diff --git a/.gitattributes b/.gitattributes
index 2e7ea331dd489cd3af22ac9392dcde433044d176..871df209b5f17c48bef638b58c9bd54d66e27704 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -59,3 +59,4 @@ phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp2_freeze_eng_top8_expert
phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-400/tokenizer.json filter=lfs diff=lfs merge=lfs -text
phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-600/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/tokenizer.json filter=lfs diff=lfs merge=lfs -text
diff --git a/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/special_tokens_map.json b/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..ac23c0aaa2434523c494330aeb79c58395378103
--- /dev/null
+++ b/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/special_tokens_map.json
@@ -0,0 +1,31 @@
+{
+ "additional_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "eos_token": {
+ "content": "<|im_end|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ "pad_token": {
+ "content": "<|endoftext|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ }
+}
diff --git a/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/tokenizer.json b/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..cd71f61a15a522601badb3dc960d800d9cb3766c
--- /dev/null
+++ b/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
+size 11422654
diff --git a/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/tokenizer_config.json b/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..ddaf69808214a44fdd26d3785b66c1367c78277a
--- /dev/null
+++ b/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/tokenizer_config.json
@@ -0,0 +1,239 @@
+{
+ "add_bos_token": false,
+ "add_prefix_space": false,
+ "added_tokens_decoder": {
+ "151643": {
+ "content": "<|endoftext|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151644": {
+ "content": "<|im_start|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151645": {
+ "content": "<|im_end|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151646": {
+ "content": "<|object_ref_start|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151647": {
+ "content": "<|object_ref_end|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151648": {
+ "content": "<|box_start|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151649": {
+ "content": "<|box_end|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151650": {
+ "content": "<|quad_start|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151651": {
+ "content": "<|quad_end|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151652": {
+ "content": "<|vision_start|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151653": {
+ "content": "<|vision_end|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151654": {
+ "content": "<|vision_pad|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151655": {
+ "content": "<|image_pad|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151656": {
+ "content": "<|video_pad|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": true
+ },
+ "151657": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ },
+ "151658": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ },
+ "151659": {
+ "content": "<|fim_prefix|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ },
+ "151660": {
+ "content": "<|fim_middle|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ },
+ "151661": {
+ "content": "<|fim_suffix|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ },
+ "151662": {
+ "content": "<|fim_pad|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ },
+ "151663": {
+ "content": "<|repo_name|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ },
+ "151664": {
+ "content": "<|file_sep|>",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ },
+ "151665": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ },
+ "151666": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ },
+ "151667": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ },
+ "151668": {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false,
+ "special": false
+ }
+ },
+ "additional_special_tokens": [
+ "<|im_start|>",
+ "<|im_end|>",
+ "<|object_ref_start|>",
+ "<|object_ref_end|>",
+ "<|box_start|>",
+ "<|box_end|>",
+ "<|quad_start|>",
+ "<|quad_end|>",
+ "<|vision_start|>",
+ "<|vision_end|>",
+ "<|vision_pad|>",
+ "<|image_pad|>",
+ "<|video_pad|>"
+ ],
+ "bos_token": null,
+ "clean_up_tokenization_spaces": false,
+ "eos_token": "<|im_end|>",
+ "errors": "replace",
+ "extra_special_tokens": {},
+ "model_max_length": 131072,
+ "pad_token": "<|endoftext|>",
+ "split_special_tokens": false,
+ "tokenizer_class": "Qwen2Tokenizer",
+ "unk_token": null
+}
diff --git a/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/trainer_state.json b/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..9bab5f280b39153a71fa6ea38edcc2859a5aa771
--- /dev/null
+++ b/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/trainer_state.json
@@ -0,0 +1,4934 @@
+{
+ "best_global_step": null,
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 0.896,
+ "eval_steps": 500,
+ "global_step": 700,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.00128,
+ "grad_norm": 0.598645806312561,
+ "learning_rate": 0.0,
+ "loss": 3.5805,
+ "step": 1
+ },
+ {
+ "epoch": 0.00256,
+ "grad_norm": 0.6911924481391907,
+ "learning_rate": 2.5000000000000004e-07,
+ "loss": 3.764,
+ "step": 2
+ },
+ {
+ "epoch": 0.00384,
+ "grad_norm": 0.6410157084465027,
+ "learning_rate": 5.000000000000001e-07,
+ "loss": 3.5077,
+ "step": 3
+ },
+ {
+ "epoch": 0.00512,
+ "grad_norm": 0.5663634538650513,
+ "learning_rate": 7.5e-07,
+ "loss": 3.4404,
+ "step": 4
+ },
+ {
+ "epoch": 0.0064,
+ "grad_norm": 0.6647990942001343,
+ "learning_rate": 1.0000000000000002e-06,
+ "loss": 3.5211,
+ "step": 5
+ },
+ {
+ "epoch": 0.00768,
+ "grad_norm": 0.6807879209518433,
+ "learning_rate": 1.25e-06,
+ "loss": 3.5386,
+ "step": 6
+ },
+ {
+ "epoch": 0.00896,
+ "grad_norm": 0.6953790783882141,
+ "learning_rate": 1.5e-06,
+ "loss": 3.7508,
+ "step": 7
+ },
+ {
+ "epoch": 0.01024,
+ "grad_norm": 0.627304196357727,
+ "learning_rate": 1.75e-06,
+ "loss": 3.6983,
+ "step": 8
+ },
+ {
+ "epoch": 0.01152,
+ "grad_norm": 0.6061170697212219,
+ "learning_rate": 2.0000000000000003e-06,
+ "loss": 3.4084,
+ "step": 9
+ },
+ {
+ "epoch": 0.0128,
+ "grad_norm": 0.7157179713249207,
+ "learning_rate": 2.25e-06,
+ "loss": 3.7005,
+ "step": 10
+ },
+ {
+ "epoch": 0.01408,
+ "grad_norm": 0.6323665976524353,
+ "learning_rate": 2.5e-06,
+ "loss": 3.6551,
+ "step": 11
+ },
+ {
+ "epoch": 0.01536,
+ "grad_norm": 0.6628862023353577,
+ "learning_rate": 2.7500000000000004e-06,
+ "loss": 3.5893,
+ "step": 12
+ },
+ {
+ "epoch": 0.01664,
+ "grad_norm": 0.7357158064842224,
+ "learning_rate": 3e-06,
+ "loss": 3.7051,
+ "step": 13
+ },
+ {
+ "epoch": 0.01792,
+ "grad_norm": 0.5685714483261108,
+ "learning_rate": 3.2500000000000002e-06,
+ "loss": 3.4995,
+ "step": 14
+ },
+ {
+ "epoch": 0.0192,
+ "grad_norm": 0.6640226244926453,
+ "learning_rate": 3.5e-06,
+ "loss": 3.6901,
+ "step": 15
+ },
+ {
+ "epoch": 0.02048,
+ "grad_norm": 0.6068924069404602,
+ "learning_rate": 3.7500000000000005e-06,
+ "loss": 3.5341,
+ "step": 16
+ },
+ {
+ "epoch": 0.02176,
+ "grad_norm": 0.7769860029220581,
+ "learning_rate": 4.000000000000001e-06,
+ "loss": 3.6584,
+ "step": 17
+ },
+ {
+ "epoch": 0.02304,
+ "grad_norm": 0.6306919455528259,
+ "learning_rate": 4.25e-06,
+ "loss": 3.5221,
+ "step": 18
+ },
+ {
+ "epoch": 0.02432,
+ "grad_norm": 0.6686413288116455,
+ "learning_rate": 4.5e-06,
+ "loss": 3.6414,
+ "step": 19
+ },
+ {
+ "epoch": 0.0256,
+ "grad_norm": 0.5317572951316833,
+ "learning_rate": 4.75e-06,
+ "loss": 3.5669,
+ "step": 20
+ },
+ {
+ "epoch": 0.02688,
+ "grad_norm": 0.5724539756774902,
+ "learning_rate": 5e-06,
+ "loss": 3.6364,
+ "step": 21
+ },
+ {
+ "epoch": 0.02816,
+ "grad_norm": 0.584823489189148,
+ "learning_rate": 5.2500000000000006e-06,
+ "loss": 3.5592,
+ "step": 22
+ },
+ {
+ "epoch": 0.02944,
+ "grad_norm": 0.5049268007278442,
+ "learning_rate": 5.500000000000001e-06,
+ "loss": 3.5221,
+ "step": 23
+ },
+ {
+ "epoch": 0.03072,
+ "grad_norm": 0.584625244140625,
+ "learning_rate": 5.75e-06,
+ "loss": 3.6199,
+ "step": 24
+ },
+ {
+ "epoch": 0.032,
+ "grad_norm": 0.5325256586074829,
+ "learning_rate": 6e-06,
+ "loss": 3.5817,
+ "step": 25
+ },
+ {
+ "epoch": 0.03328,
+ "grad_norm": 0.5040303468704224,
+ "learning_rate": 6.25e-06,
+ "loss": 3.454,
+ "step": 26
+ },
+ {
+ "epoch": 0.03456,
+ "grad_norm": 0.5283604264259338,
+ "learning_rate": 6.5000000000000004e-06,
+ "loss": 3.6028,
+ "step": 27
+ },
+ {
+ "epoch": 0.03584,
+ "grad_norm": 0.5351476073265076,
+ "learning_rate": 6.750000000000001e-06,
+ "loss": 3.5297,
+ "step": 28
+ },
+ {
+ "epoch": 0.03712,
+ "grad_norm": 0.5750259757041931,
+ "learning_rate": 7e-06,
+ "loss": 3.7047,
+ "step": 29
+ },
+ {
+ "epoch": 0.0384,
+ "grad_norm": 0.5229286551475525,
+ "learning_rate": 7.25e-06,
+ "loss": 3.503,
+ "step": 30
+ },
+ {
+ "epoch": 0.03968,
+ "grad_norm": 0.641511082649231,
+ "learning_rate": 7.500000000000001e-06,
+ "loss": 3.5712,
+ "step": 31
+ },
+ {
+ "epoch": 0.04096,
+ "grad_norm": 0.582783043384552,
+ "learning_rate": 7.75e-06,
+ "loss": 3.7143,
+ "step": 32
+ },
+ {
+ "epoch": 0.04224,
+ "grad_norm": 0.4316658079624176,
+ "learning_rate": 8.000000000000001e-06,
+ "loss": 3.4999,
+ "step": 33
+ },
+ {
+ "epoch": 0.04352,
+ "grad_norm": 0.5524796843528748,
+ "learning_rate": 8.25e-06,
+ "loss": 3.4514,
+ "step": 34
+ },
+ {
+ "epoch": 0.0448,
+ "grad_norm": 0.5018333792686462,
+ "learning_rate": 8.5e-06,
+ "loss": 3.3786,
+ "step": 35
+ },
+ {
+ "epoch": 0.04608,
+ "grad_norm": 0.45580804347991943,
+ "learning_rate": 8.750000000000001e-06,
+ "loss": 3.5256,
+ "step": 36
+ },
+ {
+ "epoch": 0.04736,
+ "grad_norm": 0.42892250418663025,
+ "learning_rate": 9e-06,
+ "loss": 3.3754,
+ "step": 37
+ },
+ {
+ "epoch": 0.04864,
+ "grad_norm": 0.5623100399971008,
+ "learning_rate": 9.250000000000001e-06,
+ "loss": 3.8133,
+ "step": 38
+ },
+ {
+ "epoch": 0.04992,
+ "grad_norm": 0.4689437747001648,
+ "learning_rate": 9.5e-06,
+ "loss": 3.4601,
+ "step": 39
+ },
+ {
+ "epoch": 0.0512,
+ "grad_norm": 0.42971721291542053,
+ "learning_rate": 9.75e-06,
+ "loss": 3.3462,
+ "step": 40
+ },
+ {
+ "epoch": 0.05248,
+ "grad_norm": 0.5345175862312317,
+ "learning_rate": 1e-05,
+ "loss": 3.6253,
+ "step": 41
+ },
+ {
+ "epoch": 0.05376,
+ "grad_norm": 0.44895467162132263,
+ "learning_rate": 9.999955184185413e-06,
+ "loss": 3.3767,
+ "step": 42
+ },
+ {
+ "epoch": 0.05504,
+ "grad_norm": 0.5316817164421082,
+ "learning_rate": 9.99982073754503e-06,
+ "loss": 3.4518,
+ "step": 43
+ },
+ {
+ "epoch": 0.05632,
+ "grad_norm": 0.48764774203300476,
+ "learning_rate": 9.999596662488987e-06,
+ "loss": 3.5663,
+ "step": 44
+ },
+ {
+ "epoch": 0.0576,
+ "grad_norm": 0.492572546005249,
+ "learning_rate": 9.999282963034126e-06,
+ "loss": 3.4972,
+ "step": 45
+ },
+ {
+ "epoch": 0.05888,
+ "grad_norm": 0.5189987421035767,
+ "learning_rate": 9.998879644803927e-06,
+ "loss": 3.646,
+ "step": 46
+ },
+ {
+ "epoch": 0.06016,
+ "grad_norm": 0.44764602184295654,
+ "learning_rate": 9.998386715028403e-06,
+ "loss": 3.5809,
+ "step": 47
+ },
+ {
+ "epoch": 0.06144,
+ "grad_norm": 0.4388003647327423,
+ "learning_rate": 9.997804182543973e-06,
+ "loss": 3.5708,
+ "step": 48
+ },
+ {
+ "epoch": 0.06272,
+ "grad_norm": 0.48791566491127014,
+ "learning_rate": 9.997132057793306e-06,
+ "loss": 3.4673,
+ "step": 49
+ },
+ {
+ "epoch": 0.064,
+ "grad_norm": 0.5207979083061218,
+ "learning_rate": 9.996370352825126e-06,
+ "loss": 3.5511,
+ "step": 50
+ },
+ {
+ "epoch": 0.06528,
+ "grad_norm": 0.4717053174972534,
+ "learning_rate": 9.995519081294009e-06,
+ "loss": 3.4461,
+ "step": 51
+ },
+ {
+ "epoch": 0.06656,
+ "grad_norm": 0.45021846890449524,
+ "learning_rate": 9.994578258460124e-06,
+ "loss": 3.2759,
+ "step": 52
+ },
+ {
+ "epoch": 0.06784,
+ "grad_norm": 0.4491179287433624,
+ "learning_rate": 9.993547901188966e-06,
+ "loss": 3.5995,
+ "step": 53
+ },
+ {
+ "epoch": 0.06912,
+ "grad_norm": 0.5132831335067749,
+ "learning_rate": 9.992428027951055e-06,
+ "loss": 3.592,
+ "step": 54
+ },
+ {
+ "epoch": 0.0704,
+ "grad_norm": 0.5232840180397034,
+ "learning_rate": 9.991218658821609e-06,
+ "loss": 3.4879,
+ "step": 55
+ },
+ {
+ "epoch": 0.07168,
+ "grad_norm": 0.47307732701301575,
+ "learning_rate": 9.989919815480166e-06,
+ "loss": 3.4639,
+ "step": 56
+ },
+ {
+ "epoch": 0.07296,
+ "grad_norm": 0.4625212550163269,
+ "learning_rate": 9.988531521210219e-06,
+ "loss": 3.5149,
+ "step": 57
+ },
+ {
+ "epoch": 0.07424,
+ "grad_norm": 0.5021692514419556,
+ "learning_rate": 9.98705380089878e-06,
+ "loss": 3.5431,
+ "step": 58
+ },
+ {
+ "epoch": 0.07552,
