Hugging Face's logo

izzcw
/
crafting_sft_fail_new_mem

Text Generation
Transformers
Safetensors
llama
llama-factory
full
Generated from Trainer
conversational
text-generation-inference
Model card Files
xet
Community
izzcw commited on
Commit
c93ca47
·
verified ·
1 Parent(s): 0636f04

Training in progress, step 500

Browse files
Files changed (6) 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 +60 -60
  6. training_args.bin +1 -1
model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f6fbc688c7e0e765cba721de8355cd93c9d722333e66c42524fbc630c3157ffe
3
  size 4976698672
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bb9561993553cfcaea2ffc00c2ec4a7c9f295680f90defcf03f76585571c5c0b
3
  size 4976698672
model-00002-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5b0082a80ff1301ecf0c05435c5645aa6125e0f7e5ef1b172541e966ffbbf58a
3
  size 4999802720
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:efe7cb6736cc3178ec7486a2ce35e2ba93feb139af75950d65d4b8de3f2421b6
3
  size 4999802720
model-00003-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6a8a7b0c3d9070d99b75a940a3f8bf8ac7bd429f7f2defc13988504bc897f04a
3
  size 4915916176
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1517c3cb05109e81d219e4e4674515ae60f46d01a537ebaa64ff3ef8c1b2a9bd
3
  size 4915916176
model-00004-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0391505b999bd6b9e66e8b6af562b9d01e73524bf2678632718701e9916086ea
3
  size 1168138808
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a797ee042a72089a83f666f06c6c049a970741071dcd9f495f53098cff45bc1
3
  size 1168138808
trainer_log.jsonl CHANGED
@@ -1,60 +1,60 @@
1
- {"current_steps": 10, "total_steps": 1548, "loss": 1.3314, "lr": 6.451612903225807e-07, "epoch": 0.006457341189765114, "percentage": 0.65, "elapsed_time": "0:02:09", "remaining_time": "5:32:35"}
2
- {"current_steps": 20, "total_steps": 1548, "loss": 1.0161, "lr": 1.2903225806451614e-06, "epoch": 0.012914682379530229, "percentage": 1.29, "elapsed_time": "0:04:15", "remaining_time": "5:25:41"}
3
- {"current_steps": 30, "total_steps": 1548, "loss": 0.7231, "lr": 1.935483870967742e-06, "epoch": 0.019372023569295342, "percentage": 1.94, "elapsed_time": "0:06:20", "remaining_time": "5:20:47"}
4
- {"current_steps": 40, "total_steps": 1548, "loss": 0.5891, "lr": 2.580645161290323e-06, "epoch": 0.025829364759060457, "percentage": 2.58, "elapsed_time": "0:08:24", "remaining_time": "5:16:59"}
5
- {"current_steps": 50, "total_steps": 1548, "loss": 0.5322, "lr": 3.225806451612903e-06, "epoch": 0.03228670594882557, "percentage": 3.23, "elapsed_time": "0:10:31", "remaining_time": "5:15:21"}
6
- {"current_steps": 50, "total_steps": 1548, "eval_loss": 0.47686967253685, "epoch": 0.03228670594882557, "percentage": 3.23, "elapsed_time": "0:12:23", "remaining_time": "6:11:21"}
7
- {"current_steps": 60, "total_steps": 1548, "loss": 0.5336, "lr": 3.870967741935484e-06, "epoch": 0.038744047138590684, "percentage": 3.88, "elapsed_time": "0:14:27", "remaining_time": "5:58:42"}
8
- {"current_steps": 70, "total_steps": 1548, "loss": 0.5416, "lr": 4.516129032258065e-06, "epoch": 0.0452013883283558, "percentage": 4.52, "elapsed_time": "0:16:33", "remaining_time": "5:49:36"}
