Training in progress, step 1000
Browse files- adapter_config.json +4 -4
- adapter_model.safetensors +1 -1
- runs/Jan16_13-31-57_pss5xry2f0aq/events.out.tfevents.1737034477.pss5xry2f0aq.31242.0 +2 -2
- runs/Jan16_18-03-50_pss5xry2f0aq/events.out.tfevents.1737050783.pss5xry2f0aq.165850.0 +3 -0
- runs/Jan16_18-37-53_pss5xry2f0aq/events.out.tfevents.1737052838.pss5xry2f0aq.173203.0 +3 -0
- trainer_log.jsonl +100 -204
- training_args.bin +1 -1
adapter_config.json
CHANGED
@@ -20,13 +20,13 @@
|
|
20 |
"rank_pattern": {},
|
21 |
"revision": null,
|
22 |
"target_modules": [
|
23 |
-
"
|
24 |
-
"up_proj",
|
25 |
"down_proj",
|
|
|
26 |
"v_proj",
|
|
|
27 |
"k_proj",
|
28 |
-
"
|
29 |
-
"o_proj"
|
30 |
],
|
31 |
"task_type": "CAUSAL_LM",
|
32 |
"use_dora": false,
|
|
|
20 |
"rank_pattern": {},
|
21 |
"revision": null,
|
22 |
"target_modules": [
|
23 |
+
"o_proj",
|
|
|
24 |
"down_proj",
|
25 |
+
"q_proj",
|
26 |
"v_proj",
|
27 |
+
"up_proj",
|
28 |
"k_proj",
|
29 |
+
"gate_proj"
|
|
|
30 |
],
|
31 |
"task_type": "CAUSAL_LM",
|
32 |
"use_dora": false,
|
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 48679352
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c4f78cd90c0f408ae5e71a25c4343b5d0880e7a8fc87ce2831eb6903c742baf0
|
3 |
size 48679352
|
runs/Jan16_13-31-57_pss5xry2f0aq/events.out.tfevents.1737034477.pss5xry2f0aq.31242.0
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1c1d163d45eb7179c8d83e9dfacddf788e21965a84abc2c53665f5f0dd399d41
|
3 |
+
size 59401
|
runs/Jan16_18-03-50_pss5xry2f0aq/events.out.tfevents.1737050783.pss5xry2f0aq.165850.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b7ac0bc769328b2584fd7037922c1f54bf683b6015de0aca2aeb0c44fb81025a
|
3 |
+
size 16229
|
runs/Jan16_18-37-53_pss5xry2f0aq/events.out.tfevents.1737052838.pss5xry2f0aq.173203.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b45e0ac709ea020919c95baab1bb09a6c6e027e62b719d1e25863642460c6a38
|
3 |
+
size 26669
|
trainer_log.jsonl
CHANGED
@@ -1,204 +1,100 @@
|
|
1 |
-
{"current_steps": 10, "total_steps":
|
2 |
-
{"current_steps": 20, "total_steps":
|
3 |
-
{"current_steps": 30, "total_steps":
|
4 |
-
{"current_steps": 40, "total_steps":
|
5 |
-
{"current_steps": 50, "total_steps":
|
6 |
-
{"current_steps": 60, "total_steps":
|
7 |
-
{"current_steps": 70, "total_steps":
|
8 |
-
{"current_steps": 80, "total_steps":
|
9 |
-
{"current_steps": 90, "total_steps":
|
10 |
-
{"current_steps": 100, "total_steps":
|
11 |
-
{"current_steps": 110, "total_steps":
|
12 |
-
{"current_steps": 120, "total_steps":
|
13 |
-
{"current_steps": 130, "total_steps":
|
14 |
-
{"current_steps": 140, "total_steps":
|
15 |
-
{"current_steps": 150, "total_steps":
|
16 |
-
{"current_steps": 160, "total_steps":
|
17 |
-
{"current_steps": 170, "total_steps":
|
18 |
-
{"current_steps": 180, "total_steps":
|
19 |
-
{"current_steps": 190, "total_steps":
|
20 |
-
{"current_steps": 200, "total_steps":
|
21 |
-
{"current_steps": 210, "total_steps":
|
22 |
-
{"current_steps": 220, "total_steps":
|
23 |
-
{"current_steps": 230, "total_steps":
|
24 |
-
{"current_steps": 240, "total_steps":
|
25 |
-
{"current_steps": 250, "total_steps":
|
26 |
-
{"current_steps": 260, "total_steps":
|
27 |
-
{"current_steps": 270, "total_steps":
|
28 |
-
{"current_steps": 280, "total_steps":
|
29 |
-
{"current_steps": 290, "total_steps":
|
30 |
-
{"current_steps": 300, "total_steps":
|
31 |
-
{"current_steps": 310, "total_steps":
|
32 |
-
{"current_steps": 320, "total_steps":
|
33 |
-
{"current_steps": 330, "total_steps":
|
34 |
-
{"current_steps": 340, "total_steps":
|
35 |
-
{"current_steps": 350, "total_steps":
|
36 |
-
{"current_steps": 360, "total_steps":
|
37 |
-
{"current_steps": 370, "total_steps":
|
38 |
-
{"current_steps": 380, "total_steps":
|
39 |
-
{"current_steps": 390, "total_steps":
|
40 |
-
{"current_steps": 400, "total_steps":
|
41 |
-
{"current_steps": 410, "total_steps":
|
42 |
-
{"current_steps": 420, "total_steps":
|
43 |
-
{"current_steps": 430, "total_steps":
|
44 |
-
{"current_steps": 440, "total_steps":
|
45 |
-
{"current_steps": 450, "total_steps":
|
46 |
-
{"current_steps": 460, "total_steps":
|
47 |
-
{"current_steps": 470, "total_steps":
|
48 |
-
{"current_steps": 480, "total_steps":
|
49 |
-
{"current_steps": 490, "total_steps":
|
50 |
-
{"current_steps": 500, "total_steps":
|
51 |
-
{"current_steps":
|
52 |
-
{"current_steps":
|
53 |
-
{"current_steps":
|
54 |
-
{"current_steps":
|
55 |
-
{"current_steps":
|
56 |
-
{"current_steps":
|
57 |
-
{"current_steps":
|
58 |
-
{"current_steps":
|
59 |
-
{"current_steps":
|
60 |
-
{"current_steps":
|
61 |
-
{"current_steps":
|
62 |
-
{"current_steps":
|
63 |
-
{"current_steps":
|
64 |
-
{"current_steps":
|
65 |
-
{"current_steps":
|
66 |
-
{"current_steps":
|
67 |
-
{"current_steps":
|
68 |
-
{"current_steps":
|
69 |
-
{"current_steps":
|
70 |
-
{"current_steps":
|
71 |
-
{"current_steps":
|
72 |
-
{"current_steps":
|
73 |
-
{"current_steps":
|
74 |
-
{"current_steps":
|
75 |
-
{"current_steps":
|
76 |
-
{"current_steps":
|
77 |
-
{"current_steps":
|
78 |
-
{"current_steps":
|
79 |
-
{"current_steps":
|
80 |
-
{"current_steps":
|
81 |
-
{"current_steps":
|
82 |
-
{"current_steps":
|
83 |
-
{"current_steps":
|
84 |
-
{"current_steps":
|
85 |
-
{"current_steps":
|
86 |
-
{"current_steps":
|
87 |
-
{"current_steps":
|
88 |
-
{"current_steps":
|
89 |
-
{"current_steps":
|
90 |
-
{"current_steps":
|
91 |
-
{"current_steps":
|
92 |
-
{"current_steps":
|
93 |
-
{"current_steps":
|
94 |
-
{"current_steps":
|
95 |
-
{"current_steps":
|
96 |
-
{"current_steps":
|
97 |
-
{"current_steps":
|
98 |
-
{"current_steps":
|
99 |
-
{"current_steps":
|
100 |
-
{"current_steps":
|
101 |
-
{"current_steps": 1000, "total_steps": 3635, "loss": 1.1462, "lr": 9.095836731737017e-05, "epoch": 0.027516372241483684, "percentage": 27.51, "elapsed_time": "1:22:15", "remaining_time": "3:36:45"}
|
102 |
-
{"current_steps": 1000, "total_steps": 3635, "eval_loss": 1.3045305013656616, "epoch": 0.027516372241483684, "percentage": 27.51, "elapsed_time": "1:56:06", "remaining_time": "5:05:56"}
