory_allocated (GB)": 50.57,
|
188 |
"step": 180,
|
|
|
190 |
},
|
191 |
{
|
192 |
"epoch": 0.7755102040816326,
|
193 |
+
"grad_norm": 2.3418948650360107,
|
194 |
"learning_rate": 8.448979591836736e-06,
|
195 |
+
"loss": 0.0586,
|
196 |
"max_memory_allocated (GB)": 57.18,
|
197 |
"memory_allocated (GB)": 50.57,
|
198 |
"step": 190,
|
|
|
200 |
},
|
201 |
{
|
202 |
"epoch": 0.8163265306122449,
|
203 |
+
"grad_norm": 0.7759063839912415,
|
204 |
"learning_rate": 8.36734693877551e-06,
|
205 |
+
"loss": 0.063,
|
206 |
"max_memory_allocated (GB)": 57.18,
|
207 |
"memory_allocated (GB)": 50.57,
|
208 |
"step": 200,
|
|
|
210 |
},
|
211 |
{
|
212 |
"epoch": 0.8571428571428571,
|
213 |
+
"grad_norm": 0.9075489044189453,
|
214 |
"learning_rate": 8.285714285714287e-06,
|
215 |
+
"loss": 0.063,
|
216 |
"max_memory_allocated (GB)": 57.18,
|
217 |
"memory_allocated (GB)": 50.57,
|
218 |
"step": 210,
|
|
|
220 |
},
|
221 |
{
|
222 |
"epoch": 0.8979591836734694,
|
223 |
+
"grad_norm": 1.0522433519363403,
|
224 |
"learning_rate": 8.204081632653062e-06,
|
225 |
+
"loss": 0.0605,
|
226 |
"max_memory_allocated (GB)": 57.18,
|
227 |
"memory_allocated (GB)": 50.57,
|
228 |
"step": 220,
|
|
|
230 |
},
|
231 |
{
|
232 |
"epoch": 0.9387755102040817,
|
233 |
+
"grad_norm": 0.7449356913566589,
|
234 |
"learning_rate": 8.122448979591837e-06,
|
235 |
+
"loss": 0.0564,
|
236 |
"max_memory_allocated (GB)": 57.18,
|
237 |
"memory_allocated (GB)": 50.57,
|
238 |
"step": 230,
|
|
|
240 |
},
|
241 |
{
|
242 |
"epoch": 0.9795918367346939,
|
243 |
+
"grad_norm": 0.878478467464447,
|
244 |
"learning_rate": 8.040816326530613e-06,
|
245 |
+
"loss": 0.0496,
|
246 |
"max_memory_allocated (GB)": 57.18,
|
247 |
"memory_allocated (GB)": 50.57,
|
248 |
"step": 240,
|
|
|
250 |
},
|
251 |
{
|
252 |
"epoch": 1.0204081632653061,
|
253 |
+
"grad_norm": 1.1043357849121094,
|
254 |
"learning_rate": 7.959183673469388e-06,
|
255 |
+
"loss": 0.0593,
|
256 |
"max_memory_allocated (GB)": 57.18,
|
257 |
"memory_allocated (GB)": 50.57,
|
258 |
"step": 250,
|
|
|
260 |
},
|
261 |
{
|
262 |
"epoch": 1.0612244897959184,
|
263 |
+
"grad_norm": 1.5091803073883057,
|
264 |
"learning_rate": 7.877551020408164e-06,
|
265 |
+
"loss": 0.0542,
|
266 |
"max_memory_allocated (GB)": 57.18,
|
267 |
"memory_allocated (GB)": 50.57,
|
268 |
"step": 260,
|
|
|
270 |
},
|
271 |
{
|
272 |
"epoch": 1.1020408163265305,
|
273 |
+
"grad_norm": 0.7768255472183228,
|
274 |
"learning_rate": 7.79591836734694e-06,
|
275 |
+
"loss": 0.0636,
|
276 |
"max_memory_allocated (GB)": 57.18,
|
277 |
"memory_allocated (GB)": 50.57,
|
278 |
"step": 270,
|
|
|
280 |
},
|
281 |
{
|
282 |
"epoch": 1.1428571428571428,
|
283 |
+
"grad_norm": 1.0384677648544312,
|
284 |
"learning_rate": 7.714285714285716e-06,
|
285 |
+
"loss": 0.0585,
|
286 |
"max_memory_allocated (GB)": 57.18,
|
287 |
"memory_allocated (GB)": 50.57,
|
288 |
"step": 280,
|
|
|
290 |
},
|
291 |
{
|
292 |
"epoch": 1.183673469387755,
|
293 |
+
"grad_norm": 1.9693238735198975,
|
294 |
"learning_rate": 7.63265306122449e-06,
|
295 |
+
"loss": 0.0512,
|
296 |
"max_memory_allocated (GB)": 57.18,
|
297 |
"memory_allocated (GB)": 50.57,
|
298 |
"step": 290,
|
|
|
300 |
},
|
301 |
{
|
302 |
"epoch": 1.2244897959183674,
|
303 |
+
"grad_norm": 1.0379549264907837,
|
304 |
"learning_rate": 7.551020408163265e-06,
|
305 |
+
"loss": 0.0647,
|
306 |
"max_memory_allocated (GB)": 57.18,
|
307 |
"memory_allocated (GB)": 50.57,
|
308 |
"step": 300,
|
|
|
310 |
},
|
311 |
{
|
312 |
"epoch": 1.2653061224489797,
|
313 |
+
"grad_norm": 1.01348078250885,
|
314 |
"learning_rate": 7.469387755102041e-06,
|
315 |
"loss": 0.0668,
|
316 |
"max_memory_allocated (GB)": 57.18,
|
|
|
320 |
},
|
321 |
{
|
322 |
"epoch": 1.306122448979592,
|
323 |
+
"grad_norm": 0.9114493727684021,
|
324 |
"learning_rate": 7.387755102040817e-06,
|
325 |
+
"loss": 0.0509,
|
326 |
"max_memory_allocated (GB)": 57.18,
|
327 |
"memory_allocated (GB)": 50.57,
|
328 |
"step": 320,
|
|
|
330 |
},
|
331 |
{
|
332 |
"epoch": 1.346938775510204,
|
333 |
+
"grad_norm": 0.5102170705795288,
|
334 |
"learning_rate": 7.306122448979592e-06,
|
335 |
+
"loss": 0.0612,
|
336 |
"max_memory_allocated (GB)": 57.18,
|
337 |
"memory_allocated (GB)": 50.57,
|
338 |
