Hugging Face's logo

mlfoundations-dev
/
oh-dcft-v1.3_no-curation_gpt-4o-mini_scale_4x

Text Generation
Transformers
Safetensors
llama
llama-factory
full
Generated from Trainer
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
sedrickkeh commited on
Commit
3614195
·
verified ·
1 Parent(s): 2174bf8

Training in progress, epoch 2

Browse files
Files changed (5) hide show
  1. model-00001-of-00004.safetensors +1 -1
  2. model-00002-of-00004.safetensors +1 -1
  3. model-00003-of-00004.safetensors +1 -1
  4. model-00004-of-00004.safetensors +1 -1
  5. trainer_log.jsonl +100 -0
model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0e5fb8ac74c8c9dc07451cdea9f6fd7a447360394a347edf230415ce73b883e5
3
  size 4976698672
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f0bfb56d265f70023f5c4dd25f6a3cf9d595d50a48c57fbe39313ade9c2ba19
3
  size 4976698672
model-00002-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6bbf17e7dad01c062572b09784a5671679e3cde686ee4c51a3fc92c9f96df3ee
3
  size 4999802720
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f6f0f0f14d532d6831ffd8493c47baf5c16b89d831baf9170f5ba92dd3b1f5f
3
  size 4999802720
model-00003-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e688dfbbeca3daa2958f04a65380c163826db6d1240f061e0bcc7302d3129536
3
  size 4915916176
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3455525d14fc4d4f812614bf8b554c708211fb547316a4e2fef4257387ace84a
3
  size 4915916176
model-00004-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:99f649ac1be9427c9518f0a58b0dd9c79fab2ef294dafb758d1f3018bcf8ff7b
3
  size 1168138808
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:521fa31ff60458549903da235e9e363eed097fdf7261dd4d2c97cda02f232b3a
3
  size 1168138808
trainer_log.jsonl CHANGED
@@ -99,3 +99,103 @@
99
  {"current_steps": 990, "total_steps": 2988, "loss": 0.7331, "lr": 5e-06, "epoch": 0.9934771700953337, "percentage": 33.13, "elapsed_time": "8:01:38", "remaining_time": "16:12:02"}
100
  {"current_steps": 996, "total_steps": 2988, "eval_loss": 0.7304001450538635, "epoch": 0.9994982438534872, "percentage": 33.33, "elapsed_time": "8:13:40", "remaining_time": "16:27:21"}
101
  {"current_steps": 1000, "total_steps": 2988, "loss": 0.754, "lr": 5e-06, "epoch": 1.0035122930255895, "percentage": 33.47, "elapsed_time": "8:16:20", "remaining_time": "16:26:42"}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
99
  {"current_steps": 990, "total_steps": 2988, "loss": 0.7331, "lr": 5e-06, "epoch": 0.9934771700953337, "percentage": 33.13, "elapsed_time": "8:01:38", "remaining_time": "16:12:02"}
100
  {"current_steps": 996, "total_steps": 2988, "eval_loss": 0.7304001450538635, "epoch": 0.9994982438534872, "percentage": 33.33, "elapsed_time": "8:13:40", "remaining_time": "16:27:21"}
101
  {"current_steps": 1000, "total_steps": 2988, "loss": 0.754, "lr": 5e-06, "epoch": 1.0035122930255895, "percentage": 33.47, "elapsed_time": "8:16:20", "remaining_time": "16:26:42"}
102
+ {"current_steps": 1010, "total_steps": 2988, "loss": 0.6938, "lr": 5e-06, "epoch": 1.0135474159558455, "percentage": 33.8, "elapsed_time": "8:21:12", "remaining_time": "16:21:35"}