|
103 |
-
{"current_steps": 1010, "total_steps": 3635, "loss": 1.1179, "lr": 9.068105067826224e-05, "epoch": 0.02779153596389852, "percentage": 27.79, "elapsed_time": "1:56:28", "remaining_time": "5:02:42"}
|
104 |
-
{"current_steps": 1020, "total_steps": 3635, "loss": 1.2901, "lr": 9.039998148058193e-05, "epoch": 0.028066699686313357, "percentage": 28.06, "elapsed_time": "1:57:00", "remaining_time": "4:59:59"}
|
105 |
-
{"current_steps": 1030, "total_steps": 3635, "loss": 1.1959, "lr": 9.011518565110861e-05, "epoch": 0.028341863408728192, "percentage": 28.34, "elapsed_time": "1:57:29", "remaining_time": "4:57:09"}
|
106 |
-
{"current_steps": 1040, "total_steps": 3635, "loss": 1.1765, "lr": 8.982668946037892e-05, "epoch": 0.02861702713114303, "percentage": 28.61, "elapsed_time": "1:57:56", "remaining_time": "4:54:18"}
|
107 |
-
{"current_steps": 1050, "total_steps": 3635, "loss": 1.1444, "lr": 8.953451952026337e-05, "epoch": 0.02889219085355787, "percentage": 28.89, "elapsed_time": "1:58:16", "remaining_time": "4:51:09"}
|
108 |
-
{"current_steps": 1060, "total_steps": 3635, "loss": 1.1443, "lr": 8.923870278151164e-05, "epoch": 0.029167354575972703, "percentage": 29.16, "elapsed_time": "1:58:50", "remaining_time": "4:48:41"}
|
109 |
-
{"current_steps": 1070, "total_steps": 3635, "loss": 1.0915, "lr": 8.893926653126654e-05, "epoch": 0.02944251829838754, "percentage": 29.44, "elapsed_time": "1:59:18", "remaining_time": "4:46:01"}
|
110 |
-
{"current_steps": 1080, "total_steps": 3635, "loss": 1.0851, "lr": 8.863623839054688e-05, "epoch": 0.029717682020802376, "percentage": 29.71, "elapsed_time": "1:59:54", "remaining_time": "4:43:40"}
|
111 |
-
{"current_steps": 1090, "total_steps": 3635, "loss": 1.1301, "lr": 8.832964631169977e-05, "epoch": 0.029992845743217214, "percentage": 29.99, "elapsed_time": "2:00:14", "remaining_time": "4:40:45"}
|
112 |
-
{"current_steps": 1100, "total_steps": 3635, "loss": 1.0452, "lr": 8.8019518575822e-05, "epoch": 0.030268009465632052, "percentage": 30.26, "elapsed_time": "2:00:35", "remaining_time": "4:37:55"}
|
113 |
-
{"current_steps": 1110, "total_steps": 3635, "loss": 1.0864, "lr": 8.770588379015148e-05, "epoch": 0.030543173188046887, "percentage": 30.54, "elapsed_time": "2:00:59", "remaining_time": "4:35:12"}
|
114 |
-
{"current_steps": 1120, "total_steps": 3635, "loss": 1.2132, "lr": 8.73887708854283e-05, "epoch": 0.030818336910461725, "percentage": 30.81, "elapsed_time": "2:01:43", "remaining_time": "4:33:20"}
|
115 |
-
{"current_steps": 1130, "total_steps": 3635, "loss": 1.1024, "lr": 8.706820911322606e-05, "epoch": 0.031093500632876563, "percentage": 31.09, "elapsed_time": "2:02:15", "remaining_time": "4:31:01"}
|
116 |
-
{"current_steps": 1140, "total_steps": 3635, "loss": 1.1621, "lr": 8.674422804325365e-05, "epoch": 0.0313686643552914, "percentage": 31.36, "elapsed_time": "2:02:49", "remaining_time": "4:28:49"}
|
117 |
-
{"current_steps": 1150, "total_steps": 3635, "loss": 1.1513, "lr": 8.641685756062767e-05, "epoch": 0.031643828077706236, "percentage": 31.64, "elapsed_time": "2:03:20", "remaining_time": "4:26:30"}
|
118 |
-
{"current_steps": 1160, "total_steps": 3635, "loss": 1.0454, "lr": 8.608612786311562e-05, "epoch": 0.031918991800121074, "percentage": 31.91, "elapsed_time": "2:03:42", "remaining_time": "4:23:56"}
|
119 |
-
{"current_steps": 1170, "total_steps": 3635, "loss": 1.2426, "lr": 8.575206945835044e-05, "epoch": 0.032194155522535906, "percentage": 32.19, "elapsed_time": "2:04:21", "remaining_time": "4:22:00"}
|
120 |
-
{"current_steps": 1180, "total_steps": 3635, "loss": 1.1782, "lr": 8.541471316101634e-05, "epoch": 0.032469319244950744, "percentage": 32.46, "elapsed_time": "2:04:38", "remaining_time": "4:19:18"}
|
121 |
-
{"current_steps": 1190, "total_steps": 3635, "loss": 1.1244, "lr": 8.50740900900064e-05, "epoch": 0.03274448296736558, "percentage": 32.74, "elapsed_time": "2:05:08", "remaining_time": "4:17:06"}
|
122 |
-
{"current_steps": 1200, "total_steps": 3635, "loss": 1.0362, "lr": 8.473023166555199e-05, "epoch": 0.03301964668978042, "percentage": 33.01, "elapsed_time": "2:05:29", "remaining_time": "4:14:38"}
|
123 |
-
{"current_steps": 1210, "total_steps": 3635, "loss": 1.1304, "lr": 8.438316960632449e-05, "epoch": 0.03329481041219526, "percentage": 33.29, "elapsed_time": "2:05:55", "remaining_time": "4:12:22"}
|
124 |
-
{"current_steps": 1220, "total_steps": 3635, "loss": 1.0598, "lr": 8.40329359265095e-05, "epoch": 0.033569974134610096, "percentage": 33.56, "elapsed_time": "2:06:17", "remaining_time": "4:09:59"}
|
125 |
-
{"current_steps": 1230, "total_steps": 3635, "loss": 1.1836, "lr": 8.367956293285362e-05, "epoch": 0.03384513785702493, "percentage": 33.84, "elapsed_time": "2:06:46", "remaining_time": "4:07:53"}
|
126 |
-
{"current_steps": 1240, "total_steps": 3635, "loss": 1.0994, "lr": 8.332308322168452e-05, "epoch": 0.034120301579439766, "percentage": 34.11, "elapsed_time": "2:07:05", "remaining_time": "4:05:28"}
|
127 |
-
{"current_steps": 1250, "total_steps": 3635, "loss": 1.1313, "lr": 8.296352967590403e-05, "epoch": 0.034395465301854604, "percentage": 34.39, "elapsed_time": "2:07:28", "remaining_time": "4:03:12"}
|
128 |
-
{"current_steps": 1260, "total_steps": 3635, "loss": 1.1368, "lr": 8.260093546195497e-05, "epoch": 0.03467062902426944, "percentage": 34.66, "elapsed_time": "2:07:55", "remaining_time": "4:01:08"}
|
129 |
-
{"current_steps": 1270, "total_steps": 3635, "loss": 1.081, "lr": 8.223533402676168e-05, "epoch": 0.03494579274668428, "percentage": 34.94, "elapsed_time": "2:08:13", "remaining_time": "3:58:46"}
|
130 |
-
{"current_steps": 1280, "total_steps": 3635, "loss": 1.2727, "lr": 8.18667590946449e-05, "epoch": 0.03522095646909911, "percentage": 35.21, "elapsed_time": "2:08:39", "remaining_time": "3:56:43"}
|
131 |
-
