Upload folder using huggingface_hub
Browse files- config.json +64 -0
- experiment_cfg/metadata.json +195 -0
- model-00001-of-00002.safetensors +3 -0
- model-00002-of-00002.safetensors +3 -0
- model.safetensors.index.json +0 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +2834 -0
config.json
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"action_dim": 32,
|
| 3 |
+
"action_head_cfg": {
|
| 4 |
+
"action_dim": 32,
|
| 5 |
+
"action_horizon": 16,
|
| 6 |
+
"add_pos_embed": true,
|
| 7 |
+
"backbone_embedding_dim": 2048,
|
| 8 |
+
"diffusion_model_cfg": {
|
| 9 |
+
"attention_head_dim": 48,
|
| 10 |
+
"cross_attention_dim": 2048,
|
| 11 |
+
"dropout": 0.2,
|
| 12 |
+
"final_dropout": true,
|
| 13 |
+
"interleave_self_attention": true,
|
| 14 |
+
"norm_type": "ada_norm",
|
| 15 |
+
"num_attention_heads": 32,
|
| 16 |
+
"num_layers": 16,
|
| 17 |
+
"output_dim": 1024,
|
| 18 |
+
"positional_embeddings": null
|
| 19 |
+
},
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"input_embedding_dim": 1536,
|
| 22 |
+
"max_action_dim": 32,
|
| 23 |
+
"max_state_dim": 64,
|
| 24 |
+
"model_dtype": "float32",
|
| 25 |
+
"noise_beta_alpha": 1.5,
|
| 26 |
+
"noise_beta_beta": 1.0,
|
| 27 |
+
"noise_s": 0.999,
|
| 28 |
+
"num_inference_timesteps": 4,
|
| 29 |
+
"num_target_vision_tokens": 32,
|
| 30 |
+
"num_timestep_buckets": 1000,
|
| 31 |
+
"tune_diffusion_model": true,
|
| 32 |
+
"tune_projector": true,
|
| 33 |
+
"use_vlln": true,
|
| 34 |
+
"vl_self_attention_cfg": {
|
| 35 |
+
"attention_head_dim": 64,
|
| 36 |
+
"dropout": 0.2,
|
| 37 |
+
"final_dropout": true,
|
| 38 |
+
"num_attention_heads": 32,
|
| 39 |
+
"num_layers": 4,
|
| 40 |
+
"positional_embeddings": null
|
| 41 |
+
}
|
| 42 |
+
},
|
| 43 |
+
"action_horizon": 16,
|
| 44 |
+
"architectures": [
|
| 45 |
+
"GR00T_N1_5"
|
| 46 |
+
],
|
| 47 |
+
"attn_implementation": null,
|
| 48 |
+
"backbone_cfg": {
|
| 49 |
+
"eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops",
|
| 50 |
+
"load_bf16": false,
|
| 51 |
+
"project_to_dim": null,
|
| 52 |
+
"reproject_vision": false,
|
| 53 |
+
"select_layer": 12,
|
| 54 |
+
"tune_llm": false,
|
| 55 |
+
"tune_visual": true,
|
| 56 |
+
"use_flash_attention": true
|
| 57 |
+
},
|
| 58 |
+
"compute_dtype": "bfloat16",
|
| 59 |
+
"hidden_size": 2048,
|
| 60 |
+
"model_dtype": "float32",
|
| 61 |
+
"model_type": "gr00t_n1_5",
|
| 62 |
+
"torch_dtype": "bfloat16",
|
| 63 |
+
"transformers_version": "4.51.3"
|
| 64 |
+
}
|
experiment_cfg/metadata.json
ADDED
|
@@ -0,0 +1,195 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"new_embodiment": {
|
| 3 |
+
"statistics": {
|
| 4 |
+
"state": {
|
| 5 |
+
"single_arm": {
|
| 6 |
+
"max": [
|
| 7 |
+
42.625370025634766,
|
| 8 |
+
59.423675537109375,
|
| 9 |
+
100.0,
|
| 10 |
+
100.0,
|
| 11 |
+
-26.07649040222168
|
| 12 |
+
],
|
| 13 |
+
"min": [
|
| 14 |
+
-37.315635681152344,
|
| 15 |
+
-99.6884765625,
|
| 16 |
+
-66.803466796875,
|
| 17 |
+
-72.73954010009766,
|
| 18 |
+
-83.41802978515625
|
| 19 |
+
],
|
| 20 |
+
"mean": [
|
| 21 |
+
1.3366721868515015,
|
| 22 |
+
-50.05180358886719,
|
| 23 |
+
62.658443450927734,
|
| 24 |
+
19.83676528930664,
|
| 25 |
+
-58.53385925292969
|
| 26 |
+
],
|
| 27 |
+
"std": [
|
| 28 |
+
12.310745239257812,
|
| 29 |
+
47.20203399658203,
|
| 30 |
+
41.652976989746094,
|
| 31 |
+
22.800580978393555,
|
| 32 |
+
7.441578388214111
|
| 33 |
+
],
|
| 34 |
+
"q01": [
|
| 35 |
+
-34.02359970092773,
|
| 36 |
+
-99.45482635498047,
|
| 37 |
+
-47.63337829589844,
|
| 38 |
+
-29.946918182373047,
|
| 39 |
+
-75.55496215820312
|
| 40 |
+
],
|
| 41 |
+
"q99": [
|
| 42 |
+
23.18289070129392,
|
| 43 |
+
49.06542205810547,
|
| 44 |
+
100.0,
|
| 45 |
+
83.53575897216797,
|
| 46 |
+
-36.36159439086918
|
| 47 |
+
]
|
| 48 |
+
},
|
| 49 |
+
"gripper": {
|
| 50 |
+
"max": [
|
| 51 |
+
99.06852722167969
|
| 52 |
+
],
|
| 53 |
+
"min": [
|
| 54 |
+
1.4637391567230225
|
| 55 |
+
],
|
| 56 |
+
"mean": [
|
| 57 |
+
22.766124725341797
|
| 58 |
+
],
|
| 59 |
+
"std": [
|
| 60 |
+
31.40079116821289
|
| 61 |
+
],
|
| 62 |
+
"q01": [
|
| 63 |
+
1.5302728414535522
|
| 64 |
+
],
|
| 65 |
+
"q99": [
|
| 66 |
+
97.07252502441406
|
| 67 |
+
]
|
| 68 |
+
}
|
| 69 |
+
},
|
| 70 |
+
"action": {
|
| 71 |
+
"single_arm": {
|
| 72 |
+
"max": [
|
| 73 |
+
43.19880294799805,
|
| 74 |
+
58.56842041015625,
|
| 75 |
+
100.0,
|
| 76 |
+
100.0,
|
| 77 |
+
-25.854921340942383
|
| 78 |
+
],
|
| 79 |
+
"min": [
|
| 80 |
+
-37.51868438720703,
|
| 81 |
+
-100.0,
|
| 82 |
+
-71.11517333984375,
|
| 83 |
+
-73.90946197509766,
|
| 84 |
+
-84.92227935791016
|
| 85 |
+
],
|
| 86 |
+
"mean": [
|
| 87 |
+
1.4406607151031494,
|
| 88 |
+
-51.057865142822266,
|
| 89 |
+
60.688201904296875,
|
| 90 |
+
19.249435424804688,
|
| 91 |
+
-58.50544738769531
|
| 92 |
+
],
|
| 93 |
+
"std": [
|
| 94 |
+
12.446574211120605,
|
| 95 |
+
46.5860481262207,
|
| 96 |
+
42.87942886352539,
|
| 97 |
+
22.993209838867188,
|
| 98 |
+
7.494980812072754
|
| 99 |
+
],
|
| 100 |
+
"q01": [
|
| 101 |
+
-34.230194091796875,
|
| 102 |
+
-100.0,
|
| 103 |
+
-52.10237503051758,
|
| 104 |
+
-31.1703702545166,
|
| 105 |
+
-75.66631896972656
|
| 106 |
+
],
|
| 107 |
+
"q99": [
|
| 108 |
+
23.943198394775365,
|
| 109 |
+
46.61052703857422,
|
| 110 |
+
100.0,
|
| 111 |
+
83.68065917968745,
|
| 112 |
+
-36.21761703491211
|
| 113 |
+
]
|
| 114 |
+
},
|
| 115 |
+
"gripper": {
|
| 116 |
+
"max": [
|
| 117 |
+
100.0
|
| 118 |
+
],
|
| 119 |
+
"min": [
|
| 120 |
+
0.0
|
| 121 |
+
],
|
| 122 |
+
"mean": [
|
| 123 |
+