+ "grad_norm": 0.42638182640075684,
+ "learning_rate": 9.98548668103595e-06,
+ "loss": 3.4137,
+ "step": 59
+ },
+ {
+ "epoch": 0.0768,
+ "grad_norm": 0.5041897892951965,
+ "learning_rate": 9.983830189714428e-06,
+ "loss": 3.5207,
+ "step": 60
+ },
+ {
+ "epoch": 0.07808,
+ "grad_norm": 0.4349015951156616,
+ "learning_rate": 9.982084356629014e-06,
+ "loss": 3.6452,
+ "step": 61
+ },
+ {
+ "epoch": 0.07936,
+ "grad_norm": 0.39008477330207825,
+ "learning_rate": 9.980249213076085e-06,
+ "loss": 3.4108,
+ "step": 62
+ },
+ {
+ "epoch": 0.08064,
+ "grad_norm": 0.4263695776462555,
+ "learning_rate": 9.97832479195302e-06,
+ "loss": 3.2657,
+ "step": 63
+ },
+ {
+ "epoch": 0.08192,
+ "grad_norm": 0.4438668191432953,
+ "learning_rate": 9.97631112775762e-06,
+ "loss": 3.4026,
+ "step": 64
+ },
+ {
+ "epoch": 0.0832,
+ "grad_norm": 0.5348494052886963,
+ "learning_rate": 9.974208256587488e-06,
+ "loss": 3.4892,
+ "step": 65
+ },
+ {
+ "epoch": 0.08448,
+ "grad_norm": 0.4377213716506958,
+ "learning_rate": 9.972016216139372e-06,
+ "loss": 3.333,
+ "step": 66
+ },
+ {
+ "epoch": 0.08576,
+ "grad_norm": 0.4178816080093384,
+ "learning_rate": 9.969735045708506e-06,
+ "loss": 3.287,
+ "step": 67
+ },
+ {
+ "epoch": 0.08704,
+ "grad_norm": 0.4345816969871521,
+ "learning_rate": 9.967364786187896e-06,
+ "loss": 3.589,
+ "step": 68
+ },
+ {
+ "epoch": 0.08832,
+ "grad_norm": 0.5251936316490173,
+ "learning_rate": 9.964905480067585e-06,
+ "loss": 3.4842,
+ "step": 69
+ },
+ {
+ "epoch": 0.0896,
+ "grad_norm": 0.4248199760913849,
+ "learning_rate": 9.962357171433895e-06,
+ "loss": 3.4109,
+ "step": 70
+ },
+ {
+ "epoch": 0.09088,
+ "grad_norm": 0.42213934659957886,
+ "learning_rate": 9.959719905968642e-06,
+ "loss": 3.2678,
+ "step": 71
+ },
+ {
+ "epoch": 0.09216,
+ "grad_norm": 0.42829281091690063,
+ "learning_rate": 9.956993730948299e-06,
+ "loss": 3.3227,
+ "step": 72
+ },
+ {
+ "epoch": 0.09344,
+ "grad_norm": 0.45543426275253296,
+ "learning_rate": 9.954178695243171e-06,
+ "loss": 3.3721,
+ "step": 73
+ },
+ {
+ "epoch": 0.09472,
+ "grad_norm": 0.40355560183525085,
+ "learning_rate": 9.951274849316505e-06,
+ "loss": 3.3787,
+ "step": 74
+ },
+ {
+ "epoch": 0.096,
+ "grad_norm": 0.4370484948158264,
+ "learning_rate": 9.94828224522359e-06,
+ "loss": 3.385,
+ "step": 75
+ },
+ {
+ "epoch": 0.09728,
+ "grad_norm": 0.4785033166408539,
+ "learning_rate": 9.945200936610821e-06,
+ "loss": 3.3991,
+ "step": 76
+ },
+ {
+ "epoch": 0.09856,
+ "grad_norm": 0.4387708008289337,
+ "learning_rate": 9.94203097871474e-06,
+ "loss": 3.324,
+ "step": 77
+ },
+ {
+ "epoch": 0.09984,
+ "grad_norm": 0.4685862362384796,
+ "learning_rate": 9.938772428361047e-06,
+ "loss": 3.5072,
+ "step": 78
+ },
+ {
+ "epoch": 0.10112,
+ "grad_norm": 0.4430760145187378,
+ "learning_rate": 9.935425343963574e-06,
+ "loss": 3.3012,
+ "step": 79
+ },
+ {
+ "epoch": 0.1024,
+ "grad_norm": 0.4138337969779968,
+ "learning_rate": 9.93198978552325e-06,
+ "loss": 3.3926,
+ "step": 80
+ },
+ {
+ "epoch": 0.10368,
+ "grad_norm": 0.42039164900779724,
+ "learning_rate": 9.928465814627016e-06,
+ "loss": 3.4073,
+ "step": 81
+ },
+ {
+ "epoch": 0.10496,
+ "grad_norm": 0.44093188643455505,
+ "learning_rate": 9.924853494446716e-06,
+ "loss": 3.237,
+ "step": 82
+ },
+ {
+ "epoch": 0.10624,
+ "grad_norm": 0.4713192582130432,
+ "learning_rate": 9.921152889737985e-06,
+ "loss": 3.5427,
+ "step": 83
+ },
+ {
+ "epoch": 0.10752,
+ "grad_norm": 0.3913152813911438,
+ "learning_rate": 9.917364066839066e-06,
+ "loss": 3.3923,
+ "step": 84
+ },
+ {
+ "epoch": 0.1088,
+ "grad_norm": 0.39092734456062317,
+ "learning_rate": 9.913487093669633e-06,
+ "loss": 3.458,
+ "step": 85
+ },
+ {
+ "epoch": 0.11008,
+ "grad_norm": 0.4117678999900818,
+ "learning_rate": 9.909522039729571e-06,
+ "loss": 3.2567,
+ "step": 86
+ },
+ {
+ "epoch": 0.11136,
+ "grad_norm": 0.35831838846206665,
+ "learning_rate": 9.90546897609773e-06,
+ "loss": 3.535,
+ "step": 87
+ },
+ {
+ "epoch": 0.11264,
+ "grad_norm": 0.41918617486953735,
+ "learning_rate": 9.901327975430645e-06,
+ "loss": 3.4221,
+ "step": 88
+ },
+ {
+ "epoch": 0.11392,
+ "grad_norm": 0.49071717262268066,
+ "learning_rate": 9.897099111961248e-06,
+ "loss": 3.4202,
+ "step": 89
+ },
+ {
+ "epoch": 0.1152,
+ "grad_norm": 0.4747026562690735,
+ "learning_rate": 9.892782461497521e-06,
+ "loss": 3.4047,
+ "step": 90
+ },
+ {
+ "epoch": 0.11648,
+ "grad_norm": 0.3721744418144226,
+ "learning_rate": 9.888378101421148e-06,
+ "loss": 3.375,
+ "step": 91
+ },
+ {
+ "epoch": 0.11776,
+ "grad_norm": 0.44745302200317383,
+ "learning_rate": 9.88388611068612e-06,
+ "loss": 3.4231,
+ "step": 92
+ },
+ {
+ "epoch": 0.11904,
+ "grad_norm": 0.45900216698646545,
+ "learning_rate": 9.879306569817331e-06,
+ "loss": 3.2258,
+ "step": 93
+ },
+ {
+ "epoch": 0.12032,
+ "grad_norm": 0.4253723621368408,
+ "learning_rate": 9.874639560909118e-06,
+ "loss": 3.3453,
+ "step": 94
+ },
+ {
+ "epoch": 0.1216,
+ "grad_norm": 0.49339723587036133,
+ "learning_rate": 9.869885167623808e-06,
+ "loss": 3.3342,
+ "step": 95
+ },
+ {
+ "epoch": 0.12288,
+ "grad_norm": 0.4880124628543854,
+ "learning_rate": 9.865043475190202e-06,
+ "loss": 3.6539,
+ "step": 96
+ },
+ {
+ "epoch": 0.12416,
+ "grad_norm": 0.5158582329750061,
+ "learning_rate": 9.860114570402055e-06,
+ "loss": 3.3565,
+ "step": 97
+ },
+ {
+ "epoch": 0.12544,
+ "grad_norm": 0.41930368542671204,
+ "learning_rate": 9.855098541616522e-06,
+ "loss": 3.4151,
+ "step": 98
+ },
+ {
+ "epoch": 0.12672,
+ "grad_norm": 0.5347989797592163,
+ "learning_rate": 9.849995478752568e-06,
+ "loss": 3.5308,
+ "step": 99
+ },
+ {
+ "epoch": 0.128,
+ "grad_norm": 0.3942601680755615,
+ "learning_rate": 9.844805473289363e-06,
+ "loss": 3.2385,
+ "step": 100
+ },
+ {
+ "epoch": 0.12928,
+ "grad_norm": 0.5181060433387756,
+ "learning_rate": 9.839528618264633e-06,
+ "loss": 3.547,
+ "step": 101
+ },
+ {
+ "epoch": 0.13056,
+ "grad_norm": 0.4099213480949402,
+ "learning_rate": 9.834165008273002e-06,
+ "loss": 3.5101,
+ "step": 102
+ },
+ {
+ "epoch": 0.13184,
+ "grad_norm": 0.48070698976516724,
+ "learning_rate": 9.82871473946429e-06,
+ "loss": 3.4874,
+ "step": 103
+ },
+ {
+ "epoch": 0.13312,
+ "grad_norm": 0.4800182580947876,
+ "learning_rate": 9.823177909541795e-06,
+ "loss": 3.3875,
+ "step": 104
+ },
+ {
+ "epoch": 0.1344,
+ "grad_norm": 0.4540846645832062,
+ "learning_rate": 9.817554617760529e-06,
+ "loss": 3.3686,
+ "step": 105
+ },
+ {
+ "epoch": 0.13568,
+ "grad_norm": 0.4731300175189972,
+ "learning_rate": 9.811844964925455e-06,
+ "loss": 3.352,
+ "step": 106
+ },
+ {
+ "epoch": 0.13696,
+ "grad_norm": 0.4138098955154419,
+ "learning_rate": 9.80604905338967e-06,
+ "loss": 3.309,
+ "step": 107
+ },
+ {
+ "epoch": 0.13824,
+ "grad_norm": 0.46461138129234314,
+ "learning_rate": 9.800166987052572e-06,
+ "loss": 3.6354,
+ "step": 108
+ },
+ {
+ "epoch": 0.13952,
+ "grad_norm": 0.44443202018737793,
+ "learning_rate": 9.794198871358002e-06,
+ "loss": 3.5193,
+ "step": 109
+ },
+ {
+ "epoch": 0.1408,
+ "grad_norm": 0.37842369079589844,
+ "learning_rate": 9.788144813292343e-06,
+ "loss": 3.3132,
+ "step": 110
+ },
+ {
+ "epoch": 0.14208,
+ "grad_norm": 0.4487096965312958,
+ "learning_rate": 9.782004921382612e-06,
+ "loss": 3.2657,
+ "step": 111
+ },
+ {
+ "epoch": 0.14336,
+ "grad_norm": 0.38858550786972046,
+ "learning_rate": 9.775779305694516e-06,
+ "loss": 3.1869,
+ "step": 112
+ },
+ {
+ "epoch": 0.14464,
+ "grad_norm": 0.4216855466365814,
+ "learning_rate": 9.769468077830467e-06,
+ "loss": 3.2485,
+ "step": 113
+ },
+ {
+ "epoch": 0.14592,
+ "grad_norm": 0.3817397356033325,
+ "learning_rate": 9.763071350927594e-06,
+ "loss": 3.2925,
+ "step": 114
+ },
+ {
+ "epoch": 0.1472,
+ "grad_norm": 0.5073004364967346,
+ "learning_rate": 9.756589239655705e-06,
+ "loss": 3.4854,
+ "step": 115
+ },
+ {
+ "epoch": 0.14848,
+ "grad_norm": 0.3787033259868622,
+ "learning_rate": 9.750021860215242e-06,
+ "loss": 3.245,
+ "step": 116
+ },
+ {
+ "epoch": 0.14976,
+ "grad_norm": 0.4737580418586731,
+ "learning_rate": 9.743369330335186e-06,
+ "loss": 3.4794,
+ "step": 117
+ },
+ {
+ "epoch": 0.15104,
+ "grad_norm": 0.47760501503944397,
+ "learning_rate": 9.736631769270958e-06,
+ "loss": 3.3043,
+ "step": 118
+ },
+ {
+ "epoch": 0.15232,
+ "grad_norm": 0.42264020442962646,
+ "learning_rate": 9.72980929780227e-06,
+ "loss": 3.3083,
+ "step": 119
+ },
+ {
+ "epoch": 0.1536,
+ "grad_norm": 0.38285794854164124,
+ "learning_rate": 9.722902038230973e-06,
+ "loss": 3.4245,
+ "step": 120
+ },
+ {
+ "epoch": 0.15488,
+ "grad_norm": 0.43436744809150696,
+ "learning_rate": 9.715910114378845e-06,
+ "loss": 3.3141,
+ "step": 121
+ },
+ {
+ "epoch": 0.15616,
+ "grad_norm": 0.3926229476928711,
+ "learning_rate": 9.708833651585402e-06,
+ "loss": 3.4586,
+ "step": 122
+ },
+ {
+ "epoch": 0.15744,
+ "grad_norm": 0.4426027238368988,
+ "learning_rate": 9.70167277670561e-06,
+ "loss": 3.4179,
+ "step": 123
+ },
+ {
+ "epoch": 0.15872,
+ "grad_norm": 0.39944279193878174,
+ "learning_rate": 9.694427618107655e-06,
+ "loss": 3.3317,
+ "step": 124
+ },
+ {
+ "epoch": 0.16,
+ "grad_norm": 0.3857992887496948,
+ "learning_rate": 9.687098305670606e-06,
+ "loss": 3.3196,
+ "step": 125
+ },
+ {
+ "epoch": 0.16128,
+ "grad_norm": 0.42690324783325195,
+ "learning_rate": 9.679684970782105e-06,
+ "loss": 3.2685,
+ "step": 126
+ },
+ {
+ "epoch": 0.16256,
+ "grad_norm": 0.3967202305793762,
+ "learning_rate": 9.672187746336012e-06,
+ "loss": 3.1782,
+ "step": 127
+ },
+ {
+ "epoch": 0.16384,
+ "grad_norm": 0.4174215495586395,
+ "learning_rate": 9.664606766730014e-06,
+ "loss": 3.3738,
+ "step": 128
+ },
+ {
+ "epoch": 0.16512,
+ "grad_norm": 0.42718198895454407,
+ "learning_rate": 9.656942167863221e-06,
+ "loss": 3.2091,
+ "step": 129
+ },
+ {
+ "epoch": 0.1664,
+ "grad_norm": 0.45092251896858215,
+ "learning_rate": 9.649194087133731e-06,
+ "loss": 3.5248,
+ "step": 130
+ },
+ {
+ "epoch": 0.16768,
+ "grad_norm": 0.3800186216831207,
+ "learning_rate": 9.641362663436162e-06,
+ "loss": 3.3337,
+ "step": 131
+ },
+ {
+ "epoch": 0.16896,
+ "grad_norm": 0.4664546847343445,
+ "learning_rate": 9.633448037159167e-06,
+ "loss": 3.482,
+ "step": 132
+ },
+ {
+ "epoch": 0.17024,
+ "grad_norm": 0.3810117840766907,
+ "learning_rate": 9.625450350182919e-06,
+ "loss": 3.3122,
+ "step": 133
+ },
+ {
+ "epoch": 0.17152,
+ "grad_norm": 0.43419206142425537,
+ "learning_rate": 9.617369745876555e-06,
+ "loss": 3.398,
+ "step": 134
+ },
+ {
+ "epoch": 0.1728,
+ "grad_norm": 0.4321575164794922,
+ "learning_rate": 9.609206369095626e-06,
+ "loss": 3.5053,
+ "step": 135
+ },
+ {
+ "epoch": 0.17408,
+ "grad_norm": 0.4542195200920105,
+ "learning_rate": 9.60096036617948e-06,
+ "loss": 3.3708,
+ "step": 136
+ },
+ {
+ "epoch": 0.17536,
+ "grad_norm": 0.4479266107082367,
+ "learning_rate": 9.592631884948654e-06,
+ "loss": 3.3495,
+ "step": 137
+ },
+ {
+ "epoch": 0.17664,
+ "grad_norm": 0.41415154933929443,
+ "learning_rate": 9.584221074702219e-06,
+ "loss": 3.342,
+ "step": 138
+ },
+ {
+ "epoch": 0.17792,
+ "grad_norm": 0.40505874156951904,
+ "learning_rate": 9.575728086215093e-06,
+ "loss": 3.3447,
+ "step": 139
+ },
+ {
+ "epoch": 0.1792,
+ "grad_norm": 0.3883638381958008,
+ "learning_rate": 9.567153071735361e-06,
+ "loss": 3.3698,
+ "step": 140
+ },
+ {
+ "epoch": 0.18048,
+ "grad_norm": 0.4226126968860626,
+ "learning_rate": 9.558496184981525e-06,
+ "loss": 3.3994,
+ "step": 141
+ },
+ {
+ "epoch": 0.18176,
+ "grad_norm": 0.5270792841911316,
+ "learning_rate": 9.549757581139758e-06,
+ "loss": 3.4491,
+ "step": 142
+ },
+ {
+ "epoch": 0.18304,
+ "grad_norm": 0.41094139218330383,
+ "learning_rate": 9.540937416861117e-06,
+ "loss": 3.3035,
+ "step": 143
+ },
+ {
+ "epoch": 0.18432,
+ "grad_norm": 0.42608124017715454,
+ "learning_rate": 9.532035850258745e-06,
+ "loss": 3.3987,
+ "step": 144
+ },
+ {
+ "epoch": 0.1856,
+ "grad_norm": 0.41302427649497986,
+ "learning_rate": 9.523053040905024e-06,
+ "loss": 3.3014,
+ "step": 145
+ },
+ {
+ "epoch": 0.18688,
+ "grad_norm": 0.42077869176864624,
+ "learning_rate": 9.513989149828718e-06,
+ "loss": 3.3705,
+ "step": 146
+ },
+ {
+ "epoch": 0.18816,
+ "grad_norm": 0.3903754651546478,
+ "learning_rate": 9.504844339512096e-06,
+ "loss": 3.2359,
+ "step": 147
+ },
+ {
+ "epoch": 0.18944,
+ "grad_norm": 0.3968545198440552,
+ "learning_rate": 9.495618773888007e-06,
+ "loss": 3.4024,
+ "step": 148
+ },
+ {
+ "epoch": 0.19072,
+ "grad_norm": 0.3992414176464081,
+ "learning_rate": 9.486312618336944e-06,
+ "loss": 3.2116,
+ "step": 149
+ },
+ {
+ "epoch": 0.192,
+ "grad_norm": 0.41804948449134827,
+ "learning_rate": 9.476926039684085e-06,
+ "loss": 3.4265,
+ "step": 150
+ },
+ {
+ "epoch": 0.19328,
+ "grad_norm": 0.43339264392852783,
+ "learning_rate": 9.467459206196298e-06,
+ "loss": 3.3954,
+ "step": 151
+ },
+ {