{"current_steps": 1290, "total_steps": 3635, "loss": 1.2116, "lr": 8.149524466421081e-05, "epoch": 0.03549612019151395, "percentage": 35.49, "elapsed_time": "2:09:20", "remaining_time": "3:55:07"}
|
132 |
-
{"current_steps": 1300, "total_steps": 3635, "loss": 1.1356, "lr": 8.11208250052149e-05, "epoch": 0.03577128391392879, "percentage": 35.76, "elapsed_time": "2:09:55", "remaining_time": "3:53:21"}
|
133 |
-
{"current_steps": 1310, "total_steps": 3635, "loss": 1.1021, "lr": 8.074353465540088e-05, "epoch": 0.036046447636343626, "percentage": 36.04, "elapsed_time": "2:10:24", "remaining_time": "3:51:27"}
|
134 |
-
{"current_steps": 1320, "total_steps": 3635, "loss": 1.0813, "lr": 8.036340841731471e-05, "epoch": 0.036321611358758464, "percentage": 36.31, "elapsed_time": "2:10:49", "remaining_time": "3:49:26"}
|
135 |
-
{"current_steps": 1330, "total_steps": 3635, "loss": 1.1339, "lr": 7.998048135509432e-05, "epoch": 0.036596775081173295, "percentage": 36.59, "elapsed_time": "2:11:16", "remaining_time": "3:47:30"}
|
136 |
-
{"current_steps": 1340, "total_steps": 3635, "loss": 1.1747, "lr": 7.959478879123529e-05, "epoch": 0.036871938803588133, "percentage": 36.86, "elapsed_time": "2:11:35", "remaining_time": "3:45:22"}
|
137 |
-
{"current_steps": 1350, "total_steps": 3635, "loss": 1.1202, "lr": 7.920636630333235e-05, "epoch": 0.03714710252600297, "percentage": 37.14, "elapsed_time": "2:11:59", "remaining_time": "3:43:24"}
|
138 |
-
{"current_steps": 1360, "total_steps": 3635, "loss": 1.1472, "lr": 7.881524972079782e-05, "epoch": 0.03742226624841781, "percentage": 37.41, "elapsed_time": "2:12:27", "remaining_time": "3:41:34"}
|
139 |
-
{"current_steps": 1370, "total_steps": 3635, "loss": 1.1254, "lr": 7.84214751215564e-05, "epoch": 0.03769742997083265, "percentage": 37.69, "elapsed_time": "2:12:58", "remaining_time": "3:39:50"}
|
140 |
-
{"current_steps": 1380, "total_steps": 3635, "loss": 1.2529, "lr": 7.802507882871737e-05, "epoch": 0.03797259369324748, "percentage": 37.96, "elapsed_time": "2:13:27", "remaining_time": "3:38:04"}
|
141 |
-
{"current_steps": 1390, "total_steps": 3635, "loss": 1.0996, "lr": 7.76260974072239e-05, "epoch": 0.03824775741566232, "percentage": 38.24, "elapsed_time": "2:13:47", "remaining_time": "3:36:05"}
|
142 |
-
{"current_steps": 1400, "total_steps": 3635, "loss": 1.1905, "lr": 7.722456766048023e-05, "epoch": 0.038522921138077156, "percentage": 38.51, "elapsed_time": "2:14:16", "remaining_time": "3:34:22"}
|
143 |
-
{"current_steps": 1410, "total_steps": 3635, "loss": 1.1614, "lr": 7.682052662695682e-05, "epoch": 0.038798084860491994, "percentage": 38.79, "elapsed_time": "2:14:33", "remaining_time": "3:32:19"}
|
144 |
-
{"current_steps": 1420, "total_steps": 3635, "loss": 1.1234, "lr": 7.641401157677376e-05, "epoch": 0.03907324858290683, "percentage": 39.06, "elapsed_time": "2:14:51", "remaining_time": "3:30:22"}
|
145 |
-
{"current_steps": 1430, "total_steps": 3635, "loss": 1.1555, "lr": 7.600506000826285e-05, "epoch": 0.03934841230532166, "percentage": 39.34, "elapsed_time": "2:15:37", "remaining_time": "3:29:07"}
|
146 |
-
{"current_steps": 1440, "total_steps": 3635, "loss": 1.1112, "lr": 7.559370964450861e-05, "epoch": 0.0396235760277365, "percentage": 39.61, "elapsed_time": "2:16:01", "remaining_time": "3:27:20"}
|
147 |
-
{"current_steps": 1450, "total_steps": 3635, "loss": 1.0284, "lr": 7.517999842986864e-05, "epoch": 0.03989873975015134, "percentage": 39.89, "elapsed_time": "2:16:30", "remaining_time": "3:25:42"}
|
148 |
-
{"current_steps": 1460, "total_steps": 3635, "loss": 1.1225, "lr": 7.476396452647345e-05, "epoch": 0.04017390347256618, "percentage": 40.17, "elapsed_time": "2:16:54", "remaining_time": "3:23:57"}
|
149 |
-
{"current_steps": 1470, "total_steps": 3635, "loss": 1.0926, "lr": 7.434564631070624e-05, "epoch": 0.040449067194981016, "percentage": 40.44, "elapsed_time": "2:17:24", "remaining_time": "3:22:22"}
|
150 |
-
{"current_steps": 1480, "total_steps": 3635, "loss": 1.1442, "lr": 7.3925082369663e-05, "epoch": 0.040724230917395854, "percentage": 40.72, "elapsed_time": "2:17:54", "remaining_time": "3:20:49"}
|
151 |
-
{"current_steps": 1490, "total_steps": 3635, "loss": 1.073, "lr": 7.350231149759302e-05, "epoch": 0.040999394639810685, "percentage": 40.99, "elapsed_time": "2:18:11", "remaining_time": "3:18:57"}
|
152 |
-
{"current_steps": 1500, "total_steps": 3635, "loss": 1.2027, "lr": 7.307737269232045e-05, "epoch": 0.04127455836222552, "percentage": 41.27, "elapsed_time": "2:18:35", "remaining_time": "3:17:15"}
|
153 |
-
{"current_steps": 1500, "total_steps": 3635, "eval_loss": 1.280773639678955, "epoch": 0.04127455836222552, "percentage": 41.27, "elapsed_time": "2:52:39", "remaining_time": "4:05:45"}
|
154 |
-
{"current_steps": 1510, "total_steps": 3635, "loss": 1.0848, "lr": 7.265030515164698e-05, "epoch": 0.04154972208464036, "percentage": 41.54, "elapsed_time": "2:53:17", "remaining_time": "4:03:51"}
|
155 |
-
{"current_steps": 1520, "total_steps": 3635, "loss": 1.1952, "lr": 7.222114826973607e-05, "epoch": 0.0418248858070552, "percentage": 41.82, "elapsed_time": "2:53:43", "remaining_time": "4:01:43"}
|
156 |
-
{"current_steps": 1530, "total_steps": 3635, "loss": 1.2237, "lr": 7.178994163347914e-05, "epoch": 0.04210004952947004, "percentage": 42.09, "elapsed_time": "2:54:26", "remaining_time": "4:00:00"}
|
157 |
-
{"current_steps": 1540, "total_steps": 3635, "loss": 1.1177, "lr": 7.135672501884392e-05, "epoch": 0.04237521325188487, "percentage": 42.37, "elapsed_time": "2:55:02", "remaining_time": "3:58:06"}
|
158 |
-
{"current_steps": 1550, "total_steps": 3635, "loss": 1.0888, "lr": 7.092153838720542e-05, "epoch": 0.04265037697429971, "percentage": 42.64, "elapsed_time": "2:55:35", "remaining_time": "3:56:11"}
|
159 |
-
{"current_steps": 1560, "total_steps": 3635, "loss": 1.0263, "lr": 7.048442188165969e-05, "epoch": 0.042925540696714545, "percentage": 42.92, "elapsed_time": "2:56:05", "remaining_time": "3:54:13"}
|
160 |
-