103
+ {"current_steps": 1020, "total_steps": 2988, "loss": 0.6963, "lr": 5e-06, "epoch": 1.0235825388861013, "percentage": 34.14, "elapsed_time": "8:26:05", "remaining_time": "16:16:26"}
104
+ {"current_steps": 1030, "total_steps": 2988, "loss": 0.6892, "lr": 5e-06, "epoch": 1.0336176618163573, "percentage": 34.47, "elapsed_time": "8:30:57", "remaining_time": "16:11:18"}
105
+ {"current_steps": 1040, "total_steps": 2988, "loss": 0.6922, "lr": 5e-06, "epoch": 1.043652784746613, "percentage": 34.81, "elapsed_time": "8:35:49", "remaining_time": "16:06:11"}
106
+ {"current_steps": 1050, "total_steps": 2988, "loss": 0.6893, "lr": 5e-06, "epoch": 1.053687907676869, "percentage": 35.14, "elapsed_time": "8:40:41", "remaining_time": "16:01:03"}
107
+ {"current_steps": 1060, "total_steps": 2988, "loss": 0.6903, "lr": 5e-06, "epoch": 1.063723030607125, "percentage": 35.48, "elapsed_time": "8:45:34", "remaining_time": "15:55:56"}
108
+ {"current_steps": 1070, "total_steps": 2988, "loss": 0.6975, "lr": 5e-06, "epoch": 1.0737581535373808, "percentage": 35.81, "elapsed_time": "8:50:27", "remaining_time": "15:50:50"}
109
+ {"current_steps": 1080, "total_steps": 2988, "loss": 0.6953, "lr": 5e-06, "epoch": 1.0837932764676368, "percentage": 36.14, "elapsed_time": "8:55:19", "remaining_time": "15:45:44"}
110
+ {"current_steps": 1090, "total_steps": 2988, "loss": 0.6924, "lr": 5e-06, "epoch": 1.0938283993978926, "percentage": 36.48, "elapsed_time": "9:00:12", "remaining_time": "15:40:39"}
111
+ {"current_steps": 1100, "total_steps": 2988, "loss": 0.6933, "lr": 5e-06, "epoch": 1.1038635223281485, "percentage": 36.81, "elapsed_time": "9:05:04", "remaining_time": "15:35:33"}
112
+ {"current_steps": 1110, "total_steps": 2988, "loss": 0.6909, "lr": 5e-06, "epoch": 1.1138986452584043, "percentage": 37.15, "elapsed_time": "9:09:57", "remaining_time": "15:30:27"}
113
+ {"current_steps": 1120, "total_steps": 2988, "loss": 0.6962, "lr": 5e-06, "epoch": 1.1239337681886603, "percentage": 37.48, "elapsed_time": "9:14:49", "remaining_time": "15:25:22"}
114
+ {"current_steps": 1130, "total_steps": 2988, "loss": 0.6909, "lr": 5e-06, "epoch": 1.1339688911189163, "percentage": 37.82, "elapsed_time": "9:19:42", "remaining_time": "15:20:18"}
115
+ {"current_steps": 1140, "total_steps": 2988, "loss": 0.6909, "lr": 5e-06, "epoch": 1.144004014049172, "percentage": 38.15, "elapsed_time": "9:24:35", "remaining_time": "15:15:14"}
116
+ {"current_steps": 1150, "total_steps": 2988, "loss": 0.6959, "lr": 5e-06, "epoch": 1.154039136979428, "percentage": 38.49, "elapsed_time": "9:29:28", "remaining_time": "15:10:10"}
117
+ {"current_steps": 1160, "total_steps": 2988, "loss": 0.6898, "lr": 5e-06, "epoch": 1.1640742599096838, "percentage": 38.82, "elapsed_time": "9:34:21", "remaining_time": "15:05:06"}
118
+ {"current_steps": 1170, "total_steps": 2988, "loss": 0.6944, "lr": 5e-06, "epoch": 1.1741093828399398, "percentage": 39.16, "elapsed_time": "9:39:13", "remaining_time": "15:00:01"}
119