+ "epoch": 0.19456,
+ "grad_norm": 0.4704466164112091,
+ "learning_rate": 9.457912287579125e-06,
+ "loss": 3.4662,
+ "step": 152
+ },
+ {
+ "epoch": 0.19584,
+ "grad_norm": 0.438205361366272,
+ "learning_rate": 9.448285454973739e-06,
+ "loss": 3.5094,
+ "step": 153
+ },
+ {
+ "epoch": 0.19712,
+ "grad_norm": 0.44826585054397583,
+ "learning_rate": 9.438578880953875e-06,
+ "loss": 3.1943,
+ "step": 154
+ },
+ {
+ "epoch": 0.1984,
+ "grad_norm": 0.46182042360305786,
+ "learning_rate": 9.428792739522747e-06,
+ "loss": 3.4228,
+ "step": 155
+ },
+ {
+ "epoch": 0.19968,
+ "grad_norm": 0.43598097562789917,
+ "learning_rate": 9.418927206109914e-06,
+ "loss": 3.3927,
+ "step": 156
+ },
+ {
+ "epoch": 0.20096,
+ "grad_norm": 0.43278130888938904,
+ "learning_rate": 9.408982457568138e-06,
+ "loss": 3.4055,
+ "step": 157
+ },
+ {
+ "epoch": 0.20224,
+ "grad_norm": 0.4054141938686371,
+ "learning_rate": 9.398958672170227e-06,
+ "loss": 3.2403,
+ "step": 158
+ },
+ {
+ "epoch": 0.20352,
+ "grad_norm": 0.40924009680747986,
+ "learning_rate": 9.388856029605821e-06,
+ "loss": 3.0961,
+ "step": 159
+ },
+ {
+ "epoch": 0.2048,
+ "grad_norm": 0.38527265191078186,
+ "learning_rate": 9.378674710978185e-06,
+ "loss": 3.2648,
+ "step": 160
+ },
+ {
+ "epoch": 0.20608,
+ "grad_norm": 0.40751633048057556,
+ "learning_rate": 9.368414898800951e-06,
+ "loss": 3.3758,
+ "step": 161
+ },
+ {
+ "epoch": 0.20736,
+ "grad_norm": 0.406125009059906,
+ "learning_rate": 9.358076776994858e-06,
+ "loss": 3.1606,
+ "step": 162
+ },
+ {
+ "epoch": 0.20864,
+ "grad_norm": 0.5410938858985901,
+ "learning_rate": 9.347660530884443e-06,
+ "loss": 3.522,
+ "step": 163
+ },
+ {
+ "epoch": 0.20992,
+ "grad_norm": 0.4553371071815491,
+ "learning_rate": 9.337166347194732e-06,
+ "loss": 3.2917,
+ "step": 164
+ },
+ {
+ "epoch": 0.2112,
+ "grad_norm": 0.4209158420562744,
+ "learning_rate": 9.326594414047877e-06,
+ "loss": 3.3555,
+ "step": 165
+ },
+ {
+ "epoch": 0.21248,
+ "grad_norm": 0.4474107027053833,
+ "learning_rate": 9.315944920959798e-06,
+ "loss": 3.3867,
+ "step": 166
+ },
+ {
+ "epoch": 0.21376,
+ "grad_norm": 0.447605699300766,
+ "learning_rate": 9.305218058836778e-06,
+ "loss": 3.4507,
+ "step": 167
+ },
+ {
+ "epoch": 0.21504,
+ "grad_norm": 0.4172501564025879,
+ "learning_rate": 9.294414019972043e-06,
+ "loss": 3.4235,
+ "step": 168
+ },
+ {
+ "epoch": 0.21632,
+ "grad_norm": 0.41198042035102844,
+ "learning_rate": 9.283532998042314e-06,
+ "loss": 3.1672,
+ "step": 169
+ },
+ {
+ "epoch": 0.2176,
+ "grad_norm": 0.41195234656333923,
+ "learning_rate": 9.272575188104335e-06,
+ "loss": 3.2836,
+ "step": 170
+ },
+ {
+ "epoch": 0.21888,
+ "grad_norm": 0.5480180978775024,
+ "learning_rate": 9.261540786591377e-06,
+ "loss": 3.2693,
+ "step": 171
+ },
+ {
+ "epoch": 0.22016,
+ "grad_norm": 0.4085384011268616,
+ "learning_rate": 9.250429991309719e-06,
+ "loss": 3.4335,
+ "step": 172
+ },
+ {
+ "epoch": 0.22144,
+ "grad_norm": 0.4334932267665863,
+ "learning_rate": 9.239243001435095e-06,
+ "loss": 3.2146,
+ "step": 173
+ },
+ {
+ "epoch": 0.22272,
+ "grad_norm": 0.5284118056297302,
+ "learning_rate": 9.22798001750913e-06,
+ "loss": 3.3151,
+ "step": 174
+ },
+ {
+ "epoch": 0.224,
+ "grad_norm": 0.4580453932285309,
+ "learning_rate": 9.216641241435748e-06,
+ "loss": 3.3858,
+ "step": 175
+ },
+ {
+ "epoch": 0.22528,
+ "grad_norm": 0.4157484471797943,
+ "learning_rate": 9.20522687647754e-06,
+ "loss": 3.1168,
+ "step": 176
+ },
+ {
+ "epoch": 0.22656,
+ "grad_norm": 0.4260411262512207,
+ "learning_rate": 9.193737127252132e-06,
+ "loss": 3.4157,
+ "step": 177
+ },
+ {
+ "epoch": 0.22784,
+ "grad_norm": 0.4258939325809479,
+ "learning_rate": 9.182172199728514e-06,
+ "loss": 3.3013,
+ "step": 178
+ },
+ {
+ "epoch": 0.22912,
+ "grad_norm": 0.4220966398715973,
+ "learning_rate": 9.170532301223343e-06,
+ "loss": 3.1636,
+ "step": 179
+ },
+ {
+ "epoch": 0.2304,
+ "grad_norm": 0.4540233910083771,
+ "learning_rate": 9.158817640397233e-06,
+ "loss": 3.1655,
+ "step": 180
+ },
+ {
+ "epoch": 0.23168,
+ "grad_norm": 0.46531087160110474,
+ "learning_rate": 9.14702842725101e-06,
+ "loss": 3.3218,
+ "step": 181
+ },
+ {
+ "epoch": 0.23296,
+ "grad_norm": 0.3994830250740051,
+ "learning_rate": 9.135164873121952e-06,
+ "loss": 3.3352,
+ "step": 182
+ },
+ {
+ "epoch": 0.23424,
+ "grad_norm": 0.38530516624450684,
+ "learning_rate": 9.123227190679995e-06,
+ "loss": 3.1435,
+ "step": 183
+ },
+ {
+ "epoch": 0.23552,
+ "grad_norm": 0.4449882209300995,
+ "learning_rate": 9.111215593923925e-06,
+ "loss": 3.1961,
+ "step": 184
+ },
+ {
+ "epoch": 0.2368,
+ "grad_norm": 0.4268646538257599,
+ "learning_rate": 9.099130298177538e-06,
+ "loss": 3.2294,
+ "step": 185
+ },
+ {
+ "epoch": 0.23808,
+ "grad_norm": 0.3653985261917114,
+ "learning_rate": 9.086971520085784e-06,
+ "loss": 3.1644,
+ "step": 186
+ },
+ {
+ "epoch": 0.23936,
+ "grad_norm": 0.43704408407211304,
+ "learning_rate": 9.07473947761088e-06,
+ "loss": 3.2679,
+ "step": 187
+ },
+ {
+ "epoch": 0.24064,
+ "grad_norm": 0.3928852677345276,
+ "learning_rate": 9.062434390028407e-06,
+ "loss": 3.2232,
+ "step": 188
+ },
+ {
+ "epoch": 0.24192,
+ "grad_norm": 0.41391831636428833,
+ "learning_rate": 9.050056477923373e-06,
+ "loss": 3.1693,
+ "step": 189
+ },
+ {
+ "epoch": 0.2432,
+ "grad_norm": 0.4765762984752655,
+ "learning_rate": 9.037605963186265e-06,
+ "loss": 3.3771,
+ "step": 190
+ },
+ {
+ "epoch": 0.24448,
+ "grad_norm": 0.45245447754859924,
+ "learning_rate": 9.025083069009065e-06,
+ "loss": 3.3205,
+ "step": 191
+ },
+ {
+ "epoch": 0.24576,
+ "grad_norm": 0.47386234998703003,
+ "learning_rate": 9.012488019881253e-06,
+ "loss": 3.3884,
+ "step": 192
+ },
+ {
+ "epoch": 0.24704,
+ "grad_norm": 0.4492136538028717,
+ "learning_rate": 8.999821041585788e-06,
+ "loss": 3.2707,
+ "step": 193
+ },
+ {
+ "epoch": 0.24832,
+ "grad_norm": 0.48031893372535706,
+ "learning_rate": 8.987082361195047e-06,
+ "loss": 3.3092,
+ "step": 194
+ },
+ {
+ "epoch": 0.2496,
+ "grad_norm": 0.43227964639663696,
+ "learning_rate": 8.974272207066767e-06,
+ "loss": 3.2356,
+ "step": 195
+ },
+ {
+ "epoch": 0.25088,
+ "grad_norm": 0.41915279626846313,
+ "learning_rate": 8.961390808839945e-06,
+ "loss": 3.1538,
+ "step": 196
+ },
+ {
+ "epoch": 0.25216,
+ "grad_norm": 0.4137171506881714,
+ "learning_rate": 8.948438397430721e-06,
+ "loss": 3.2144,
+ "step": 197
+ },
+ {
+ "epoch": 0.25344,
+ "grad_norm": 0.40594160556793213,
+ "learning_rate": 8.935415205028243e-06,
+ "loss": 3.3202,
+ "step": 198
+ },
+ {
+ "epoch": 0.25472,
+ "grad_norm": 0.4788910746574402,
+ "learning_rate": 8.922321465090504e-06,
+ "loss": 3.292,
+ "step": 199
+ },
+ {
+ "epoch": 0.256,
+ "grad_norm": 0.41766247153282166,
+ "learning_rate": 8.90915741234015e-06,
+ "loss": 3.2817,
+ "step": 200
+ },
+ {
+ "epoch": 0.25728,
+ "grad_norm": 0.4632660150527954,
+ "learning_rate": 8.89592328276028e-06,
+ "loss": 3.3346,
+ "step": 201
+ },
+ {
+ "epoch": 0.25856,
+ "grad_norm": 0.4152847230434418,
+ "learning_rate": 8.882619313590212e-06,
+ "loss": 3.2362,
+ "step": 202
+ },
+ {
+ "epoch": 0.25984,
+ "grad_norm": 0.43168312311172485,
+ "learning_rate": 8.869245743321235e-06,
+ "loss": 3.4433,
+ "step": 203
+ },
+ {
+ "epoch": 0.26112,
+ "grad_norm": 0.4464055299758911,
+ "learning_rate": 8.855802811692327e-06,
+ "loss": 3.3869,
+ "step": 204
+ },
+ {
+ "epoch": 0.2624,
+ "grad_norm": 0.410875529050827,
+ "learning_rate": 8.842290759685857e-06,
+ "loss": 3.2912,
+ "step": 205
+ },
+ {
+ "epoch": 0.26368,
+ "grad_norm": 0.46046632528305054,
+ "learning_rate": 8.828709829523277e-06,
+ "loss": 3.2882,
+ "step": 206
+ },
+ {
+ "epoch": 0.26496,
+ "grad_norm": 0.4462791085243225,
+ "learning_rate": 8.815060264660759e-06,
+ "loss": 3.2814,
+ "step": 207
+ },
+ {
+ "epoch": 0.26624,
+ "grad_norm": 0.3402050733566284,
+ "learning_rate": 8.80134230978486e-06,
+ "loss": 3.2469,
+ "step": 208
+ },
+ {
+ "epoch": 0.26752,
+ "grad_norm": 0.4195862114429474,
+ "learning_rate": 8.787556210808101e-06,
+ "loss": 3.1108,
+ "step": 209
+ },
+ {
+ "epoch": 0.2688,
+ "grad_norm": 0.5287924408912659,
+ "learning_rate": 8.773702214864587e-06,
+ "loss": 3.3165,
+ "step": 210
+ },
+ {
+ "epoch": 0.27008,
+ "grad_norm": 0.4569048285484314,
+ "learning_rate": 8.759780570305565e-06,
+ "loss": 3.2329,
+ "step": 211
+ },
+ {
+ "epoch": 0.27136,
+ "grad_norm": 0.37840911746025085,
+ "learning_rate": 8.745791526694969e-06,
+ "loss": 3.384,
+ "step": 212
+ },
+ {
+ "epoch": 0.27264,
+ "grad_norm": 0.37828516960144043,
+ "learning_rate": 8.731735334804953e-06,
+ "loss": 3.1662,
+ "step": 213
+ },
+ {
+ "epoch": 0.27392,
+ "grad_norm": 0.45762836933135986,
+ "learning_rate": 8.717612246611395e-06,
+ "loss": 3.2593,
+ "step": 214
+ },
+ {
+ "epoch": 0.2752,
+ "grad_norm": 0.4416005313396454,
+ "learning_rate": 8.703422515289374e-06,
+ "loss": 3.3551,
+ "step": 215
+ },
+ {
+ "epoch": 0.27648,
+ "grad_norm": 0.43728742003440857,
+ "learning_rate": 8.689166395208638e-06,
+ "loss": 3.3224,
+ "step": 216
+ },
+ {
+ "epoch": 0.27776,
+ "grad_norm": 0.4207445979118347,
+ "learning_rate": 8.67484414192904e-06,
+ "loss": 3.2076,
+ "step": 217
+ },
+ {
+ "epoch": 0.27904,
+ "grad_norm": 0.38989484310150146,
+ "learning_rate": 8.660456012195957e-06,
+ "loss": 3.207,
+ "step": 218
+ },
+ {
+ "epoch": 0.28032,
+ "grad_norm": 0.38946497440338135,
+ "learning_rate": 8.646002263935695e-06,
+ "loss": 3.3566,
+ "step": 219
+ },
+ {
+ "epoch": 0.2816,
+ "grad_norm": 0.4172903001308441,
+ "learning_rate": 8.631483156250855e-06,
+ "loss": 3.3409,
+ "step": 220
+ },
+ {
+ "epoch": 0.28288,
+ "grad_norm": 0.4878745973110199,
+ "learning_rate": 8.616898949415688e-06,
+ "loss": 3.2783,
+ "step": 221
+ },
+ {
+ "epoch": 0.28416,
+ "grad_norm": 0.41896191239356995,
+ "learning_rate": 8.60224990487144e-06,
+ "loss": 3.3567,
+ "step": 222
+ },
+ {
+ "epoch": 0.28544,
+ "grad_norm": 0.3486616015434265,
+ "learning_rate": 8.587536285221656e-06,
+ "loss": 3.1591,
+ "step": 223
+ },
+ {
+ "epoch": 0.28672,
+ "grad_norm": 0.40318334102630615,
+ "learning_rate": 8.572758354227476e-06,
+ "loss": 3.366,
+ "step": 224
+ },
+ {
+ "epoch": 0.288,
+ "grad_norm": 0.4261138439178467,
+ "learning_rate": 8.55791637680291e-06,
+ "loss": 3.2326,
+ "step": 225
+ },
+ {
+ "epoch": 0.28928,
+ "grad_norm": 0.45052197575569153,
+ "learning_rate": 8.543010619010077e-06,
+ "loss": 3.2843,
+ "step": 226
+ },
+ {
+ "epoch": 0.29056,
+ "grad_norm": 0.3807830214500427,
+ "learning_rate": 8.528041348054448e-06,
+ "loss": 3.2061,
+ "step": 227
+ },
+ {
+ "epoch": 0.29184,
+ "grad_norm": 0.5273177027702332,
+ "learning_rate": 8.513008832280054e-06,
+ "loss": 3.3204,
+ "step": 228
+ },
+ {
+ "epoch": 0.29312,
+ "grad_norm": 0.4441533386707306,
+ "learning_rate": 8.49791334116467e-06,
+ "loss": 3.2058,
+ "step": 229
+ },
+ {
+ "epoch": 0.2944,
+ "grad_norm": 0.3720758259296417,
+ "learning_rate": 8.482755145314987e-06,
+ "loss": 3.3038,
+ "step": 230
+ },
+ {
+ "epoch": 0.29568,
+ "grad_norm": 0.46793773770332336,
+ "learning_rate": 8.467534516461763e-06,
+ "loss": 3.3257,
+ "step": 231
+ },
+ {
+ "epoch": 0.29696,
+ "grad_norm": 0.483468621969223,
+ "learning_rate": 8.452251727454953e-06,
+ "loss": 3.228,
+ "step": 232
+ },
+ {
+ "epoch": 0.29824,
+ "grad_norm": 0.4611411690711975,
+ "learning_rate": 8.436907052258809e-06,
+ "loss": 3.2971,
+ "step": 233
+ },
+ {
+ "epoch": 0.29952,
+ "grad_norm": 0.4505979120731354,
+ "learning_rate": 8.421500765946984e-06,
+ "loss": 3.2283,
+ "step": 234
+ },
+ {
+ "epoch": 0.3008,
+ "grad_norm": 0.3875080943107605,
+ "learning_rate": 8.406033144697579e-06,
+ "loss": 3.3503,
+ "step": 235
+ },
+ {
+ "epoch": 0.30208,
+ "grad_norm": 0.4378403425216675,
+ "learning_rate": 8.390504465788217e-06,
+ "loss": 3.3522,
+ "step": 236
+ },
+ {
+ "epoch": 0.30336,
+ "grad_norm": 0.47747403383255005,
+ "learning_rate": 8.374915007591053e-06,
+ "loss": 3.2618,
+ "step": 237
+ },
+ {
+ "epoch": 0.30464,
+ "grad_norm": 0.39179787039756775,
+ "learning_rate": 8.359265049567798e-06,
+ "loss": 3.2309,
+ "step": 238
+ },
+ {
+ "epoch": 0.30592,
+ "grad_norm": 0.5158241987228394,
+ "learning_rate": 8.343554872264697e-06,
+ "loss": 3.4973,
+ "step": 239
+ },
+ {
+ "epoch": 0.3072,
+ "grad_norm": 0.4149444103240967,
+ "learning_rate": 8.327784757307506e-06,
+ "loss": 3.2623,
+ "step": 240
+ },
+ {
+ "epoch": 0.30848,
+ "grad_norm": 0.3934272527694702,
+ "learning_rate": 8.311954987396444e-06,
+ "loss": 3.2344,
+ "step": 241
+ },
+ {
+ "epoch": 0.30976,
+ "grad_norm": 0.3814394474029541,
+ "learning_rate": 8.296065846301129e-06,
+ "loss": 3.3489,
+ "step": 242
+ },
+ {
+ "epoch": 0.31104,
+ "grad_norm": 0.4408077895641327,
+ "learning_rate": 8.280117618855475e-06,
+ "loss": 3.2045,
+ "step": 243
+ },
+ {
+ "epoch": 0.31232,
+ "grad_norm": 0.35254621505737305,