"step": 330,
|
|
|
340 |
},
|
341 |
{
|
342 |
"epoch": 1.3877551020408163,
|
343 |
+
"grad_norm": 1.097835659980774,
|
344 |
"learning_rate": 7.224489795918368e-06,
|
345 |
"loss": 0.055,
|
346 |
"max_memory_allocated (GB)": 57.18,
|
|
|
350 |
},
|
351 |
{
|
352 |
"epoch": 1.4285714285714286,
|
353 |
+
"grad_norm": 0.5733208656311035,
|
354 |
"learning_rate": 7.1428571428571436e-06,
|
355 |
+
"loss": 0.05,
|
356 |
"max_memory_allocated (GB)": 57.18,
|
357 |
"memory_allocated (GB)": 50.57,
|
358 |
"step": 350,
|
|
|
360 |
},
|
361 |
{
|
362 |
"epoch": 1.469387755102041,
|
363 |
+
"grad_norm": 2.428065538406372,
|
364 |
"learning_rate": 7.061224489795919e-06,
|
365 |
+
"loss": 0.0458,
|
366 |
"max_memory_allocated (GB)": 57.18,
|
367 |
"memory_allocated (GB)": 50.57,
|
368 |
"step": 360,
|
|
|
370 |
},
|
371 |
{
|
372 |
"epoch": 1.510204081632653,
|
373 |
+
"grad_norm": 0.9827563166618347,
|
374 |
"learning_rate": 6.979591836734695e-06,
|
375 |
+
"loss": 0.0588,
|
376 |
"max_memory_allocated (GB)": 57.18,
|
377 |
"memory_allocated (GB)": 50.57,
|
378 |
"step": 370,
|
|
|
380 |
},
|
381 |
{
|
382 |
"epoch": 1.5510204081632653,
|
383 |
+
"grad_norm": 0.5563956499099731,
|
384 |
"learning_rate": 6.8979591836734705e-06,
|
385 |
+
"loss": 0.0473,
|
386 |
"max_memory_allocated (GB)": 57.18,
|
387 |
"memory_allocated (GB)": 50.57,
|
388 |
"step": 380,
|
|
|
390 |
},
|
391 |
{
|
392 |
"epoch": 1.5918367346938775,
|
393 |
+
"grad_norm": 0.9855480194091797,
|
394 |
"learning_rate": 6.816326530612245e-06,
|
395 |
+
"loss": 0.0653,
|
396 |
"max_memory_allocated (GB)": 57.18,
|
397 |
"memory_allocated (GB)": 50.57,
|
398 |
"step": 390,
|
|
|
400 |
},
|
401 |
{
|
402 |
"epoch": 1.6326530612244898,
|
403 |
+
"grad_norm": 0.7717217803001404,
|
404 |
"learning_rate": 6.734693877551021e-06,
|
405 |
+
"loss": 0.0618,
|
406 |
"max_memory_allocated (GB)": 57.18,
|
407 |
"memory_allocated (GB)": 50.57,
|
408 |
"step": 400,
|
|
|
410 |
},
|
411 |
{
|
412 |
"epoch": 1.6734693877551021,
|
413 |
+
"grad_norm": 0.5391837358474731,
|
414 |
"learning_rate": 6.653061224489797e-06,
|
415 |
+
"loss": 0.061,
|
416 |
"max_memory_allocated (GB)": 57.18,
|
417 |
"memory_allocated (GB)": 50.57,
|
418 |
"step": 410,
|
|
|
420 |
},
|
421 |
{
|
422 |
"epoch": 1.7142857142857144,
|
423 |
+
"grad_norm": 1.2298884391784668,
|
424 |
"learning_rate": 6.571428571428572e-06,
|
425 |
+
"loss": 0.0534,
|
426 |
"max_memory_allocated (GB)": 57.18,
|
427 |
"memory_allocated (GB)": 50.57,
|
428 |
"step": 420,
|
|
|
430 |
},
|
431 |
{
|
432 |
"epoch": 1.7551020408163265,
|
433 |
+
"grad_norm": 0.22150574624538422,
|
434 |
"learning_rate": 6.489795918367348e-06,
|
435 |
+
"loss": 0.0681,
|
436 |
"max_memory_allocated (GB)": 57.18,
|
437 |
"memory_allocated (GB)": 50.57,
|
438 |
"step": 430,
|
|
|
440 |
},
|
441 |
{
|
442 |
"epoch": 1.7959183673469388,
|
443 |
+
"grad_norm": 0.8992190957069397,
|
444 |
"learning_rate": 6.408163265306124e-06,
|
445 |
"loss": 0.0606,
|
446 |
"max_memory_allocated (GB)": 57.18,
|
|
|
450 |
},
|
451 |
{
|
452 |
"epoch": 1.836734693877551,
|
453 |
+
"grad_norm": 0.9748144149780273,
|
454 |
"learning_rate": 6.326530612244899e-06,
|
455 |
+
"loss": 0.0436,
|
456 |
"max_memory_allocated (GB)": 57.18,
|
457 |
"memory_allocated (GB)": 50.57,
|
458 |
"step": 450,
|
|
|
460 |
},
|
461 |
{
|
462 |
"epoch": 1.8775510204081631,
|
463 |
+
"grad_norm": 0.3272276520729065,
|
464 |
"learning_rate": 6.244897959183675e-06,
|
465 |
+
"loss": 0.0545,
|
466 |
"max_memory_allocated (GB)": 57.18,
|
467 |
"memory_allocated (GB)": 50.57,
|
468 |
"step": 460,
|
|
|
470 |
},
|
471 |
{
|
472 |
"epoch": 1.9183673469387754,
|
473 |
+
"grad_norm": 0.7515161633491516,
|
474 |
"learning_rate": 6.163265306122449e-06,
|
475 |
"loss": 0.0427,
|
476 |
"max_memory_allocated (GB)": 57.18,
|
|
|
480 |
},
|
481 |
{
|
482 |
"epoch": 1.9591836734693877,
|
483 |
+
"grad_norm": 0.35405218601226807,
|
484 |
"learning_rate": 6.0816326530612245e-06,
|
485 |
+
"loss": 0.0501,
|
486 |
"max_memory_allocated (GB)": 57.18,
|
487 |
"memory_allocated (GB)": 50.57,
|
488 |
"step": 480,
|
|
|
490 |
},
|
491 |
{
|
492 |
"epoch": 2.0,
|
493 |
+
"grad_norm": 0.9796536564826965,
|
494 |
"learning_rate": 6e-06,
|
495 |
+
"loss": 0.0641,
|
496 |