22.644290924072266
|
| 124 |
+
],
|
| 125 |
+
"std": [
|
| 126 |
+
37.26095199584961
|
| 127 |
+
],
|
| 128 |
+
"q01": [
|
| 129 |
+
0.0
|
| 130 |
+
],
|
| 131 |
+
"q99": [
|
| 132 |
+
100.0
|
| 133 |
+
]
|
| 134 |
+
}
|
| 135 |
+
}
|
| 136 |
+
},
|
| 137 |
+
"modalities": {
|
| 138 |
+
"video": {
|
| 139 |
+
"front": {
|
| 140 |
+
"resolution": [
|
| 141 |
+
640,
|
| 142 |
+
480
|
| 143 |
+
],
|
| 144 |
+
"channels": 3,
|
| 145 |
+
"fps": 30.0
|
| 146 |
+
},
|
| 147 |
+
"wrist": {
|
| 148 |
+
"resolution": [
|
| 149 |
+
640,
|
| 150 |
+
480
|
| 151 |
+
],
|
| 152 |
+
"channels": 3,
|
| 153 |
+
"fps": 30.0
|
| 154 |
+
}
|
| 155 |
+
},
|
| 156 |
+
"state": {
|
| 157 |
+
"single_arm": {
|
| 158 |
+
"absolute": true,
|
| 159 |
+
"rotation_type": null,
|
| 160 |
+
"shape": [
|
| 161 |
+
5
|
| 162 |
+
],
|
| 163 |
+
"continuous": true
|
| 164 |
+
},
|
| 165 |
+
"gripper": {
|
| 166 |
+
"absolute": true,
|
| 167 |
+
"rotation_type": null,
|
| 168 |
+
"shape": [
|
| 169 |
+
1
|
| 170 |
+
],
|
| 171 |
+
"continuous": true
|
| 172 |
+
}
|
| 173 |
+
},
|
| 174 |
+
"action": {
|
| 175 |
+
"single_arm": {
|
| 176 |
+
"absolute": true,
|
| 177 |
+
"rotation_type": null,
|
| 178 |
+
"shape": [
|
| 179 |
+
5
|
| 180 |
+
],
|
| 181 |
+
"continuous": true
|
| 182 |
+
},
|
| 183 |
+
"gripper": {
|
| 184 |
+
"absolute": true,
|
| 185 |
+
"rotation_type": null,
|
| 186 |
+
"shape": [
|
| 187 |
+
1
|
| 188 |
+
],
|
| 189 |
+
"continuous": true
|
| 190 |
+
}
|
| 191 |
+
}
|
| 192 |
+
},
|
| 193 |
+
"embodiment_tag": "new_embodiment"
|
| 194 |
+
}
|
| 195 |
+
}
|
model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0e65895a7b6c052e58b679bdd68bff54a68694a81f38e23d3e03a92059b3600b
|
| 3 |
+
size 4999367032
|
model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bbf940e8342c555de101c098a2515888e7ac011469c4f15ac759a7386da06f13
|
| 3 |
+
size 2586705312
|
model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:915bb59986c55d5389d7a3d0191ea195363f00a77417af4aed5feeced94a0a16
|
| 3 |
+
size 4147424698
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7df854773850354e3478f76f3d198aca7f62d98ff10ad26ac5a66498f2515f2c
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7211b192985f3cf213c0fe5d877c55ab1774647baa7fcf0d4d0c00aad611b2fd
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2834 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 15.444015444015443,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 4000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.03861003861003861,
|
| 14 |
+
"grad_norm": 9.622601509094238,
|
| 15 |
+
"learning_rate": 1.8e-06,
|
| 16 |
+
"loss": 0.931,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.07722007722007722,
|
| 21 |
+
"grad_norm": 5.078718185424805,
|
| 22 |
+
"learning_rate": 3.8e-06,
|
| 23 |
+
"loss": 0.8664,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.11583011583011583,
|
| 28 |
+
"grad_norm": 1.835941195487976,
|
| 29 |
+
"learning_rate": 5.8e-06,
|
| 30 |
+
"loss": 0.5242,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.15444015444015444,
|
| 35 |
+
"grad_norm": 2.551701307296753,
|
| 36 |
+
"learning_rate": 7.8e-06,
|
| 37 |
+
"loss": 0.3725,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.19305019305019305,
|
| 42 |
+
"grad_norm": 1.223149299621582,
|
| 43 |
+
"learning_rate": 9.800000000000001e-06,
|
| 44 |
+
"loss": 0.3039,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.23166023166023167,
|
| 49 |
+
"grad_norm": 1.3134554624557495,
|
| 50 |
+
"learning_rate": 1.18e-05,
|
| 51 |
+
"loss": 0.288,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.2702702702702703,
|
| 56 |
+
"grad_norm": 1.1588878631591797,
|
| 57 |
+
"learning_rate": 1.3800000000000002e-05,
|
| 58 |
+
"loss": 0.2424,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.3088803088803089,
|
| 63 |
+
"grad_norm": 1.2676962614059448,
|
| 64 |
+
"learning_rate": 1.58e-05,
|
| 65 |
+
"loss": 0.2169,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.3474903474903475,
|
| 70 |
+
"grad_norm": 1.6042144298553467,
|
| 71 |
+
"learning_rate": 1.78e-05,
|
| 72 |
+
"loss": 0.1882,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.3861003861003861,
|
| 77 |
+
"grad_norm": 0.8347752690315247,
|
| 78 |
+
"learning_rate": 1.9800000000000004e-05,
|
| 79 |
+
"loss": 0.1924,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.4247104247104247,
|
| 84 |
+
"grad_norm": 0.7770773768424988,
|
| 85 |
+
"learning_rate": 2.18e-05,
|
| 86 |
+
"loss": 0.1731,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.46332046332046334,
|
| 91 |
+
"grad_norm": 0.7227001190185547,
|
| 92 |
+
"learning_rate": 2.38e-05,
|
| 93 |
+
"loss": 0.1668,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.5019305019305019,
|
| 98 |
+
"grad_norm": 1.5397839546203613,
|
| 99 |
+
"learning_rate": 2.58e-05,
|
| 100 |
+
"loss": 0.1628,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.5405405405405406,
|
| 105 |
+
"grad_norm": 1.3684794902801514,
|
| 106 |
+
"learning_rate": 2.7800000000000005e-05,
|
| 107 |
+
"loss": 0.1614,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.5791505791505791,
|
| 112 |
+
"grad_norm": 1.1917550563812256,
|
| 113 |
+
"learning_rate": 2.98e-05,
|
| 114 |
+
"loss": 0.1521,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.6177606177606177,
|
| 119 |
+
"grad_norm": 1.1747092008590698,
|
| 120 |
+
"learning_rate": 3.18e-05,
|
| 121 |
+
"loss": 0.154,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.6563706563706564,
|
| 126 |
+
"grad_norm": 1.8071014881134033,