+ "learning_rate": 8.264110590952609e-06,
+ "loss": 3.0956,
+ "step": 244
+ },
+ {
+ "epoch": 0.3136,
+ "grad_norm": 0.36275210976600647,
+ "learning_rate": 8.248045049539726e-06,
+ "loss": 3.2131,
+ "step": 245
+ },
+ {
+ "epoch": 0.31488,
+ "grad_norm": 0.4215599000453949,
+ "learning_rate": 8.231921282612957e-06,
+ "loss": 3.288,
+ "step": 246
+ },
+ {
+ "epoch": 0.31616,
+ "grad_norm": 0.4736253321170807,
+ "learning_rate": 8.2157395792122e-06,
+ "loss": 3.38,
+ "step": 247
+ },
+ {
+ "epoch": 0.31744,
+ "grad_norm": 0.4226548373699188,
+ "learning_rate": 8.199500229415946e-06,
+ "loss": 3.268,
+ "step": 248
+ },
+ {
+ "epoch": 0.31872,
+ "grad_norm": 0.45304951071739197,
+ "learning_rate": 8.183203524336067e-06,
+ "loss": 3.3091,
+ "step": 249
+ },
+ {
+ "epoch": 0.32,
+ "grad_norm": 0.4109630286693573,
+ "learning_rate": 8.166849756112612e-06,
+ "loss": 3.173,
+ "step": 250
+ },
+ {
+ "epoch": 0.32128,
+ "grad_norm": 0.4195004403591156,
+ "learning_rate": 8.150439217908557e-06,
+ "loss": 3.1425,
+ "step": 251
+ },
+ {
+ "epoch": 0.32256,
+ "grad_norm": 0.4661792814731598,
+ "learning_rate": 8.133972203904555e-06,
+ "loss": 3.3558,
+ "step": 252
+ },
+ {
+ "epoch": 0.32384,
+ "grad_norm": 0.41877481341362,
+ "learning_rate": 8.117449009293668e-06,
+ "loss": 3.2049,
+ "step": 253
+ },
+ {
+ "epoch": 0.32512,
+ "grad_norm": 0.43276724219322205,
+ "learning_rate": 8.100869930276065e-06,
+ "loss": 3.2679,
+ "step": 254
+ },
+ {
+ "epoch": 0.3264,
+ "grad_norm": 0.36627158522605896,
+ "learning_rate": 8.08423526405372e-06,
+ "loss": 3.2026,
+ "step": 255
+ },
+ {
+ "epoch": 0.32768,
+ "grad_norm": 0.373998761177063,
+ "learning_rate": 8.067545308825075e-06,
+ "loss": 3.3428,
+ "step": 256
+ },
+ {
+ "epoch": 0.32896,
+ "grad_norm": 0.4387235939502716,
+ "learning_rate": 8.050800363779711e-06,
+ "loss": 3.2708,
+ "step": 257
+ },
+ {
+ "epoch": 0.33024,
+ "grad_norm": 0.43042445182800293,
+ "learning_rate": 8.034000729092967e-06,
+ "loss": 3.2476,
+ "step": 258
+ },
+ {
+ "epoch": 0.33152,
+ "grad_norm": 0.3883137106895447,
+ "learning_rate": 8.01714670592057e-06,
+ "loss": 3.155,
+ "step": 259
+ },
+ {
+ "epoch": 0.3328,
+ "grad_norm": 0.4483623802661896,
+ "learning_rate": 8.000238596393226e-06,
+ "loss": 3.2673,
+ "step": 260
+ },
+ {
+ "epoch": 0.33408,
+ "grad_norm": 0.3685246706008911,
+ "learning_rate": 7.98327670361122e-06,
+ "loss": 3.2708,
+ "step": 261
+ },
+ {
+ "epoch": 0.33536,
+ "grad_norm": 0.3615477979183197,
+ "learning_rate": 7.96626133163897e-06,
+ "loss": 3.3,
+ "step": 262
+ },
+ {
+ "epoch": 0.33664,
+ "grad_norm": 0.40237149596214294,
+ "learning_rate": 7.949192785499573e-06,
+ "loss": 3.1329,
+ "step": 263
+ },
+ {
+ "epoch": 0.33792,
+ "grad_norm": 0.45421358942985535,
+ "learning_rate": 7.932071371169353e-06,
+ "loss": 3.3781,
+ "step": 264
+ },
+ {
+ "epoch": 0.3392,
+ "grad_norm": 0.35313281416893005,
+ "learning_rate": 7.914897395572362e-06,
+ "loss": 3.3714,
+ "step": 265
+ },
+ {
+ "epoch": 0.34048,
+ "grad_norm": 0.42352741956710815,
+ "learning_rate": 7.897671166574878e-06,
+ "loss": 3.2673,
+ "step": 266
+ },
+ {
+ "epoch": 0.34176,
+ "grad_norm": 0.4240070879459381,
+ "learning_rate": 7.8803929929799e-06,
+ "loss": 3.3597,
+ "step": 267
+ },
+ {
+ "epoch": 0.34304,
+ "grad_norm": 0.4785767197608948,
+ "learning_rate": 7.863063184521596e-06,
+ "loss": 3.1689,
+ "step": 268
+ },
+ {
+ "epoch": 0.34432,
+ "grad_norm": 0.37650686502456665,
+ "learning_rate": 7.845682051859757e-06,
+ "loss": 3.1712,
+ "step": 269
+ },
+ {
+ "epoch": 0.3456,
+ "grad_norm": 0.401387095451355,
+ "learning_rate": 7.828249906574235e-06,
+ "loss": 3.1763,
+ "step": 270
+ },
+ {
+ "epoch": 0.34688,
+ "grad_norm": 0.4616657793521881,
+ "learning_rate": 7.810767061159341e-06,
+ "loss": 3.3826,
+ "step": 271
+ },
+ {
+ "epoch": 0.34816,
+ "grad_norm": 0.48199304938316345,
+ "learning_rate": 7.793233829018263e-06,
+ "loss": 3.076,
+ "step": 272
+ },
+ {
+ "epoch": 0.34944,
+ "grad_norm": 0.4462188184261322,
+ "learning_rate": 7.775650524457431e-06,
+ "loss": 3.2589,
+ "step": 273
+ },
+ {
+ "epoch": 0.35072,
+ "grad_norm": 0.364860862493515,
+ "learning_rate": 7.758017462680893e-06,
+ "loss": 3.1881,
+ "step": 274
+ },
+ {
+ "epoch": 0.352,
+ "grad_norm": 0.41849344968795776,
+ "learning_rate": 7.740334959784656e-06,
+ "loss": 3.2993,
+ "step": 275
+ },
+ {
+ "epoch": 0.35328,
+ "grad_norm": 0.4385211169719696,
+ "learning_rate": 7.722603332751035e-06,
+ "loss": 3.2685,
+ "step": 276
+ },
+ {
+ "epoch": 0.35456,
+ "grad_norm": 0.38513725996017456,
+ "learning_rate": 7.70482289944295e-06,
+ "loss": 3.2552,
+ "step": 277
+ },
+ {
+ "epoch": 0.35584,
+ "grad_norm": 0.3823578655719757,
+ "learning_rate": 7.68699397859824e-06,
+ "loss": 3.3631,
+ "step": 278
+ },
+ {
+ "epoch": 0.35712,
+ "grad_norm": 0.5122019052505493,
+ "learning_rate": 7.669116889823955e-06,
+ "loss": 3.2389,
+ "step": 279
+ },
+ {
+ "epoch": 0.3584,
+ "grad_norm": 0.378545343875885,
+ "learning_rate": 7.651191953590606e-06,
+ "loss": 3.2334,
+ "step": 280
+ },
+ {
+ "epoch": 0.35968,
+ "grad_norm": 0.39015185832977295,
+ "learning_rate": 7.63321949122645e-06,
+ "loss": 3.2224,
+ "step": 281
+ },
+ {
+ "epoch": 0.36096,
+ "grad_norm": 0.4863089323043823,
+ "learning_rate": 7.615199824911694e-06,
+ "loss": 3.4202,
+ "step": 282
+ },
+ {
+ "epoch": 0.36224,
+ "grad_norm": 0.3806248605251312,
+ "learning_rate": 7.5971332776727525e-06,
+ "loss": 3.3495,
+ "step": 283
+ },
+ {
+ "epoch": 0.36352,
+ "grad_norm": 0.4714275896549225,
+ "learning_rate": 7.579020173376439e-06,
+ "loss": 3.197,
+ "step": 284
+ },
+ {
+ "epoch": 0.3648,
+ "grad_norm": 0.44943785667419434,
+ "learning_rate": 7.5608608367241595e-06,
+ "loss": 3.1311,
+ "step": 285
+ },
+ {
+ "epoch": 0.36608,
+ "grad_norm": 0.4081331193447113,
+ "learning_rate": 7.542655593246103e-06,
+ "loss": 3.241,
+ "step": 286
+ },
+ {
+ "epoch": 0.36736,
+ "grad_norm": 0.4430176317691803,
+ "learning_rate": 7.5244047692953944e-06,
+ "loss": 3.3312,
+ "step": 287
+ },
+ {
+ "epoch": 0.36864,
+ "grad_norm": 0.448420912027359,
+ "learning_rate": 7.506108692042251e-06,
+ "loss": 3.1125,
+ "step": 288
+ },
+ {
+ "epoch": 0.36992,
+ "grad_norm": 0.4760783016681671,
+ "learning_rate": 7.487767689468114e-06,
+ "loss": 3.4506,
+ "step": 289
+ },
+ {
+ "epoch": 0.3712,
+ "grad_norm": 0.4651552736759186,
+ "learning_rate": 7.469382090359774e-06,
+ "loss": 3.3007,
+ "step": 290
+ },
+ {
+ "epoch": 0.37248,
+ "grad_norm": 0.4438134729862213,
+ "learning_rate": 7.45095222430347e-06,
+ "loss": 3.3165,
+ "step": 291
+ },
+ {
+ "epoch": 0.37376,
+ "grad_norm": 0.38648146390914917,
+ "learning_rate": 7.432478421678983e-06,
+ "loss": 3.1762,
+ "step": 292
+ },
+ {
+ "epoch": 0.37504,
+ "grad_norm": 0.44642776250839233,
+ "learning_rate": 7.413961013653725e-06,
+ "loss": 3.1242,
+ "step": 293
+ },
+ {
+ "epoch": 0.37632,
+ "grad_norm": 0.3903616666793823,
+ "learning_rate": 7.395400332176781e-06,
+ "loss": 3.2325,
+ "step": 294
+ },
+ {
+ "epoch": 0.3776,
+ "grad_norm": 0.4252457618713379,
+ "learning_rate": 7.376796709972975e-06,
+ "loss": 3.0806,
+ "step": 295
+ },
+ {
+ "epoch": 0.37888,
+ "grad_norm": 0.4216412603855133,
+ "learning_rate": 7.358150480536904e-06,
+ "loss": 3.3427,
+ "step": 296
+ },
+ {
+ "epoch": 0.38016,
+ "grad_norm": 0.3958504796028137,
+ "learning_rate": 7.339461978126947e-06,
+ "loss": 3.3271,
+ "step": 297
+ },
+ {
+ "epoch": 0.38144,
+ "grad_norm": 0.417348712682724,
+ "learning_rate": 7.3207315377592935e-06,
+ "loss": 3.2639,
+ "step": 298
+ },
+ {
+ "epoch": 0.38272,
+ "grad_norm": 0.41278159618377686,
+ "learning_rate": 7.301959495201915e-06,
+ "loss": 3.3004,
+ "step": 299
+ },
+ {
+ "epoch": 0.384,
+ "grad_norm": 0.41572651267051697,
+ "learning_rate": 7.283146186968566e-06,
+ "loss": 3.1847,
+ "step": 300
+ },
+ {
+ "epoch": 0.38528,
+ "grad_norm": 0.4208621382713318,
+ "learning_rate": 7.2642919503127405e-06,
+ "loss": 3.3485,
+ "step": 301
+ },
+ {
+ "epoch": 0.38656,
+ "grad_norm": 0.4609411954879761,
+ "learning_rate": 7.2453971232216244e-06,
+ "loss": 3.4132,
+ "step": 302
+ },
+ {
+ "epoch": 0.38784,
+ "grad_norm": 0.3979472815990448,
+ "learning_rate": 7.2264620444100505e-06,
+ "loss": 3.1796,
+ "step": 303
+ },
+ {
+ "epoch": 0.38912,
+ "grad_norm": 0.3755399286746979,
+ "learning_rate": 7.207487053314408e-06,
+ "loss": 3.259,
+ "step": 304
+ },
+ {
+ "epoch": 0.3904,
+ "grad_norm": 0.4647941291332245,
+ "learning_rate": 7.188472490086569e-06,
+ "loss": 3.3306,
+ "step": 305
+ },
+ {
+ "epoch": 0.39168,
+ "grad_norm": 0.3828791081905365,
+ "learning_rate": 7.169418695587791e-06,
+ "loss": 3.1996,
+ "step": 306
+ },
+ {
+ "epoch": 0.39296,
+ "grad_norm": 0.4810822606086731,
+ "learning_rate": 7.1503260113826035e-06,
+ "loss": 3.2503,
+ "step": 307
+ },
+ {
+ "epoch": 0.39424,
+ "grad_norm": 0.43079525232315063,
+ "learning_rate": 7.131194779732682e-06,
+ "loss": 3.3527,
+ "step": 308
+ },
+ {
+ "epoch": 0.39552,
+ "grad_norm": 0.3956846296787262,
+ "learning_rate": 7.112025343590721e-06,
+ "loss": 3.1649,
+ "step": 309
+ },
+ {
+ "epoch": 0.3968,
+ "grad_norm": 0.39624664187431335,
+ "learning_rate": 7.092818046594279e-06,
+ "loss": 3.1066,
+ "step": 310
+ },
+ {
+ "epoch": 0.39808,
+ "grad_norm": 0.38523128628730774,
+ "learning_rate": 7.073573233059618e-06,
+ "loss": 3.1582,
+ "step": 311
+ },
+ {
+ "epoch": 0.39936,
+ "grad_norm": 0.40635234117507935,
+ "learning_rate": 7.0542912479755385e-06,
+ "loss": 3.28,
+ "step": 312
+ },
+ {
+ "epoch": 0.40064,
+ "grad_norm": 0.37109220027923584,
+ "learning_rate": 7.034972436997185e-06,
+ "loss": 3.3248,
+ "step": 313
+ },
+ {
+ "epoch": 0.40192,
+ "grad_norm": 0.40318062901496887,
+ "learning_rate": 7.015617146439863e-06,
+ "loss": 3.0695,
+ "step": 314
+ },
+ {
+ "epoch": 0.4032,
+ "grad_norm": 0.37492793798446655,
+ "learning_rate": 6.996225723272812e-06,
+ "loss": 3.1437,
+ "step": 315
+ },
+ {
+ "epoch": 0.40448,
+ "grad_norm": 0.47444018721580505,
+ "learning_rate": 6.976798515113003e-06,
+ "loss": 3.2307,
+ "step": 316
+ },
+ {
+ "epoch": 0.40576,
+ "grad_norm": 0.45195314288139343,
+ "learning_rate": 6.957335870218906e-06,
+ "loss": 3.3534,
+ "step": 317
+ },
+ {
+ "epoch": 0.40704,
+ "grad_norm": 0.4104592204093933,
+ "learning_rate": 6.937838137484226e-06,
+ "loss": 3.3513,
+ "step": 318
+ },
+ {
+ "epoch": 0.40832,
+ "grad_norm": 0.41414302587509155,
+ "learning_rate": 6.918305666431678e-06,
+ "loss": 3.2582,
+ "step": 319
+ },
+ {
+ "epoch": 0.4096,
+ "grad_norm": 0.3929477035999298,
+ "learning_rate": 6.898738807206702e-06,
+ "loss": 3.2648,
+ "step": 320
+ },
+ {
+ "epoch": 0.41088,
+ "grad_norm": 0.36228302121162415,
+ "learning_rate": 6.879137910571191e-06,
+ "loss": 3.2399,
+ "step": 321
+ },
+ {
+ "epoch": 0.41216,
+ "grad_norm": 0.4036976099014282,
+ "learning_rate": 6.859503327897207e-06,
+ "loss": 3.1497,
+ "step": 322
+ },
+ {
+ "epoch": 0.41344,
+ "grad_norm": 0.42336946725845337,
+ "learning_rate": 6.839835411160674e-06,
+ "loss": 3.1844,
+ "step": 323
+ },
+ {
+ "epoch": 0.41472,
+ "grad_norm": 0.4051893353462219,
+ "learning_rate": 6.820134512935076e-06,
+ "loss": 3.2465,
+ "step": 324
+ },
+ {
+ "epoch": 0.416,
+ "grad_norm": 0.4318195879459381,
+ "learning_rate": 6.800400986385136e-06,
+ "loss": 3.1324,
+ "step": 325
+ },
+ {
+ "epoch": 0.41728,
+ "grad_norm": 0.43580979108810425,
+ "learning_rate": 6.780635185260479e-06,
+ "loss": 3.2357,
+ "step": 326
+ },
+ {
+ "epoch": 0.41856,
+ "grad_norm": 0.3950352966785431,
+ "learning_rate": 6.760837463889297e-06,
+ "loss": 3.3189,
+ "step": 327
+ },
+ {
+ "epoch": 0.41984,
+ "grad_norm": 0.39430123567581177,
+ "learning_rate": 6.741008177171995e-06,
+ "loss": 3.2364,
+ "step": 328
+ },
+ {
+ "epoch": 0.42112,
+ "grad_norm": 0.49953415989875793,
+ "learning_rate": 6.7211476805748254e-06,
+ "loss": 3.1676,
+ "step": 329
+ },
+ {
+ "epoch": 0.4224,
+ "grad_norm": 0.38956841826438904,
+ "learning_rate": 6.701256330123523e-06,
+ "loss": 3.1134,
+ "step": 330
+ },
+ {
+ "epoch": 0.42368,
+ "grad_norm": 0.47153738141059875,
+ "learning_rate": 6.68133448239692e-06,
+ "loss": 3.1834,
+ "step": 331
+ },
+ {
+ "epoch": 0.42496,
+ "grad_norm": 0.4385981261730194,
+ "learning_rate": 6.661382494520543e-06,
+ "loss": 3.1893,
+ "step": 332
+ },
+ {
+ "epoch": 0.42624,
+ "grad_norm": 0.40008580684661865,
+ "learning_rate": 6.641400724160235e-06,
+ "loss": 3.3213,
+ "step": 333
+ },
+ {
+ "epoch": 0.42752,
+ "grad_norm": 0.41795697808265686,
+ "learning_rate": 6.621389529515718e-06,
+ "loss": 3.145,
+ "step": 334
+ },
+ {
+ "epoch": 0.4288,
+ "grad_norm": 0.48342013359069824,
+ "learning_rate": 6.601349269314188e-06,
+ "loss": 3.2096,
+ "step": 335
+ },
+ {
+ "epoch": 0.43008,
+ "grad_norm": 0.38955411314964294,
+ "learning_rate": 6.581280302803882e-06,
+ "loss": 3.1825,
+ "step": 336
+ },
+ {