+
"max_memory_allocated (GB)": 60.52,
|
497 |
"memory_allocated (GB)": 50.57,
|
498 |
"step": 490,
|
499 |
"total_memory_available (GB)": 94.62
|
500 |
},
|
501 |
{
|
502 |
"epoch": 2.0408163265306123,
|
503 |
+
"grad_norm": 0.639869213104248,
|
504 |
"learning_rate": 5.918367346938776e-06,
|
505 |
+
"loss": 0.0592,
|
506 |
+
"max_memory_allocated (GB)": 60.52,
|
507 |
"memory_allocated (GB)": 50.57,
|
508 |
"step": 500,
|
509 |
"total_memory_available (GB)": 94.62
|
510 |
},
|
511 |
{
|
512 |
"epoch": 2.0816326530612246,
|
513 |
+
"grad_norm": 0.8377614617347717,
|
514 |
"learning_rate": 5.8367346938775515e-06,
|
515 |
+
"loss": 0.0497,
|
516 |
+
"max_memory_allocated (GB)": 60.52,
|
517 |
"memory_allocated (GB)": 50.57,
|
518 |
"step": 510,
|
519 |
"total_memory_available (GB)": 94.62
|
520 |
},
|
521 |
{
|
522 |
"epoch": 2.122448979591837,
|
523 |
+
"grad_norm": 1.07124662399292,
|
524 |
"learning_rate": 5.755102040816327e-06,
|
525 |
+
"loss": 0.047,
|
526 |
+
"max_memory_allocated (GB)": 60.52,
|
527 |
"memory_allocated (GB)": 50.57,
|
528 |
"step": 520,
|
529 |
"total_memory_available (GB)": 94.62
|
530 |
},
|
531 |
{
|
532 |
"epoch": 2.163265306122449,
|
533 |
+
"grad_norm": 1.0354053974151611,
|
534 |
"learning_rate": 5.673469387755103e-06,
|
535 |
+
"loss": 0.0636,
|
536 |
+
"max_memory_allocated (GB)": 60.52,
|
537 |
"memory_allocated (GB)": 50.57,
|
538 |
"step": 530,
|
539 |
"total_memory_available (GB)": 94.62
|
540 |
},
|
541 |
{
|
542 |
"epoch": 2.204081632653061,
|
543 |
+
"grad_norm": 0.5396034717559814,
|
544 |
"learning_rate": 5.591836734693878e-06,
|
545 |
+
"loss": 0.0602,
|
546 |
+
"max_memory_allocated (GB)": 60.52,
|
547 |
"memory_allocated (GB)": 50.57,
|
548 |
"step": 540,
|
549 |
"total_memory_available (GB)": 94.62
|
550 |
},
|
551 |
{
|
552 |
"epoch": 2.2448979591836733,
|
553 |
+
"grad_norm": 1.1461529731750488,
|
554 |
"learning_rate": 5.510204081632653e-06,
|
555 |
+
"loss": 0.0629,
|
556 |
+
"max_memory_allocated (GB)": 60.52,
|
557 |
"memory_allocated (GB)": 50.57,
|
558 |
"step": 550,
|
559 |
"total_memory_available (GB)": 94.62
|
560 |
},
|
561 |
{
|
562 |
"epoch": 2.2857142857142856,
|
563 |
+
"grad_norm": 1.732596516609192,
|
564 |
"learning_rate": 5.428571428571429e-06,
|
565 |
+
"loss": 0.0575,
|
566 |
+
"max_memory_allocated (GB)": 60.52,
|
567 |
"memory_allocated (GB)": 50.57,
|
568 |
"step": 560,
|
569 |
"total_memory_available (GB)": 94.62
|
570 |
},
|
571 |
{
|
572 |
"epoch": 2.326530612244898,
|
573 |
+
"grad_norm": 0.6468518376350403,
|
574 |
"learning_rate": 5.3469387755102045e-06,
|
575 |
+
"loss": 0.0581,
|
576 |
+
"max_memory_allocated (GB)": 60.52,
|
577 |
"memory_allocated (GB)": 50.57,
|
578 |
"step": 570,
|
579 |
"total_memory_available (GB)": 94.62
|
580 |
},
|
581 |
{
|
582 |
"epoch": 2.36734693877551,
|
583 |
+
"grad_norm": 1.7525995969772339,
|
584 |
"learning_rate": 5.26530612244898e-06,
|
585 |
+
"loss": 0.0678,
|
586 |
+
"max_memory_allocated (GB)": 60.52,
|
587 |
"memory_allocated (GB)": 50.57,
|
588 |
"step": 580,
|
589 |
"total_memory_available (GB)": 94.62
|
590 |
},
|
591 |
{
|
592 |
"epoch": 2.4081632653061225,
|
593 |
+
"grad_norm": 0.5057170391082764,
|
594 |
"learning_rate": 5.183673469387756e-06,
|
595 |
+
"loss": 0.0533,
|
596 |
+
"max_memory_allocated (GB)": 60.52,
|
597 |
"memory_allocated (GB)": 50.57,
|
598 |
"step": 590,
|
599 |
"total_memory_available (GB)": 94.62
|
600 |
},
|
601 |
{
|
602 |
"epoch": 2.4489795918367347,
|
603 |
+
"grad_norm": 1.2042871713638306,
|
604 |
"learning_rate": 5.1020408163265315e-06,
|
605 |
+
"loss": 0.0444,
|
606 |
+
"max_memory_allocated (GB)": 60.52,
|
607 |
"memory_allocated (GB)": 50.57,
|
608 |
"step": 600,
|
609 |
"total_memory_available (GB)": 94.62
|
610 |
},
|
611 |
{
|
612 |
"epoch": 2.489795918367347,
|
613 |
+
"grad_norm": 0.20784775912761688,
|
614 |
"learning_rate": 5.020408163265307e-06,
|
615 |
+
"loss": 0.0413,
|
616 |
+
"max_memory_allocated (GB)": 60.52,
|
617 |
"memory_allocated (GB)": 50.57,
|
618 |
"step": 610,
|
619 |
"total_memory_available (GB)": 94.62
|
620 |
},
|
621 |
{
|
622 |
"epoch": 2.5306122448979593,
|
623 |
+
"grad_norm": 5.924367427825928,
|
624 |
"learning_rate": 4.938775510204082e-06,
|
625 |
+
"loss": 0.0424,