9
- {"current_steps": 80, "total_steps": 1548, "loss": 0.5361, "lr": 5.161290322580646e-06, "epoch": 0.051658729518120915, "percentage": 5.17, "elapsed_time": "0:18:39", "remaining_time": "5:42:20"}
10
- {"current_steps": 90, "total_steps": 1548, "loss": 0.4883, "lr": 5.806451612903226e-06, "epoch": 0.058116070707886026, "percentage": 5.81, "elapsed_time": "0:20:44", "remaining_time": "5:35:53"}
11
- {"current_steps": 100, "total_steps": 1548, "loss": 0.5257, "lr": 6.451612903225806e-06, "epoch": 0.06457341189765115, "percentage": 6.46, "elapsed_time": "0:22:50", "remaining_time": "5:30:42"}
12
- {"current_steps": 100, "total_steps": 1548, "eval_loss": 0.4827158451080322, "epoch": 0.06457341189765115, "percentage": 6.46, "elapsed_time": "0:24:42", "remaining_time": "5:57:44"}
13
- {"current_steps": 110, "total_steps": 1548, "loss": 0.519, "lr": 7.096774193548388e-06, "epoch": 0.07103075308741626, "percentage": 7.11, "elapsed_time": "0:26:48", "remaining_time": "5:50:21"}
14
- {"current_steps": 120, "total_steps": 1548, "loss": 0.5215, "lr": 7.741935483870968e-06, "epoch": 0.07748809427718137, "percentage": 7.75, "elapsed_time": "0:28:53", "remaining_time": "5:43:51"}
15
- {"current_steps": 130, "total_steps": 1548, "loss": 0.5459, "lr": 8.387096774193549e-06, "epoch": 0.08394543546694648, "percentage": 8.4, "elapsed_time": "0:30:58", "remaining_time": "5:37:56"}
16
- {"current_steps": 140, "total_steps": 1548, "loss": 0.5442, "lr": 9.03225806451613e-06, "epoch": 0.0904027766567116, "percentage": 9.04, "elapsed_time": "0:33:04", "remaining_time": "5:32:42"}
17
- {"current_steps": 150, "total_steps": 1548, "loss": 0.5101, "lr": 9.67741935483871e-06, "epoch": 0.09686011784647672, "percentage": 9.69, "elapsed_time": "0:35:10", "remaining_time": "5:27:53"}
18
- {"current_steps": 150, "total_steps": 1548, "eval_loss": 0.49281740188598633, "epoch": 0.09686011784647672, "percentage": 9.69, "elapsed_time": "0:37:02", "remaining_time": "5:45:17"}
19
- {"current_steps": 160, "total_steps": 1548, "loss": 0.5419, "lr": 9.999682112882774e-06, "epoch": 0.10331745903624183, "percentage": 10.34, "elapsed_time": "0:39:07", "remaining_time": "5:39:28"}
20
- {"current_steps": 170, "total_steps": 1548, "loss": 0.5354, "lr": 9.997139258465151e-06, "epoch": 0.10977480022600694, "percentage": 10.98, "elapsed_time": "0:41:15", "remaining_time": "5:34:24"}
21
- {"current_steps": 180, "total_steps": 1548, "loss": 0.5437, "lr": 9.992054842933846e-06, "epoch": 0.11623214141577205, "percentage": 11.63, "elapsed_time": "0:43:19", "remaining_time": "5:29:13"}
22
- {"current_steps": 190, "total_steps": 1548, "loss": 0.5211, "lr": 9.984431452238968e-06, "epoch": 0.12268948260553716, "percentage": 12.27, "elapsed_time": "0:45:24", "remaining_time": "5:24:35"}
23
- {"current_steps": 200, "total_steps": 1548, "loss": 0.5272, "lr": 9.974272963661561e-06, "epoch": 0.1291468237953023, "percentage": 12.92, "elapsed_time": "0:47:29", "remaining_time": "5:20:08"}
24
- {"current_steps": 200, "total_steps": 1548, "eval_loss": 0.4881571829319, "epoch": 0.1291468237953023, "percentage": 12.92, "elapsed_time": "0:49:21", "remaining_time": "5:32:43"}
25