+ "epoch": 0.43136,
+ "grad_norm": 0.39123156666755676,
+ "learning_rate": 6.561182989747629e-06,
+ "loss": 3.3196,
+ "step": 337
+ },
+ {
+ "epoch": 0.43264,
+ "grad_norm": 0.40691521763801575,
+ "learning_rate": 6.5410576904164145e-06,
+ "loss": 3.1102,
+ "step": 338
+ },
+ {
+ "epoch": 0.43392,
+ "grad_norm": 0.4109227657318115,
+ "learning_rate": 6.520904765582909e-06,
+ "loss": 3.1896,
+ "step": 339
+ },
+ {
+ "epoch": 0.4352,
+ "grad_norm": 0.4398311376571655,
+ "learning_rate": 6.500724576515012e-06,
+ "loss": 3.2177,
+ "step": 340
+ },
+ {
+ "epoch": 0.43648,
+ "grad_norm": 0.41105419397354126,
+ "learning_rate": 6.480517484969366e-06,
+ "loss": 3.2674,
+ "step": 341
+ },
+ {
+ "epoch": 0.43776,
+ "grad_norm": 0.43113186955451965,
+ "learning_rate": 6.46028385318488e-06,
+ "loss": 3.2567,
+ "step": 342
+ },
+ {
+ "epoch": 0.43904,
+ "grad_norm": 0.5099824666976929,
+ "learning_rate": 6.44002404387623e-06,
+ "loss": 3.4293,
+ "step": 343
+ },
+ {
+ "epoch": 0.44032,
+ "grad_norm": 0.5957097411155701,
+ "learning_rate": 6.419738420227357e-06,
+ "loss": 3.3156,
+ "step": 344
+ },
+ {
+ "epoch": 0.4416,
+ "grad_norm": 0.4592932462692261,
+ "learning_rate": 6.399427345884964e-06,
+ "loss": 3.2337,
+ "step": 345
+ },
+ {
+ "epoch": 0.44288,
+ "grad_norm": 0.4420956075191498,
+ "learning_rate": 6.379091184951984e-06,
+ "loss": 3.2589,
+ "step": 346
+ },
+ {
+ "epoch": 0.44416,
+ "grad_norm": 0.4118671417236328,
+ "learning_rate": 6.358730301981066e-06,
+ "loss": 3.2683,
+ "step": 347
+ },
+ {
+ "epoch": 0.44544,
+ "grad_norm": 0.4201429486274719,
+ "learning_rate": 6.338345061968033e-06,
+ "loss": 3.1371,
+ "step": 348
+ },
+ {
+ "epoch": 0.44672,
+ "grad_norm": 0.3861488997936249,
+ "learning_rate": 6.3179358303453386e-06,
+ "loss": 3.1778,
+ "step": 349
+ },
+ {
+ "epoch": 0.448,
+ "grad_norm": 0.42568036913871765,
+ "learning_rate": 6.29750297297552e-06,
+ "loss": 3.2517,
+ "step": 350
+ },
+ {
+ "epoch": 0.44928,
+ "grad_norm": 0.39109253883361816,
+ "learning_rate": 6.277046856144633e-06,
+ "loss": 3.1012,
+ "step": 351
+ },
+ {
+ "epoch": 0.45056,
+ "grad_norm": 0.43290814757347107,
+ "learning_rate": 6.256567846555699e-06,
+ "loss": 3.2777,
+ "step": 352
+ },
+ {
+ "epoch": 0.45184,
+ "grad_norm": 0.4072764813899994,
+ "learning_rate": 6.236066311322111e-06,
+ "loss": 3.2811,
+ "step": 353
+ },
+ {
+ "epoch": 0.45312,
+ "grad_norm": 0.45435935258865356,
+ "learning_rate": 6.2155426179610715e-06,
+ "loss": 3.4914,
+ "step": 354
+ },
+ {
+ "epoch": 0.4544,
+ "grad_norm": 0.4328829348087311,
+ "learning_rate": 6.1949971343869975e-06,
+ "loss": 3.2827,
+ "step": 355
+ },
+ {
+ "epoch": 0.45568,
+ "grad_norm": 0.4269879162311554,
+ "learning_rate": 6.17443022890492e-06,
+ "loss": 3.2118,
+ "step": 356
+ },
+ {
+ "epoch": 0.45696,
+ "grad_norm": 0.42866218090057373,
+ "learning_rate": 6.153842270203887e-06,
+ "loss": 3.25,
+ "step": 357
+ },
+ {
+ "epoch": 0.45824,
+ "grad_norm": 0.43354228138923645,
+ "learning_rate": 6.133233627350356e-06,
+ "loss": 3.1804,
+ "step": 358
+ },
+ {
+ "epoch": 0.45952,
+ "grad_norm": 0.5081104636192322,
+ "learning_rate": 6.112604669781572e-06,
+ "loss": 3.4438,
+ "step": 359
+ },
+ {
+ "epoch": 0.4608,
+ "grad_norm": 0.4651986360549927,
+ "learning_rate": 6.091955767298954e-06,
+ "loss": 3.3727,
+ "step": 360
+ },
+ {
+ "epoch": 0.46208,
+ "grad_norm": 0.3870481252670288,
+ "learning_rate": 6.0712872900614515e-06,
+ "loss": 3.1145,
+ "step": 361
+ },
+ {
+ "epoch": 0.46336,
+ "grad_norm": 0.4193924367427826,
+ "learning_rate": 6.050599608578925e-06,
+ "loss": 3.2524,
+ "step": 362
+ },
+ {
+ "epoch": 0.46464,
+ "grad_norm": 0.42114901542663574,
+ "learning_rate": 6.029893093705492e-06,
+ "loss": 3.1527,
+ "step": 363
+ },
+ {
+ "epoch": 0.46592,
+ "grad_norm": 0.39673611521720886,
+ "learning_rate": 6.009168116632888e-06,
+ "loss": 3.0491,
+ "step": 364
+ },
+ {
+ "epoch": 0.4672,
+ "grad_norm": 0.3314211368560791,
+ "learning_rate": 5.988425048883799e-06,
+ "loss": 3.057,
+ "step": 365
+ },
+ {
+ "epoch": 0.46848,
+ "grad_norm": 0.4453369081020355,
+ "learning_rate": 5.967664262305221e-06,
+ "loss": 3.2038,
+ "step": 366
+ },
+ {
+ "epoch": 0.46976,
+ "grad_norm": 0.3986211121082306,
+ "learning_rate": 5.946886129061777e-06,
+ "loss": 3.3501,
+ "step": 367
+ },
+ {
+ "epoch": 0.47104,
+ "grad_norm": 0.41228437423706055,
+ "learning_rate": 5.926091021629056e-06,
+ "loss": 3.2064,
+ "step": 368
+ },
+ {
+ "epoch": 0.47232,
+ "grad_norm": 0.4700061082839966,
+ "learning_rate": 5.905279312786928e-06,
+ "loss": 3.2372,
+ "step": 369
+ },
+ {
+ "epoch": 0.4736,
+ "grad_norm": 0.42562106251716614,
+ "learning_rate": 5.884451375612865e-06,
+ "loss": 3.212,
+ "step": 370
+ },
+ {
+ "epoch": 0.47488,
+ "grad_norm": 0.4857390224933624,
+ "learning_rate": 5.863607583475259e-06,
+ "loss": 3.278,
+ "step": 371
+ },
+ {
+ "epoch": 0.47616,
+ "grad_norm": 0.3687186539173126,
+ "learning_rate": 5.842748310026717e-06,
+ "loss": 3.1253,
+ "step": 372
+ },
+ {
+ "epoch": 0.47744,
+ "grad_norm": 0.38398629426956177,
+ "learning_rate": 5.821873929197371e-06,
+ "loss": 3.2098,
+ "step": 373
+ },
+ {
+ "epoch": 0.47872,
+ "grad_norm": 0.38533806800842285,
+ "learning_rate": 5.8009848151881765e-06,
+ "loss": 3.2778,
+ "step": 374
+ },
+ {
+ "epoch": 0.48,
+ "grad_norm": 0.45727336406707764,
+ "learning_rate": 5.780081342464195e-06,
+ "loss": 3.3698,
+ "step": 375
+ },
+ {
+ "epoch": 0.48128,
+ "grad_norm": 0.41655293107032776,
+ "learning_rate": 5.75916388574789e-06,
+ "loss": 3.1184,
+ "step": 376
+ },
+ {
+ "epoch": 0.48256,
+ "grad_norm": 0.4045466184616089,
+ "learning_rate": 5.738232820012407e-06,
+ "loss": 3.3063,
+ "step": 377
+ },
+ {
+ "epoch": 0.48384,
+ "grad_norm": 0.379459947347641,
+ "learning_rate": 5.717288520474849e-06,
+ "loss": 3.0753,
+ "step": 378
+ },
+ {
+ "epoch": 0.48512,
+ "grad_norm": 0.4356738328933716,
+ "learning_rate": 5.696331362589557e-06,
+ "loss": 3.1975,
+ "step": 379
+ },
+ {
+ "epoch": 0.4864,
+ "grad_norm": 0.3822133541107178,
+ "learning_rate": 5.675361722041368e-06,
+ "loss": 3.1793,
+ "step": 380
+ },
+ {
+ "epoch": 0.48768,
+ "grad_norm": 0.5050751566886902,
+ "learning_rate": 5.654379974738893e-06,
+ "loss": 3.2048,
+ "step": 381
+ },
+ {
+ "epoch": 0.48896,
+ "grad_norm": 0.39909955859184265,
+ "learning_rate": 5.6333864968077716e-06,
+ "loss": 3.2516,
+ "step": 382
+ },
+ {
+ "epoch": 0.49024,
+ "grad_norm": 0.4170902967453003,
+ "learning_rate": 5.6123816645839295e-06,
+ "loss": 3.2045,
+ "step": 383
+ },
+ {
+ "epoch": 0.49152,
+ "grad_norm": 0.4631553292274475,
+ "learning_rate": 5.591365854606829e-06,
+ "loss": 3.3919,
+ "step": 384
+ },
+ {
+ "epoch": 0.4928,
+ "grad_norm": 0.44511735439300537,
+ "learning_rate": 5.570339443612734e-06,
+ "loss": 3.1974,
+ "step": 385
+ },
+ {
+ "epoch": 0.49408,
+ "grad_norm": 0.4099262058734894,
+ "learning_rate": 5.549302808527936e-06,
+ "loss": 3.1686,
+ "step": 386
+ },
+ {
+ "epoch": 0.49536,
+ "grad_norm": 0.402769535779953,
+ "learning_rate": 5.528256326462007e-06,
+ "loss": 3.2271,
+ "step": 387
+ },
+ {
+ "epoch": 0.49664,
+ "grad_norm": 0.3749559223651886,
+ "learning_rate": 5.507200374701048e-06,
+ "loss": 3.1808,
+ "step": 388
+ },
+ {
+ "epoch": 0.49792,
+ "grad_norm": 0.3516296148300171,
+ "learning_rate": 5.486135330700906e-06,
+ "loss": 3.033,
+ "step": 389
+ },
+ {
+ "epoch": 0.4992,
+ "grad_norm": 0.5268987417221069,
+ "learning_rate": 5.465061572080429e-06,
+ "loss": 3.2837,
+ "step": 390
+ },
+ {
+ "epoch": 0.50048,
+ "grad_norm": 0.4084585905075073,
+ "learning_rate": 5.443979476614674e-06,
+ "loss": 3.285,
+ "step": 391
+ },
+ {
+ "epoch": 0.50176,
+ "grad_norm": 0.5457457900047302,
+ "learning_rate": 5.4228894222281595e-06,
+ "loss": 3.3747,
+ "step": 392
+ },
+ {
+ "epoch": 0.50304,
+ "grad_norm": 0.43756410479545593,
+ "learning_rate": 5.401791786988068e-06,
+ "loss": 3.127,
+ "step": 393
+ },
+ {
+ "epoch": 0.50432,
+ "grad_norm": 0.4421897530555725,
+ "learning_rate": 5.380686949097486e-06,
+ "loss": 3.3014,
+ "step": 394
+ },
+ {
+ "epoch": 0.5056,
+ "grad_norm": 0.4444248378276825,
+ "learning_rate": 5.359575286888613e-06,
+ "loss": 3.1109,
+ "step": 395
+ },
+ {
+ "epoch": 0.50688,
+ "grad_norm": 0.38180235028266907,
+ "learning_rate": 5.338457178815984e-06,
+ "loss": 3.2503,
+ "step": 396
+ },
+ {
+ "epoch": 0.50816,
+ "grad_norm": 0.49522870779037476,
+ "learning_rate": 5.3173330034496875e-06,
+ "loss": 3.2582,
+ "step": 397
+ },
+ {
+ "epoch": 0.50944,
+ "grad_norm": 0.39769619703292847,
+ "learning_rate": 5.296203139468572e-06,
+ "loss": 3.0916,
+ "step": 398
+ },
+ {
+ "epoch": 0.51072,
+ "grad_norm": 0.4254739284515381,
+ "learning_rate": 5.275067965653465e-06,
+ "loss": 3.3287,
+ "step": 399
+ },
+ {
+ "epoch": 0.512,
+ "grad_norm": 0.43100857734680176,
+ "learning_rate": 5.2539278608803786e-06,
+ "loss": 3.3448,
+ "step": 400
+ },
+ {
+ "epoch": 0.51328,
+ "grad_norm": 0.3966229557991028,
+ "learning_rate": 5.23278320411372e-06,
+ "loss": 3.1516,
+ "step": 401
+ },
+ {
+ "epoch": 0.51456,
+ "grad_norm": 0.3804360330104828,
+ "learning_rate": 5.211634374399496e-06,
+ "loss": 3.1992,
+ "step": 402
+ },
+ {
+ "epoch": 0.51584,
+ "grad_norm": 0.402805358171463,
+ "learning_rate": 5.190481750858517e-06,
+ "loss": 3.1752,
+ "step": 403
+ },
+ {
+ "epoch": 0.51712,
+ "grad_norm": 0.4411153197288513,
+ "learning_rate": 5.169325712679608e-06,
+ "loss": 3.2063,
+ "step": 404
+ },
+ {
+ "epoch": 0.5184,
+ "grad_norm": 0.411283403635025,
+ "learning_rate": 5.148166639112799e-06,
+ "loss": 3.1938,
+ "step": 405
+ },
+ {
+ "epoch": 0.51968,
+ "grad_norm": 0.43547821044921875,
+ "learning_rate": 5.127004909462541e-06,
+ "loss": 3.2911,
+ "step": 406
+ },
+ {
+ "epoch": 0.52096,
+ "grad_norm": 0.3784047067165375,
+ "learning_rate": 5.1058409030808935e-06,
+ "loss": 3.0922,
+ "step": 407
+ },
+ {
+ "epoch": 0.52224,
+ "grad_norm": 0.37176749110221863,
+ "learning_rate": 5.0846749993607296e-06,
+ "loss": 3.1972,
+ "step": 408
+ },
+ {
+ "epoch": 0.52352,
+ "grad_norm": 0.39474594593048096,
+ "learning_rate": 5.063507577728938e-06,
+ "loss": 3.2388,
+ "step": 409
+ },
+ {
+ "epoch": 0.5248,
+ "grad_norm": 0.44064849615097046,
+ "learning_rate": 5.042339017639614e-06,
+ "loss": 3.2396,
+ "step": 410
+ },
+ {
+ "epoch": 0.52608,
+ "grad_norm": 0.3827720284461975,
+ "learning_rate": 5.021169698567264e-06,
+ "loss": 3.2039,
+ "step": 411
+ },
+ {
+ "epoch": 0.52736,
+ "grad_norm": 0.44552549719810486,
+ "learning_rate": 5e-06,
+ "loss": 3.1445,
+ "step": 412
+ },
+ {
+ "epoch": 0.52864,
+ "grad_norm": 0.4261825680732727,
+ "learning_rate": 4.978830301432738e-06,
+ "loss": 3.1208,
+ "step": 413
+ },
+ {
+ "epoch": 0.52992,
+ "grad_norm": 0.47567233443260193,
+ "learning_rate": 4.957660982360386e-06,
+ "loss": 3.1194,
+ "step": 414
+ },
+ {
+ "epoch": 0.5312,
+ "grad_norm": 0.41242220997810364,
+ "learning_rate": 4.936492422271064e-06,
+ "loss": 3.3059,
+ "step": 415
+ },
+ {
+ "epoch": 0.53248,
+ "grad_norm": 0.3963751792907715,
+ "learning_rate": 4.915325000639271e-06,
+ "loss": 3.1593,
+ "step": 416
+ },
+ {
+ "epoch": 0.53376,
+ "grad_norm": 0.508144199848175,
+ "learning_rate": 4.894159096919109e-06,
+ "loss": 3.345,
+ "step": 417
+ },
+ {
+ "epoch": 0.53504,
+ "grad_norm": 0.4466751515865326,
+ "learning_rate": 4.87299509053746e-06,
+ "loss": 3.166,
+ "step": 418
+ },
+ {
+ "epoch": 0.53632,
+ "grad_norm": 0.42744773626327515,
+ "learning_rate": 4.8518333608872015e-06,
+ "loss": 3.2012,
+ "step": 419
+ },
+ {
+ "epoch": 0.5376,
+ "grad_norm": 0.4188246726989746,
+ "learning_rate": 4.830674287320395e-06,
+ "loss": 3.1217,
+ "step": 420
+ },
+ {
+ "epoch": 0.53888,
+ "grad_norm": 0.4558735191822052,
+ "learning_rate": 4.809518249141484e-06,
+ "loss": 3.2641,
+ "step": 421
+ },
+ {
+ "epoch": 0.54016,
+ "grad_norm": 0.4426221251487732,
+ "learning_rate": 4.788365625600505e-06,
+ "loss": 3.1585,
+ "step": 422
+ },
+ {
+ "epoch": 0.54144,
+ "grad_norm": 0.3888532519340515,
+ "learning_rate": 4.767216795886281e-06,
+ "loss": 3.1475,
+ "step": 423
+ },
+ {
+ "epoch": 0.54272,
+ "grad_norm": 0.4421764314174652,
+ "learning_rate": 4.746072139119622e-06,
+ "loss": 3.3551,
+ "step": 424
+ },
+ {
+ "epoch": 0.544,
+ "grad_norm": 0.4097662568092346,
+ "learning_rate": 4.724932034346536e-06,
+ "loss": 3.1878,
+ "step": 425
+ },
+ {
+ "epoch": 0.54528,
+ "grad_norm": 0.40844854712486267,
+ "learning_rate": 4.703796860531429e-06,
+ "loss": 3.1557,
+ "step": 426
+ },
+ {
+ "epoch": 0.54656,
+ "grad_norm": 0.3919289708137512,
+ "learning_rate": 4.682666996550314e-06,
+ "loss": 3.2407,
+ "step": 427
+ },
+ {
+ "epoch": 0.54784,
+ "grad_norm": 0.3589138388633728,
+ "learning_rate": 4.661542821184016e-06,
+ "loss": 3.1488,
+ "step": 428
+ },
+ {
+ "epoch": 0.54912,
+ "grad_norm": 0.4772014617919922,
+ "learning_rate": 4.640424713111389e-06,