|
| 127 |
+
"learning_rate": 3.38e-05,
|
| 128 |
+
"loss": 0.1483,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.694980694980695,
|
| 133 |
+
"grad_norm": 0.7885197997093201,
|
| 134 |
+
"learning_rate": 3.58e-05,
|
| 135 |
+
"loss": 0.1494,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.7335907335907336,
|
| 140 |
+
"grad_norm": 1.5388290882110596,
|
| 141 |
+
"learning_rate": 3.7800000000000004e-05,
|
| 142 |
+
"loss": 0.1316,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.7722007722007722,
|
| 147 |
+
"grad_norm": 0.9852861762046814,
|
| 148 |
+
"learning_rate": 3.9800000000000005e-05,
|
| 149 |
+
"loss": 0.1293,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.8108108108108109,
|
| 154 |
+
"grad_norm": 1.3038679361343384,
|
| 155 |
+
"learning_rate": 4.18e-05,
|
| 156 |
+
"loss": 0.1384,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.8494208494208494,
|
| 161 |
+
"grad_norm": 0.8152418732643127,
|
| 162 |
+
"learning_rate": 4.38e-05,
|
| 163 |
+
"loss": 0.1319,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.888030888030888,
|
| 168 |
+
"grad_norm": 1.0062090158462524,
|
| 169 |
+
"learning_rate": 4.58e-05,
|
| 170 |
+
"loss": 0.1302,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.9266409266409267,
|
| 175 |
+
"grad_norm": 1.2621272802352905,
|
| 176 |
+
"learning_rate": 4.78e-05,
|
| 177 |
+
"loss": 0.133,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.9652509652509652,
|
| 182 |
+
"grad_norm": 0.8130956292152405,
|
| 183 |
+
"learning_rate": 4.9800000000000004e-05,
|
| 184 |
+
"loss": 0.1218,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.0038610038610039,
|
| 189 |
+
"grad_norm": 1.1164008378982544,
|
| 190 |
+
"learning_rate": 5.1800000000000005e-05,
|
| 191 |
+
"loss": 0.1159,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 1.0424710424710424,
|
| 196 |
+
"grad_norm": 1.2107709646224976,
|
| 197 |
+
"learning_rate": 5.380000000000001e-05,
|
| 198 |
+
"loss": 0.1398,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 1.0810810810810811,
|
| 203 |
+
"grad_norm": 1.32729971408844,
|
| 204 |
+
"learning_rate": 5.580000000000001e-05,
|
| 205 |
+
"loss": 0.1229,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.1196911196911197,
|
| 210 |
+
"grad_norm": 0.9606643915176392,
|
| 211 |
+
"learning_rate": 5.7799999999999995e-05,
|
| 212 |
+
"loss": 0.112,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 1.1583011583011582,
|
| 217 |
+
"grad_norm": 1.4731770753860474,
|
| 218 |
+
"learning_rate": 5.9800000000000003e-05,
|
| 219 |
+
"loss": 0.1133,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 1.196911196911197,
|
| 224 |
+
"grad_norm": 0.6345296502113342,
|
| 225 |
+
"learning_rate": 6.18e-05,
|
| 226 |
+
"loss": 0.1142,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.2355212355212355,
|
| 231 |
+
"grad_norm": 1.0257564783096313,
|
| 232 |
+
"learning_rate": 6.38e-05,
|
| 233 |
+
"loss": 0.1204,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 1.2741312741312742,
|
| 238 |
+
"grad_norm": 1.2207335233688354,
|
| 239 |
+
"learning_rate": 6.58e-05,
|
| 240 |
+
"loss": 0.1245,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 1.3127413127413128,
|
| 245 |
+
"grad_norm": 0.8870983719825745,
|
| 246 |
+
"learning_rate": 6.780000000000001e-05,
|
| 247 |
+
"loss": 0.1037,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 1.3513513513513513,
|
| 252 |
+
"grad_norm": 0.5262231826782227,
|
| 253 |
+
"learning_rate": 6.98e-05,
|
| 254 |
+
"loss": 0.1069,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 1.3899613899613898,
|
| 259 |
+
"grad_norm": 0.7265033721923828,
|
| 260 |
+
"learning_rate": 7.18e-05,
|
| 261 |
+
"loss": 0.1,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 1.4285714285714286,
|
| 266 |
+
"grad_norm": 1.1064072847366333,
|
| 267 |
+
"learning_rate": 7.38e-05,
|
| 268 |
+
"loss": 0.1098,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.4671814671814671,
|
| 273 |
+
"grad_norm": 0.9277971982955933,
|
| 274 |
+
"learning_rate": 7.58e-05,
|
| 275 |
+
"loss": 0.1032,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 1.505791505791506,
|
| 280 |
+
"grad_norm": 1.124891996383667,
|
| 281 |
+
"learning_rate": 7.780000000000001e-05,
|
| 282 |
+
"loss": 0.1144,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 1.5444015444015444,
|
| 287 |
+
"grad_norm": 0.719609260559082,
|
| 288 |
+
"learning_rate": 7.98e-05,
|
| 289 |
+
"loss": 0.1004,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.583011583011583,
|
| 294 |
+
"grad_norm": 0.7683860659599304,
|
| 295 |
+
"learning_rate": 8.18e-05,
|
| 296 |
+
"loss": 0.1122,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 1.6216216216216215,
|
| 301 |
+
"grad_norm": 0.7830708622932434,
|
| 302 |
+
"learning_rate": 8.38e-05,
|
| 303 |
+
"loss": 0.1002,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 1.6602316602316602,
|
| 308 |
+
"grad_norm": 1.1331608295440674,
|
| 309 |
+
"learning_rate": 8.58e-05,
|
| 310 |
+
"loss": 0.1163,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.698841698841699,
|
| 315 |
+
"grad_norm": 0.9713950157165527,
|
| 316 |
+
"learning_rate": 8.78e-05,
|
| 317 |
+
"loss": 0.1091,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.7374517374517375,
|
| 322 |
+
"grad_norm": 0.9027305245399475,
|
| 323 |
+
"learning_rate": 8.98e-05,
|
| 324 |
+
"loss": 0.1083,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 1.776061776061776,
|
| 329 |
+
"grad_norm": 0.7454614639282227,
|
| 330 |
+
"learning_rate": 9.180000000000001e-05,