|
626 |
+
"max_memory_allocated (GB)": 60.52,
|
627 |
"memory_allocated (GB)": 50.57,
|
628 |
"step": 620,
|
629 |
"total_memory_available (GB)": 94.62
|
630 |
},
|
631 |
{
|
632 |
"epoch": 2.571428571428571,
|
633 |
+
"grad_norm": 0.40434470772743225,
|
634 |
"learning_rate": 4.857142857142858e-06,
|
635 |
+
"loss": 0.0815,
|
636 |
+
"max_memory_allocated (GB)": 60.52,
|
637 |
"memory_allocated (GB)": 50.57,
|
638 |
"step": 630,
|
639 |
"total_memory_available (GB)": 94.62
|
640 |
},
|
641 |
{
|
642 |
"epoch": 2.612244897959184,
|
643 |
+
"grad_norm": 0.7878668904304504,
|
644 |
"learning_rate": 4.775510204081633e-06,
|
645 |
+
"loss": 0.0495,
|
646 |
+
"max_memory_allocated (GB)": 60.52,
|
647 |
"memory_allocated (GB)": 50.57,
|
648 |
"step": 640,
|
649 |
"total_memory_available (GB)": 94.62
|
650 |
},
|
651 |
{
|
652 |
"epoch": 2.6530612244897958,
|
653 |
+
"grad_norm": 0.5937454700469971,
|
654 |
"learning_rate": 4.693877551020409e-06,
|
655 |
+
"loss": 0.0471,
|
656 |
+
"max_memory_allocated (GB)": 60.52,
|
657 |
"memory_allocated (GB)": 50.57,
|
658 |
"step": 650,
|
659 |
"total_memory_available (GB)": 94.62
|
660 |
},
|
661 |
{
|
662 |
"epoch": 2.693877551020408,
|
663 |
+
"grad_norm": 0.9869357347488403,
|
664 |
"learning_rate": 4.612244897959184e-06,
|
665 |
+
"loss": 0.0472,
|
666 |
+
"max_memory_allocated (GB)": 60.52,
|
667 |
"memory_allocated (GB)": 50.57,
|
668 |
"step": 660,
|
669 |
"total_memory_available (GB)": 94.62
|
670 |
},
|
671 |
{
|
672 |
"epoch": 2.7346938775510203,
|
673 |
+
"grad_norm": 0.7606731653213501,
|
674 |
"learning_rate": 4.530612244897959e-06,
|
675 |
+
"loss": 0.0864,
|
676 |
+
"max_memory_allocated (GB)": 60.52,
|
677 |
"memory_allocated (GB)": 50.57,
|
678 |
"step": 670,
|
679 |
"total_memory_available (GB)": 94.62
|
680 |
},
|
681 |
{
|
682 |
"epoch": 2.7755102040816326,
|
683 |
+
"grad_norm": 1.5359723567962646,
|
684 |
"learning_rate": 4.448979591836735e-06,
|
685 |
+
"loss": 0.0482,
|
686 |
+
"max_memory_allocated (GB)": 60.52,
|
687 |
"memory_allocated (GB)": 50.57,
|
688 |
"step": 680,
|
689 |
"total_memory_available (GB)": 94.62
|
690 |
},
|
691 |
{
|
692 |
"epoch": 2.816326530612245,
|
693 |
+
"grad_norm": 0.370749294757843,
|
694 |
"learning_rate": 4.367346938775511e-06,
|
695 |
+
"loss": 0.0447,
|
696 |
+
"max_memory_allocated (GB)": 60.52,
|
697 |
"memory_allocated (GB)": 50.57,
|
698 |
"step": 690,
|
699 |
"total_memory_available (GB)": 94.62
|
700 |
},
|
701 |
{
|
702 |
"epoch": 2.857142857142857,
|
703 |
+
"grad_norm": 0.633791983127594,
|
704 |
"learning_rate": 4.2857142857142855e-06,
|
705 |
+
"loss": 0.044,
|
706 |
+
"max_memory_allocated (GB)": 60.52,
|
707 |
"memory_allocated (GB)": 50.57,
|
708 |
"step": 700,
|
709 |
"total_memory_available (GB)": 94.62
|
710 |
},
|
711 |
{
|
712 |
"epoch": 2.8979591836734695,
|
713 |
+
"grad_norm": 1.1335562467575073,
|
714 |
"learning_rate": 4.204081632653061e-06,
|
715 |
+
"loss": 0.0552,
|
716 |
+
"max_memory_allocated (GB)": 60.52,
|
717 |
"memory_allocated (GB)": 50.57,
|
718 |
"step": 710,
|
719 |
"total_memory_available (GB)": 94.62
|
720 |
},
|
721 |
{
|
722 |
"epoch": 2.938775510204082,
|
723 |
+
"grad_norm": 1.5698113441467285,
|
724 |
"learning_rate": 4.122448979591837e-06,
|
725 |
+
"loss": 0.0576,
|
726 |
+
"max_memory_allocated (GB)": 60.52,
|
727 |
"memory_allocated (GB)": 50.57,
|
728 |
"step": 720,
|
729 |
"total_memory_available (GB)": 94.62
|
730 |
},
|
731 |
{
|
732 |
"epoch": 2.979591836734694,
|
733 |
+
"grad_norm": 0.20579344034194946,
|
734 |
"learning_rate": 4.040816326530612e-06,
|
735 |
+
"loss": 0.0336,
|
736 |
+
"max_memory_allocated (GB)": 60.52,
|
737 |
"memory_allocated (GB)": 50.57,
|
738 |
"step": 730,
|
739 |
"total_memory_available (GB)": 94.62
|
740 |
},
|
741 |
{
|
742 |
"epoch": 3.020408163265306,
|
743 |
+
"grad_norm": 1.3172132968902588,
|
744 |
"learning_rate": 3.959183673469388e-06,
|
745 |
+
"loss": 0.0534,
|
746 |
+
"max_memory_allocated (GB)": 60.52,
|
747 |
"memory_allocated (GB)": 50.57,
|
748 |
"step": 740,
|
749 |
"total_memory_available (GB)": 94.62
|
750 |
},
|
751 |
{
|
752 |
"epoch": 3.061224489795918,
|
753 |
+
"grad_norm": 1.5193818807601929,
|
754 |
"learning_rate": 3.877551020408164e-06,
|