{"current_steps": 1570, "total_steps": 3635, "loss": 1.0503, "lr": 7.004541582332091e-05, "epoch": 0.043200704419129383, "percentage": 43.19, "elapsed_time": "2:56:49", "remaining_time": "3:52:33"}
|
161 |
-
{"current_steps": 1580, "total_steps": 3635, "loss": 1.1327, "lr": 6.960456070760207e-05, "epoch": 0.04347586814154422, "percentage": 43.47, "elapsed_time": "2:57:34", "remaining_time": "3:50:57"}
|
162 |
-
{"current_steps": 1590, "total_steps": 3635, "loss": 1.1632, "lr": 6.91618972004795e-05, "epoch": 0.04375103186395905, "percentage": 43.74, "elapsed_time": "2:57:58", "remaining_time": "3:48:54"}
|
163 |
-
{"current_steps": 1600, "total_steps": 3635, "loss": 1.0534, "lr": 6.87174661347417e-05, "epoch": 0.04402619558637389, "percentage": 44.02, "elapsed_time": "2:58:26", "remaining_time": "3:46:57"}
|
164 |
-
{"current_steps": 1610, "total_steps": 3635, "loss": 1.0609, "lr": 6.827130850622276e-05, "epoch": 0.04430135930878873, "percentage": 44.29, "elapsed_time": "2:58:43", "remaining_time": "3:44:47"}
|
165 |
-
{"current_steps": 1620, "total_steps": 3635, "loss": 1.1634, "lr": 6.782346547002086e-05, "epoch": 0.04457652303120357, "percentage": 44.57, "elapsed_time": "2:59:15", "remaining_time": "3:42:57"}
|
166 |
-
{"current_steps": 1630, "total_steps": 3635, "loss": 1.0655, "lr": 6.737397833670191e-05, "epoch": 0.044851686753618406, "percentage": 44.84, "elapsed_time": "2:59:40", "remaining_time": "3:41:01"}
|
167 |
-
{"current_steps": 1640, "total_steps": 3635, "loss": 1.0159, "lr": 6.692288856848888e-05, "epoch": 0.04512685047603324, "percentage": 45.12, "elapsed_time": "3:00:07", "remaining_time": "3:39:07"}
|
168 |
-
{"current_steps": 1650, "total_steps": 3635, "loss": 1.0922, "lr": 6.647023777543731e-05, "epoch": 0.045402014198448075, "percentage": 45.39, "elapsed_time": "3:00:32", "remaining_time": "3:37:11"}
|
169 |
-
{"current_steps": 1660, "total_steps": 3635, "loss": 1.2094, "lr": 6.60160677115969e-05, "epoch": 0.04567717792086291, "percentage": 45.67, "elapsed_time": "3:01:01", "remaining_time": "3:35:22"}
|
170 |
-
{"current_steps": 1670, "total_steps": 3635, "loss": 1.0698, "lr": 6.556042027116013e-05, "epoch": 0.04595234164327775, "percentage": 45.94, "elapsed_time": "3:01:19", "remaining_time": "3:33:21"}
|
171 |
-
{"current_steps": 1680, "total_steps": 3635, "loss": 1.0657, "lr": 6.510333748459771e-05, "epoch": 0.04622750536569259, "percentage": 46.22, "elapsed_time": "3:01:40", "remaining_time": "3:31:24"}
|
172 |
-
{"current_steps": 1690, "total_steps": 3635, "loss": 1.1707, "lr": 6.464486151478155e-05, "epoch": 0.04650266908810742, "percentage": 46.49, "elapsed_time": "3:02:17", "remaining_time": "3:29:47"}
|
173 |
-
{"current_steps": 1700, "total_steps": 3635, "loss": 1.225, "lr": 6.418503465309551e-05, "epoch": 0.04677783281052226, "percentage": 46.77, "elapsed_time": "3:02:52", "remaining_time": "3:28:09"}
|
174 |
-
{"current_steps": 1710, "total_steps": 3635, "loss": 1.2481, "lr": 6.372389931553434e-05, "epoch": 0.0470529965329371, "percentage": 47.04, "elapsed_time": "3:03:18", "remaining_time": "3:26:21"}
|
175 |
-
{"current_steps": 1720, "total_steps": 3635, "loss": 1.0919, "lr": 6.326149803879106e-05, "epoch": 0.047328160255351935, "percentage": 47.32, "elapsed_time": "3:03:40", "remaining_time": "3:24:30"}
|
176 |
-
{"current_steps": 1730, "total_steps": 3635, "loss": 1.2227, "lr": 6.279787347633322e-05, "epoch": 0.04760332397776677, "percentage": 47.59, "elapsed_time": "3:04:13", "remaining_time": "3:22:52"}
|
177 |
-
{"current_steps": 1740, "total_steps": 3635, "loss": 1.0782, "lr": 6.233306839446838e-05, "epoch": 0.04787848770018161, "percentage": 47.87, "elapsed_time": "3:04:41", "remaining_time": "3:21:08"}
|
178 |
-
{"current_steps": 1750, "total_steps": 3635, "loss": 1.1113, "lr": 6.186712566839927e-05, "epoch": 0.04815365142259644, "percentage": 48.14, "elapsed_time": "3:05:09", "remaining_time": "3:19:26"}
|
179 |
-
{"current_steps": 1760, "total_steps": 3635, "loss": 1.1447, "lr": 6.140008827826877e-05, "epoch": 0.04842881514501128, "percentage": 48.42, "elapsed_time": "3:05:39", "remaining_time": "3:17:47"}
|
180 |
-
{"current_steps": 1770, "total_steps": 3635, "loss": 1.0916, "lr": 6.093199930519531e-05, "epoch": 0.04870397886742612, "percentage": 48.69, "elapsed_time": "3:06:05", "remaining_time": "3:16:04"}
|
181 |
-
{"current_steps": 1780, "total_steps": 3635, "loss": 1.1642, "lr": 6.046290192729893e-05, "epoch": 0.04897914258984096, "percentage": 48.97, "elapsed_time": "3:06:39", "remaining_time": "3:14:31"}
|
182 |
-
{"current_steps": 1790, "total_steps": 3635, "loss": 1.1882, "lr": 5.999283941571831e-05, "epoch": 0.049254306312255795, "percentage": 49.24, "elapsed_time": "3:07:11", "remaining_time": "3:12:56"}
|
183 |
-
{"current_steps": 1800, "total_steps": 3635, "loss": 1.0697, "lr": 5.952185513061937e-05, "epoch": 0.04952947003467063, "percentage": 49.52, "elapsed_time": "3:07:37", "remaining_time": "3:11:16"}
|
184 |
-
{"current_steps": 1810, "total_steps": 3635, "loss": 1.2385, "lr": 5.90499925171956e-05, "epoch": 0.049804633757085465, "percentage": 49.79, "elapsed_time": "3:08:19", "remaining_time": "3:09:53"}
|
185 |
-
{"current_steps": 1820, "total_steps": 3635, "loss": 1.1276, "lr": 5.85772951016604e-05, "epoch": 0.0500797974795003, "percentage": 50.07, "elapsed_time": "3:08:47", "remaining_time": "3:08:16"}
|
186 |
-
{"current_steps": 1830, "total_steps": 3635, "loss": 1.1327, "lr": 5.810380648723222e-05, "epoch": 0.05035496120191514, "percentage": 50.34, "elapsed_time": "3:09:10", "remaining_time": "3:06:35"}
|
187 |
-
{"current_steps": 1840, "total_steps": 3635, "loss": 1.0801, "lr": 5.762957035011238e-05, "epoch": 0.05063012492432998, "percentage": 50.62, "elapsed_time": "3:09:42", "remaining_time": "3:05:03"}
|
188 |
-
{"current_steps": 1850, "total_steps": 3635, "loss": 1.2177, "lr": 5.715463043545622e-05, "epoch": 0.05090528864674481, "percentage": 50.89, "elapsed_time": "3:10:22", "remaining_time": "3:03:41"}
|
189 |
-