|
| 331 |
+
"loss": 0.1024,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.8146718146718146,
|
| 336 |
+
"grad_norm": 0.8494850397109985,
|
| 337 |
+
"learning_rate": 9.38e-05,
|
| 338 |
+
"loss": 0.1083,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.8532818532818531,
|
| 343 |
+
"grad_norm": 1.129940152168274,
|
| 344 |
+
"learning_rate": 9.58e-05,
|
| 345 |
+
"loss": 0.0981,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 1.8918918918918919,
|
| 350 |
+
"grad_norm": 1.074065089225769,
|
| 351 |
+
"learning_rate": 9.78e-05,
|
| 352 |
+
"loss": 0.1071,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.9305019305019306,
|
| 357 |
+
"grad_norm": 0.9429806470870972,
|
| 358 |
+
"learning_rate": 9.98e-05,
|
| 359 |
+
"loss": 0.0902,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.9691119691119692,
|
| 364 |
+
"grad_norm": 0.6448252201080322,
|
| 365 |
+
"learning_rate": 9.9999778549206e-05,
|
| 366 |
+
"loss": 0.0975,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 2.0077220077220077,
|
| 371 |
+
"grad_norm": 0.7670938968658447,
|
| 372 |
+
"learning_rate": 9.999901304280685e-05,
|
| 373 |
+
"loss": 0.0956,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.0463320463320462,
|
| 378 |
+
"grad_norm": 0.7558152675628662,
|
| 379 |
+
"learning_rate": 9.999770075521164e-05,
|
| 380 |
+
"loss": 0.1038,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 2.0849420849420848,
|
| 385 |
+
"grad_norm": 0.695389449596405,
|
| 386 |
+
"learning_rate": 9.99958417007713e-05,
|
| 387 |
+
"loss": 0.1041,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 2.1235521235521237,
|
| 392 |
+
"grad_norm": 0.5272794365882874,
|
| 393 |
+
"learning_rate": 9.999343589981615e-05,
|
| 394 |
+
"loss": 0.0934,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 2.1621621621621623,
|
| 399 |
+
"grad_norm": 0.6446290612220764,
|
| 400 |
+
"learning_rate": 9.999048337865568e-05,
|
| 401 |
+
"loss": 0.0887,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 2.200772200772201,
|
| 406 |
+
"grad_norm": 0.6806206107139587,
|
| 407 |
+
"learning_rate": 9.998698416957815e-05,
|
| 408 |
+
"loss": 0.0914,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 2.2393822393822393,
|
| 413 |
+
"grad_norm": 0.7006375193595886,
|
| 414 |
+
"learning_rate": 9.998293831085037e-05,
|
| 415 |
+
"loss": 0.1032,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 2.277992277992278,
|
| 420 |
+
"grad_norm": 1.029902696609497,
|
| 421 |
+
"learning_rate": 9.997834584671719e-05,
|
| 422 |
+
"loss": 0.0975,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 2.3166023166023164,
|
| 427 |
+
"grad_norm": 1.0552887916564941,
|
| 428 |
+
"learning_rate": 9.997320682740107e-05,
|
| 429 |
+
"loss": 0.0878,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 2.3552123552123554,
|
| 434 |
+
"grad_norm": 1.2659938335418701,
|
| 435 |
+
"learning_rate": 9.996752130910149e-05,
|
| 436 |
+
"loss": 0.0948,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 2.393822393822394,
|
| 441 |
+
"grad_norm": 1.136674404144287,
|
| 442 |
+
"learning_rate": 9.99612893539944e-05,
|
| 443 |
+
"loss": 0.0871,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 2.4324324324324325,
|
| 448 |
+
"grad_norm": 0.6282244324684143,
|
| 449 |
+
"learning_rate": 9.995451103023144e-05,
|
| 450 |
+
"loss": 0.085,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 2.471042471042471,
|
| 455 |
+
"grad_norm": 0.7048323750495911,
|
| 456 |
+
"learning_rate": 9.994718641193928e-05,
|
| 457 |
+
"loss": 0.0746,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 2.5096525096525095,
|
| 462 |
+
"grad_norm": 0.6798334717750549,
|
| 463 |
+
"learning_rate": 9.993931557921874e-05,
|
| 464 |
+
"loss": 0.0902,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 2.5482625482625485,
|
| 469 |
+
"grad_norm": 0.58338862657547,
|
| 470 |
+
"learning_rate": 9.993089861814402e-05,
|
| 471 |
+
"loss": 0.0753,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 2.586872586872587,
|
| 476 |
+
"grad_norm": 0.7592474818229675,
|
| 477 |
+
"learning_rate": 9.992193562076166e-05,
|
| 478 |
+
"loss": 0.0869,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 2.6254826254826256,
|
| 483 |
+
"grad_norm": 0.9833205938339233,
|
| 484 |
+
"learning_rate": 9.991242668508954e-05,
|
| 485 |
+
"loss": 0.0868,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 2.664092664092664,
|
| 490 |
+
"grad_norm": 0.9708911180496216,
|
| 491 |
+
"learning_rate": 9.990237191511587e-05,
|
| 492 |
+
"loss": 0.0878,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 2.7027027027027026,
|
| 497 |
+
"grad_norm": 0.5875493884086609,
|
| 498 |
+
"learning_rate": 9.989177142079802e-05,
|
| 499 |
+
"loss": 0.0852,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.741312741312741,
|
| 504 |
+
"grad_norm": 0.6512062549591064,
|
| 505 |
+
"learning_rate": 9.988062531806126e-05,
|
| 506 |
+
"loss": 0.0877,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 2.7799227799227797,
|
| 511 |
+
"grad_norm": 1.0817633867263794,
|
| 512 |
+
"learning_rate": 9.986893372879762e-05,
|
| 513 |
+
"loss": 0.0868,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 2.8185328185328187,
|
| 518 |
+
"grad_norm": 0.8996541500091553,
|
| 519 |
+
"learning_rate": 9.985669678086443e-05,
|
| 520 |
+
"loss": 0.0837,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 2.857142857142857,
|
| 525 |
+
"grad_norm": 0.5964193344116211,
|
| 526 |
+
"learning_rate": 9.984391460808298e-05,
|
| 527 |
+