755 |
+
"loss": 0.0605,
|
756 |
+
"max_memory_allocated (GB)": 60.52,
|
757 |
"memory_allocated (GB)": 50.57,
|
758 |
"step": 750,
|
759 |
"total_memory_available (GB)": 94.62
|
760 |
},
|
761 |
{
|
762 |
"epoch": 3.1020408163265305,
|
763 |
+
"grad_norm": 0.7448021769523621,
|
764 |
"learning_rate": 3.795918367346939e-06,
|
765 |
+
"loss": 0.0524,
|
766 |
+
"max_memory_allocated (GB)": 60.52,
|
767 |
"memory_allocated (GB)": 50.57,
|
768 |
"step": 760,
|
769 |
"total_memory_available (GB)": 94.62
|
770 |
},
|
771 |
{
|
772 |
"epoch": 3.142857142857143,
|
773 |
+
"grad_norm": 0.6830514073371887,
|
774 |
"learning_rate": 3.7142857142857146e-06,
|
775 |
+
"loss": 0.0687,
|
776 |
+
"max_memory_allocated (GB)": 60.52,
|
777 |
"memory_allocated (GB)": 50.57,
|
778 |
"step": 770,
|
779 |
"total_memory_available (GB)": 94.62
|
780 |
},
|
781 |
{
|
782 |
"epoch": 3.183673469387755,
|
783 |
+
"grad_norm": 0.4188525378704071,
|
784 |
"learning_rate": 3.6326530612244903e-06,
|
785 |
+
"loss": 0.0646,
|
786 |
+
"max_memory_allocated (GB)": 60.52,
|
787 |
"memory_allocated (GB)": 50.57,
|
788 |
"step": 780,
|
789 |
"total_memory_available (GB)": 94.62
|
790 |
},
|
791 |
{
|
792 |
"epoch": 3.2244897959183674,
|
793 |
+
"grad_norm": 0.562319278717041,
|
794 |
"learning_rate": 3.5510204081632655e-06,
|
795 |
+
"loss": 0.0523,
|
796 |
+
"max_memory_allocated (GB)": 60.52,
|
797 |
"memory_allocated (GB)": 50.57,
|
798 |
"step": 790,
|
799 |
"total_memory_available (GB)": 94.62
|
800 |
},
|
801 |
{
|
802 |
"epoch": 3.2653061224489797,
|
803 |
+
"grad_norm": 0.383428156375885,
|
804 |
"learning_rate": 3.469387755102041e-06,
|
805 |
+
"loss": 0.0425,
|
806 |
+
"max_memory_allocated (GB)": 60.52,
|
807 |
"memory_allocated (GB)": 50.57,
|
808 |
"step": 800,
|
809 |
"total_memory_available (GB)": 94.62
|
810 |
},
|
811 |
{
|
812 |
"epoch": 3.306122448979592,
|
813 |
+
"grad_norm": 0.6743464469909668,
|
814 |
"learning_rate": 3.3877551020408168e-06,
|
815 |
+
"loss": 0.0546,
|
816 |
+
"max_memory_allocated (GB)": 60.52,
|
817 |
"memory_allocated (GB)": 50.57,
|
818 |
"step": 810,
|
819 |
"total_memory_available (GB)": 94.62
|
820 |
},
|
821 |
{
|
822 |
"epoch": 3.3469387755102042,
|
823 |
+
"grad_norm": 0.7104524970054626,
|
824 |
"learning_rate": 3.3061224489795924e-06,
|
825 |
+
"loss": 0.0448,
|
826 |
+
"max_memory_allocated (GB)": 60.52,
|
827 |
"memory_allocated (GB)": 50.57,
|
828 |
"step": 820,
|
829 |
"total_memory_available (GB)": 94.62
|
830 |
},
|
831 |
{
|
832 |
"epoch": 3.387755102040816,
|
833 |
+
"grad_norm": 0.5083824396133423,
|
834 |
"learning_rate": 3.2244897959183672e-06,
|
835 |
+
"loss": 0.048,
|
836 |
+
"max_memory_allocated (GB)": 60.52,
|
837 |
"memory_allocated (GB)": 50.57,
|
838 |
"step": 830,
|
839 |
"total_memory_available (GB)": 94.62
|
840 |
},
|
841 |
{
|
842 |
"epoch": 3.4285714285714284,
|
843 |
+
"grad_norm": 0.49159926176071167,
|
844 |
"learning_rate": 3.142857142857143e-06,
|
845 |
+
"loss": 0.0521,
|
846 |
+
"max_memory_allocated (GB)": 60.52,
|
847 |
"memory_allocated (GB)": 50.57,
|
848 |
"step": 840,
|
849 |
"total_memory_available (GB)": 94.62
|
850 |
},
|
851 |
{
|
852 |
"epoch": 3.4693877551020407,
|
853 |
+
"grad_norm": 0.635971188545227,
|
854 |
"learning_rate": 3.0612244897959185e-06,
|
855 |
+
"loss": 0.081,
|
856 |
+
"max_memory_allocated (GB)": 60.52,
|
857 |
"memory_allocated (GB)": 50.57,
|
858 |
"step": 850,
|
859 |
"total_memory_available (GB)": 94.62
|
860 |
},
|
861 |
{
|
862 |
"epoch": 3.510204081632653,
|
863 |
+
"grad_norm": 5.446665287017822,
|
864 |
"learning_rate": 2.979591836734694e-06,
|
865 |
+
"loss": 0.0449,
|
866 |
+
"max_memory_allocated (GB)": 60.52,
|
867 |
"memory_allocated (GB)": 50.57,
|
868 |
"step": 860,
|
869 |
"total_memory_available (GB)": 94.62
|
870 |
},
|
871 |
{
|
872 |
"epoch": 3.5510204081632653,
|
873 |
+
"grad_norm": 0.3760354816913605,
|
874 |
"learning_rate": 2.8979591836734694e-06,
|
875 |
+
"loss": 0.064,
|
876 |
+
"max_memory_allocated (GB)": 60.52,
|
877 |
"memory_allocated (GB)": 50.57,
|
878 |
"step": 870,
|
879 |
"total_memory_available (GB)": 94.62
|
880 |
},
|
881 |
{
|
882 |
"epoch": 3.5918367346938775,
|
883 |
+
"grad_norm": 1.0110430717468262,
|
884 |
"learning_rate": 2.816326530612245e-06,