+ "loss": 3.2059,
+ "step": 429
+ },
+ {
+ "epoch": 0.5504,
+ "grad_norm": 0.4213854968547821,
+ "learning_rate": 4.619313050902515e-06,
+ "loss": 3.183,
+ "step": 430
+ },
+ {
+ "epoch": 0.55168,
+ "grad_norm": 0.42899787425994873,
+ "learning_rate": 4.598208213011934e-06,
+ "loss": 3.2579,
+ "step": 431
+ },
+ {
+ "epoch": 0.55296,
+ "grad_norm": 0.42090630531311035,
+ "learning_rate": 4.577110577771842e-06,
+ "loss": 3.1706,
+ "step": 432
+ },
+ {
+ "epoch": 0.55424,
+ "grad_norm": 0.4222717881202698,
+ "learning_rate": 4.556020523385326e-06,
+ "loss": 3.3265,
+ "step": 433
+ },
+ {
+ "epoch": 0.55552,
+ "grad_norm": 0.3918984532356262,
+ "learning_rate": 4.534938427919574e-06,
+ "loss": 3.2802,
+ "step": 434
+ },
+ {
+ "epoch": 0.5568,
+ "grad_norm": 0.4150027930736542,
+ "learning_rate": 4.513864669299094e-06,
+ "loss": 3.1546,
+ "step": 435
+ },
+ {
+ "epoch": 0.55808,
+ "grad_norm": 0.35899603366851807,
+ "learning_rate": 4.492799625298953e-06,
+ "loss": 3.3594,
+ "step": 436
+ },
+ {
+ "epoch": 0.55936,
+ "grad_norm": 0.41200870275497437,
+ "learning_rate": 4.471743673537994e-06,
+ "loss": 3.2704,
+ "step": 437
+ },
+ {
+ "epoch": 0.56064,
+ "grad_norm": 0.48116815090179443,
+ "learning_rate": 4.450697191472068e-06,
+ "loss": 3.2729,
+ "step": 438
+ },
+ {
+ "epoch": 0.56192,
+ "grad_norm": 0.37491047382354736,
+ "learning_rate": 4.4296605563872675e-06,
+ "loss": 3.1086,
+ "step": 439
+ },
+ {
+ "epoch": 0.5632,
+ "grad_norm": 0.40018004179000854,
+ "learning_rate": 4.408634145393172e-06,
+ "loss": 3.2337,
+ "step": 440
+ },
+ {
+ "epoch": 0.56448,
+ "grad_norm": 0.44275936484336853,
+ "learning_rate": 4.387618335416073e-06,
+ "loss": 3.2297,
+ "step": 441
+ },
+ {
+ "epoch": 0.56576,
+ "grad_norm": 0.42288705706596375,
+ "learning_rate": 4.366613503192229e-06,
+ "loss": 3.3318,
+ "step": 442
+ },
+ {
+ "epoch": 0.56704,
+ "grad_norm": 0.4355139434337616,
+ "learning_rate": 4.345620025261108e-06,
+ "loss": 3.2496,
+ "step": 443
+ },
+ {
+ "epoch": 0.56832,
+ "grad_norm": 0.41569557785987854,
+ "learning_rate": 4.324638277958634e-06,
+ "loss": 3.3101,
+ "step": 444
+ },
+ {
+ "epoch": 0.5696,
+ "grad_norm": 0.4655461311340332,
+ "learning_rate": 4.303668637410444e-06,
+ "loss": 3.3135,
+ "step": 445
+ },
+ {
+ "epoch": 0.57088,
+ "grad_norm": 0.47526422142982483,
+ "learning_rate": 4.2827114795251514e-06,
+ "loss": 3.318,
+ "step": 446
+ },
+ {
+ "epoch": 0.57216,
+ "grad_norm": 0.42340806126594543,
+ "learning_rate": 4.261767179987595e-06,
+ "loss": 3.1391,
+ "step": 447
+ },
+ {
+ "epoch": 0.57344,
+ "grad_norm": 0.3571900725364685,
+ "learning_rate": 4.240836114252112e-06,
+ "loss": 3.2225,
+ "step": 448
+ },
+ {
+ "epoch": 0.57472,
+ "grad_norm": 0.3726104199886322,
+ "learning_rate": 4.219918657535806e-06,
+ "loss": 3.0541,
+ "step": 449
+ },
+ {
+ "epoch": 0.576,
+ "grad_norm": 0.4537498652935028,
+ "learning_rate": 4.199015184811825e-06,
+ "loss": 3.175,
+ "step": 450
+ },
+ {
+ "epoch": 0.57728,
+ "grad_norm": 0.460414856672287,
+ "learning_rate": 4.17812607080263e-06,
+ "loss": 3.0558,
+ "step": 451
+ },
+ {
+ "epoch": 0.57856,
+ "grad_norm": 0.4168579876422882,
+ "learning_rate": 4.1572516899732845e-06,
+ "loss": 3.2629,
+ "step": 452
+ },
+ {
+ "epoch": 0.57984,
+ "grad_norm": 0.4084390699863434,
+ "learning_rate": 4.136392416524742e-06,
+ "loss": 3.2282,
+ "step": 453
+ },
+ {
+ "epoch": 0.58112,
+ "grad_norm": 0.37755584716796875,
+ "learning_rate": 4.115548624387136e-06,
+ "loss": 3.132,
+ "step": 454
+ },
+ {
+ "epoch": 0.5824,
+ "grad_norm": 0.4635026752948761,
+ "learning_rate": 4.094720687213075e-06,
+ "loss": 3.1166,
+ "step": 455
+ },
+ {
+ "epoch": 0.58368,
+ "grad_norm": 0.4304349720478058,
+ "learning_rate": 4.073908978370945e-06,
+ "loss": 3.2584,
+ "step": 456
+ },
+ {
+ "epoch": 0.58496,
+ "grad_norm": 0.333839476108551,
+ "learning_rate": 4.053113870938224e-06,
+ "loss": 3.0712,
+ "step": 457
+ },
+ {
+ "epoch": 0.58624,
+ "grad_norm": 0.3698170781135559,
+ "learning_rate": 4.032335737694781e-06,
+ "loss": 3.1703,
+ "step": 458
+ },
+ {
+ "epoch": 0.58752,
+ "grad_norm": 0.41348758339881897,
+ "learning_rate": 4.011574951116202e-06,
+ "loss": 3.1395,
+ "step": 459
+ },
+ {
+ "epoch": 0.5888,
+ "grad_norm": 0.41458842158317566,
+ "learning_rate": 3.990831883367114e-06,
+ "loss": 3.3513,
+ "step": 460
+ },
+ {
+ "epoch": 0.59008,
+ "grad_norm": 0.4611389636993408,
+ "learning_rate": 3.970106906294509e-06,
+ "loss": 3.272,
+ "step": 461
+ },
+ {
+ "epoch": 0.59136,
+ "grad_norm": 0.4636837840080261,
+ "learning_rate": 3.949400391421076e-06,
+ "loss": 3.096,
+ "step": 462
+ },
+ {
+ "epoch": 0.59264,
+ "grad_norm": 0.45109736919403076,
+ "learning_rate": 3.9287127099385485e-06,
+ "loss": 3.2003,
+ "step": 463
+ },
+ {
+ "epoch": 0.59392,
+ "grad_norm": 0.38678988814353943,
+ "learning_rate": 3.9080442327010486e-06,
+ "loss": 3.2991,
+ "step": 464
+ },
+ {
+ "epoch": 0.5952,
+ "grad_norm": 0.4240047335624695,
+ "learning_rate": 3.887395330218429e-06,
+ "loss": 3.1706,
+ "step": 465
+ },
+ {
+ "epoch": 0.59648,
+ "grad_norm": 0.46216750144958496,
+ "learning_rate": 3.866766372649645e-06,
+ "loss": 3.1204,
+ "step": 466
+ },
+ {
+ "epoch": 0.59776,
+ "grad_norm": 0.47889336943626404,
+ "learning_rate": 3.846157729796115e-06,
+ "loss": 3.299,
+ "step": 467
+ },
+ {
+ "epoch": 0.59904,
+ "grad_norm": 0.45589450001716614,
+ "learning_rate": 3.825569771095082e-06,
+ "loss": 3.3222,
+ "step": 468
+ },
+ {
+ "epoch": 0.60032,
+ "grad_norm": 0.4282165765762329,
+ "learning_rate": 3.8050028656130046e-06,
+ "loss": 3.2342,
+ "step": 469
+ },
+ {
+ "epoch": 0.6016,
+ "grad_norm": 0.3705621361732483,
+ "learning_rate": 3.7844573820389285e-06,
+ "loss": 3.2554,
+ "step": 470
+ },
+ {
+ "epoch": 0.60288,
+ "grad_norm": 0.36508941650390625,
+ "learning_rate": 3.76393368867789e-06,
+ "loss": 3.1425,
+ "step": 471
+ },
+ {
+ "epoch": 0.60416,
+ "grad_norm": 0.38813576102256775,
+ "learning_rate": 3.743432153444304e-06,
+ "loss": 3.2299,
+ "step": 472
+ },
+ {
+ "epoch": 0.60544,
+ "grad_norm": 0.4241071343421936,
+ "learning_rate": 3.7229531438553666e-06,
+ "loss": 3.3988,
+ "step": 473
+ },
+ {
+ "epoch": 0.60672,
+ "grad_norm": 0.3721230626106262,
+ "learning_rate": 3.702497027024481e-06,
+ "loss": 3.1511,
+ "step": 474
+ },
+ {
+ "epoch": 0.608,
+ "grad_norm": 0.3681172728538513,
+ "learning_rate": 3.682064169654663e-06,
+ "loss": 3.0434,
+ "step": 475
+ },
+ {
+ "epoch": 0.60928,
+ "grad_norm": 0.43913304805755615,
+ "learning_rate": 3.661654938031968e-06,
+ "loss": 3.1311,
+ "step": 476
+ },
+ {
+ "epoch": 0.61056,
+ "grad_norm": 0.4249042868614197,
+ "learning_rate": 3.641269698018934e-06,
+ "loss": 3.3137,
+ "step": 477
+ },
+ {
+ "epoch": 0.61184,
+ "grad_norm": 0.44223716855049133,
+ "learning_rate": 3.6209088150480177e-06,
+ "loss": 3.3463,
+ "step": 478
+ },
+ {
+ "epoch": 0.61312,
+ "grad_norm": 0.3802657127380371,
+ "learning_rate": 3.6005726541150377e-06,
+ "loss": 3.2276,
+ "step": 479
+ },
+ {
+ "epoch": 0.6144,
+ "grad_norm": 0.506656289100647,
+ "learning_rate": 3.5802615797726426e-06,
+ "loss": 3.2908,
+ "step": 480
+ },
+ {
+ "epoch": 0.61568,
+ "grad_norm": 0.4581313729286194,
+ "learning_rate": 3.5599759561237725e-06,
+ "loss": 3.2354,
+ "step": 481
+ },
+ {
+ "epoch": 0.61696,
+ "grad_norm": 0.40165674686431885,
+ "learning_rate": 3.539716146815122e-06,
+ "loss": 3.3592,
+ "step": 482
+ },
+ {
+ "epoch": 0.61824,
+ "grad_norm": 0.380220890045166,
+ "learning_rate": 3.5194825150306365e-06,
+ "loss": 3.1187,
+ "step": 483
+ },
+ {
+ "epoch": 0.61952,
+ "grad_norm": 0.34094783663749695,
+ "learning_rate": 3.4992754234849903e-06,
+ "loss": 3.0677,
+ "step": 484
+ },
+ {
+ "epoch": 0.6208,
+ "grad_norm": 0.4550640881061554,
+ "learning_rate": 3.4790952344170927e-06,
+ "loss": 3.0467,
+ "step": 485
+ },
+ {
+ "epoch": 0.62208,
+ "grad_norm": 0.35535815358161926,
+ "learning_rate": 3.4589423095835884e-06,
+ "loss": 3.2373,
+ "step": 486
+ },
+ {
+ "epoch": 0.62336,
+ "grad_norm": 0.4261285364627838,
+ "learning_rate": 3.4388170102523723e-06,
+ "loss": 3.3266,
+ "step": 487
+ },
+ {
+ "epoch": 0.62464,
+ "grad_norm": 0.44516661763191223,
+ "learning_rate": 3.418719697196119e-06,
+ "loss": 3.2289,
+ "step": 488
+ },
+ {
+ "epoch": 0.62592,
+ "grad_norm": 0.39649543166160583,
+ "learning_rate": 3.398650730685813e-06,
+ "loss": 3.1891,
+ "step": 489
+ },
+ {
+ "epoch": 0.6272,
+ "grad_norm": 0.4533661901950836,
+ "learning_rate": 3.3786104704842837e-06,
+ "loss": 3.2072,
+ "step": 490
+ },
+ {
+ "epoch": 0.62848,
+ "grad_norm": 0.3850015699863434,
+ "learning_rate": 3.3585992758397658e-06,
+ "loss": 3.2083,
+ "step": 491
+ },
+ {
+ "epoch": 0.62976,
+ "grad_norm": 0.41455599665641785,
+ "learning_rate": 3.338617505479458e-06,
+ "loss": 3.2415,
+ "step": 492
+ },
+ {
+ "epoch": 0.63104,
+ "grad_norm": 0.38948673009872437,
+ "learning_rate": 3.318665517603083e-06,
+ "loss": 3.1053,
+ "step": 493
+ },
+ {
+ "epoch": 0.63232,
+ "grad_norm": 0.4419487714767456,
+ "learning_rate": 3.2987436698764764e-06,
+ "loss": 3.2768,
+ "step": 494
+ },
+ {
+ "epoch": 0.6336,
+ "grad_norm": 0.4058285057544708,
+ "learning_rate": 3.278852319425176e-06,
+ "loss": 2.9477,
+ "step": 495
+ },
+ {
+ "epoch": 0.63488,
+ "grad_norm": 0.42230239510536194,
+ "learning_rate": 3.258991822828007e-06,
+ "loss": 3.1691,
+ "step": 496
+ },
+ {
+ "epoch": 0.63616,
+ "grad_norm": 0.40991345047950745,
+ "learning_rate": 3.2391625361107033e-06,
+ "loss": 3.1256,
+ "step": 497
+ },
+ {
+ "epoch": 0.63744,
+ "grad_norm": 0.44282329082489014,
+ "learning_rate": 3.2193648147395225e-06,
+ "loss": 3.2324,
+ "step": 498
+ },
+ {
+ "epoch": 0.63872,
+ "grad_norm": 0.3513530492782593,
+ "learning_rate": 3.199599013614866e-06,
+ "loss": 3.033,
+ "step": 499
+ },
+ {
+ "epoch": 0.64,
+ "grad_norm": 0.32818418741226196,
+ "learning_rate": 3.179865487064926e-06,
+ "loss": 3.137,
+ "step": 500
+ },
+ {
+ "epoch": 0.64128,
+ "grad_norm": 0.3866097331047058,
+ "learning_rate": 3.160164588839328e-06,
+ "loss": 3.1927,
+ "step": 501
+ },
+ {
+ "epoch": 0.64256,
+ "grad_norm": 0.3529679775238037,
+ "learning_rate": 3.140496672102795e-06,
+ "loss": 3.1195,
+ "step": 502
+ },
+ {
+ "epoch": 0.64384,
+ "grad_norm": 0.4259391725063324,
+ "learning_rate": 3.1208620894288105e-06,
+ "loss": 3.2652,
+ "step": 503
+ },
+ {
+ "epoch": 0.64512,
+ "grad_norm": 0.5365630388259888,
+ "learning_rate": 3.1012611927932977e-06,
+ "loss": 3.5371,
+ "step": 504
+ },
+ {
+ "epoch": 0.6464,
+ "grad_norm": 0.4127902090549469,
+ "learning_rate": 3.081694333568323e-06,
+ "loss": 3.3195,
+ "step": 505
+ },
+ {
+ "epoch": 0.64768,
+ "grad_norm": 0.3627055883407593,
+ "learning_rate": 3.062161862515775e-06,
+ "loss": 3.2351,
+ "step": 506
+ },
+ {
+ "epoch": 0.64896,
+ "grad_norm": 0.4089221656322479,
+ "learning_rate": 3.0426641297810977e-06,
+ "loss": 3.0563,
+ "step": 507
+ },
+ {
+ "epoch": 0.65024,
+ "grad_norm": 0.3662814795970917,
+ "learning_rate": 3.023201484886996e-06,
+ "loss": 3.1082,
+ "step": 508
+ },
+ {
+ "epoch": 0.65152,
+ "grad_norm": 0.4775957763195038,
+ "learning_rate": 3.0037742767271894e-06,
+ "loss": 3.3031,
+ "step": 509
+ },
+ {
+ "epoch": 0.6528,
+ "grad_norm": 0.4358830451965332,
+ "learning_rate": 2.98438285356014e-06,
+ "loss": 3.2255,
+ "step": 510
+ },
+ {
+ "epoch": 0.65408,
+ "grad_norm": 0.36608263850212097,
+ "learning_rate": 2.965027563002815e-06,
+ "loss": 3.0036,
+ "step": 511
+ },
+ {
+ "epoch": 0.65536,
+ "grad_norm": 0.4147147536277771,
+ "learning_rate": 2.945708752024463e-06,
+ "loss": 3.1673,
+ "step": 512
+ },
+ {
+ "epoch": 0.65664,
+ "grad_norm": 0.423662006855011,
+ "learning_rate": 2.926426766940384e-06,
+ "loss": 3.2939,
+ "step": 513
+ },
+ {
+ "epoch": 0.65792,
+ "grad_norm": 0.3562603294849396,
+ "learning_rate": 2.9071819534057232e-06,
+ "loss": 3.1551,
+ "step": 514
+ },
+ {
+ "epoch": 0.6592,
+ "grad_norm": 0.4852513372898102,
+ "learning_rate": 2.8879746564092792e-06,
+ "loss": 3.1389,
+ "step": 515
+ },
+ {
+ "epoch": 0.66048,
+ "grad_norm": 0.41819432377815247,
+ "learning_rate": 2.8688052202673188e-06,
+ "loss": 3.2727,
+ "step": 516
+ },
+ {
+ "epoch": 0.66176,
+ "grad_norm": 0.38178080320358276,
+ "learning_rate": 2.8496739886173994e-06,
+ "loss": 3.195,
+ "step": 517
+ },
+ {
+ "epoch": 0.66304,
+ "grad_norm": 0.45543479919433594,
+ "learning_rate": 2.83058130441221e-06,
+ "loss": 3.2634,
+ "step": 518
+ },
+ {
+ "epoch": 0.66432,
+ "grad_norm": 0.4460856318473816,
+ "learning_rate": 2.811527509913431e-06,
+ "loss": 3.1912,
+ "step": 519
+ },
+ {
+ "epoch": 0.6656,
+ "grad_norm": 0.4400303065776825,
+ "learning_rate": 2.7925129466855937e-06,
+ "loss": 3.2751,
+ "step": 520
+ },
+ {
+ "epoch": 0.66688,
+ "grad_norm": 0.3979838788509369,
+ "learning_rate": 2.773537955589951e-06,
+ "loss": 3.1445,