- {"current_steps": 210, "total_steps": 1548, "loss": 0.5408, "lr": 9.961584543841625e-06, "epoch": 0.1356041649850674, "percentage": 13.57, "elapsed_time": "0:51:27", "remaining_time": "5:27:54"}
26
- {"current_steps": 220, "total_steps": 1548, "loss": 0.5263, "lr": 9.946372646150322e-06, "epoch": 0.1420615061748325, "percentage": 14.21, "elapsed_time": "0:53:35", "remaining_time": "5:23:30"}
27
- {"current_steps": 230, "total_steps": 1548, "loss": 0.533, "lr": 9.928645007407791e-06, "epoch": 0.14851884736459764, "percentage": 14.86, "elapsed_time": "0:55:40", "remaining_time": "5:19:05"}
28
- {"current_steps": 240, "total_steps": 1548, "loss": 0.5044, "lr": 9.908410643948162e-06, "epoch": 0.15497618855436274, "percentage": 15.5, "elapsed_time": "0:57:47", "remaining_time": "5:14:58"}
29
- {"current_steps": 250, "total_steps": 1548, "loss": 0.5228, "lr": 9.885679847033832e-06, "epoch": 0.16143352974412786, "percentage": 16.15, "elapsed_time": "0:59:54", "remaining_time": "5:11:02"}
30
- {"current_steps": 250, "total_steps": 1548, "eval_loss": 0.46039626002311707, "epoch": 0.16143352974412786, "percentage": 16.15, "elapsed_time": "1:01:46", "remaining_time": "5:20:44"}
31
- {"current_steps": 260, "total_steps": 1548, "loss": 0.5046, "lr": 9.860464177621286e-06, "epoch": 0.16789087093389296, "percentage": 16.8, "elapsed_time": "1:03:51", "remaining_time": "5:16:20"}
32
- {"current_steps": 270, "total_steps": 1548, "loss": 0.4923, "lr": 9.832776460481162e-06, "epoch": 0.1743482121236581, "percentage": 17.44, "elapsed_time": "1:05:56", "remaining_time": "5:12:07"}
33
- {"current_steps": 280, "total_steps": 1548, "loss": 0.5036, "lr": 9.802630777675528e-06, "epoch": 0.1808055533134232, "percentage": 18.09, "elapsed_time": "1:08:01", "remaining_time": "5:08:02"}
34
- {"current_steps": 290, "total_steps": 1548, "loss": 0.5151, "lr": 9.770042461395705e-06, "epoch": 0.1872628945031883, "percentage": 18.73, "elapsed_time": "1:10:08", "remaining_time": "5:04:16"}
35
- {"current_steps": 300, "total_steps": 1548, "loss": 0.5103, "lr": 9.735028086164265e-06, "epoch": 0.19372023569295344, "percentage": 19.38, "elapsed_time": "1:12:13", "remaining_time": "5:00:25"}
36
- {"current_steps": 300, "total_steps": 1548, "eval_loss": 0.444090336561203, "epoch": 0.19372023569295344, "percentage": 19.38, "elapsed_time": "1:14:05", "remaining_time": "5:08:11"}
37
- {"current_steps": 310, "total_steps": 1548, "loss": 0.4909, "lr": 9.697605460405162e-06, "epoch": 0.20017757688271853, "percentage": 20.03, "elapsed_time": "1:16:10", "remaining_time": "5:04:11"}
38
- {"current_steps": 320, "total_steps": 1548, "loss": 0.5065, "lr": 9.657793617386337e-06, "epoch": 0.20663491807248366, "percentage": 20.67, "elapsed_time": "1:18:14", "remaining_time": "5:00:16"}
39
- {"current_steps": 330, "total_steps": 1548, "loss": 0.5045, "lr": 9.615612805539305e-06, "epoch": 0.21309225926224876, "percentage": 21.32, "elapsed_time": "1:20:21", "remaining_time": "4:56:35"}
40
- {"current_steps": 340, "total_steps": 1548, "loss": 0.4849, "lr": 9.571084478160763e-06, "epoch": 0.21954960045201388, "percentage": 21.96, "elapsed_time": "1:22:25", "remaining_time": "4:52:51"}
41