{"current_steps": 1860, "total_steps": 3635, "loss": 1.1432, "lr": 5.667903055333794e-05, "epoch": 0.05118045236915965, "percentage": 51.17, "elapsed_time": "3:10:50", "remaining_time": "3:02:07"}
|
190 |
-
{"current_steps": 1870, "total_steps": 3635, "loss": 1.0714, "lr": 5.620281457470935e-05, "epoch": 0.05145561609157449, "percentage": 51.44, "elapsed_time": "3:11:24", "remaining_time": "3:00:39"}
|
191 |
-
{"current_steps": 1880, "total_steps": 3635, "loss": 1.0537, "lr": 5.5726026427353126e-05, "epoch": 0.051730779813989325, "percentage": 51.72, "elapsed_time": "3:11:47", "remaining_time": "2:59:01"}
|
192 |
-
{"current_steps": 1890, "total_steps": 3635, "loss": 1.0393, "lr": 5.524871009183074e-05, "epoch": 0.05200594353640416, "percentage": 51.99, "elapsed_time": "3:12:04", "remaining_time": "2:57:20"}
|
193 |
-
{"current_steps": 1900, "total_steps": 3635, "loss": 1.1002, "lr": 5.47709095974255e-05, "epoch": 0.052281107258818994, "percentage": 52.27, "elapsed_time": "3:12:43", "remaining_time": "2:55:59"}
|
194 |
-
{"current_steps": 1910, "total_steps": 3635, "loss": 1.2777, "lr": 5.429266901808121e-05, "epoch": 0.05255627098123383, "percentage": 52.54, "elapsed_time": "3:13:23", "remaining_time": "2:54:39"}
|
195 |
-
{"current_steps": 1920, "total_steps": 3635, "loss": 1.1843, "lr": 5.3814032468336617e-05, "epoch": 0.05283143470364867, "percentage": 52.82, "elapsed_time": "3:13:50", "remaining_time": "2:53:08"}
|
196 |
-
{"current_steps": 1930, "total_steps": 3635, "loss": 1.1452, "lr": 5.333504409925609e-05, "epoch": 0.05310659842606351, "percentage": 53.09, "elapsed_time": "3:14:28", "remaining_time": "2:51:47"}
|
197 |
-
{"current_steps": 1940, "total_steps": 3635, "loss": 1.1351, "lr": 5.285574809435704e-05, "epoch": 0.05338176214847835, "percentage": 53.37, "elapsed_time": "3:14:51", "remaining_time": "2:50:14"}
|
198 |
-
{"current_steps": 1950, "total_steps": 3635, "loss": 1.091, "lr": 5.237618866553423e-05, "epoch": 0.05365692587089318, "percentage": 53.65, "elapsed_time": "3:15:16", "remaining_time": "2:48:44"}
|
199 |
-
{"current_steps": 1960, "total_steps": 3635, "loss": 1.0485, "lr": 5.189641004898157e-05, "epoch": 0.053932089593308016, "percentage": 53.92, "elapsed_time": "3:15:50", "remaining_time": "2:47:22"}
|
200 |
-
{"current_steps": 1970, "total_steps": 3635, "loss": 1.1641, "lr": 5.1416456501111554e-05, "epoch": 0.054207253315722854, "percentage": 54.2, "elapsed_time": "3:16:11", "remaining_time": "2:45:49"}
|
201 |
-
{"current_steps": 1980, "total_steps": 3635, "loss": 1.1219, "lr": 5.0936372294472975e-05, "epoch": 0.05448241703813769, "percentage": 54.47, "elapsed_time": "3:16:30", "remaining_time": "2:44:15"}
|
202 |
-
{"current_steps": 1990, "total_steps": 3635, "loss": 1.1912, "lr": 5.045620171366701e-05, "epoch": 0.05475758076055253, "percentage": 54.75, "elapsed_time": "3:17:08", "remaining_time": "2:42:57"}
|
203 |
-
{"current_steps": 2000, "total_steps": 3635, "loss": 1.1671, "lr": 4.997598905126229e-05, "epoch": 0.05503274448296737, "percentage": 55.02, "elapsed_time": "3:17:33", "remaining_time": "2:41:30"}
|
204 |
-
{"current_steps": 2000, "total_steps": 3635, "eval_loss": 1.2716114521026611, "epoch": 0.05503274448296737, "percentage": 55.02, "elapsed_time": "3:51:30", "remaining_time": "3:09:15"}
|
|
|
1 |
+
{"current_steps": 10, "total_steps": 3170, "loss": 1.3533, "lr": 3.1545741324921135e-06, "epoch": 0.0003155071777882947, "percentage": 0.32, "elapsed_time": "0:00:19", "remaining_time": "1:40:20"}
|
2 |
+
{"current_steps": 20, "total_steps": 3170, "loss": 1.3656, "lr": 6.309148264984227e-06, "epoch": 0.0006310143555765894, "percentage": 0.63, "elapsed_time": "0:00:54", "remaining_time": "2:22:39"}
|
3 |
+
{"current_steps": 30, "total_steps": 3170, "loss": 1.3685, "lr": 9.46372239747634e-06, "epoch": 0.000946521533364884, "percentage": 0.95, "elapsed_time": "0:01:19", "remaining_time": "2:19:11"}
|
4 |
+
{"current_steps": 40, "total_steps": 3170, "loss": 1.2711, "lr": 1.2618296529968454e-05, "epoch": 0.0012620287111531788, "percentage": 1.26, "elapsed_time": "0:01:49", "remaining_time": "2:22:34"}
|
5 |
+
{"current_steps": 50, "total_steps": 3170, "loss": 1.1823, "lr": 1.577287066246057e-05, "epoch": 0.0015775358889414735, "percentage": 1.58, "elapsed_time": "0:02:13", "remaining_time": "2:18:46"}
|
6 |
+
{"current_steps": 60, "total_steps": 3170, "loss": 1.3238, "lr": 1.892744479495268e-05, "epoch": 0.001893043066729768, "percentage": 1.89, "elapsed_time": "0:02:38", "remaining_time": "2:17:09"}
|
7 |
+
{"current_steps": 70, "total_steps": 3170, "loss": 1.2858, "lr": 2.2082018927444796e-05, "epoch": 0.0022085502445180626, "percentage": 2.21, "elapsed_time": "0:03:11", "remaining_time": "2:21:11"}
|
8 |
+
{"current_steps": 80, "total_steps": 3170, "loss": 1.2527, "lr": 2.5236593059936908e-05, "epoch": 0.0025240574223063575, "percentage": 2.52, "elapsed_time": "0:03:47", "remaining_time": "2:26:44"}
|
9 |
+
{"current_steps": 90, "total_steps": 3170, "loss": 1.2617, "lr": 2.8391167192429026e-05, "epoch": 0.002839564600094652, "percentage": 2.84, "elapsed_time": "0:04:11", "remaining_time": "2:23:14"}
|
10 |
+
{"current_steps": 100, "total_steps": 3170, "loss": 1.145, "lr": 3.154574132492114e-05, "epoch": 0.003155071777882947, "percentage": 3.15, "elapsed_time": "0:04:32", "remaining_time": "2:19:12"}
|
11 |
+
{"current_steps": 110, "total_steps": 3170, "loss": 1.1987, "lr": 3.470031545741325e-05, "epoch": 0.0034705789556712416, "percentage": 3.47, "elapsed_time": "0:04:55", "remaining_time": "2:16:52"}
|
12 |
+
{"current_steps": 120, "total_steps": 3170, "loss": 1.2743, "lr": 3.785488958990536e-05, "epoch": 0.003786086133459536, "percentage": 3.79, "elapsed_time": "0:05:38", "remaining_time": "2:23:18"}
|
13 |
+
{"current_steps": 130, "total_steps": 3170, "loss": 1.1172, "lr": 4.1009463722397477e-05, "epoch": 0.004101593311247831, "percentage": 4.1, "elapsed_time": "0:06:03", "remaining_time": "2:21:31"}
|
14 |
+