"loss": 0.078,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 2.8957528957528957,
|
| 532 |
+
"grad_norm": 0.49110063910484314,
|
| 533 |
+
"learning_rate": 9.983058735023709e-05,
|
| 534 |
+
"loss": 0.0745,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 2.9343629343629343,
|
| 539 |
+
"grad_norm": 0.853044867515564,
|
| 540 |
+
"learning_rate": 9.98167151530715e-05,
|
| 541 |
+
"loss": 0.0752,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 2.972972972972973,
|
| 546 |
+
"grad_norm": 0.6450267434120178,
|
| 547 |
+
"learning_rate": 9.980229816829034e-05,
|
| 548 |
+
"loss": 0.0843,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 3.011583011583012,
|
| 553 |
+
"grad_norm": 0.5127630829811096,
|
| 554 |
+
"learning_rate": 9.978733655355544e-05,
|
| 555 |
+
"loss": 0.0773,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 3.0501930501930503,
|
| 560 |
+
"grad_norm": 0.7203671336174011,
|
| 561 |
+
"learning_rate": 9.977183047248464e-05,
|
| 562 |
+
"loss": 0.072,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 3.088803088803089,
|
| 567 |
+
"grad_norm": 0.6123097538948059,
|
| 568 |
+
"learning_rate": 9.975578009464992e-05,
|
| 569 |
+
"loss": 0.071,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 3.1274131274131274,
|
| 574 |
+
"grad_norm": 0.491828054189682,
|
| 575 |
+
"learning_rate": 9.97391855955757e-05,
|
| 576 |
+
"loss": 0.0718,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 3.166023166023166,
|
| 581 |
+
"grad_norm": 0.7841641306877136,
|
| 582 |
+
"learning_rate": 9.972204715673669e-05,
|
| 583 |
+
"loss": 0.0668,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 3.2046332046332044,
|
| 588 |
+
"grad_norm": 0.6048414707183838,
|
| 589 |
+
"learning_rate": 9.970436496555617e-05,
|
| 590 |
+
"loss": 0.0625,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 3.2432432432432434,
|
| 595 |
+
"grad_norm": 0.5147990584373474,
|
| 596 |
+
"learning_rate": 9.968613921540373e-05,
|
| 597 |
+
"loss": 0.0732,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 3.281853281853282,
|
| 602 |
+
"grad_norm": 0.6365183591842651,
|
| 603 |
+
"learning_rate": 9.966737010559326e-05,
|
| 604 |
+
"loss": 0.0777,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 3.3204633204633205,
|
| 609 |
+
"grad_norm": 0.6734189987182617,
|
| 610 |
+
"learning_rate": 9.964805784138072e-05,
|
| 611 |
+
"loss": 0.0707,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 3.359073359073359,
|
| 616 |
+
"grad_norm": 0.9035069346427917,
|
| 617 |
+
"learning_rate": 9.962820263396195e-05,
|
| 618 |
+
"loss": 0.0819,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 3.3976833976833976,
|
| 623 |
+
"grad_norm": 0.5493319630622864,
|
| 624 |
+
"learning_rate": 9.960780470047033e-05,
|
| 625 |
+
"loss": 0.0788,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 3.436293436293436,
|
| 630 |
+
"grad_norm": 0.7141504883766174,
|
| 631 |
+
"learning_rate": 9.958686426397437e-05,
|
| 632 |
+
"loss": 0.0795,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 3.474903474903475,
|
| 637 |
+
"grad_norm": 0.8821272253990173,
|
| 638 |
+
"learning_rate": 9.956538155347534e-05,
|
| 639 |
+
"loss": 0.072,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 3.5135135135135136,
|
| 644 |
+
"grad_norm": 0.6582593321800232,
|
| 645 |
+
"learning_rate": 9.95433568039047e-05,
|
| 646 |
+
"loss": 0.0692,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 3.552123552123552,
|
| 651 |
+
"grad_norm": 0.514739453792572,
|
| 652 |
+
"learning_rate": 9.952079025612162e-05,
|
| 653 |
+
"loss": 0.0701,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 3.5907335907335907,
|
| 658 |
+
"grad_norm": 0.9509376883506775,
|
| 659 |
+
"learning_rate": 9.949768215691022e-05,
|
| 660 |
+
"loss": 0.0669,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 3.629343629343629,
|
| 665 |
+
"grad_norm": 0.7678855657577515,
|
| 666 |
+
"learning_rate": 9.9474032758977e-05,
|
| 667 |
+
"loss": 0.0734,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 3.667953667953668,
|
| 672 |
+
"grad_norm": 0.6348609924316406,
|
| 673 |
+
"learning_rate": 9.944984232094794e-05,
|
| 674 |
+
"loss": 0.0815,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 3.7065637065637067,
|
| 679 |
+
"grad_norm": 0.42085355520248413,
|
| 680 |
+
"learning_rate": 9.942511110736584e-05,
|
| 681 |
+
"loss": 0.0752,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 3.7451737451737452,
|
| 686 |
+
"grad_norm": 0.8175306916236877,
|
| 687 |
+
"learning_rate": 9.939983938868726e-05,
|
| 688 |
+
"loss": 0.0682,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 3.7837837837837838,
|
| 693 |
+
"grad_norm": 0.6957929730415344,
|
| 694 |
+
"learning_rate": 9.93740274412797e-05,
|
| 695 |
+
"loss": 0.0661,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 3.8223938223938223,
|
| 700 |
+
"grad_norm": 0.47698843479156494,
|
| 701 |
+
"learning_rate": 9.934767554741846e-05,
|
| 702 |
+
"loss": 0.0711,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 3.861003861003861,
|
| 707 |
+
"grad_norm": 0.7313933372497559,
|
| 708 |
+
"learning_rate": 9.932078399528361e-05,
|
| 709 |
+
"loss": 0.072,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 3.8996138996138994,
|
| 714 |
+
"grad_norm": 0.5685924291610718,
|
| 715 |
+
"learning_rate": 9.929335307895689e-05,
|
| 716 |
+
"loss": 0.065,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 3.9382239382239383,
|
| 721 |
+
"grad_norm": 0.7497240900993347,
|
| 722 |
+
"learning_rate": 9.926538309841839e-05,