- {"current_steps": 350, "total_steps": 1548, "loss": 0.4746, "lr": 9.524231282501362e-06, "epoch": 0.226006941641779, "percentage": 22.61, "elapsed_time": "1:24:32", "remaining_time": "4:49:21"}
42
- {"current_steps": 350, "total_steps": 1548, "eval_loss": 0.4407587945461273, "epoch": 0.226006941641779, "percentage": 22.61, "elapsed_time": "1:26:24", "remaining_time": "4:55:44"}
43
- {"current_steps": 360, "total_steps": 1548, "loss": 0.5119, "lr": 9.475077048247243e-06, "epoch": 0.2324642828315441, "percentage": 23.26, "elapsed_time": "1:28:31", "remaining_time": "4:52:08"}
44
- {"current_steps": 370, "total_steps": 1548, "loss": 0.4989, "lr": 9.423646775400186e-06, "epoch": 0.23892162402130923, "percentage": 23.9, "elapsed_time": "1:30:36", "remaining_time": "4:48:28"}
45
- {"current_steps": 380, "total_steps": 1548, "loss": 0.4775, "lr": 9.369966621562525e-06, "epoch": 0.24537896521107433, "percentage": 24.55, "elapsed_time": "1:32:42", "remaining_time": "4:44:57"}
46
- {"current_steps": 390, "total_steps": 1548, "loss": 0.5011, "lr": 9.314063888633312e-06, "epoch": 0.2518363064008394, "percentage": 25.19, "elapsed_time": "1:34:47", "remaining_time": "4:41:26"}
47
- {"current_steps": 400, "total_steps": 1548, "loss": 0.4763, "lr": 9.255967008922475e-06, "epoch": 0.2582936475906046, "percentage": 25.84, "elapsed_time": "1:36:51", "remaining_time": "4:37:59"}
48
- {"current_steps": 400, "total_steps": 1548, "eval_loss": 0.4318215548992157, "epoch": 0.2582936475906046, "percentage": 25.84, "elapsed_time": "1:38:43", "remaining_time": "4:43:21"}
49
- {"current_steps": 410, "total_steps": 1548, "loss": 0.4848, "lr": 9.19570553069006e-06, "epoch": 0.2647509887803697, "percentage": 26.49, "elapsed_time": "1:40:49", "remaining_time": "4:39:51"}
50
- {"current_steps": 420, "total_steps": 1548, "loss": 0.4629, "lr": 9.133310103117891e-06, "epoch": 0.2712083299701348, "percentage": 27.13, "elapsed_time": "1:42:55", "remaining_time": "4:36:25"}
51
- {"current_steps": 430, "total_steps": 1548, "loss": 0.4801, "lr": 9.068812460721296e-06, "epoch": 0.27766567115989993, "percentage": 27.78, "elapsed_time": "1:45:01", "remaining_time": "4:33:02"}
52
- {"current_steps": 440, "total_steps": 1548, "loss": 0.4797, "lr": 9.002245407208831e-06, "epoch": 0.284123012349665, "percentage": 28.42, "elapsed_time": "1:47:07", "remaining_time": "4:29:45"}
53
- {"current_steps": 450, "total_steps": 1548, "loss": 0.4793, "lr": 8.9336427987982e-06, "epoch": 0.2905803535394301, "percentage": 29.07, "elapsed_time": "1:49:13", "remaining_time": "4:26:31"}
54
- {"current_steps": 450, "total_steps": 1548, "eval_loss": 0.42118436098098755, "epoch": 0.2905803535394301, "percentage": 29.07, "elapsed_time": "1:51:05", "remaining_time": "4:31:05"}
55
- {"current_steps": 460, "total_steps": 1548, "loss": 0.4712, "lr": 8.863039526996891e-06, "epoch": 0.2970376947291953, "percentage": 29.72, "elapsed_time": "1:53:12", "remaining_time": "4:27:45"}
56
- {"current_steps": 470, "total_steps": 1548, "loss": 0.4765, "lr": 8.790471500856229e-06, "epoch": 0.3034950359189604, "percentage": 30.36, "elapsed_time": "1:55:18", "remaining_time": "4:24:28"}
57
- {"current_steps": 480, "total_steps": 1548, "loss": 0.4744, "lr": 8.715975628707917e-06, "epoch": 0.3099523771087255, "percentage": 31.01, "elapsed_time": "1:57:24", "remaining_time": "4:21:14"}