+ {"current_steps": 1180, "total_steps": 2988, "loss": 0.697, "lr": 5e-06, "epoch": 1.1841445057701956, "percentage": 39.49, "elapsed_time": "9:44:05", "remaining_time": "14:54:57"}
120
+ {"current_steps": 1190, "total_steps": 2988, "loss": 0.6983, "lr": 5e-06, "epoch": 1.1941796287004516, "percentage": 39.83, "elapsed_time": "9:48:58", "remaining_time": "14:49:53"}
121
+ {"current_steps": 1200, "total_steps": 2988, "loss": 0.6917, "lr": 5e-06, "epoch": 1.2042147516307076, "percentage": 40.16, "elapsed_time": "9:53:51", "remaining_time": "14:44:50"}
122
+ {"current_steps": 1210, "total_steps": 2988, "loss": 0.6931, "lr": 5e-06, "epoch": 1.2142498745609633, "percentage": 40.5, "elapsed_time": "9:58:43", "remaining_time": "14:39:47"}
123
+ {"current_steps": 1220, "total_steps": 2988, "loss": 0.6959, "lr": 5e-06, "epoch": 1.2242849974912193, "percentage": 40.83, "elapsed_time": "10:03:36", "remaining_time": "14:34:44"}
124
+ {"current_steps": 1230, "total_steps": 2988, "loss": 0.6909, "lr": 5e-06, "epoch": 1.234320120421475, "percentage": 41.16, "elapsed_time": "10:08:29", "remaining_time": "14:29:42"}
125
+ {"current_steps": 1240, "total_steps": 2988, "loss": 0.6928, "lr": 5e-06, "epoch": 1.244355243351731, "percentage": 41.5, "elapsed_time": "10:13:22", "remaining_time": "14:24:40"}
126
+ {"current_steps": 1250, "total_steps": 2988, "loss": 0.6947, "lr": 5e-06, "epoch": 1.2543903662819869, "percentage": 41.83, "elapsed_time": "10:18:15", "remaining_time": "14:19:37"}
127
+ {"current_steps": 1260, "total_steps": 2988, "loss": 0.6948, "lr": 5e-06, "epoch": 1.2644254892122428, "percentage": 42.17, "elapsed_time": "10:23:08", "remaining_time": "14:14:35"}
128
+ {"current_steps": 1270, "total_steps": 2988, "loss": 0.6948, "lr": 5e-06, "epoch": 1.2744606121424988, "percentage": 42.5, "elapsed_time": "10:28:01", "remaining_time": "14:09:33"}
129
+ {"current_steps": 1280, "total_steps": 2988, "loss": 0.691, "lr": 5e-06, "epoch": 1.2844957350727546, "percentage": 42.84, "elapsed_time": "10:32:53", "remaining_time": "14:04:31"}
130
+ {"current_steps": 1290, "total_steps": 2988, "loss": 0.6867, "lr": 5e-06, "epoch": 1.2945308580030106, "percentage": 43.17, "elapsed_time": "10:37:46", "remaining_time": "13:59:29"}
131
+ {"current_steps": 1300, "total_steps": 2988, "loss": 0.6959, "lr": 5e-06, "epoch": 1.3045659809332664, "percentage": 43.51, "elapsed_time": "10:42:39", "remaining_time": "13:54:27"}
132
+ {"current_steps": 1310, "total_steps": 2988, "loss": 0.6965, "lr": 5e-06, "epoch": 1.3146011038635224, "percentage": 43.84, "elapsed_time": "10:47:32", "remaining_time": "13:49:26"}
133
+ {"current_steps": 1320, "total_steps": 2988, "loss": 0.6916, "lr": 5e-06, "epoch": 1.3246362267937783, "percentage": 44.18, "elapsed_time": "10:52:25", "remaining_time": "13:44:25"}
134
+ {"current_steps": 1330, "total_steps": 2988, "loss": 0.6879, "lr": 5e-06, "epoch": 1.3346713497240341, "percentage": 44.51, "elapsed_time": "10:57:18", "remaining_time": "13:39:24"}
135
+ {"current_steps": 1340, "total_steps": 2988, "loss": 0.6941, "lr": 5e-06, "epoch": 1.3447064726542899, "percentage": 44.85, "elapsed_time": "11:02:10", "remaining_time": "13:34:22"}
