Hugging Face's logo

mlfoundations-dev
/
hp_ablations_gemma_scheduler_constant

Text Generation
Transformers
Safetensors
gemma2
llama-factory
full
Generated from Trainer
text-generation-inference
Model card Files
xet
 Community
sedrickkeh commited on
Commit
36c39a6
·
verified ·
1 Parent(s): 55ec6ad

Training in progress, epoch 1

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 +45 -0
model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f815f1b37c9df64ddb15ca678fa6daa9fbafecf980d429a1c7606af5ed0434be
3
  size 4903351912
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b921ae986d40c57377351ab0759cd728f44691d226e2beffd2ea996055ff3af0
3
  size 4903351912
model-00002-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8937331a2e86d20eeb33352c3f7a5e6ba1d317f649ff008160e673e8041bd3e8
3
  size 4947570872
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e25ac706ebd7934760853351d456ed3f9ba96ff18515c244e7daff06204d40e5
3
  size 4947570872
model-00003-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bf882095df62c0e0d93a08bd6116bb5533f7159f419616012ded3411d51b82ac
3
  size 4962221464
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e2f83ee63677e5e544bd1c60faba7df6c87eaea3bab6f9e93daebfe1c0d68dfb
3
  size 4962221464
model-00004-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:78e06ca3e36c7fe1e2c47a55cd9899ec9e0d2d618f20c9391582c1393109e490
3
  size 3670322200
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:443e48154fadd0f4740f7db341cb8993d9835db9f6c54a8fde3f9ac271c2abc2
3
  size 3670322200
trainer_log.jsonl CHANGED
@@ -43,3 +43,48 @@
43
  {"current_steps": 430, "total_steps": 1329, "loss": 0.5886, "learning_rate": 5e-06, "epoch": 0.9703808180535967, "percentage": 32.36, "elapsed_time": "10:34:34", "remaining_time": "22:06:42"}
44
  {"current_steps": 440, "total_steps": 1329, "loss": 0.5934, "learning_rate": 5e-06, "epoch": 0.9929478138222849, "percentage": 33.11, "elapsed_time": "10:49:20", "remaining_time": "21:51:57"}
45
  {"current_steps": 443, "total_steps": 1329, "eval_loss": 0.5912777781486511, "epoch": 0.9997179125528914, "percentage": 33.33, "elapsed_time": "11:05:20", "remaining_time": "22:10:40"}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
43
  {"current_steps": 430, "total_steps": 1329, "loss": 0.5886, "learning_rate": 5e-06, "epoch": 0.9703808180535967, "percentage": 32.36, "elapsed_time": "10:34:34", "remaining_time": "22:06:42"}
44
  {"current_steps": 440, "total_steps": 1329, "loss": 0.5934, "learning_rate": 5e-06, "epoch": 0.9929478138222849, "percentage": 33.11, "elapsed_time": "10:49:20", "remaining_time": "21:51:57"}
45
  {"current_steps": 443, "total_steps": 1329, "eval_loss": 0.5912777781486511, "epoch": 0.9997179125528914, "percentage": 33.33, "elapsed_time": "11:05:20", "remaining_time": "22:10:40"}
46
+ {"current_steps": 450, "total_steps": 1329, "loss": 0.6008, "learning_rate": 5e-06, "epoch": 1.0155148095909732, "percentage": 33.86, "elapsed_time": "11:16:48", "remaining_time": "22:02:01"}
47
+ {"current_steps": 460, "total_steps": 1329, "loss": 0.528, "learning_rate": 5e-06, "epoch": 1.0380818053596614, "percentage": 34.61, "elapsed_time": "11:31:34", "remaining_time": "21:46:28"}
48
+ {"current_steps": 470, "total_steps": 1329, "loss": 0.5429, "learning_rate": 5e-06, "epoch": 1.0606488011283497, "percentage": 35.36, "elapsed_time": "11:46:20", "remaining_time": "21:30:56"}
49
+ {"current_steps": 480, "total_steps": 1329, "loss": 0.5308, "learning_rate": 5e-06, "epoch": 1.0832157968970382, "percentage": 36.12, "elapsed_time": "12:01:05", "remaining_time": "21:15:26"}
50
+ {"current_steps": 490, "total_steps": 1329, "loss": 0.5408, "learning_rate": 5e-06, "epoch": 1.1057827926657264, "percentage": 36.87, "elapsed_time": "12:15:51", "remaining_time": "20:59:58"}
51