58
- {"current_steps": 490, "total_steps": 1548, "loss": 0.4593, "lr": 8.63958979939235e-06, "epoch": 0.31640971829849057, "percentage": 31.65, "elapsed_time": "1:59:30", "remaining_time": "4:18:01"}
59
- {"current_steps": 500, "total_steps": 1548, "loss": 0.452, "lr": 8.56135286298822e-06, "epoch": 0.3228670594882557, "percentage": 32.3, "elapsed_time": "2:01:35", "remaining_time": "4:14:51"}
60
- {"current_steps": 500, "total_steps": 1548, "eval_loss": 0.4149642884731293, "epoch": 0.3228670594882557, "percentage": 32.3, "elapsed_time": "2:03:27", "remaining_time": "4:18:46"}
 
1
+ {"current_steps": 10, "total_steps": 1548, "loss": 1.3322, "lr": 6.451612903225807e-07, "epoch": 0.006457341189765114, "percentage": 0.65, "elapsed_time": "0:02:10", "remaining_time": "5:33:49"}
2
+ {"current_steps": 20, "total_steps": 1548, "loss": 1.0135, "lr": 1.2903225806451614e-06, "epoch": 0.012914682379530229, "percentage": 1.29, "elapsed_time": "0:04:17", "remaining_time": "5:27:33"}
3
+ {"current_steps": 30, "total_steps": 1548, "loss": 0.7236, "lr": 1.935483870967742e-06, "epoch": 0.019372023569295342, "percentage": 1.94, "elapsed_time": "0:06:22", "remaining_time": "5:22:48"}
4
+ {"current_steps": 40, "total_steps": 1548, "loss": 0.5902, "lr": 2.580645161290323e-06, "epoch": 0.025829364759060457, "percentage": 2.58, "elapsed_time": "0:08:28", "remaining_time": "5:19:17"}
5
+ {"current_steps": 50, "total_steps": 1548, "loss": 0.5341, "lr": 3.225806451612903e-06, "epoch": 0.03228670594882557, "percentage": 3.23, "elapsed_time": "0:10:35", "remaining_time": "5:17:33"}
6
+ {"current_steps": 50, "total_steps": 1548, "eval_loss": 0.4792882204055786, "epoch": 0.03228670594882557, "percentage": 3.23, "elapsed_time": "0:12:29", "remaining_time": "6:14:04"}
7
+ {"current_steps": 60, "total_steps": 1548, "loss": 0.5369, "lr": 3.870967741935484e-06, "epoch": 0.038744047138590684, "percentage": 3.88, "elapsed_time": "0:14:34", "remaining_time": "6:01:25"}
8
+ {"current_steps": 70, "total_steps": 1548, "loss": 0.5446, "lr": 4.516129032258065e-06, "epoch": 0.0452013883283558, "percentage": 4.52, "elapsed_time": "0:16:41", "remaining_time": "5:52:16"}
9
+ {"current_steps": 80, "total_steps": 1548, "loss": 0.5406, "lr": 5.161290322580646e-06, "epoch": 0.051658729518120915, "percentage": 5.17, "elapsed_time": "0:18:47", "remaining_time": "5:44:54"}
10
+ {"current_steps": 90, "total_steps": 1548, "loss": 0.4874, "lr": 5.806451612903226e-06, "epoch": 0.058116070707886026, "percentage": 5.81, "elapsed_time": "0:20:53", "remaining_time": "5:38:25"}
11
+ {"current_steps": 100, "total_steps": 1548, "loss": 0.5261, "lr": 6.451612903225806e-06, "epoch": 0.06457341189765115, "percentage": 6.46, "elapsed_time": "0:23:00", "remaining_time": "5:33:12"}
12
+ {"current_steps": 100, "total_steps": 1548, "eval_loss": 0.4858162999153137, "epoch": 0.06457341189765115, "percentage": 6.46, "elapsed_time": "0:24:53", "remaining_time": "6:00:29"}
13
+ {"current_steps": 110, "total_steps": 1548, "loss": 0.5207, "lr": 7.096774193548388e-06, "epoch": 0.07103075308741626, "percentage": 7.11, "elapsed_time": "0:27:00", "remaining_time": "5:53:07"}