136
+ {"current_steps": 1350, "total_steps": 2988, "loss": 0.6881, "lr": 5e-06, "epoch": 1.3547415955845459, "percentage": 45.18, "elapsed_time": "11:07:02", "remaining_time": "13:29:20"}
137
+ {"current_steps": 1360, "total_steps": 2988, "loss": 0.6963, "lr": 5e-06, "epoch": 1.3647767185148019, "percentage": 45.52, "elapsed_time": "11:11:55", "remaining_time": "13:24:19"}
138
+ {"current_steps": 1370, "total_steps": 2988, "loss": 0.6848, "lr": 5e-06, "epoch": 1.3748118414450576, "percentage": 45.85, "elapsed_time": "11:16:47", "remaining_time": "13:19:18"}
139
+ {"current_steps": 1380, "total_steps": 2988, "loss": 0.6933, "lr": 5e-06, "epoch": 1.3848469643753136, "percentage": 46.18, "elapsed_time": "11:21:39", "remaining_time": "13:14:16"}
140
+ {"current_steps": 1390, "total_steps": 2988, "loss": 0.6956, "lr": 5e-06, "epoch": 1.3948820873055694, "percentage": 46.52, "elapsed_time": "11:26:31", "remaining_time": "13:09:15"}
141
+ {"current_steps": 1400, "total_steps": 2988, "loss": 0.6913, "lr": 5e-06, "epoch": 1.4049172102358254, "percentage": 46.85, "elapsed_time": "11:31:23", "remaining_time": "13:04:14"}
142
+ {"current_steps": 1410, "total_steps": 2988, "loss": 0.6905, "lr": 5e-06, "epoch": 1.4149523331660814, "percentage": 47.19, "elapsed_time": "11:36:16", "remaining_time": "12:59:13"}
143
+ {"current_steps": 1420, "total_steps": 2988, "loss": 0.6913, "lr": 5e-06, "epoch": 1.4249874560963371, "percentage": 47.52, "elapsed_time": "11:41:08", "remaining_time": "12:54:13"}
144
+ {"current_steps": 1430, "total_steps": 2988, "loss": 0.6944, "lr": 5e-06, "epoch": 1.4350225790265931, "percentage": 47.86, "elapsed_time": "11:46:01", "remaining_time": "12:49:13"}
145
+ {"current_steps": 1440, "total_steps": 2988, "loss": 0.6906, "lr": 5e-06, "epoch": 1.445057701956849, "percentage": 48.19, "elapsed_time": "11:50:53", "remaining_time": "12:44:12"}
146
+ {"current_steps": 1450, "total_steps": 2988, "loss": 0.6915, "lr": 5e-06, "epoch": 1.455092824887105, "percentage": 48.53, "elapsed_time": "11:55:46", "remaining_time": "12:39:12"}
147
+ {"current_steps": 1460, "total_steps": 2988, "loss": 0.6911, "lr": 5e-06, "epoch": 1.4651279478173609, "percentage": 48.86, "elapsed_time": "12:00:38", "remaining_time": "12:34:12"}
148
+ {"current_steps": 1470, "total_steps": 2988, "loss": 0.692, "lr": 5e-06, "epoch": 1.4751630707476167, "percentage": 49.2, "elapsed_time": "12:05:31", "remaining_time": "12:29:12"}
149
+ {"current_steps": 1480, "total_steps": 2988, "loss": 0.6884, "lr": 5e-06, "epoch": 1.4851981936778724, "percentage": 49.53, "elapsed_time": "12:10:23", "remaining_time": "12:24:12"}
150
+ {"current_steps": 1490, "total_steps": 2988, "loss": 0.6886, "lr": 5e-06, "epoch": 1.4952333166081284, "percentage": 49.87, "elapsed_time": "12:15:15", "remaining_time": "12:19:12"}
151
+ {"current_steps": 1500, "total_steps": 2988, "loss": 0.695, "lr": 5e-06, "epoch": 1.5052684395383844, "percentage": 50.2, "elapsed_time": "12:20:08", "remaining_time": "12:14:13"}
152