|
885 |
+
"loss": 0.0433,
|
886 |
+
"max_memory_allocated (GB)": 60.52,
|
887 |
"memory_allocated (GB)": 50.57,
|
888 |
"step": 880,
|
889 |
"total_memory_available (GB)": 94.62
|
890 |
},
|
891 |
{
|
892 |
"epoch": 3.63265306122449,
|
893 |
+
"grad_norm": 0.1704891324043274,
|
894 |
"learning_rate": 2.7346938775510207e-06,
|
895 |
+
"loss": 0.0282,
|
896 |
+
"max_memory_allocated (GB)": 60.52,
|
897 |
"memory_allocated (GB)": 50.57,
|
898 |
"step": 890,
|
899 |
"total_memory_available (GB)": 94.62
|
900 |
},
|
901 |
{
|
902 |
"epoch": 3.673469387755102,
|
903 |
+
"grad_norm": 0.8289852142333984,
|
904 |
"learning_rate": 2.6530612244897964e-06,
|
905 |
+
"loss": 0.0518,
|
906 |
+
"max_memory_allocated (GB)": 60.52,
|
907 |
"memory_allocated (GB)": 50.57,
|
908 |
"step": 900,
|
909 |
"total_memory_available (GB)": 94.62
|
910 |
},
|
911 |
{
|
912 |
"epoch": 3.7142857142857144,
|
913 |
+
"grad_norm": 0.812251091003418,
|
914 |
"learning_rate": 2.571428571428571e-06,
|
915 |
+
"loss": 0.0598,
|
916 |
+
"max_memory_allocated (GB)": 60.52,
|
917 |
"memory_allocated (GB)": 50.57,
|
918 |
"step": 910,
|
919 |
"total_memory_available (GB)": 94.62
|
920 |
},
|
921 |
{
|
922 |
"epoch": 3.7551020408163263,
|
923 |
+
"grad_norm": 0.14789150655269623,
|
924 |
"learning_rate": 2.489795918367347e-06,
|
925 |
+
"loss": 0.0542,
|
926 |
+
"max_memory_allocated (GB)": 60.52,
|
927 |
"memory_allocated (GB)": 50.57,
|
928 |
"step": 920,
|
929 |
"total_memory_available (GB)": 94.62
|
930 |
},
|
931 |
{
|
932 |
"epoch": 3.795918367346939,
|
933 |
+
"grad_norm": 0.8937461972236633,
|
934 |
"learning_rate": 2.4081632653061225e-06,
|
935 |
+
"loss": 0.0499,
|
936 |
+
"max_memory_allocated (GB)": 60.52,
|
937 |
"memory_allocated (GB)": 50.57,
|
938 |
"step": 930,
|
939 |
"total_memory_available (GB)": 94.62
|
940 |
},
|
941 |
{
|
942 |
"epoch": 3.836734693877551,
|
943 |
+
"grad_norm": 6.832267761230469,
|
944 |
"learning_rate": 2.326530612244898e-06,
|
945 |
"loss": 0.0489,
|
946 |
+
"max_memory_allocated (GB)": 60.52,
|
947 |
"memory_allocated (GB)": 50.57,
|
948 |
"step": 940,
|
949 |
"total_memory_available (GB)": 94.62
|
950 |
},
|
951 |
{
|
952 |
"epoch": 3.877551020408163,
|
953 |
+
"grad_norm": 0.7444981336593628,
|
954 |
"learning_rate": 2.244897959183674e-06,
|
955 |
+
"loss": 0.0591,
|
956 |
+
"max_memory_allocated (GB)": 60.52,
|
957 |
"memory_allocated (GB)": 50.57,
|
958 |
"step": 950,
|
959 |
"total_memory_available (GB)": 94.62
|
960 |
},
|
961 |
{
|
962 |
"epoch": 3.9183673469387754,
|
963 |
+
"grad_norm": 0.38545939326286316,
|
964 |
"learning_rate": 2.1632653061224495e-06,
|
965 |
+
"loss": 0.0499,
|
966 |
+
"max_memory_allocated (GB)": 60.52,
|
967 |
"memory_allocated (GB)": 50.57,
|
968 |
"step": 960,
|
969 |
"total_memory_available (GB)": 94.62
|
970 |
},
|
971 |
{
|
972 |
"epoch": 3.9591836734693877,
|
973 |
+
"grad_norm": 1.0209099054336548,
|
974 |
"learning_rate": 2.0816326530612247e-06,
|
975 |
+
"loss": 0.0736,
|
976 |
+
"max_memory_allocated (GB)": 60.52,
|
977 |
"memory_allocated (GB)": 50.57,
|
978 |
"step": 970,
|
979 |
"total_memory_available (GB)": 94.62
|
980 |
},
|
981 |
{
|
982 |
"epoch": 4.0,
|
983 |
+
"grad_norm": 0.6384561657905579,
|
984 |
"learning_rate": 2.0000000000000003e-06,
|
985 |
+
"loss": 0.0705,
|
986 |
+
"max_memory_allocated (GB)": 60.52,
|
987 |
"memory_allocated (GB)": 50.57,
|
988 |
"step": 980,
|
989 |
"total_memory_available (GB)": 94.62
|
990 |
},
|
991 |
{
|
992 |
"epoch": 4.040816326530612,
|
993 |
+
"grad_norm": 0.6284443736076355,
|
994 |
"learning_rate": 1.9183673469387756e-06,
|
995 |
"loss": 0.0483,
|
996 |
+
"max_memory_allocated (GB)": 60.52,
|
997 |
"memory_allocated (GB)": 50.57,
|
998 |
"step": 990,
|
999 |
"total_memory_available (GB)": 94.62
|
1000 |
},
|
1001 |
{
|
1002 |
"epoch": 4.081632653061225,
|
1003 |
+
"grad_norm": 1.0421397686004639,
|
1004 |
"learning_rate": 1.8367346938775512e-06,
|
1005 |
+
"loss": 0.0627,
|
1006 |
+
"max_memory_allocated (GB)": 60.52,
|
1007 |
"memory_allocated (GB)": 50.57,
|
1008 |
"step": 1000,
|
1009 |
"total_memory_available (GB)": 94.62
|
1010 |
},
|
1011 |
{
|
1012 |
"epoch": 4.122448979591836,
|
1013 |
+
"grad_norm": 0.6968585252761841,
|
1014 |