14
+ {"current_steps": 120, "total_steps": 1548, "loss": 0.5235, "lr": 7.741935483870968e-06, "epoch": 0.07748809427718137, "percentage": 7.75, "elapsed_time": "0:29:07", "remaining_time": "5:46:37"}
15
+ {"current_steps": 130, "total_steps": 1548, "loss": 0.55, "lr": 8.387096774193549e-06, "epoch": 0.08394543546694648, "percentage": 8.4, "elapsed_time": "0:31:14", "remaining_time": "5:40:42"}
16
+ {"current_steps": 140, "total_steps": 1548, "loss": 0.545, "lr": 9.03225806451613e-06, "epoch": 0.0904027766567116, "percentage": 9.04, "elapsed_time": "0:33:21", "remaining_time": "5:35:26"}
17
+ {"current_steps": 150, "total_steps": 1548, "loss": 0.5103, "lr": 9.67741935483871e-06, "epoch": 0.09686011784647672, "percentage": 9.69, "elapsed_time": "0:35:28", "remaining_time": "5:30:36"}
18
+ {"current_steps": 150, "total_steps": 1548, "eval_loss": 0.4932478964328766, "epoch": 0.09686011784647672, "percentage": 9.69, "elapsed_time": "0:37:21", "remaining_time": "5:48:10"}
19
+ {"current_steps": 160, "total_steps": 1548, "loss": 0.5344, "lr": 9.999682112882774e-06, "epoch": 0.10331745903624183, "percentage": 10.34, "elapsed_time": "0:39:27", "remaining_time": "5:42:19"}
20
+ {"current_steps": 170, "total_steps": 1548, "loss": 0.5323, "lr": 9.997139258465151e-06, "epoch": 0.10977480022600694, "percentage": 10.98, "elapsed_time": "0:41:36", "remaining_time": "5:37:15"}
21
+ {"current_steps": 180, "total_steps": 1548, "loss": 0.5393, "lr": 9.992054842933846e-06, "epoch": 0.11623214141577205, "percentage": 11.63, "elapsed_time": "0:43:41", "remaining_time": "5:32:02"}
22
+ {"current_steps": 190, "total_steps": 1548, "loss": 0.5208, "lr": 9.984431452238968e-06, "epoch": 0.12268948260553716, "percentage": 12.27, "elapsed_time": "0:45:48", "remaining_time": "5:27:22"}
23
+ {"current_steps": 200, "total_steps": 1548, "loss": 0.5236, "lr": 9.974272963661561e-06, "epoch": 0.1291468237953023, "percentage": 12.92, "elapsed_time": "0:47:54", "remaining_time": "5:22:55"}
24
+ {"current_steps": 200, "total_steps": 1548, "eval_loss": 0.48204439878463745, "epoch": 0.1291468237953023, "percentage": 12.92, "elapsed_time": "0:49:47", "remaining_time": "5:35:37"}
25
+ {"current_steps": 210, "total_steps": 1548, "loss": 0.5369, "lr": 9.961584543841625e-06, "epoch": 0.1356041649850674, "percentage": 13.57, "elapsed_time": "0:51:54", "remaining_time": "5:30:45"}
26
+ {"current_steps": 220, "total_steps": 1548, "loss": 0.5244, "lr": 9.946372646150322e-06, "epoch": 0.1420615061748325, "percentage": 14.21, "elapsed_time": "0:54:03", "remaining_time": "5:26:19"}
27
+ {"current_steps": 230, "total_steps": 1548, "loss": 0.5336, "lr": 9.928645007407791e-06, "epoch": 0.14851884736459764, "percentage": 14.86, "elapsed_time": "0:56:09", "remaining_time": "5:21:51"}
28
+ {"current_steps": 240, "total_steps": 1548, "loss": 0.5015, "lr": 9.908410643948162e-06, "epoch": 0.15497618855436274, "percentage": 15.5, "elapsed_time": "0:58:17", "remaining_time": "5:17:42"}
29
+ {"current_steps": 250, "total_steps": 1548, "loss": 0.524, "lr": 9.885679847033832e-06, "epoch": 0.16143352974412786, "percentage": 16.15, "elapsed_time": "1:00:25", "remaining_time": "5:13:44"}