{"current_steps": 140, "total_steps": 3170, "loss": 1.2585, "lr": 4.416403785488959e-05, "epoch": 0.004417100489036125, "percentage": 4.42, "elapsed_time": "0:06:33", "remaining_time": "2:21:58"}
|
15 |
+
{"current_steps": 150, "total_steps": 3170, "loss": 1.2598, "lr": 4.731861198738171e-05, "epoch": 0.00473260766682442, "percentage": 4.73, "elapsed_time": "0:06:47", "remaining_time": "2:16:49"}
|
16 |
+
{"current_steps": 160, "total_steps": 3170, "loss": 1.2826, "lr": 5.0473186119873815e-05, "epoch": 0.005048114844612715, "percentage": 5.05, "elapsed_time": "0:07:16", "remaining_time": "2:16:57"}
|
17 |
+
{"current_steps": 170, "total_steps": 3170, "loss": 1.2544, "lr": 5.362776025236593e-05, "epoch": 0.00536362202240101, "percentage": 5.36, "elapsed_time": "0:07:45", "remaining_time": "2:16:58"}
|
18 |
+
{"current_steps": 180, "total_steps": 3170, "loss": 1.1758, "lr": 5.678233438485805e-05, "epoch": 0.005679129200189304, "percentage": 5.68, "elapsed_time": "0:08:25", "remaining_time": "2:19:54"}
|
19 |
+
{"current_steps": 190, "total_steps": 3170, "loss": 1.3851, "lr": 5.993690851735017e-05, "epoch": 0.005994636377977599, "percentage": 5.99, "elapsed_time": "0:09:03", "remaining_time": "2:22:08"}
|
20 |
+
{"current_steps": 200, "total_steps": 3170, "loss": 1.0848, "lr": 6.309148264984228e-05, "epoch": 0.006310143555765894, "percentage": 6.31, "elapsed_time": "0:09:34", "remaining_time": "2:22:16"}
|
21 |
+
{"current_steps": 210, "total_steps": 3170, "loss": 1.3279, "lr": 6.624605678233438e-05, "epoch": 0.006625650733554188, "percentage": 6.62, "elapsed_time": "0:10:09", "remaining_time": "2:23:05"}
|
22 |
+
{"current_steps": 220, "total_steps": 3170, "loss": 1.0995, "lr": 6.94006309148265e-05, "epoch": 0.006941157911342483, "percentage": 6.94, "elapsed_time": "0:10:26", "remaining_time": "2:20:03"}
|
23 |
+
{"current_steps": 230, "total_steps": 3170, "loss": 1.1639, "lr": 7.255520504731861e-05, "epoch": 0.007256665089130778, "percentage": 7.26, "elapsed_time": "0:10:54", "remaining_time": "2:19:28"}
|
24 |
+
{"current_steps": 240, "total_steps": 3170, "loss": 1.255, "lr": 7.570977917981072e-05, "epoch": 0.007572172266919072, "percentage": 7.57, "elapsed_time": "0:11:31", "remaining_time": "2:20:38"}
|
25 |
+
{"current_steps": 250, "total_steps": 3170, "loss": 1.1427, "lr": 7.886435331230284e-05, "epoch": 0.007887679444707366, "percentage": 7.89, "elapsed_time": "0:12:04", "remaining_time": "2:21:07"}
|
26 |
+
{"current_steps": 260, "total_steps": 3170, "loss": 1.3347, "lr": 8.201892744479495e-05, "epoch": 0.008203186622495662, "percentage": 8.2, "elapsed_time": "0:12:37", "remaining_time": "2:21:22"}
|
27 |
+
{"current_steps": 270, "total_steps": 3170, "loss": 1.1372, "lr": 8.517350157728708e-05, "epoch": 0.008518693800283956, "percentage": 8.52, "elapsed_time": "0:13:07", "remaining_time": "2:21:01"}
|
28 |
+
{"current_steps": 280, "total_steps": 3170, "loss": 1.2528, "lr": 8.832807570977918e-05, "epoch": 0.00883420097807225, "percentage": 8.83, "elapsed_time": "0:13:50", "remaining_time": "2:22:54"}
|
29 |
+
{"current_steps": 290, "total_steps": 3170, "loss": 1.2139, "lr": 9.148264984227129e-05, "epoch": 0.009149708155860546, "percentage": 9.15, "elapsed_time": "0:14:23", "remaining_time": "2:22:59"}
|
30 |
+
{"current_steps": 300, "total_steps": 3170, "loss": 1.3012, "lr": 9.463722397476341e-05, "epoch": 0.00946521533364884, "percentage": 9.46, "elapsed_time": "0:15:01", "remaining_time": "2:23:42"}
|
31 |
+
{"current_steps": 310, "total_steps": 3170, "loss": 1.1328, "lr": 9.779179810725552e-05, "epoch": 0.009780722511437134, "percentage": 9.78, "elapsed_time": "0:15:27", "remaining_time": "2:22:37"}
|
32 |
+
{"current_steps": 320, "total_steps": 3170, "loss": 1.1742, "lr": 9.999972717866783e-05, "epoch": 0.01009622968922543, "percentage": 10.09, "elapsed_time": "0:16:08", "remaining_time": "2:23:48"}
|
33 |
+
{"current_steps": 330, "total_steps": 3170, "loss": 1.1614, "lr": 9.99948771044751e-05, "epoch": 0.010411736867013724, "percentage": 10.41, "elapsed_time": "0:16:38", "remaining_time": "2:23:12"}
|
34 |
+
{"current_steps": 340, "total_steps": 3170, "loss": 1.1259, "lr": 9.998396501092419e-05, "epoch": 0.01072724404480202, "percentage": 10.73, "elapsed_time": "0:16:58", "remaining_time": "2:21:19"}
|
35 |
+
{"current_steps": 350, "total_steps": 3170, "loss": 1.1327, "lr": 9.996699222113709e-05, "epoch": 0.011042751222590314, "percentage": 11.04, "elapsed_time": "0:17:19", "remaining_time": "2:19:35"}
|
36 |
+
{"current_steps": 360, "total_steps": 3170, "loss": 1.2165, "lr": 9.994396079311228e-05, "epoch": 0.011358258400378608, "percentage": 11.36, "elapsed_time": "0:17:50", "remaining_time": "2:19:16"}
|
37 |
+
{"current_steps": 370, "total_steps": 3170, "loss": 1.1041, "lr": 9.991487351947513e-05, "epoch": 0.011673765578166904, "percentage": 11.67, "elapsed_time": "0:18:22", "remaining_time": "2:19:03"}
|
38 |
+
{"current_steps": 380, "total_steps": 3170, "loss": 1.3298, "lr": 9.987973392713932e-05, "epoch": 0.011989272755955198, "percentage": 11.99, "elapsed_time": "0:19:16", "remaining_time": "2:21:29"}
|
39 |
+
{"current_steps": 390, "total_steps": 3170, "loss": 1.236, "lr": 9.983854627687918e-05, "epoch": 0.012304779933743492, "percentage": 12.3, "elapsed_time": "0:19:46", "remaining_time": "2:20:56"}
|
40 |
+
{"current_steps": 400, "total_steps": 3170, "loss": 1.1544, "lr": 9.979131556281304e-05, "epoch": 0.012620287111531788, "percentage": 12.62, "elapsed_time": "0:20:27", "remaining_time": "2:21:39"}
|
41 |
+
{"current_steps": 410, "total_steps": 3170, "loss": 1.2401, "lr": 9.973804751179779e-05, "epoch": 0.012935794289320082, "percentage": 12.93, "elapsed_time": "0:21:04", "remaining_time": "2:21:53"}
|
42 |
+
{"current_steps": 420, "total_steps": 3170, "loss": 1.2296, "lr": 9.967874858273423e-05, "epoch": 0.013251301467108376, "percentage": 13.25, "elapsed_time": "0:21:57", "remaining_time": "2:23:43"}
|
43 |
+
{"current_steps": 430, "total_steps": 3170, "loss": 1.0889, "lr": 9.961342596578422e-05, "epoch": 0.013566808644896672, "percentage": 13.56, "elapsed_time": "0:22:15", "remaining_time": "2:21:51"}