|
| 723 |
+
"loss": 0.0613,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 3.976833976833977,
|
| 728 |
+
"grad_norm": 0.6621657013893127,
|
| 729 |
+
"learning_rate": 9.923687435954334e-05,
|
| 730 |
+
"loss": 0.0583,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 4.015444015444015,
|
| 735 |
+
"grad_norm": 0.8020884990692139,
|
| 736 |
+
"learning_rate": 9.920782717409873e-05,
|
| 737 |
+
"loss": 0.0652,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 4.054054054054054,
|
| 742 |
+
"grad_norm": 0.43511030077934265,
|
| 743 |
+
"learning_rate": 9.917824185973994e-05,
|
| 744 |
+
"loss": 0.0633,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 4.0926640926640925,
|
| 749 |
+
"grad_norm": 0.489058256149292,
|
| 750 |
+
"learning_rate": 9.914811874000723e-05,
|
| 751 |
+
"loss": 0.061,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 4.1312741312741315,
|
| 756 |
+
"grad_norm": 0.5102558135986328,
|
| 757 |
+
"learning_rate": 9.911745814432218e-05,
|
| 758 |
+
"loss": 0.0702,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 4.1698841698841695,
|
| 763 |
+
"grad_norm": 0.588080644607544,
|
| 764 |
+
"learning_rate": 9.90862604079842e-05,
|
| 765 |
+
"loss": 0.0566,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 4.2084942084942085,
|
| 770 |
+
"grad_norm": 0.498820960521698,
|
| 771 |
+
"learning_rate": 9.90545258721667e-05,
|
| 772 |
+
"loss": 0.056,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 4.2471042471042475,
|
| 777 |
+
"grad_norm": 0.45086774230003357,
|
| 778 |
+
"learning_rate": 9.90222548839135e-05,
|
| 779 |
+
"loss": 0.0553,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 4.285714285714286,
|
| 784 |
+
"grad_norm": 0.6507774591445923,
|
| 785 |
+
"learning_rate": 9.898944779613495e-05,
|
| 786 |
+
"loss": 0.064,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 4.324324324324325,
|
| 791 |
+
"grad_norm": 0.4545491337776184,
|
| 792 |
+
"learning_rate": 9.89561049676041e-05,
|
| 793 |
+
"loss": 0.0617,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 4.362934362934363,
|
| 798 |
+
"grad_norm": 0.6638012528419495,
|
| 799 |
+
"learning_rate": 9.89222267629528e-05,
|
| 800 |
+
"loss": 0.0797,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 4.401544401544402,
|
| 805 |
+
"grad_norm": 0.7237333059310913,
|
| 806 |
+
"learning_rate": 9.888781355266763e-05,
|
| 807 |
+
"loss": 0.0683,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 4.440154440154441,
|
| 812 |
+
"grad_norm": 0.6219819784164429,
|
| 813 |
+
"learning_rate": 9.885286571308598e-05,
|
| 814 |
+
"loss": 0.0856,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 4.478764478764479,
|
| 819 |
+
"grad_norm": 0.4631660580635071,
|
| 820 |
+
"learning_rate": 9.881738362639182e-05,
|
| 821 |
+
"loss": 0.0742,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 4.517374517374518,
|
| 826 |
+
"grad_norm": 0.6822670698165894,
|
| 827 |
+
"learning_rate": 9.878136768061154e-05,
|
| 828 |
+
"loss": 0.0683,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 4.555984555984556,
|
| 833 |
+
"grad_norm": 0.5832715630531311,
|
| 834 |
+
"learning_rate": 9.874481826960979e-05,
|
| 835 |
+
"loss": 0.0605,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 4.594594594594595,
|
| 840 |
+
"grad_norm": 0.7674006819725037,
|
| 841 |
+
"learning_rate": 9.870773579308503e-05,
|
| 842 |
+
"loss": 0.0672,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 4.633204633204633,
|
| 847 |
+
"grad_norm": 0.5156298279762268,
|
| 848 |
+
"learning_rate": 9.867012065656533e-05,
|
| 849 |
+
"loss": 0.0714,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 4.671814671814672,
|
| 854 |
+
"grad_norm": 0.5317661166191101,
|
| 855 |
+
"learning_rate": 9.863197327140376e-05,
|
| 856 |
+
"loss": 0.0607,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 4.710424710424711,
|
| 861 |
+
"grad_norm": 0.7597309350967407,
|
| 862 |
+
"learning_rate": 9.859329405477403e-05,
|
| 863 |
+
"loss": 0.0631,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 4.749034749034749,
|
| 868 |
+
"grad_norm": 0.5964063405990601,
|
| 869 |
+
"learning_rate": 9.855408342966585e-05,
|
| 870 |
+
"loss": 0.0638,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 4.787644787644788,
|
| 875 |
+
"grad_norm": 0.5615619421005249,
|
| 876 |
+
"learning_rate": 9.851434182488033e-05,
|
| 877 |
+
"loss": 0.0611,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 4.826254826254826,
|
| 882 |
+
"grad_norm": 0.5075325965881348,
|
| 883 |
+
"learning_rate": 9.84740696750253e-05,
|
| 884 |
+
"loss": 0.0613,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 4.864864864864865,
|
| 889 |
+
"grad_norm": 0.5688664317131042,
|
| 890 |
+
"learning_rate": 9.843326742051055e-05,
|
| 891 |
+
"loss": 0.0587,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 4.903474903474903,
|
| 896 |
+
"grad_norm": 0.49671441316604614,
|
| 897 |
+
"learning_rate": 9.839193550754297e-05,
|
| 898 |
+
"loss": 0.0549,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 4.942084942084942,
|
| 903 |
+
"grad_norm": 0.7952184677124023,
|
| 904 |
+
"learning_rate": 9.835007438812177e-05,
|
| 905 |
+
"loss": 0.061,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 4.980694980694981,
|
| 910 |
+
"grad_norm": 0.4357011914253235,
|
| 911 |
+
"learning_rate": 9.830768452003341e-05,
|
| 912 |
+
"loss": 0.0595,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 5.019305019305019,
|
| 917 |
+
"grad_norm": 0.6598787903785706,
|
| 918 |
+