30
+ {"current_steps": 250, "total_steps": 1548, "eval_loss": 0.462333619594574, "epoch": 0.16143352974412786, "percentage": 16.15, "elapsed_time": "1:02:18", "remaining_time": "5:23:31"}
31
+ {"current_steps": 260, "total_steps": 1548, "loss": 0.5043, "lr": 9.860464177621286e-06, "epoch": 0.16789087093389296, "percentage": 16.8, "elapsed_time": "1:04:24", "remaining_time": "5:19:04"}
32
+ {"current_steps": 270, "total_steps": 1548, "loss": 0.4924, "lr": 9.832776460481162e-06, "epoch": 0.1743482121236581, "percentage": 17.44, "elapsed_time": "1:06:30", "remaining_time": "5:14:49"}
33
+ {"current_steps": 280, "total_steps": 1548, "loss": 0.5015, "lr": 9.802630777675528e-06, "epoch": 0.1808055533134232, "percentage": 18.09, "elapsed_time": "1:08:36", "remaining_time": "5:10:42"}
34
+ {"current_steps": 290, "total_steps": 1548, "loss": 0.5159, "lr": 9.770042461395705e-06, "epoch": 0.1872628945031883, "percentage": 18.73, "elapsed_time": "1:10:45", "remaining_time": "5:06:54"}
35
+ {"current_steps": 300, "total_steps": 1548, "loss": 0.5107, "lr": 9.735028086164265e-06, "epoch": 0.19372023569295344, "percentage": 19.38, "elapsed_time": "1:12:50", "remaining_time": "5:03:02"}
36
+ {"current_steps": 300, "total_steps": 1548, "eval_loss": 0.44542330503463745, "epoch": 0.19372023569295344, "percentage": 19.38, "elapsed_time": "1:14:43", "remaining_time": "5:10:53"}
37
+ {"current_steps": 310, "total_steps": 1548, "loss": 0.4875, "lr": 9.697605460405162e-06, "epoch": 0.20017757688271853, "percentage": 20.03, "elapsed_time": "1:16:50", "remaining_time": "5:06:50"}
38
+ {"current_steps": 320, "total_steps": 1548, "loss": 0.5043, "lr": 9.657793617386337e-06, "epoch": 0.20663491807248366, "percentage": 20.67, "elapsed_time": "1:18:55", "remaining_time": "5:02:53"}
39
+ {"current_steps": 330, "total_steps": 1548, "loss": 0.5056, "lr": 9.615612805539305e-06, "epoch": 0.21309225926224876, "percentage": 21.32, "elapsed_time": "1:21:03", "remaining_time": "4:59:10"}
40
+ {"current_steps": 340, "total_steps": 1548, "loss": 0.4821, "lr": 9.571084478160763e-06, "epoch": 0.21954960045201388, "percentage": 21.96, "elapsed_time": "1:23:08", "remaining_time": "4:55:24"}
41
+ {"current_steps": 350, "total_steps": 1548, "loss": 0.4723, "lr": 9.524231282501362e-06, "epoch": 0.226006941641779, "percentage": 22.61, "elapsed_time": "1:25:16", "remaining_time": "4:51:51"}
42
+ {"current_steps": 350, "total_steps": 1548, "eval_loss": 0.43804243206977844, "epoch": 0.226006941641779, "percentage": 22.61, "elapsed_time": "1:27:09", "remaining_time": "4:58:18"}
43
+ {"current_steps": 360, "total_steps": 1548, "loss": 0.5259, "lr": 9.475077048247243e-06, "epoch": 0.2324642828315441, "percentage": 23.26, "elapsed_time": "1:29:17", "remaining_time": "4:54:40"}
44
+ {"current_steps": 370, "total_steps": 1548, "loss": 0.4965, "lr": 9.423646775400186e-06, "epoch": 0.23892162402130923, "percentage": 23.9, "elapsed_time": "1:31:23", "remaining_time": "4:50:58"}
45
+ {"current_steps": 380, "total_steps": 1548, "loss": 0.4769, "lr": 9.369966621562525e-06, "epoch": 0.24537896521107433, "percentage": 24.55, "elapsed_time": "1:33:30", "remaining_time": "4:47:26"}
46