+ {"current_steps": 1510, "total_steps": 2988, "loss": 0.6962, "lr": 5e-06, "epoch": 1.5153035624686404, "percentage": 50.54, "elapsed_time": "12:25:01", "remaining_time": "12:09:13"}
153
+ {"current_steps": 1520, "total_steps": 2988, "loss": 0.6953, "lr": 5e-06, "epoch": 1.5253386853988962, "percentage": 50.87, "elapsed_time": "12:29:53", "remaining_time": "12:04:14"}
154
+ {"current_steps": 1530, "total_steps": 2988, "loss": 0.6912, "lr": 5e-06, "epoch": 1.535373808329152, "percentage": 51.2, "elapsed_time": "12:34:46", "remaining_time": "11:59:15"}
155
+ {"current_steps": 1540, "total_steps": 2988, "loss": 0.6881, "lr": 5e-06, "epoch": 1.545408931259408, "percentage": 51.54, "elapsed_time": "12:39:38", "remaining_time": "11:54:15"}
156
+ {"current_steps": 1550, "total_steps": 2988, "loss": 0.6874, "lr": 5e-06, "epoch": 1.555444054189664, "percentage": 51.87, "elapsed_time": "12:44:30", "remaining_time": "11:49:16"}
157
+ {"current_steps": 1560, "total_steps": 2988, "loss": 0.6923, "lr": 5e-06, "epoch": 1.5654791771199197, "percentage": 52.21, "elapsed_time": "12:49:23", "remaining_time": "11:44:17"}
158
+ {"current_steps": 1570, "total_steps": 2988, "loss": 0.6896, "lr": 5e-06, "epoch": 1.5755143000501755, "percentage": 52.54, "elapsed_time": "12:54:15", "remaining_time": "11:39:18"}
159
+ {"current_steps": 1580, "total_steps": 2988, "loss": 0.6903, "lr": 5e-06, "epoch": 1.5855494229804314, "percentage": 52.88, "elapsed_time": "12:59:08", "remaining_time": "11:34:19"}
160
+ {"current_steps": 1590, "total_steps": 2988, "loss": 0.6917, "lr": 5e-06, "epoch": 1.5955845459106874, "percentage": 53.21, "elapsed_time": "13:04:00", "remaining_time": "11:29:20"}
161
+ {"current_steps": 1600, "total_steps": 2988, "loss": 0.6913, "lr": 5e-06, "epoch": 1.6056196688409434, "percentage": 53.55, "elapsed_time": "13:08:53", "remaining_time": "11:24:21"}
162
+ {"current_steps": 1610, "total_steps": 2988, "loss": 0.6855, "lr": 5e-06, "epoch": 1.6156547917711992, "percentage": 53.88, "elapsed_time": "13:13:46", "remaining_time": "11:19:23"}
163
+ {"current_steps": 1620, "total_steps": 2988, "loss": 0.689, "lr": 5e-06, "epoch": 1.625689914701455, "percentage": 54.22, "elapsed_time": "13:18:38", "remaining_time": "11:14:24"}
164
+ {"current_steps": 1630, "total_steps": 2988, "loss": 0.6917, "lr": 5e-06, "epoch": 1.635725037631711, "percentage": 54.55, "elapsed_time": "13:23:31", "remaining_time": "11:09:26"}
165
+ {"current_steps": 1640, "total_steps": 2988, "loss": 0.6844, "lr": 5e-06, "epoch": 1.645760160561967, "percentage": 54.89, "elapsed_time": "13:28:23", "remaining_time": "11:04:27"}
166
+ {"current_steps": 1650, "total_steps": 2988, "loss": 0.6894, "lr": 5e-06, "epoch": 1.655795283492223, "percentage": 55.22, "elapsed_time": "13:33:16", "remaining_time": "10:59:29"}
167
+ {"current_steps": 1660, "total_steps": 2988, "loss": 0.6985, "lr": 5e-06, "epoch": 1.6658304064224787, "percentage": 55.56, "elapsed_time": "13:38:08", "remaining_time": "10:54:30"}
168
+ {"current_steps": 1670, "total_steps": 2988, "loss": 0.6899, "lr": 5e-06, "epoch": 1.6758655293527345, "percentage": 55.89, "elapsed_time": "13:43:00", "remaining_time": "10:49:32"}