+ "step": 521
+ },
+ {
+ "epoch": 0.66816,
+ "grad_norm": 0.4822862446308136,
+ "learning_rate": 2.754602876778375e-06,
+ "loss": 3.2145,
+ "step": 522
+ },
+ {
+ "epoch": 0.66944,
+ "grad_norm": 0.4274682402610779,
+ "learning_rate": 2.7357080496872624e-06,
+ "loss": 3.2048,
+ "step": 523
+ },
+ {
+ "epoch": 0.67072,
+ "grad_norm": 0.44275274872779846,
+ "learning_rate": 2.716853813031435e-06,
+ "loss": 3.3277,
+ "step": 524
+ },
+ {
+ "epoch": 0.672,
+ "grad_norm": 0.34875813126564026,
+ "learning_rate": 2.6980405047980853e-06,
+ "loss": 3.2432,
+ "step": 525
+ },
+ {
+ "epoch": 0.67328,
+ "grad_norm": 0.41443300247192383,
+ "learning_rate": 2.679268462240708e-06,
+ "loss": 3.2418,
+ "step": 526
+ },
+ {
+ "epoch": 0.67456,
+ "grad_norm": 0.3756672441959381,
+ "learning_rate": 2.660538021873054e-06,
+ "loss": 2.9299,
+ "step": 527
+ },
+ {
+ "epoch": 0.67584,
+ "grad_norm": 0.3978712558746338,
+ "learning_rate": 2.6418495194630996e-06,
+ "loss": 3.1615,
+ "step": 528
+ },
+ {
+ "epoch": 0.67712,
+ "grad_norm": 0.45214176177978516,
+ "learning_rate": 2.6232032900270254e-06,
+ "loss": 3.1528,
+ "step": 529
+ },
+ {
+ "epoch": 0.6784,
+ "grad_norm": 0.5182440280914307,
+ "learning_rate": 2.60459966782322e-06,
+ "loss": 3.3659,
+ "step": 530
+ },
+ {
+ "epoch": 0.67968,
+ "grad_norm": 0.45991212129592896,
+ "learning_rate": 2.5860389863462765e-06,
+ "loss": 3.2059,
+ "step": 531
+ },
+ {
+ "epoch": 0.68096,
+ "grad_norm": 0.42178404331207275,
+ "learning_rate": 2.5675215783210163e-06,
+ "loss": 3.1966,
+ "step": 532
+ },
+ {
+ "epoch": 0.68224,
+ "grad_norm": 0.36444970965385437,
+ "learning_rate": 2.549047775696533e-06,
+ "loss": 3.238,
+ "step": 533
+ },
+ {
+ "epoch": 0.68352,
+ "grad_norm": 0.4043339788913727,
+ "learning_rate": 2.530617909640228e-06,
+ "loss": 3.1494,
+ "step": 534
+ },
+ {
+ "epoch": 0.6848,
+ "grad_norm": 0.42447343468666077,
+ "learning_rate": 2.5122323105318867e-06,
+ "loss": 3.3635,
+ "step": 535
+ },
+ {
+ "epoch": 0.68608,
+ "grad_norm": 0.41279226541519165,
+ "learning_rate": 2.4938913079577498e-06,
+ "loss": 3.3071,
+ "step": 536
+ },
+ {
+ "epoch": 0.68736,
+ "grad_norm": 0.35708916187286377,
+ "learning_rate": 2.475595230704607e-06,
+ "loss": 3.1745,
+ "step": 537
+ },
+ {
+ "epoch": 0.68864,
+ "grad_norm": 0.38239094614982605,
+ "learning_rate": 2.457344406753899e-06,
+ "loss": 3.1061,
+ "step": 538
+ },
+ {
+ "epoch": 0.68992,
+ "grad_norm": 0.43517130613327026,
+ "learning_rate": 2.439139163275841e-06,
+ "loss": 3.256,
+ "step": 539
+ },
+ {
+ "epoch": 0.6912,
+ "grad_norm": 0.36539337038993835,
+ "learning_rate": 2.4209798266235633e-06,
+ "loss": 3.0907,
+ "step": 540
+ },
+ {
+ "epoch": 0.69248,
+ "grad_norm": 0.4229903221130371,
+ "learning_rate": 2.402866722327248e-06,
+ "loss": 3.3267,
+ "step": 541
+ },
+ {
+ "epoch": 0.69376,
+ "grad_norm": 0.46307510137557983,
+ "learning_rate": 2.3848001750883078e-06,
+ "loss": 3.2101,
+ "step": 542
+ },
+ {
+ "epoch": 0.69504,
+ "grad_norm": 0.409217894077301,
+ "learning_rate": 2.366780508773552e-06,
+ "loss": 3.1863,
+ "step": 543
+ },
+ {
+ "epoch": 0.69632,
+ "grad_norm": 0.4121105670928955,
+ "learning_rate": 2.3488080464093943e-06,
+ "loss": 3.1563,
+ "step": 544
+ },
+ {
+ "epoch": 0.6976,
+ "grad_norm": 0.3696381151676178,
+ "learning_rate": 2.330883110176049e-06,
+ "loss": 3.2673,
+ "step": 545
+ },
+ {
+ "epoch": 0.69888,
+ "grad_norm": 0.38089385628700256,
+ "learning_rate": 2.3130060214017607e-06,
+ "loss": 3.123,
+ "step": 546
+ },
+ {
+ "epoch": 0.70016,
+ "grad_norm": 0.43925783038139343,
+ "learning_rate": 2.2951771005570516e-06,
+ "loss": 3.2132,
+ "step": 547
+ },
+ {
+ "epoch": 0.70144,
+ "grad_norm": 0.44682785868644714,
+ "learning_rate": 2.2773966672489666e-06,
+ "loss": 3.3306,
+ "step": 548
+ },
+ {
+ "epoch": 0.70272,
+ "grad_norm": 0.45180270075798035,
+ "learning_rate": 2.2596650402153437e-06,
+ "loss": 3.2011,
+ "step": 549
+ },
+ {
+ "epoch": 0.704,
+ "grad_norm": 0.39848238229751587,
+ "learning_rate": 2.2419825373191095e-06,
+ "loss": 3.1815,
+ "step": 550
+ },
+ {
+ "epoch": 0.70528,
+ "grad_norm": 0.44619321823120117,
+ "learning_rate": 2.2243494755425717e-06,
+ "loss": 3.3634,
+ "step": 551
+ },
+ {
+ "epoch": 0.70656,
+ "grad_norm": 0.42492371797561646,
+ "learning_rate": 2.2067661709817384e-06,
+ "loss": 3.2077,
+ "step": 552
+ },
+ {
+ "epoch": 0.70784,
+ "grad_norm": 0.3781604766845703,
+ "learning_rate": 2.1892329388406586e-06,
+ "loss": 3.1012,
+ "step": 553
+ },
+ {
+ "epoch": 0.70912,
+ "grad_norm": 0.4202871024608612,
+ "learning_rate": 2.171750093425767e-06,
+ "loss": 3.2434,
+ "step": 554
+ },
+ {
+ "epoch": 0.7104,
+ "grad_norm": 0.46861398220062256,
+ "learning_rate": 2.1543179481402443e-06,
+ "loss": 3.2651,
+ "step": 555
+ },
+ {
+ "epoch": 0.71168,
+ "grad_norm": 0.4059511721134186,
+ "learning_rate": 2.1369368154784055e-06,
+ "loss": 3.1735,
+ "step": 556
+ },
+ {
+ "epoch": 0.71296,
+ "grad_norm": 0.3523228168487549,
+ "learning_rate": 2.1196070070200998e-06,
+ "loss": 3.1192,
+ "step": 557
+ },
+ {
+ "epoch": 0.71424,
+ "grad_norm": 0.42841845750808716,
+ "learning_rate": 2.1023288334251225e-06,
+ "loss": 3.2601,
+ "step": 558
+ },
+ {
+ "epoch": 0.71552,
+ "grad_norm": 0.45266565680503845,
+ "learning_rate": 2.0851026044276405e-06,
+ "loss": 3.2528,
+ "step": 559
+ },
+ {
+ "epoch": 0.7168,
+ "grad_norm": 0.47882166504859924,
+ "learning_rate": 2.0679286288306473e-06,
+ "loss": 3.2212,
+ "step": 560
+ },
+ {
+ "epoch": 0.71808,
+ "grad_norm": 0.38681042194366455,
+ "learning_rate": 2.0508072145004284e-06,
+ "loss": 3.2236,
+ "step": 561
+ },
+ {
+ "epoch": 0.71936,
+ "grad_norm": 0.43691524863243103,
+ "learning_rate": 2.0337386683610315e-06,
+ "loss": 3.0605,
+ "step": 562
+ },
+ {
+ "epoch": 0.72064,
+ "grad_norm": 0.37310537695884705,
+ "learning_rate": 2.016723296388779e-06,
+ "loss": 3.1873,
+ "step": 563
+ },
+ {
+ "epoch": 0.72192,
+ "grad_norm": 0.3997083604335785,
+ "learning_rate": 1.9997614036067743e-06,
+ "loss": 3.1234,
+ "step": 564
+ },
+ {
+ "epoch": 0.7232,
+ "grad_norm": 0.44470348954200745,
+ "learning_rate": 1.9828532940794325e-06,
+ "loss": 3.2318,
+ "step": 565
+ },
+ {
+ "epoch": 0.72448,
+ "grad_norm": 0.36877819895744324,
+ "learning_rate": 1.9659992709070346e-06,
+ "loss": 3.0443,
+ "step": 566
+ },
+ {
+ "epoch": 0.72576,
+ "grad_norm": 0.4897530674934387,
+ "learning_rate": 1.94919963622029e-06,
+ "loss": 3.1786,
+ "step": 567
+ },
+ {
+ "epoch": 0.72704,
+ "grad_norm": 0.5083633661270142,
+ "learning_rate": 1.932454691174925e-06,
+ "loss": 3.2537,
+ "step": 568
+ },
+ {
+ "epoch": 0.72832,
+ "grad_norm": 0.41880008578300476,
+ "learning_rate": 1.9157647359462828e-06,
+ "loss": 3.2255,
+ "step": 569
+ },
+ {
+ "epoch": 0.7296,
+ "grad_norm": 0.41470640897750854,
+ "learning_rate": 1.899130069723935e-06,
+ "loss": 3.2095,
+ "step": 570
+ },
+ {
+ "epoch": 0.73088,
+ "grad_norm": 0.3552654981613159,
+ "learning_rate": 1.8825509907063328e-06,
+ "loss": 3.1227,
+ "step": 571
+ },
+ {
+ "epoch": 0.73216,
+ "grad_norm": 0.4061950445175171,
+ "learning_rate": 1.8660277960954465e-06,
+ "loss": 3.2754,
+ "step": 572
+ },
+ {
+ "epoch": 0.73344,
+ "grad_norm": 0.38475826382637024,
+ "learning_rate": 1.8495607820914451e-06,
+ "loss": 3.087,
+ "step": 573
+ },
+ {
+ "epoch": 0.73472,
+ "grad_norm": 0.38132384419441223,
+ "learning_rate": 1.8331502438873883e-06,
+ "loss": 3.1652,
+ "step": 574
+ },
+ {
+ "epoch": 0.736,
+ "grad_norm": 0.3933321535587311,
+ "learning_rate": 1.8167964756639334e-06,
+ "loss": 3.0539,
+ "step": 575
+ },
+ {
+ "epoch": 0.73728,
+ "grad_norm": 0.44638434052467346,
+ "learning_rate": 1.8004997705840565e-06,
+ "loss": 3.2883,
+ "step": 576
+ },
+ {
+ "epoch": 0.73856,
+ "grad_norm": 0.43572214245796204,
+ "learning_rate": 1.7842604207878006e-06,
+ "loss": 3.2324,
+ "step": 577
+ },
+ {
+ "epoch": 0.73984,
+ "grad_norm": 0.4043799638748169,
+ "learning_rate": 1.7680787173870456e-06,
+ "loss": 3.2391,
+ "step": 578
+ },
+ {
+ "epoch": 0.74112,
+ "grad_norm": 0.39593952894210815,
+ "learning_rate": 1.7519549504602751e-06,
+ "loss": 3.1998,
+ "step": 579
+ },
+ {
+ "epoch": 0.7424,
+ "grad_norm": 0.46171334385871887,
+ "learning_rate": 1.7358894090473928e-06,
+ "loss": 3.3026,
+ "step": 580
+ },
+ {
+ "epoch": 0.74368,
+ "grad_norm": 0.43642088770866394,
+ "learning_rate": 1.7198823811445254e-06,
+ "loss": 3.158,
+ "step": 581
+ },
+ {
+ "epoch": 0.74496,
+ "grad_norm": 0.4814632534980774,
+ "learning_rate": 1.7039341536988736e-06,
+ "loss": 3.2261,
+ "step": 582
+ },
+ {
+ "epoch": 0.74624,
+ "grad_norm": 0.3887886703014374,
+ "learning_rate": 1.6880450126035576e-06,
+ "loss": 3.2097,
+ "step": 583
+ },
+ {
+ "epoch": 0.74752,
+ "grad_norm": 0.386666476726532,
+ "learning_rate": 1.6722152426924959e-06,
+ "loss": 3.2009,
+ "step": 584
+ },
+ {
+ "epoch": 0.7488,
+ "grad_norm": 0.3606075048446655,
+ "learning_rate": 1.6564451277353045e-06,
+ "loss": 3.4273,
+ "step": 585
+ },
+ {
+ "epoch": 0.75008,
+ "grad_norm": 0.44366732239723206,
+ "learning_rate": 1.6407349504322034e-06,
+ "loss": 3.217,
+ "step": 586
+ },
+ {
+ "epoch": 0.75136,
+ "grad_norm": 0.4136437177658081,
+ "learning_rate": 1.6250849924089485e-06,
+ "loss": 3.2606,
+ "step": 587
+ },
+ {
+ "epoch": 0.75264,
+ "grad_norm": 0.3872526288032532,
+ "learning_rate": 1.6094955342117852e-06,
+ "loss": 3.2858,
+ "step": 588
+ },
+ {
+ "epoch": 0.75392,
+ "grad_norm": 0.4162059724330902,
+ "learning_rate": 1.5939668553024234e-06,
+ "loss": 3.3008,
+ "step": 589
+ },
+ {
+ "epoch": 0.7552,
+ "grad_norm": 0.5008651614189148,
+ "learning_rate": 1.5784992340530187e-06,
+ "loss": 3.2958,
+ "step": 590
+ },
+ {
+ "epoch": 0.75648,
+ "grad_norm": 0.4136720299720764,
+ "learning_rate": 1.5630929477411899e-06,
+ "loss": 3.1658,
+ "step": 591
+ },
+ {
+ "epoch": 0.75776,
+ "grad_norm": 0.47601282596588135,
+ "learning_rate": 1.5477482725450481e-06,
+ "loss": 3.2161,
+ "step": 592
+ },
+ {
+ "epoch": 0.75904,
+ "grad_norm": 0.38817092776298523,
+ "learning_rate": 1.5324654835382386e-06,
+ "loss": 3.0832,
+ "step": 593
+ },
+ {
+ "epoch": 0.76032,
+ "grad_norm": 0.35280418395996094,
+ "learning_rate": 1.5172448546850166e-06,
+ "loss": 3.0775,
+ "step": 594
+ },
+ {
+ "epoch": 0.7616,
+ "grad_norm": 0.47483620047569275,
+ "learning_rate": 1.5020866588353334e-06,
+ "loss": 3.1745,
+ "step": 595
+ },
+ {
+ "epoch": 0.76288,
+ "grad_norm": 0.42937204241752625,
+ "learning_rate": 1.4869911677199479e-06,
+ "loss": 3.2806,
+ "step": 596
+ },
+ {
+ "epoch": 0.76416,
+ "grad_norm": 0.4562806487083435,
+ "learning_rate": 1.4719586519455536e-06,
+ "loss": 3.2842,
+ "step": 597
+ },
+ {
+ "epoch": 0.76544,
+ "grad_norm": 0.3967139720916748,
+ "learning_rate": 1.4569893809899244e-06,
+ "loss": 3.1516,
+ "step": 598
+ },
+ {
+ "epoch": 0.76672,
+ "grad_norm": 0.47894346714019775,
+ "learning_rate": 1.4420836231970914e-06,
+ "loss": 3.3556,
+ "step": 599
+ },
+ {
+ "epoch": 0.768,
+ "grad_norm": 0.41101381182670593,
+ "learning_rate": 1.427241645772523e-06,
+ "loss": 3.3777,
+ "step": 600
+ },
+ {
+ "epoch": 0.76928,
+ "grad_norm": 0.4031635820865631,
+ "learning_rate": 1.4124637147783431e-06,
+ "loss": 3.1111,
+ "step": 601
+ },
+ {
+ "epoch": 0.77056,
+ "grad_norm": 0.3950492739677429,
+ "learning_rate": 1.397750095128561e-06,
+ "loss": 3.1866,
+ "step": 602
+ },
+ {
+ "epoch": 0.77184,
+ "grad_norm": 0.31918975710868835,
+ "learning_rate": 1.383101050584314e-06,
+ "loss": 3.1123,
+ "step": 603
+ },
+ {
+ "epoch": 0.77312,
+ "grad_norm": 0.38897478580474854,
+ "learning_rate": 1.3685168437491476e-06,
+ "loss": 3.4152,
+ "step": 604
+ },
+ {
+ "epoch": 0.7744,
+ "grad_norm": 0.47953012585639954,
+ "learning_rate": 1.3539977360643054e-06,
+ "loss": 3.2003,
+ "step": 605
+ },
+ {
+ "epoch": 0.77568,
+ "grad_norm": 0.3375909924507141,
+ "learning_rate": 1.3395439878040423e-06,
+ "loss": 3.0269,
+ "step": 606
+ },
+ {
+ "epoch": 0.77696,
+ "grad_norm": 0.3962901830673218,
+ "learning_rate": 1.3251558580709623e-06,
+ "loss": 3.1389,
+ "step": 607
+ },
+ {
+ "epoch": 0.77824,
+ "grad_norm": 0.4136311709880829,
+ "learning_rate": 1.3108336047913633e-06,
+ "loss": 3.2957,
+ "step": 608
+ },
+ {
+ "epoch": 0.77952,
+ "grad_norm": 0.437147855758667,
+ "learning_rate": 1.2965774847106273e-06,
+ "loss": 3.1701,
+ "step": 609
+ },
+ {
+ "epoch": 0.7808,
+ "grad_norm": 0.4272133708000183,
+ "learning_rate": 1.282387753388607e-06,
+ "loss": 3.108,
+ "step": 610
+ },
+ {
+ "epoch": 0.78208,
+ "grad_norm": 0.47195637226104736,
+ "learning_rate": 1.2682646651950476e-06,
+ "loss": 3.1458,
+ "step": 611
+ },
+ {
+ "epoch": 0.78336,
+ "grad_norm": 0.44117170572280884,
+ "learning_rate": 1.2542084733050313e-06,
+ "loss": 3.0431,
+ "step": 612
+ },
+ {
+ "epoch": 0.78464,
+ "grad_norm": 0.3714393675327301,
+ "learning_rate": 1.2402194296944364e-06,