+ {"current_steps": 390, "total_steps": 1548, "loss": 0.4993, "lr": 9.314063888633312e-06, "epoch": 0.2518363064008394, "percentage": 25.19, "elapsed_time": "1:35:36", "remaining_time": "4:43:53"}
47
+ {"current_steps": 400, "total_steps": 1548, "loss": 0.4771, "lr": 9.255967008922475e-06, "epoch": 0.2582936475906046, "percentage": 25.84, "elapsed_time": "1:37:42", "remaining_time": "4:40:25"}
48
+ {"current_steps": 400, "total_steps": 1548, "eval_loss": 0.4323236048221588, "epoch": 0.2582936475906046, "percentage": 25.84, "elapsed_time": "1:39:35", "remaining_time": "4:45:50"}
49
+ {"current_steps": 410, "total_steps": 1548, "loss": 0.4848, "lr": 9.19570553069006e-06, "epoch": 0.2647509887803697, "percentage": 26.49, "elapsed_time": "1:41:42", "remaining_time": "4:42:18"}
50
+ {"current_steps": 420, "total_steps": 1548, "loss": 0.4608, "lr": 9.133310103117891e-06, "epoch": 0.2712083299701348, "percentage": 27.13, "elapsed_time": "1:43:49", "remaining_time": "4:38:49"}
51
+ {"current_steps": 430, "total_steps": 1548, "loss": 0.4828, "lr": 9.068812460721296e-06, "epoch": 0.27766567115989993, "percentage": 27.78, "elapsed_time": "1:45:56", "remaining_time": "4:35:25"}
52
+ {"current_steps": 440, "total_steps": 1548, "loss": 0.4789, "lr": 9.002245407208831e-06, "epoch": 0.284123012349665, "percentage": 28.42, "elapsed_time": "1:48:03", "remaining_time": "4:32:06"}
53
+ {"current_steps": 450, "total_steps": 1548, "loss": 0.4835, "lr": 8.9336427987982e-06, "epoch": 0.2905803535394301, "percentage": 29.07, "elapsed_time": "1:50:11", "remaining_time": "4:28:51"}
54
+ {"current_steps": 450, "total_steps": 1548, "eval_loss": 0.4248904883861542, "epoch": 0.2905803535394301, "percentage": 29.07, "elapsed_time": "1:52:04", "remaining_time": "4:33:27"}
55
+ {"current_steps": 460, "total_steps": 1548, "loss": 0.4727, "lr": 8.863039526996891e-06, "epoch": 0.2970376947291953, "percentage": 29.72, "elapsed_time": "1:54:12", "remaining_time": "4:30:06"}
56
+ {"current_steps": 470, "total_steps": 1548, "loss": 0.4808, "lr": 8.790471500856229e-06, "epoch": 0.3034950359189604, "percentage": 30.36, "elapsed_time": "1:56:19", "remaining_time": "4:26:47"}
57
+ {"current_steps": 480, "total_steps": 1548, "loss": 0.4775, "lr": 8.715975628707917e-06, "epoch": 0.3099523771087255, "percentage": 31.01, "elapsed_time": "1:58:26", "remaining_time": "4:23:31"}
58
+ {"current_steps": 490, "total_steps": 1548, "loss": 0.4683, "lr": 8.63958979939235e-06, "epoch": 0.31640971829849057, "percentage": 31.65, "elapsed_time": "2:00:32", "remaining_time": "4:20:16"}
59
+ {"current_steps": 500, "total_steps": 1548, "loss": 0.455, "lr": 8.56135286298822e-06, "epoch": 0.3228670594882557, "percentage": 32.3, "elapsed_time": "2:02:39", "remaining_time": "4:17:04"}
60
+ {"current_steps": 500, "total_steps": 1548, "eval_loss": 0.4204798936843872, "epoch": 0.3228670594882557, "percentage": 32.3, "elapsed_time": "2:04:32", "remaining_time": "4:21:02"}
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d2da57172588c9ee0aaf74d75f02529ec5abc6f1e761797c0db231c227e803c1
3
  size 7544
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a5b6378023639756555e4cd06e6b9c5ee88b1e86a7bd81bc670c1ce7e056882
3
  size 7544