"learning_rate": 1.7551020408163267e-06,
|
1015 |
+
"loss": 0.0632,
|
1016 |
+
"max_memory_allocated (GB)": 60.52,
|
1017 |
"memory_allocated (GB)": 50.57,
|
1018 |
"step": 1010,
|
1019 |
"total_memory_available (GB)": 94.62
|
1020 |
},
|
1021 |
{
|
1022 |
"epoch": 4.163265306122449,
|
1023 |
+
"grad_norm": 0.35346081852912903,
|
1024 |
"learning_rate": 1.6734693877551023e-06,
|
1025 |
+
"loss": 0.0402,
|
1026 |
+
"max_memory_allocated (GB)": 60.52,
|
1027 |
"memory_allocated (GB)": 50.57,
|
1028 |
"step": 1020,
|
1029 |
"total_memory_available (GB)": 94.62
|
1030 |
},
|
1031 |
{
|
1032 |
"epoch": 4.204081632653061,
|
1033 |
+
"grad_norm": 0.8111013174057007,
|
1034 |
"learning_rate": 1.5918367346938775e-06,
|
1035 |
+
"loss": 0.0528,
|
1036 |
+
"max_memory_allocated (GB)": 60.52,
|
1037 |
"memory_allocated (GB)": 50.57,
|
1038 |
"step": 1030,
|
1039 |
"total_memory_available (GB)": 94.62
|
1040 |
},
|
1041 |
{
|
1042 |
"epoch": 4.244897959183674,
|
1043 |
+
"grad_norm": 0.5146018862724304,
|
1044 |
"learning_rate": 1.5102040816326532e-06,
|
1045 |
+
"loss": 0.0396,
|
1046 |
+
"max_memory_allocated (GB)": 60.52,
|
1047 |
"memory_allocated (GB)": 50.57,
|
1048 |
"step": 1040,
|
1049 |
"total_memory_available (GB)": 94.62
|
1050 |
},
|
1051 |
{
|
1052 |
"epoch": 4.285714285714286,
|
1053 |
+
"grad_norm": 0.6711630821228027,
|
1054 |
"learning_rate": 1.4285714285714286e-06,
|
1055 |
+
"loss": 0.0493,
|
1056 |
+
"max_memory_allocated (GB)": 60.52,
|
1057 |
"memory_allocated (GB)": 50.57,
|
1058 |
"step": 1050,
|
1059 |
"total_memory_available (GB)": 94.62
|
1060 |
},
|
1061 |
{
|
1062 |
"epoch": 4.326530612244898,
|
1063 |
+
"grad_norm": 0.3221682012081146,
|
1064 |
"learning_rate": 1.3469387755102043e-06,
|
1065 |
+
"loss": 0.0454,
|
1066 |
+
"max_memory_allocated (GB)": 60.52,
|
1067 |
"memory_allocated (GB)": 50.57,
|
1068 |
"step": 1060,
|
1069 |
"total_memory_available (GB)": 94.62
|
1070 |
},
|
1071 |
{
|
1072 |
"epoch": 4.36734693877551,
|
1073 |
+
"grad_norm": 0.9966460466384888,
|
1074 |
"learning_rate": 1.2653061224489795e-06,
|
1075 |
+
"loss": 0.0485,
|
1076 |
+
"max_memory_allocated (GB)": 60.52,
|
1077 |
"memory_allocated (GB)": 50.57,
|
1078 |
"step": 1070,
|
1079 |
"total_memory_available (GB)": 94.62
|
1080 |
},
|
1081 |
{
|
1082 |
"epoch": 4.408163265306122,
|
1083 |
+
"grad_norm": 0.45073866844177246,
|
1084 |
"learning_rate": 1.1836734693877552e-06,
|
1085 |
+
"loss": 0.052,
|
1086 |
+
"max_memory_allocated (GB)": 60.52,
|
1087 |
"memory_allocated (GB)": 50.57,
|
1088 |
"step": 1080,
|
1089 |
"total_memory_available (GB)": 94.62
|
1090 |
},
|
1091 |
{
|
1092 |
"epoch": 4.448979591836735,
|
1093 |
+
"grad_norm": 0.7129877805709839,
|
1094 |
"learning_rate": 1.1020408163265308e-06,
|
1095 |
+
"loss": 0.0511,
|
1096 |
+
"max_memory_allocated (GB)": 60.52,
|
1097 |
"memory_allocated (GB)": 50.57,
|
1098 |
"step": 1090,
|
1099 |
"total_memory_available (GB)": 94.62
|
1100 |
},
|
1101 |
{
|
1102 |
"epoch": 4.489795918367347,
|
1103 |
+
"grad_norm": 0.333344966173172,
|
1104 |
"learning_rate": 1.0204081632653063e-06,
|
1105 |
+
"loss": 0.0352,
|
1106 |
+
"max_memory_allocated (GB)": 60.52,
|
1107 |
"memory_allocated (GB)": 50.57,
|
1108 |
"step": 1100,
|
1109 |
"total_memory_available (GB)": 94.62
|
1110 |
},
|
1111 |
{
|
1112 |
"epoch": 4.530612244897959,
|
1113 |
+
"grad_norm": 0.11522499471902847,
|
1114 |
"learning_rate": 9.387755102040817e-07,
|
1115 |
+
"loss": 0.0457,
|
1116 |
+
"max_memory_allocated (GB)": 60.52,
|
1117 |
"memory_allocated (GB)": 50.57,
|
1118 |
"step": 1110,
|
1119 |
"total_memory_available (GB)": 94.62
|
1120 |
},
|
1121 |
{
|
1122 |
"epoch": 4.571428571428571,
|
1123 |
+
"grad_norm": 0.3113747239112854,
|
1124 |
"learning_rate": 8.571428571428572e-07,
|
1125 |
+
"loss": 0.0402,
|
1126 |
+
"max_memory_allocated (GB)": 60.52,
|
1127 |
"memory_allocated (GB)": 50.57,
|
1128 |
"step": 1120,
|
1129 |
"total_memory_available (GB)": 94.62
|
1130 |
},
|
1131 |
{
|
1132 |
"epoch": 4.612244897959184,
|
1133 |
+
"grad_norm": 0.8939142823219299,
|
1134 |
"learning_rate": 7.755102040816327e-07,
|
1135 |
+
"loss": 0.0536,
|
1136 |
+
"max_memory_allocated (GB)": 60.52,
|
1137 |
"memory_allocated (GB)": 50.57,
|
1138 |
"step": 1130,