+ {"current_steps": 500, "total_steps": 1329, "loss": 0.5396, "learning_rate": 5e-06, "epoch": 1.1283497884344147, "percentage": 37.62, "elapsed_time": "12:30:37", "remaining_time": "20:44:32"}
52
+ {"current_steps": 510, "total_steps": 1329, "loss": 0.5407, "learning_rate": 5e-06, "epoch": 1.150916784203103, "percentage": 38.37, "elapsed_time": "12:45:23", "remaining_time": "20:29:07"}
53
+ {"current_steps": 520, "total_steps": 1329, "loss": 0.5333, "learning_rate": 5e-06, "epoch": 1.1734837799717912, "percentage": 39.13, "elapsed_time": "13:00:08", "remaining_time": "20:13:43"}
54
+ {"current_steps": 530, "total_steps": 1329, "loss": 0.5395, "learning_rate": 5e-06, "epoch": 1.1960507757404795, "percentage": 39.88, "elapsed_time": "13:14:54", "remaining_time": "19:58:21"}
55
+ {"current_steps": 540, "total_steps": 1329, "loss": 0.5371, "learning_rate": 5e-06, "epoch": 1.2186177715091677, "percentage": 40.63, "elapsed_time": "13:29:40", "remaining_time": "19:43:00"}
56
+ {"current_steps": 550, "total_steps": 1329, "loss": 0.5419, "learning_rate": 5e-06, "epoch": 1.2411847672778562, "percentage": 41.38, "elapsed_time": "13:44:25", "remaining_time": "19:27:41"}
57
+ {"current_steps": 560, "total_steps": 1329, "loss": 0.5438, "learning_rate": 5e-06, "epoch": 1.2637517630465445, "percentage": 42.14, "elapsed_time": "13:59:11", "remaining_time": "19:12:23"}
58
+ {"current_steps": 570, "total_steps": 1329, "loss": 0.541, "learning_rate": 5e-06, "epoch": 1.2863187588152327, "percentage": 42.89, "elapsed_time": "14:13:57", "remaining_time": "18:57:07"}
59
+ {"current_steps": 580, "total_steps": 1329, "loss": 0.5409, "learning_rate": 5e-06, "epoch": 1.308885754583921, "percentage": 43.64, "elapsed_time": "14:28:43", "remaining_time": "18:41:51"}
60
+ {"current_steps": 590, "total_steps": 1329, "loss": 0.5411, "learning_rate": 5e-06, "epoch": 1.3314527503526092, "percentage": 44.39, "elapsed_time": "14:43:29", "remaining_time": "18:26:36"}
61
+ {"current_steps": 600, "total_steps": 1329, "loss": 0.5463, "learning_rate": 5e-06, "epoch": 1.3540197461212977, "percentage": 45.15, "elapsed_time": "14:58:15", "remaining_time": "18:11:22"}
62
+ {"current_steps": 610, "total_steps": 1329, "loss": 0.5391, "learning_rate": 5e-06, "epoch": 1.376586741889986, "percentage": 45.9, "elapsed_time": "15:13:01", "remaining_time": "17:56:09"}
63
+ {"current_steps": 620, "total_steps": 1329, "loss": 0.5433, "learning_rate": 5e-06, "epoch": 1.3991537376586742, "percentage": 46.65, "elapsed_time": "15:27:46", "remaining_time": "17:40:57"}
64
+ {"current_steps": 630, "total_steps": 1329, "loss": 0.5413, "learning_rate": 5e-06, "epoch": 1.4217207334273625, "percentage": 47.4, "elapsed_time": "15:42:32", "remaining_time": "17:25:46"}
65
+ {"current_steps": 640, "total_steps": 1329, "loss": 0.5409, "learning_rate": 5e-06, "epoch": 1.4442877291960508, "percentage": 48.16, "elapsed_time": "15:57:18", "remaining_time": "17:10:36"}
66
+ {"current_steps": 650, "total_steps": 1329, "loss": 0.5418, "learning_rate": 5e-06, "epoch": 1.466854724964739, "percentage": 48.91, "elapsed_time": "16:12:04", "remaining_time": "16:55:26"}
67
+ {"current_steps": 660, "total_steps": 1329, "loss": 0.5456, "learning_rate": 5e-06, "epoch": 1.4894217207334273, "percentage": 49.66, "elapsed_time": "16:26:50", "remaining_time": "16:40:17"}
68
+ {"current_steps": 670, "total_steps": 1329, "loss": 0.5403, "learning_rate": 5e-06, "epoch": 1.5119887165021155, "percentage": 50.41, "elapsed_time": "16:41:36", "remaining_time": "16:25:10"}
69
+ {"current_steps": 680, "total_steps": 1329, "loss": 0.5344, "learning_rate": 5e-06, "epoch": 1.5345557122708038, "percentage": 51.17, "elapsed_time": "16:56:23", "remaining_time": "16:10:03"}
70
+ {"current_steps": 690, "total_steps": 1329, "loss": 0.5366, "learning_rate": 5e-06, "epoch": 1.5571227080394923, "percentage": 51.92, "elapsed_time": "17:11:10", "remaining_time": "15:54:57"}
71