169
+ {"current_steps": 1680, "total_steps": 2988, "loss": 0.6901, "lr": 5e-06, "epoch": 1.6859006522829905, "percentage": 56.22, "elapsed_time": "13:47:53", "remaining_time": "10:44:34"}
170
+ {"current_steps": 1690, "total_steps": 2988, "loss": 0.6926, "lr": 5e-06, "epoch": 1.6959357752132465, "percentage": 56.56, "elapsed_time": "13:52:45", "remaining_time": "10:39:35"}
171
+ {"current_steps": 1700, "total_steps": 2988, "loss": 0.6889, "lr": 5e-06, "epoch": 1.7059708981435022, "percentage": 56.89, "elapsed_time": "13:57:38", "remaining_time": "10:34:37"}
172
+ {"current_steps": 1710, "total_steps": 2988, "loss": 0.6938, "lr": 5e-06, "epoch": 1.716006021073758, "percentage": 57.23, "elapsed_time": "14:02:30", "remaining_time": "10:29:39"}
173
+ {"current_steps": 1720, "total_steps": 2988, "loss": 0.6892, "lr": 5e-06, "epoch": 1.726041144004014, "percentage": 57.56, "elapsed_time": "14:07:23", "remaining_time": "10:24:41"}
174
+ {"current_steps": 1730, "total_steps": 2988, "loss": 0.6876, "lr": 5e-06, "epoch": 1.73607626693427, "percentage": 57.9, "elapsed_time": "14:12:15", "remaining_time": "10:19:44"}
175
+ {"current_steps": 1740, "total_steps": 2988, "loss": 0.6914, "lr": 5e-06, "epoch": 1.746111389864526, "percentage": 58.23, "elapsed_time": "14:17:08", "remaining_time": "10:14:46"}
176
+ {"current_steps": 1750, "total_steps": 2988, "loss": 0.6936, "lr": 5e-06, "epoch": 1.7561465127947817, "percentage": 58.57, "elapsed_time": "14:22:00", "remaining_time": "10:09:48"}
177
+ {"current_steps": 1760, "total_steps": 2988, "loss": 0.6884, "lr": 5e-06, "epoch": 1.7661816357250375, "percentage": 58.9, "elapsed_time": "14:26:53", "remaining_time": "10:04:50"}
178
+ {"current_steps": 1770, "total_steps": 2988, "loss": 0.6861, "lr": 5e-06, "epoch": 1.7762167586552935, "percentage": 59.24, "elapsed_time": "14:31:45", "remaining_time": "9:59:53"}
179
+ {"current_steps": 1780, "total_steps": 2988, "loss": 0.6884, "lr": 5e-06, "epoch": 1.7862518815855495, "percentage": 59.57, "elapsed_time": "14:36:37", "remaining_time": "9:54:55"}
180
+ {"current_steps": 1790, "total_steps": 2988, "loss": 0.696, "lr": 5e-06, "epoch": 1.7962870045158055, "percentage": 59.91, "elapsed_time": "14:41:30", "remaining_time": "9:49:58"}
181
+ {"current_steps": 1800, "total_steps": 2988, "loss": 0.6936, "lr": 5e-06, "epoch": 1.8063221274460612, "percentage": 60.24, "elapsed_time": "14:46:22", "remaining_time": "9:45:00"}
182
+ {"current_steps": 1810, "total_steps": 2988, "loss": 0.6894, "lr": 5e-06, "epoch": 1.816357250376317, "percentage": 60.58, "elapsed_time": "14:51:14", "remaining_time": "9:40:02"}
183
+ {"current_steps": 1820, "total_steps": 2988, "loss": 0.6901, "lr": 5e-06, "epoch": 1.826392373306573, "percentage": 60.91, "elapsed_time": "14:56:06", "remaining_time": "9:35:05"}
184
+ {"current_steps": 1830, "total_steps": 2988, "loss": 0.6878, "lr": 5e-06, "epoch": 1.836427496236829, "percentage": 61.24, "elapsed_time": "15:00:59", "remaining_time": "9:30:07"}
185