"learning_rate": 9.826476636684671e-05,
|
| 919 |
+
"loss": 0.0667,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 5.057915057915058,
|
| 924 |
+
"grad_norm": 0.6508802771568298,
|
| 925 |
+
"learning_rate": 9.822132039790773e-05,
|
| 926 |
+
"loss": 0.057,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 5.096525096525096,
|
| 931 |
+
"grad_norm": 0.8480905294418335,
|
| 932 |
+
"learning_rate": 9.817734708833461e-05,
|
| 933 |
+
"loss": 0.0622,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 5.135135135135135,
|
| 938 |
+
"grad_norm": 0.6605005860328674,
|
| 939 |
+
"learning_rate": 9.813284691901243e-05,
|
| 940 |
+
"loss": 0.0516,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 5.173745173745174,
|
| 945 |
+
"grad_norm": 0.5409937500953674,
|
| 946 |
+
"learning_rate": 9.808782037658792e-05,
|
| 947 |
+
"loss": 0.0606,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 5.212355212355212,
|
| 952 |
+
"grad_norm": 0.4284970462322235,
|
| 953 |
+
"learning_rate": 9.804226795346411e-05,
|
| 954 |
+
"loss": 0.0539,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 5.250965250965251,
|
| 959 |
+
"grad_norm": 0.5258792042732239,
|
| 960 |
+
"learning_rate": 9.799619014779503e-05,
|
| 961 |
+
"loss": 0.0549,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 5.289575289575289,
|
| 966 |
+
"grad_norm": 0.50562983751297,
|
| 967 |
+
"learning_rate": 9.794958746348013e-05,
|
| 968 |
+
"loss": 0.0567,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 5.328185328185328,
|
| 973 |
+
"grad_norm": 0.5888584852218628,
|
| 974 |
+
"learning_rate": 9.790246041015896e-05,
|
| 975 |
+
"loss": 0.0597,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 5.366795366795367,
|
| 980 |
+
"grad_norm": 0.6059539318084717,
|
| 981 |
+
"learning_rate": 9.785480950320538e-05,
|
| 982 |
+
"loss": 0.0569,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 5.405405405405405,
|
| 987 |
+
"grad_norm": 0.6896811723709106,
|
| 988 |
+
"learning_rate": 9.78066352637221e-05,
|
| 989 |
+
"loss": 0.0584,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 5.444015444015444,
|
| 994 |
+
"grad_norm": 0.48738232254981995,
|
| 995 |
+
"learning_rate": 9.775793821853488e-05,
|
| 996 |
+
"loss": 0.0461,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 5.482625482625482,
|
| 1001 |
+
"grad_norm": 0.4102442264556885,
|
| 1002 |
+
"learning_rate": 9.77087189001868e-05,
|
| 1003 |
+
"loss": 0.0586,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 5.521235521235521,
|
| 1008 |
+
"grad_norm": 0.4332749545574188,
|
| 1009 |
+
"learning_rate": 9.765897784693243e-05,
|
| 1010 |
+
"loss": 0.05,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 5.559845559845559,
|
| 1015 |
+
"grad_norm": 0.3397570550441742,
|
| 1016 |
+
"learning_rate": 9.760871560273197e-05,
|
| 1017 |
+
"loss": 0.0547,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 5.598455598455598,
|
| 1022 |
+
"grad_norm": 0.4303004741668701,
|
| 1023 |
+
"learning_rate": 9.755793271724526e-05,
|
| 1024 |
+
"loss": 0.0529,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 5.637065637065637,
|
| 1029 |
+
"grad_norm": 0.4703856110572815,
|
| 1030 |
+
"learning_rate": 9.750662974582584e-05,
|
| 1031 |
+
"loss": 0.052,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 5.675675675675675,
|
| 1036 |
+
"grad_norm": 0.5116140246391296,
|
| 1037 |
+
"learning_rate": 9.745480724951473e-05,
|
| 1038 |
+
"loss": 0.0584,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 5.714285714285714,
|
| 1043 |
+
"grad_norm": 0.4973282516002655,
|
| 1044 |
+
"learning_rate": 9.740246579503447e-05,
|
| 1045 |
+
"loss": 0.0608,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 5.752895752895753,
|
| 1050 |
+
"grad_norm": 0.4521777033805847,
|
| 1051 |
+
"learning_rate": 9.734960595478284e-05,
|
| 1052 |
+
"loss": 0.0539,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 5.7915057915057915,
|
| 1057 |
+
"grad_norm": 0.47332489490509033,
|
| 1058 |
+
"learning_rate": 9.729622830682657e-05,
|
| 1059 |
+
"loss": 0.0584,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 5.8301158301158305,
|
| 1064 |
+
"grad_norm": 0.663093626499176,
|
| 1065 |
+
"learning_rate": 9.724233343489504e-05,
|
| 1066 |
+
"loss": 0.0589,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 5.8687258687258685,
|
| 1071 |
+
"grad_norm": 0.4286026954650879,
|
| 1072 |
+
"learning_rate": 9.718792192837396e-05,
|
| 1073 |
+
"loss": 0.0591,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 5.9073359073359075,
|
| 1078 |
+
"grad_norm": 0.6002220511436462,
|
| 1079 |
+
"learning_rate": 9.713299438229886e-05,
|
| 1080 |
+
"loss": 0.0571,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 5.945945945945946,
|
| 1085 |
+
"grad_norm": 0.6186395287513733,
|
| 1086 |
+
"learning_rate": 9.707755139734855e-05,
|
| 1087 |
+
"loss": 0.0537,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 5.984555984555985,
|
| 1092 |
+
"grad_norm": 0.5132899284362793,
|
| 1093 |
+
"learning_rate": 9.702159357983866e-05,
|
| 1094 |
+
"loss": 0.0611,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 6.023166023166024,
|
| 1099 |
+
"grad_norm": 0.6831052303314209,
|
| 1100 |
+
"learning_rate": 9.696512154171492e-05,
|
| 1101 |
+
"loss": 0.054,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 6.061776061776062,
|
| 1106 |
+
"grad_norm": 0.4993661940097809,
|
| 1107 |
+
"learning_rate": 9.690813590054645e-05,
|
| 1108 |
+
"loss": 0.0481,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 6.100386100386101,
|