+ "loss": 3.2251,
+ "step": 613
+ },
+ {
+ "epoch": 0.78592,
+ "grad_norm": 0.3848955035209656,
+ "learning_rate": 1.226297785135414e-06,
+ "loss": 3.112,
+ "step": 614
+ },
+ {
+ "epoch": 0.7872,
+ "grad_norm": 0.4060964584350586,
+ "learning_rate": 1.2124437891918995e-06,
+ "loss": 3.127,
+ "step": 615
+ },
+ {
+ "epoch": 0.78848,
+ "grad_norm": 0.4776317775249481,
+ "learning_rate": 1.198657690215142e-06,
+ "loss": 3.2744,
+ "step": 616
+ },
+ {
+ "epoch": 0.78976,
+ "grad_norm": 0.3607068359851837,
+ "learning_rate": 1.18493973533924e-06,
+ "loss": 3.0847,
+ "step": 617
+ },
+ {
+ "epoch": 0.79104,
+ "grad_norm": 0.40119075775146484,
+ "learning_rate": 1.1712901704767254e-06,
+ "loss": 3.1508,
+ "step": 618
+ },
+ {
+ "epoch": 0.79232,
+ "grad_norm": 0.48490265011787415,
+ "learning_rate": 1.1577092403141431e-06,
+ "loss": 3.3611,
+ "step": 619
+ },
+ {
+ "epoch": 0.7936,
+ "grad_norm": 0.4562755227088928,
+ "learning_rate": 1.1441971883076751e-06,
+ "loss": 3.1808,
+ "step": 620
+ },
+ {
+ "epoch": 0.79488,
+ "grad_norm": 0.3519897162914276,
+ "learning_rate": 1.130754256678766e-06,
+ "loss": 3.1289,
+ "step": 621
+ },
+ {
+ "epoch": 0.79616,
+ "grad_norm": 0.48656830191612244,
+ "learning_rate": 1.1173806864097885e-06,
+ "loss": 3.2322,
+ "step": 622
+ },
+ {
+ "epoch": 0.79744,
+ "grad_norm": 0.3918210566043854,
+ "learning_rate": 1.104076717239721e-06,
+ "loss": 3.1986,
+ "step": 623
+ },
+ {
+ "epoch": 0.79872,
+ "grad_norm": 0.4006650447845459,
+ "learning_rate": 1.0908425876598512e-06,
+ "loss": 3.2758,
+ "step": 624
+ },
+ {
+ "epoch": 0.8,
+ "grad_norm": 0.5226495862007141,
+ "learning_rate": 1.0776785349094975e-06,
+ "loss": 3.3191,
+ "step": 625
+ },
+ {
+ "epoch": 0.80128,
+ "grad_norm": 0.4087540805339813,
+ "learning_rate": 1.064584794971757e-06,
+ "loss": 3.2615,
+ "step": 626
+ },
+ {
+ "epoch": 0.80256,
+ "grad_norm": 0.3686673045158386,
+ "learning_rate": 1.0515616025692803e-06,
+ "loss": 3.0902,
+ "step": 627
+ },
+ {
+ "epoch": 0.80384,
+ "grad_norm": 0.42244529724121094,
+ "learning_rate": 1.0386091911600565e-06,
+ "loss": 3.1176,
+ "step": 628
+ },
+ {
+ "epoch": 0.80512,
+ "grad_norm": 0.40941327810287476,
+ "learning_rate": 1.0257277929332332e-06,
+ "loss": 3.2902,
+ "step": 629
+ },
+ {
+ "epoch": 0.8064,
+ "grad_norm": 0.3845454156398773,
+ "learning_rate": 1.012917638804954e-06,
+ "loss": 3.1606,
+ "step": 630
+ },
+ {
+ "epoch": 0.80768,
+ "grad_norm": 0.41114914417266846,
+ "learning_rate": 1.000178958414214e-06,
+ "loss": 3.2061,
+ "step": 631
+ },
+ {
+ "epoch": 0.80896,
+ "grad_norm": 0.435870498418808,
+ "learning_rate": 9.875119801187495e-07,
+ "loss": 3.2187,
+ "step": 632
+ },
+ {
+ "epoch": 0.81024,
+ "grad_norm": 0.4186806082725525,
+ "learning_rate": 9.749169309909385e-07,
+ "loss": 3.3124,
+ "step": 633
+ },
+ {
+ "epoch": 0.81152,
+ "grad_norm": 0.39340364933013916,
+ "learning_rate": 9.62394036813737e-07,
+ "loss": 3.2878,
+ "step": 634
+ },
+ {
+ "epoch": 0.8128,
+ "grad_norm": 0.4385184943675995,
+ "learning_rate": 9.499435220766284e-07,
+ "loss": 3.0919,
+ "step": 635
+ },
+ {
+ "epoch": 0.81408,
+ "grad_norm": 0.41075342893600464,
+ "learning_rate": 9.375656099715935e-07,
+ "loss": 3.1427,
+ "step": 636
+ },
+ {
+ "epoch": 0.81536,
+ "grad_norm": 0.42290058732032776,
+ "learning_rate": 9.252605223891209e-07,
+ "loss": 3.291,
+ "step": 637
+ },
+ {
+ "epoch": 0.81664,
+ "grad_norm": 0.35646817088127136,
+ "learning_rate": 9.130284799142181e-07,
+ "loss": 3.1174,
+ "step": 638
+ },
+ {
+ "epoch": 0.81792,
+ "grad_norm": 0.47008344531059265,
+ "learning_rate": 9.008697018224638e-07,
+ "loss": 3.1031,
+ "step": 639
+ },
+ {
+ "epoch": 0.8192,
+ "grad_norm": 0.4934890568256378,
+ "learning_rate": 8.887844060760759e-07,
+ "loss": 3.2072,
+ "step": 640
+ },
+ {
+ "epoch": 0.82048,
+ "grad_norm": 0.3626819849014282,
+ "learning_rate": 8.767728093200061e-07,
+ "loss": 3.1176,
+ "step": 641
+ },
+ {
+ "epoch": 0.82176,
+ "grad_norm": 0.45377063751220703,
+ "learning_rate": 8.648351268780497e-07,
+ "loss": 3.236,
+ "step": 642
+ },
+ {
+ "epoch": 0.82304,
+ "grad_norm": 0.3831605017185211,
+ "learning_rate": 8.529715727489912e-07,
+ "loss": 3.0761,
+ "step": 643
+ },
+ {
+ "epoch": 0.82432,
+ "grad_norm": 0.3722161650657654,
+ "learning_rate": 8.411823596027685e-07,
+ "loss": 3.1599,
+ "step": 644
+ },
+ {
+ "epoch": 0.8256,
+ "grad_norm": 0.4182044267654419,
+ "learning_rate": 8.294676987766587e-07,
+ "loss": 3.1269,
+ "step": 645
+ },
+ {
+ "epoch": 0.82688,
+ "grad_norm": 0.44710206985473633,
+ "learning_rate": 8.178278002714879e-07,
+ "loss": 3.3266,
+ "step": 646
+ },
+ {
+ "epoch": 0.82816,
+ "grad_norm": 0.43425288796424866,
+ "learning_rate": 8.062628727478683e-07,
+ "loss": 3.2249,
+ "step": 647
+ },
+ {
+ "epoch": 0.82944,
+ "grad_norm": 0.45978471636772156,
+ "learning_rate": 7.947731235224615e-07,
+ "loss": 3.2364,
+ "step": 648
+ },
+ {
+ "epoch": 0.83072,
+ "grad_norm": 0.4050600230693817,
+ "learning_rate": 7.833587585642527e-07,
+ "loss": 3.1364,
+ "step": 649
+ },
+ {
+ "epoch": 0.832,
+ "grad_norm": 0.4046768248081207,
+ "learning_rate": 7.720199824908692e-07,
+ "loss": 3.3118,
+ "step": 650
+ },
+ {
+ "epoch": 0.83328,
+ "grad_norm": 0.38230207562446594,
+ "learning_rate": 7.607569985649066e-07,
+ "loss": 3.1475,
+ "step": 651
+ },
+ {
+ "epoch": 0.83456,
+ "grad_norm": 0.40083029866218567,
+ "learning_rate": 7.49570008690283e-07,
+ "loss": 3.142,
+ "step": 652
+ },
+ {
+ "epoch": 0.83584,
+ "grad_norm": 0.39003580808639526,
+ "learning_rate": 7.384592134086233e-07,
+ "loss": 3.1758,
+ "step": 653
+ },
+ {
+ "epoch": 0.83712,
+ "grad_norm": 0.4197646677494049,
+ "learning_rate": 7.274248118956667e-07,
+ "loss": 3.1587,
+ "step": 654
+ },
+ {
+ "epoch": 0.8384,
+ "grad_norm": 0.43639469146728516,
+ "learning_rate": 7.164670019576869e-07,
+ "loss": 3.126,
+ "step": 655
+ },
+ {
+ "epoch": 0.83968,
+ "grad_norm": 0.33914485573768616,
+ "learning_rate": 7.055859800279579e-07,
+ "loss": 3.0862,
+ "step": 656
+ },
+ {
+ "epoch": 0.84096,
+ "grad_norm": 0.44340065121650696,
+ "learning_rate": 6.947819411632223e-07,
+ "loss": 3.2378,
+ "step": 657
+ },
+ {
+ "epoch": 0.84224,
+ "grad_norm": 0.42810869216918945,
+ "learning_rate": 6.840550790402029e-07,
+ "loss": 3.2728,
+ "step": 658
+ },
+ {
+ "epoch": 0.84352,
+ "grad_norm": 0.3873375952243805,
+ "learning_rate": 6.734055859521249e-07,
+ "loss": 3.2419,
+ "step": 659
+ },
+ {
+ "epoch": 0.8448,
+ "grad_norm": 0.45877885818481445,
+ "learning_rate": 6.628336528052693e-07,
+ "loss": 3.2233,
+ "step": 660
+ },
+ {
+ "epoch": 0.84608,
+ "grad_norm": 0.37361663579940796,
+ "learning_rate": 6.523394691155566e-07,
+ "loss": 3.1335,
+ "step": 661
+ },
+ {
+ "epoch": 0.84736,
+ "grad_norm": 0.3814260959625244,
+ "learning_rate": 6.419232230051431e-07,
+ "loss": 3.1061,
+ "step": 662
+ },
+ {
+ "epoch": 0.84864,
+ "grad_norm": 0.5441914200782776,
+ "learning_rate": 6.315851011990499e-07,
+ "loss": 3.3743,
+ "step": 663
+ },
+ {
+ "epoch": 0.84992,
+ "grad_norm": 0.3911764919757843,
+ "learning_rate": 6.213252890218163e-07,
+ "loss": 3.2657,
+ "step": 664
+ },
+ {
+ "epoch": 0.8512,
+ "grad_norm": 0.3666767477989197,
+ "learning_rate": 6.111439703941796e-07,
+ "loss": 3.1252,
+ "step": 665
+ },
+ {
+ "epoch": 0.85248,
+ "grad_norm": 0.4568954408168793,
+ "learning_rate": 6.010413278297738e-07,
+ "loss": 3.1785,
+ "step": 666
+ },
+ {
+ "epoch": 0.85376,
+ "grad_norm": 0.48718690872192383,
+ "learning_rate": 5.91017542431862e-07,
+ "loss": 3.2092,
+ "step": 667
+ },
+ {
+ "epoch": 0.85504,
+ "grad_norm": 0.4052923619747162,
+ "learning_rate": 5.810727938900879e-07,
+ "loss": 3.2578,
+ "step": 668
+ },
+ {
+ "epoch": 0.85632,
+ "grad_norm": 0.34738919138908386,
+ "learning_rate": 5.712072604772539e-07,
+ "loss": 3.0401,
+ "step": 669
+ },
+ {
+ "epoch": 0.8576,
+ "grad_norm": 0.4067544639110565,
+ "learning_rate": 5.614211190461267e-07,
+ "loss": 3.1373,
+ "step": 670
+ },
+ {
+ "epoch": 0.85888,
+ "grad_norm": 0.4251026511192322,
+ "learning_rate": 5.517145450262639e-07,
+ "loss": 3.1457,
+ "step": 671
+ },
+ {
+ "epoch": 0.86016,
+ "grad_norm": 0.3959572911262512,
+ "learning_rate": 5.420877124208768e-07,
+ "loss": 3.2423,
+ "step": 672
+ },
+ {
+ "epoch": 0.86144,
+ "grad_norm": 0.4151482880115509,
+ "learning_rate": 5.32540793803703e-07,
+ "loss": 3.1768,
+ "step": 673
+ },
+ {
+ "epoch": 0.86272,
+ "grad_norm": 0.41975468397140503,
+ "learning_rate": 5.230739603159151e-07,
+ "loss": 3.1767,
+ "step": 674
+ },
+ {
+ "epoch": 0.864,
+ "grad_norm": 0.45273515582084656,
+ "learning_rate": 5.136873816630572e-07,
+ "loss": 3.3572,
+ "step": 675
+ },
+ {
+ "epoch": 0.86528,
+ "grad_norm": 0.3809858560562134,
+ "learning_rate": 5.043812261119951e-07,
+ "loss": 3.1807,
+ "step": 676
+ },
+ {
+ "epoch": 0.86656,
+ "grad_norm": 0.4853816032409668,
+ "learning_rate": 4.951556604879049e-07,
+ "loss": 3.1771,
+ "step": 677
+ },
+ {
+ "epoch": 0.86784,
+ "grad_norm": 0.40455958247184753,
+ "learning_rate": 4.860108501712824e-07,
+ "loss": 3.2435,
+ "step": 678
+ },
+ {
+ "epoch": 0.86912,
+ "grad_norm": 0.4055866599082947,
+ "learning_rate": 4.769469590949783e-07,
+ "loss": 3.2621,
+ "step": 679
+ },
+ {
+ "epoch": 0.8704,
+ "grad_norm": 0.3917360305786133,
+ "learning_rate": 4.6796414974125625e-07,
+ "loss": 3.1705,
+ "step": 680
+ },
+ {
+ "epoch": 0.87168,
+ "grad_norm": 0.48204123973846436,
+ "learning_rate": 4.590625831388834e-07,
+ "loss": 3.254,
+ "step": 681
+ },
+ {
+ "epoch": 0.87296,
+ "grad_norm": 0.414188951253891,
+ "learning_rate": 4.502424188602439e-07,
+ "loss": 3.2597,
+ "step": 682
+ },
+ {
+ "epoch": 0.87424,
+ "grad_norm": 0.49144065380096436,
+ "learning_rate": 4.4150381501847585e-07,
+ "loss": 3.201,
+ "step": 683
+ },
+ {
+ "epoch": 0.87552,
+ "grad_norm": 0.4012136161327362,
+ "learning_rate": 4.3284692826464024e-07,
+ "loss": 3.1826,
+ "step": 684
+ },
+ {
+ "epoch": 0.8768,
+ "grad_norm": 0.3911590278148651,
+ "learning_rate": 4.242719137849077e-07,
+ "loss": 3.297,
+ "step": 685
+ },
+ {
+ "epoch": 0.87808,
+ "grad_norm": 0.4054816663265228,
+ "learning_rate": 4.1577892529778396e-07,
+ "loss": 3.1724,
+ "step": 686
+ },
+ {
+ "epoch": 0.87936,
+ "grad_norm": 0.36779817938804626,
+ "learning_rate": 4.07368115051347e-07,
+ "loss": 3.0552,
+ "step": 687
+ },
+ {
+ "epoch": 0.88064,
+ "grad_norm": 0.3602275550365448,
+ "learning_rate": 3.9903963382052044e-07,
+ "loss": 3.1554,
+ "step": 688
+ },
+ {
+ "epoch": 0.88192,
+ "grad_norm": 0.36117613315582275,
+ "learning_rate": 3.907936309043753e-07,
+ "loss": 3.2799,
+ "step": 689
+ },
+ {
+ "epoch": 0.8832,
+ "grad_norm": 0.380859375,
+ "learning_rate": 3.826302541234456e-07,
+ "loss": 3.1231,
+ "step": 690
+ },
+ {
+ "epoch": 0.88448,
+ "grad_norm": 0.45788511633872986,
+ "learning_rate": 3.745496498170831e-07,
+ "loss": 3.0921,
+ "step": 691
+ },
+ {
+ "epoch": 0.88576,
+ "grad_norm": 0.3664979934692383,
+ "learning_rate": 3.665519628408332e-07,
+ "loss": 3.25,
+ "step": 692
+ },
+ {
+ "epoch": 0.88704,
+ "grad_norm": 0.4517773389816284,
+ "learning_rate": 3.586373365638385e-07,
+ "loss": 3.1777,
+ "step": 693
+ },
+ {
+ "epoch": 0.88832,
+ "grad_norm": 0.433430016040802,
+ "learning_rate": 3.5080591286626996e-07,
+ "loss": 3.1743,
+ "step": 694
+ },
+ {
+ "epoch": 0.8896,
+ "grad_norm": 0.3954312205314636,
+ "learning_rate": 3.430578321367789e-07,
+ "loss": 3.2248,
+ "step": 695
+ },
+ {
+ "epoch": 0.89088,
+ "grad_norm": 0.42478951811790466,
+ "learning_rate": 3.3539323326998663e-07,
+ "loss": 3.1824,
+ "step": 696
+ },
+ {
+ "epoch": 0.89216,
+ "grad_norm": 0.4399225115776062,
+ "learning_rate": 3.2781225366398885e-07,
+ "loss": 3.2669,
+ "step": 697
+ },
+ {
+ "epoch": 0.89344,
+ "grad_norm": 0.4366867244243622,
+ "learning_rate": 3.203150292178953e-07,
+ "loss": 3.1167,
+ "step": 698
+ },
+ {
+ "epoch": 0.89472,
+ "grad_norm": 0.4125209152698517,
+ "learning_rate": 3.1290169432939556e-07,
+ "loss": 3.3727,
+ "step": 699
+ },
+ {
+ "epoch": 0.896,
+ "grad_norm": 0.4972803592681885,
+ "learning_rate": 3.0557238189234615e-07,
+ "loss": 3.3066,
+ "step": 700
+ }
+ ],
+ "logging_steps": 1,
+ "max_steps": 782,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 1,
+ "save_steps": 100,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": false
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 4.7832958380068045e+17,
+ "train_batch_size": 1,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/training_args.bin b/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..9e22b0644079b6913db532603be2442c22e80651
--- /dev/null
+++ b/phase2_full/checkpoints/marco_nano_base/dan_Latn_50k_exp3_train_only_dan_top8_experts/checkpoint-700/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f027f9806ef546c21ad1922c3546053e56d62a2646ee7af743accedc3dfb9be1
+size 6289