|
44 |
+
{"current_steps": 440, "total_steps": 3170, "loss": 1.2241, "lr": 9.954208758149867e-05, "epoch": 0.013882315822684966, "percentage": 13.88, "elapsed_time": "0:22:44", "remaining_time": "2:21:05"}
|
45 |
+
{"current_steps": 450, "total_steps": 3170, "loss": 1.2967, "lr": 9.946474207985713e-05, "epoch": 0.01419782300047326, "percentage": 14.2, "elapsed_time": "0:23:05", "remaining_time": "2:19:33"}
|
46 |
+
{"current_steps": 460, "total_steps": 3170, "loss": 1.2981, "lr": 9.938139883921906e-05, "epoch": 0.014513330178261556, "percentage": 14.51, "elapsed_time": "0:23:35", "remaining_time": "2:18:58"}
|
47 |
+
{"current_steps": 470, "total_steps": 3170, "loss": 1.2427, "lr": 9.929206796518662e-05, "epoch": 0.01482883735604985, "percentage": 14.83, "elapsed_time": "0:24:12", "remaining_time": "2:19:02"}
|
48 |
+
{"current_steps": 480, "total_steps": 3170, "loss": 1.1139, "lr": 9.919676028937936e-05, "epoch": 0.015144344533838144, "percentage": 15.14, "elapsed_time": "0:24:37", "remaining_time": "2:17:58"}
|
49 |
+
{"current_steps": 490, "total_steps": 3170, "loss": 1.2958, "lr": 9.909548736812085e-05, "epoch": 0.01545985171162644, "percentage": 15.46, "elapsed_time": "0:25:18", "remaining_time": "2:18:22"}
|
50 |
+
{"current_steps": 500, "total_steps": 3170, "loss": 1.1714, "lr": 9.898826148103737e-05, "epoch": 0.015775358889414733, "percentage": 15.77, "elapsed_time": "0:25:46", "remaining_time": "2:17:37"}
|
51 |
+
{"current_steps": 510, "total_steps": 3170, "loss": 1.1652, "lr": 9.887509562956911e-05, "epoch": 0.01609086606720303, "percentage": 16.09, "elapsed_time": "0:26:14", "remaining_time": "2:16:51"}
|
52 |
+
{"current_steps": 520, "total_steps": 3170, "loss": 1.0727, "lr": 9.875600353539363e-05, "epoch": 0.016406373244991324, "percentage": 16.4, "elapsed_time": "0:26:36", "remaining_time": "2:15:35"}
|
53 |
+
{"current_steps": 530, "total_steps": 3170, "loss": 1.2942, "lr": 9.863099963876201e-05, "epoch": 0.01672188042277962, "percentage": 16.72, "elapsed_time": "0:27:18", "remaining_time": "2:16:00"}
|
54 |
+
{"current_steps": 540, "total_steps": 3170, "loss": 1.157, "lr": 9.850009909674807e-05, "epoch": 0.017037387600567912, "percentage": 17.03, "elapsed_time": "0:27:55", "remaining_time": "2:15:58"}
|
55 |
+
{"current_steps": 550, "total_steps": 3170, "loss": 1.2506, "lr": 9.83633177814104e-05, "epoch": 0.017352894778356207, "percentage": 17.35, "elapsed_time": "0:28:23", "remaining_time": "2:15:15"}
|
56 |
+
{"current_steps": 560, "total_steps": 3170, "loss": 1.1519, "lr": 9.822067227786793e-05, "epoch": 0.0176684019561445, "percentage": 17.67, "elapsed_time": "0:29:03", "remaining_time": "2:15:26"}
|
57 |
+
{"current_steps": 570, "total_steps": 3170, "loss": 1.0709, "lr": 9.807217988228884e-05, "epoch": 0.017983909133932798, "percentage": 17.98, "elapsed_time": "0:29:35", "remaining_time": "2:14:57"}
|
58 |
+
{"current_steps": 580, "total_steps": 3170, "loss": 1.1671, "lr": 9.791785859979344e-05, "epoch": 0.018299416311721092, "percentage": 18.3, "elapsed_time": "0:30:06", "remaining_time": "2:14:27"}
|
59 |
+
{"current_steps": 590, "total_steps": 3170, "loss": 1.1114, "lr": 9.775772714227096e-05, "epoch": 0.018614923489509386, "percentage": 18.61, "elapsed_time": "0:30:31", "remaining_time": "2:13:27"}
|
60 |
+
{"current_steps": 600, "total_steps": 3170, "loss": 1.2766, "lr": 9.759180492611065e-05, "epoch": 0.01893043066729768, "percentage": 18.93, "elapsed_time": "0:31:19", "remaining_time": "2:14:12"}
|
61 |
+
{"current_steps": 610, "total_steps": 3170, "loss": 1.1934, "lr": 9.742011206984756e-05, "epoch": 0.019245937845085975, "percentage": 19.24, "elapsed_time": "0:31:52", "remaining_time": "2:13:47"}
|
62 |
+
{"current_steps": 620, "total_steps": 3170, "loss": 1.0602, "lr": 9.724266939172302e-05, "epoch": 0.01956144502287427, "percentage": 19.56, "elapsed_time": "0:32:23", "remaining_time": "2:13:11"}
|
63 |
+
{"current_steps": 630, "total_steps": 3170, "loss": 1.3037, "lr": 9.705949840716047e-05, "epoch": 0.019876952200662566, "percentage": 19.87, "elapsed_time": "0:33:03", "remaining_time": "2:13:17"}
|
64 |
+
{"current_steps": 640, "total_steps": 3170, "loss": 1.1023, "lr": 9.687062132615658e-05, "epoch": 0.02019245937845086, "percentage": 20.19, "elapsed_time": "0:33:41", "remaining_time": "2:13:11"}
|
65 |
+
{"current_steps": 650, "total_steps": 3170, "loss": 1.1474, "lr": 9.667606105058828e-05, "epoch": 0.020507966556239154, "percentage": 20.5, "elapsed_time": "0:34:10", "remaining_time": "2:12:30"}
|
66 |
+
{"current_steps": 660, "total_steps": 3170, "loss": 1.2257, "lr": 9.647584117143576e-05, "epoch": 0.02082347373402745, "percentage": 20.82, "elapsed_time": "0:34:44", "remaining_time": "2:12:08"}
|
67 |
+
{"current_steps": 670, "total_steps": 3170, "loss": 1.0884, "lr": 9.626998596592209e-05, "epoch": 0.021138980911815743, "percentage": 21.14, "elapsed_time": "0:35:22", "remaining_time": "2:12:00"}
|
68 |
+
{"current_steps": 680, "total_steps": 3170, "loss": 1.1375, "lr": 9.605852039456947e-05, "epoch": 0.02145448808960404, "percentage": 21.45, "elapsed_time": "0:35:45", "remaining_time": "2:10:56"}
|
69 |
+
{"current_steps": 690, "total_steps": 3170, "loss": 1.1861, "lr": 9.584147009817277e-05, "epoch": 0.021769995267392334, "percentage": 21.77, "elapsed_time": "0:36:26", "remaining_time": "2:10:59"}
|
70 |
+
{"current_steps": 700, "total_steps": 3170, "loss": 1.2251, "lr": 9.561886139469045e-05, "epoch": 0.02208550244518063, "percentage": 22.08, "elapsed_time": "0:36:57", "remaining_time": "2:10:24"}
|
71 |
+
{"current_steps": 710, "total_steps": 3170, "loss": 1.2597, "lr": 9.539072127605342e-05, "epoch": 0.022401009622968922, "percentage": 22.4, "elapsed_time": "0:37:36", "remaining_time": "2:10:17"}
|
72 |
+
{"current_steps": 720, "total_steps": 3170, "loss": 1.2133, "lr": 9.515707740489224e-05, "epoch": 0.022716516800757217, "percentage": 22.71, "elapsed_time": "0:38:01", "remaining_time": "2:09:23"}