+ {"current_steps": 1840, "total_steps": 2988, "loss": 0.691, "lr": 5e-06, "epoch": 1.8464626191670848, "percentage": 61.58, "elapsed_time": "15:05:51", "remaining_time": "9:25:10"}
186
+ {"current_steps": 1850, "total_steps": 2988, "loss": 0.6925, "lr": 5e-06, "epoch": 1.8564977420973405, "percentage": 61.91, "elapsed_time": "15:10:43", "remaining_time": "9:20:13"}
187
+ {"current_steps": 1860, "total_steps": 2988, "loss": 0.6864, "lr": 5e-06, "epoch": 1.8665328650275965, "percentage": 62.25, "elapsed_time": "15:15:35", "remaining_time": "9:15:15"}
188
+ {"current_steps": 1870, "total_steps": 2988, "loss": 0.6885, "lr": 5e-06, "epoch": 1.8765679879578525, "percentage": 62.58, "elapsed_time": "15:20:27", "remaining_time": "9:10:18"}
189
+ {"current_steps": 1880, "total_steps": 2988, "loss": 0.6861, "lr": 5e-06, "epoch": 1.8866031108881085, "percentage": 62.92, "elapsed_time": "15:25:19", "remaining_time": "9:05:21"}
190
+ {"current_steps": 1890, "total_steps": 2988, "loss": 0.6895, "lr": 5e-06, "epoch": 1.8966382338183643, "percentage": 63.25, "elapsed_time": "15:30:11", "remaining_time": "9:00:24"}
191
+ {"current_steps": 1900, "total_steps": 2988, "loss": 0.6891, "lr": 5e-06, "epoch": 1.90667335674862, "percentage": 63.59, "elapsed_time": "15:35:04", "remaining_time": "8:55:27"}
192
+ {"current_steps": 1910, "total_steps": 2988, "loss": 0.6867, "lr": 5e-06, "epoch": 1.916708479678876, "percentage": 63.92, "elapsed_time": "15:39:56", "remaining_time": "8:50:29"}
193
+ {"current_steps": 1920, "total_steps": 2988, "loss": 0.6936, "lr": 5e-06, "epoch": 1.926743602609132, "percentage": 64.26, "elapsed_time": "15:44:47", "remaining_time": "8:45:32"}
194
+ {"current_steps": 1930, "total_steps": 2988, "loss": 0.6882, "lr": 5e-06, "epoch": 1.936778725539388, "percentage": 64.59, "elapsed_time": "15:49:39", "remaining_time": "8:40:35"}
195
+ {"current_steps": 1940, "total_steps": 2988, "loss": 0.6875, "lr": 5e-06, "epoch": 1.9468138484696438, "percentage": 64.93, "elapsed_time": "15:54:31", "remaining_time": "8:35:38"}
196
+ {"current_steps": 1950, "total_steps": 2988, "loss": 0.6879, "lr": 5e-06, "epoch": 1.9568489713998996, "percentage": 65.26, "elapsed_time": "15:59:24", "remaining_time": "8:30:41"}
197
+ {"current_steps": 1960, "total_steps": 2988, "loss": 0.687, "lr": 5e-06, "epoch": 1.9668840943301555, "percentage": 65.6, "elapsed_time": "16:04:16", "remaining_time": "8:25:45"}
198
+ {"current_steps": 1970, "total_steps": 2988, "loss": 0.6891, "lr": 5e-06, "epoch": 1.9769192172604115, "percentage": 65.93, "elapsed_time": "16:09:08", "remaining_time": "8:20:48"}
199
+ {"current_steps": 1980, "total_steps": 2988, "loss": 0.688, "lr": 5e-06, "epoch": 1.9869543401906673, "percentage": 66.27, "elapsed_time": "16:14:01", "remaining_time": "8:15:51"}
200
+ {"current_steps": 1990, "total_steps": 2988, "loss": 0.6916, "lr": 5e-06, "epoch": 1.996989463120923, "percentage": 66.6, "elapsed_time": "16:18:53", "remaining_time": "8:10:55"}
201
+ {"current_steps": 1993, "total_steps": 2988, "eval_loss": 0.7170566320419312, "epoch": 2.0, "percentage": 66.7, "elapsed_time": "16:29:13", "remaining_time": "8:13:51"}