|
1139 |
"total_memory_available (GB)": 94.62
|
1140 |
},
|
1141 |
{
|
1142 |
"epoch": 4.653061224489796,
|
1143 |
+
"grad_norm": 0.7175144553184509,
|
1144 |
"learning_rate": 6.938775510204082e-07,
|
1145 |
+
"loss": 0.0724,
|
1146 |
+
"max_memory_allocated (GB)": 60.52,
|
1147 |
"memory_allocated (GB)": 50.57,
|
1148 |
"step": 1140,
|
1149 |
"total_memory_available (GB)": 94.62
|
1150 |
},
|
1151 |
{
|
1152 |
"epoch": 4.6938775510204085,
|
1153 |
+
"grad_norm": 0.6586973071098328,
|
1154 |
"learning_rate": 6.122448979591837e-07,
|
1155 |
+
"loss": 0.0514,
|
1156 |
+
"max_memory_allocated (GB)": 60.52,
|
1157 |
"memory_allocated (GB)": 50.57,
|
1158 |
"step": 1150,
|
1159 |
"total_memory_available (GB)": 94.62
|
1160 |
},
|
1161 |
{
|
1162 |
"epoch": 4.73469387755102,
|
1163 |
+
"grad_norm": 1.190521478652954,
|
1164 |
"learning_rate": 5.306122448979592e-07,
|
1165 |
+
"loss": 0.0608,
|
1166 |
+
"max_memory_allocated (GB)": 60.52,
|
1167 |
"memory_allocated (GB)": 50.57,
|
1168 |
"step": 1160,
|
1169 |
"total_memory_available (GB)": 94.62
|
1170 |
},
|
1171 |
{
|
1172 |
"epoch": 4.775510204081632,
|
1173 |
+
"grad_norm": 0.47001194953918457,
|
1174 |
"learning_rate": 4.489795918367347e-07,
|
1175 |
+
"loss": 0.0548,
|
1176 |
+
"max_memory_allocated (GB)": 60.52,
|
1177 |
"memory_allocated (GB)": 50.57,
|
1178 |
"step": 1170,
|
1179 |
"total_memory_available (GB)": 94.62
|
1180 |
},
|
1181 |
{
|
1182 |
"epoch": 4.816326530612245,
|
1183 |
+
"grad_norm": 0.9076634049415588,
|
1184 |
"learning_rate": 3.6734693877551025e-07,
|
1185 |
+
"loss": 0.0438,
|
1186 |
+
"max_memory_allocated (GB)": 60.52,
|
1187 |
"memory_allocated (GB)": 50.57,
|
1188 |
"step": 1180,
|
1189 |
"total_memory_available (GB)": 94.62
|
1190 |
},
|
1191 |
{
|
1192 |
"epoch": 4.857142857142857,
|
1193 |
+
"grad_norm": 1.4096022844314575,
|
1194 |
"learning_rate": 2.8571428571428575e-07,
|
1195 |
+
"loss": 0.0542,
|
1196 |
+
"max_memory_allocated (GB)": 60.52,
|
1197 |
"memory_allocated (GB)": 50.57,
|
1198 |
"step": 1190,
|
1199 |
"total_memory_available (GB)": 94.62
|
1200 |
},
|
1201 |
{
|
1202 |
"epoch": 4.8979591836734695,
|
1203 |
+
"grad_norm": 0.7559374570846558,
|
1204 |
"learning_rate": 2.0408163265306121e-07,
|
1205 |
+
"loss": 0.0715,
|
1206 |
+
"max_memory_allocated (GB)": 60.52,
|
1207 |
"memory_allocated (GB)": 50.57,
|
1208 |
"step": 1200,
|
1209 |
"total_memory_available (GB)": 94.62
|
1210 |
},
|
1211 |
{
|
1212 |
"epoch": 4.938775510204081,
|
1213 |
+
"grad_norm": 0.6551031470298767,
|
1214 |
"learning_rate": 1.2244897959183673e-07,
|
1215 |
+
"loss": 0.0647,
|
1216 |
+
"max_memory_allocated (GB)": 60.52,
|
1217 |
"memory_allocated (GB)": 50.57,
|
1218 |
"step": 1210,
|
1219 |
"total_memory_available (GB)": 94.62
|
1220 |
},
|
1221 |
{
|
1222 |
"epoch": 4.979591836734694,
|
1223 |
+
"grad_norm": 0.9617031812667847,
|
1224 |
"learning_rate": 4.0816326530612253e-08,
|
1225 |
+
"loss": 0.0418,
|
1226 |
+
"max_memory_allocated (GB)": 60.52,
|
1227 |
"memory_allocated (GB)": 50.57,
|
1228 |
"step": 1220,
|
1229 |
"total_memory_available (GB)": 94.62
|
1230 |
},
|
1231 |
{
|
1232 |
"epoch": 5.0,
|
1233 |
+
"max_memory_allocated (GB)": 60.52,
|
1234 |
"memory_allocated (GB)": 50.57,
|
1235 |
"step": 1225,
|
1236 |
"total_flos": 3.0598946525952e+16,
|
1237 |
"total_memory_available (GB)": 94.62,
|
1238 |
+
"train_loss": 0.06082177159737567,
|
1239 |
+
"train_runtime": 1079.1005,
|
1240 |
+
"train_samples_per_second": 52.128,
|
1241 |
+
"train_steps_per_second": 1.304
|
1242 |
}
|
1243 |
],
|
1244 |
"logging_steps": 10,
|
validation_results.json
CHANGED
@@ -1,10 +1,10 @@
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
-
"eval_loss": 0.
|
4 |
-
"eval_runtime":
|
5 |
-
"eval_samples_per_second":
|
6 |
-
"eval_steps_per_second":
|
7 |
-
"max_memory_allocated (GB)":
|
8 |
"memory_allocated (GB)": 51.27,
|
9 |
"total_memory_available (GB)": 94.62
|
10 |
}
|
|
|
1 |
{
|
2 |
"epoch": 5.0,
|
3 |
+
"eval_loss": 0.12328128516674042,
|
4 |
+
"eval_runtime": 24.5719,
|
5 |
+
"eval_samples_per_second": 36.301,
|
6 |
+
"eval_steps_per_second": 2.33,
|
7 |
+
"max_memory_allocated (GB)": 60.52,
|
8 |
"memory_allocated (GB)": 51.27,
|
9 |
"total_memory_available (GB)": 94.62
|
10 |
}
|