+ {"current_steps": 700, "total_steps": 1329, "loss": 0.5408, "learning_rate": 5e-06, "epoch": 1.5796897038081805, "percentage": 52.67, "elapsed_time": "17:25:56", "remaining_time": "15:39:50"}
72
+ {"current_steps": 710, "total_steps": 1329, "loss": 0.5435, "learning_rate": 5e-06, "epoch": 1.6022566995768688, "percentage": 53.42, "elapsed_time": "17:40:42", "remaining_time": "15:24:45"}
73
+ {"current_steps": 720, "total_steps": 1329, "loss": 0.5518, "learning_rate": 5e-06, "epoch": 1.6248236953455573, "percentage": 54.18, "elapsed_time": "17:55:27", "remaining_time": "15:09:39"}
74
+ {"current_steps": 730, "total_steps": 1329, "loss": 0.535, "learning_rate": 5e-06, "epoch": 1.6473906911142455, "percentage": 54.93, "elapsed_time": "18:10:13", "remaining_time": "14:54:35"}
75
+ {"current_steps": 740, "total_steps": 1329, "loss": 0.5508, "learning_rate": 5e-06, "epoch": 1.6699576868829338, "percentage": 55.68, "elapsed_time": "18:24:59", "remaining_time": "14:39:30"}
76
+ {"current_steps": 750, "total_steps": 1329, "loss": 0.5422, "learning_rate": 5e-06, "epoch": 1.692524682651622, "percentage": 56.43, "elapsed_time": "18:39:46", "remaining_time": "14:24:27"}
77
+ {"current_steps": 760, "total_steps": 1329, "loss": 0.541, "learning_rate": 5e-06, "epoch": 1.7150916784203103, "percentage": 57.19, "elapsed_time": "18:54:32", "remaining_time": "14:09:24"}
78
+ {"current_steps": 770, "total_steps": 1329, "loss": 0.5417, "learning_rate": 5e-06, "epoch": 1.7376586741889986, "percentage": 57.94, "elapsed_time": "19:09:18", "remaining_time": "13:54:22"}
79
+ {"current_steps": 780, "total_steps": 1329, "loss": 0.5375, "learning_rate": 5e-06, "epoch": 1.7602256699576868, "percentage": 58.69, "elapsed_time": "19:24:04", "remaining_time": "13:39:19"}
80
+ {"current_steps": 790, "total_steps": 1329, "loss": 0.541, "learning_rate": 5e-06, "epoch": 1.782792665726375, "percentage": 59.44, "elapsed_time": "19:38:50", "remaining_time": "13:24:18"}
81
+ {"current_steps": 800, "total_steps": 1329, "loss": 0.536, "learning_rate": 5e-06, "epoch": 1.8053596614950633, "percentage": 60.2, "elapsed_time": "19:53:36", "remaining_time": "13:09:16"}
82
+ {"current_steps": 810, "total_steps": 1329, "loss": 0.5468, "learning_rate": 5e-06, "epoch": 1.8279266572637518, "percentage": 60.95, "elapsed_time": "20:08:22", "remaining_time": "12:54:15"}
83
+ {"current_steps": 820, "total_steps": 1329, "loss": 0.5341, "learning_rate": 5e-06, "epoch": 1.85049365303244, "percentage": 61.7, "elapsed_time": "20:23:09", "remaining_time": "12:39:15"}
84
+ {"current_steps": 830, "total_steps": 1329, "loss": 0.5462, "learning_rate": 5e-06, "epoch": 1.8730606488011283, "percentage": 62.45, "elapsed_time": "20:37:55", "remaining_time": "12:24:14"}
85
+ {"current_steps": 840, "total_steps": 1329, "loss": 0.5461, "learning_rate": 5e-06, "epoch": 1.8956276445698168, "percentage": 63.21, "elapsed_time": "20:52:41", "remaining_time": "12:09:14"}
86
+ {"current_steps": 850, "total_steps": 1329, "loss": 0.534, "learning_rate": 5e-06, "epoch": 1.918194640338505, "percentage": 63.96, "elapsed_time": "21:07:27", "remaining_time": "11:54:15"}
87
+ {"current_steps": 860, "total_steps": 1329, "loss": 0.539, "learning_rate": 5e-06, "epoch": 1.9407616361071933, "percentage": 64.71, "elapsed_time": "21:22:14", "remaining_time": "11:39:15"}
88
+ {"current_steps": 870, "total_steps": 1329, "loss": 0.5456, "learning_rate": 5e-06, "epoch": 1.9633286318758816, "percentage": 65.46, "elapsed_time": "21:37:00", "remaining_time": "11:24:16"}
89
+ {"current_steps": 880, "total_steps": 1329, "loss": 0.5439, "learning_rate": 5e-06, "epoch": 1.9858956276445698, "percentage": 66.22, "elapsed_time": "21:51:46", "remaining_time": "11:09:18"}
90
+ {"current_steps": 886, "total_steps": 1329, "eval_loss": 0.5872690677642822, "epoch": 1.9994358251057829, "percentage": 66.67, "elapsed_time": "22:12:26", "remaining_time": "11:06:13"}