|
73 |
+
{"current_steps": 730, "total_steps": 3170, "loss": 1.175, "lr": 9.491795811118296e-05, "epoch": 0.02303202397854551, "percentage": 23.03, "elapsed_time": "0:38:26", "remaining_time": "2:08:28"}
|
74 |
+
{"current_steps": 740, "total_steps": 3170, "loss": 1.0637, "lr": 9.467339238881199e-05, "epoch": 0.023347531156333808, "percentage": 23.34, "elapsed_time": "0:38:42", "remaining_time": "2:07:08"}
|
75 |
+
{"current_steps": 750, "total_steps": 3170, "loss": 1.0974, "lr": 9.442340989206047e-05, "epoch": 0.023663038334122102, "percentage": 23.66, "elapsed_time": "0:39:12", "remaining_time": "2:06:30"}
|
76 |
+
{"current_steps": 760, "total_steps": 3170, "loss": 1.2166, "lr": 9.416804093200874e-05, "epoch": 0.023978545511910396, "percentage": 23.97, "elapsed_time": "0:39:47", "remaining_time": "2:06:10"}
|
77 |
+
{"current_steps": 770, "total_steps": 3170, "loss": 1.25, "lr": 9.390731647286089e-05, "epoch": 0.02429405268969869, "percentage": 24.29, "elapsed_time": "0:40:31", "remaining_time": "2:06:19"}
|
78 |
+
{"current_steps": 780, "total_steps": 3170, "loss": 1.0963, "lr": 9.364126812819038e-05, "epoch": 0.024609559867486985, "percentage": 24.61, "elapsed_time": "0:41:11", "remaining_time": "2:06:11"}
|
79 |
+
{"current_steps": 790, "total_steps": 3170, "loss": 1.1501, "lr": 9.336992815710668e-05, "epoch": 0.02492506704527528, "percentage": 24.92, "elapsed_time": "0:41:51", "remaining_time": "2:06:05"}
|
80 |
+
{"current_steps": 800, "total_steps": 3170, "loss": 1.1365, "lr": 9.309332946034392e-05, "epoch": 0.025240574223063576, "percentage": 25.24, "elapsed_time": "0:42:26", "remaining_time": "2:05:43"}
|
81 |
+
{"current_steps": 810, "total_steps": 3170, "loss": 1.1154, "lr": 9.281150557627143e-05, "epoch": 0.02555608140085187, "percentage": 25.55, "elapsed_time": "0:42:54", "remaining_time": "2:04:59"}
|
82 |
+
{"current_steps": 820, "total_steps": 3170, "loss": 1.1984, "lr": 9.252449067682721e-05, "epoch": 0.025871588578640164, "percentage": 25.87, "elapsed_time": "0:43:25", "remaining_time": "2:04:25"}
|
83 |
+
{"current_steps": 830, "total_steps": 3170, "loss": 1.1346, "lr": 9.22323195633745e-05, "epoch": 0.02618709575642846, "percentage": 26.18, "elapsed_time": "0:43:48", "remaining_time": "2:03:31"}
|
84 |
+
{"current_steps": 840, "total_steps": 3170, "loss": 1.1597, "lr": 9.19350276624819e-05, "epoch": 0.026502602934216753, "percentage": 26.5, "elapsed_time": "0:44:11", "remaining_time": "2:02:34"}
|
85 |
+
{"current_steps": 850, "total_steps": 3170, "loss": 1.1, "lr": 9.163265102162794e-05, "epoch": 0.026818110112005047, "percentage": 26.81, "elapsed_time": "0:44:39", "remaining_time": "2:01:52"}
|
86 |
+
{"current_steps": 860, "total_steps": 3170, "loss": 1.0557, "lr": 9.132522630483017e-05, "epoch": 0.027133617289793344, "percentage": 27.13, "elapsed_time": "0:45:02", "remaining_time": "2:00:59"}
|
87 |
+
{"current_steps": 870, "total_steps": 3170, "loss": 1.146, "lr": 9.101279078819949e-05, "epoch": 0.02744912446758164, "percentage": 27.44, "elapsed_time": "0:45:23", "remaining_time": "1:59:59"}
|
88 |
+
{"current_steps": 880, "total_steps": 3170, "loss": 1.1513, "lr": 9.069538235542037e-05, "epoch": 0.027764631645369933, "percentage": 27.76, "elapsed_time": "0:45:50", "remaining_time": "1:59:18"}
|
89 |
+
{"current_steps": 890, "total_steps": 3170, "loss": 1.1684, "lr": 9.037303949315736e-05, "epoch": 0.028080138823158227, "percentage": 28.08, "elapsed_time": "0:46:26", "remaining_time": "1:58:59"}
|
90 |
+
{"current_steps": 900, "total_steps": 3170, "loss": 1.1054, "lr": 9.004580128638846e-05, "epoch": 0.02839564600094652, "percentage": 28.39, "elapsed_time": "0:46:51", "remaining_time": "1:58:11"}
|
91 |
+
{"current_steps": 910, "total_steps": 3170, "loss": 1.2186, "lr": 8.971370741366592e-05, "epoch": 0.028711153178734815, "percentage": 28.71, "elapsed_time": "0:47:16", "remaining_time": "1:57:23"}
|
92 |
+
{"current_steps": 920, "total_steps": 3170, "loss": 1.2054, "lr": 8.937679814230517e-05, "epoch": 0.029026660356523112, "percentage": 29.02, "elapsed_time": "0:47:46", "remaining_time": "1:56:50"}
|
93 |
+
{"current_steps": 930, "total_steps": 3170, "loss": 1.1217, "lr": 8.903511432350221e-05, "epoch": 0.029342167534311406, "percentage": 29.34, "elapsed_time": "0:48:07", "remaining_time": "1:55:54"}
|
94 |
+
{"current_steps": 940, "total_steps": 3170, "loss": 1.0987, "lr": 8.868869738738038e-05, "epoch": 0.0296576747120997, "percentage": 29.65, "elapsed_time": "0:48:39", "remaining_time": "1:55:25"}
|
95 |
+
{"current_steps": 950, "total_steps": 3170, "loss": 1.1457, "lr": 8.833758933796678e-05, "epoch": 0.029973181889887995, "percentage": 29.97, "elapsed_time": "0:49:06", "remaining_time": "1:54:46"}
|
96 |
+
{"current_steps": 960, "total_steps": 3170, "loss": 1.2404, "lr": 8.798183274809917e-05, "epoch": 0.03028868906767629, "percentage": 30.28, "elapsed_time": "0:49:44", "remaining_time": "1:54:29"}
|
97 |
+
{"current_steps": 970, "total_steps": 3170, "loss": 1.0919, "lr": 8.762147075426392e-05, "epoch": 0.030604196245464583, "percentage": 30.6, "elapsed_time": "0:50:15", "remaining_time": "1:54:00"}
|
98 |
+
{"current_steps": 980, "total_steps": 3170, "loss": 1.1757, "lr": 8.725654705136558e-05, "epoch": 0.03091970342325288, "percentage": 30.91, "elapsed_time": "0:50:49", "remaining_time": "1:53:35"}
|
99 |
+
{"current_steps": 990, "total_steps": 3170, "loss": 1.25, "lr": 8.688710588742872e-05, "epoch": 0.031235210601041175, "percentage": 31.23, "elapsed_time": "0:51:34", "remaining_time": "1:53:34"}
|
100 |
+
{"current_steps": 1000, "total_steps": 3170, "loss": 1.1356, "lr": 8.651319205823278e-05, "epoch": 0.031550717778829465, "percentage": 31.55, "elapsed_time": "0:52:18", "remaining_time": "1:53:30"}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5496
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7dca573d339daaf5799b1658860820b296cc8b2a228efea9a70adcab36040347
|
3 |
size 5496
|