| 1113 |
+
"grad_norm": 0.3774958550930023,
|
| 1114 |
+
"learning_rate": 9.685063727951914e-05,
|
| 1115 |
+
"loss": 0.0486,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 6.138996138996139,
|
| 1120 |
+
"grad_norm": 0.6719356775283813,
|
| 1121 |
+
"learning_rate": 9.679262630742865e-05,
|
| 1122 |
+
"loss": 0.0597,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 6.177606177606178,
|
| 1127 |
+
"grad_norm": 0.43270793557167053,
|
| 1128 |
+
"learning_rate": 9.673410361867373e-05,
|
| 1129 |
+
"loss": 0.0517,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 6.216216216216216,
|
| 1134 |
+
"grad_norm": 0.4987122416496277,
|
| 1135 |
+
"learning_rate": 9.667506985324909e-05,
|
| 1136 |
+
"loss": 0.0527,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 6.254826254826255,
|
| 1141 |
+
"grad_norm": 0.43240949511528015,
|
| 1142 |
+
"learning_rate": 9.661552565673855e-05,
|
| 1143 |
+
"loss": 0.0517,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 6.293436293436294,
|
| 1148 |
+
"grad_norm": 0.357299268245697,
|
| 1149 |
+
"learning_rate": 9.655547168030789e-05,
|
| 1150 |
+
"loss": 0.0495,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 6.332046332046332,
|
| 1155 |
+
"grad_norm": 0.5696622133255005,
|
| 1156 |
+
"learning_rate": 9.649490858069777e-05,
|
| 1157 |
+
"loss": 0.0574,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 6.370656370656371,
|
| 1162 |
+
"grad_norm": 0.4387001693248749,
|
| 1163 |
+
"learning_rate": 9.643383702021658e-05,
|
| 1164 |
+
"loss": 0.0608,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 6.409266409266409,
|
| 1169 |
+
"grad_norm": 0.40567219257354736,
|
| 1170 |
+
"learning_rate": 9.637225766673307e-05,
|
| 1171 |
+
"loss": 0.0496,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 6.447876447876448,
|
| 1176 |
+
"grad_norm": 0.3879441022872925,
|
| 1177 |
+
"learning_rate": 9.631017119366922e-05,
|
| 1178 |
+
"loss": 0.0585,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 6.486486486486487,
|
| 1183 |
+
"grad_norm": 0.42737478017807007,
|
| 1184 |
+
"learning_rate": 9.624757827999273e-05,
|
| 1185 |
+
"loss": 0.05,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 6.525096525096525,
|
| 1190 |
+
"grad_norm": 0.3754231631755829,
|
| 1191 |
+
"learning_rate": 9.618447961020971e-05,
|
| 1192 |
+
"loss": 0.0556,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 6.563706563706564,
|
| 1197 |
+
"grad_norm": 0.5022825002670288,
|
| 1198 |
+
"learning_rate": 9.612087587435707e-05,
|
| 1199 |
+
"loss": 0.0464,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 6.602316602316602,
|
| 1204 |
+
"grad_norm": 0.6352974772453308,
|
| 1205 |
+
"learning_rate": 9.605676776799508e-05,
|
| 1206 |
+
"loss": 0.0554,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 6.640926640926641,
|
| 1211 |
+
"grad_norm": 0.5270296931266785,
|
| 1212 |
+
"learning_rate": 9.599215599219973e-05,
|
| 1213 |
+
"loss": 0.0583,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 6.67953667953668,
|
| 1218 |
+
"grad_norm": 0.36621153354644775,
|
| 1219 |
+
"learning_rate": 9.592704125355505e-05,
|
| 1220 |
+
"loss": 0.0557,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 6.718146718146718,
|
| 1225 |
+
"grad_norm": 0.5127571821212769,
|
| 1226 |
+
"learning_rate": 9.586142426414538e-05,
|
| 1227 |
+
"loss": 0.0512,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 6.756756756756757,
|
| 1232 |
+
"grad_norm": 0.5686606168746948,
|
| 1233 |
+
"learning_rate": 9.57953057415476e-05,
|
| 1234 |
+
"loss": 0.0453,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 6.795366795366795,
|
| 1239 |
+
"grad_norm": 0.3864609897136688,
|
| 1240 |
+
"learning_rate": 9.572868640882328e-05,
|
| 1241 |
+
"loss": 0.0534,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 6.833976833976834,
|
| 1246 |
+
"grad_norm": 0.5303792953491211,
|
| 1247 |
+
"learning_rate": 9.56615669945108e-05,
|
| 1248 |
+
"loss": 0.0534,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 6.872586872586872,
|
| 1253 |
+
"grad_norm": 0.3972107172012329,
|
| 1254 |
+
"learning_rate": 9.55939482326173e-05,
|
| 1255 |
+
"loss": 0.0506,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 6.911196911196911,
|
| 1260 |
+
"grad_norm": 0.40220698714256287,
|
| 1261 |
+
"learning_rate": 9.552583086261069e-05,
|
| 1262 |
+
"loss": 0.0487,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 6.94980694980695,
|
| 1267 |
+
"grad_norm": 0.4272628128528595,
|
| 1268 |
+
"learning_rate": 9.545721562941168e-05,
|
| 1269 |
+
"loss": 0.0462,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 6.988416988416988,
|
| 1274 |
+
"grad_norm": 0.37570661306381226,
|
| 1275 |
+
"learning_rate": 9.538810328338543e-05,
|
| 1276 |
+
"loss": 0.0553,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 7.027027027027027,
|
| 1281 |
+
"grad_norm": 0.3967061936855316,
|
| 1282 |
+
"learning_rate": 9.531849458033349e-05,
|
| 1283 |
+
"loss": 0.0518,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 7.065637065637065,
|
| 1288 |
+
"grad_norm": 0.47626838088035583,
|
| 1289 |
+
"learning_rate": 9.524839028148547e-05,
|
| 1290 |
+
"loss": 0.0533,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 7.104247104247104,
|
| 1295 |
+
"grad_norm": 0.29281380772590637,
|
| 1296 |
+
"learning_rate": 9.517779115349077e-05,
|
| 1297 |
+
"loss": 0.0463,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 7.142857142857143,
|
| 1302 |
+
"grad_norm": 0.3861166834831238,
|
| 1303 |
+
"learning_rate": 9.510669796841014e-05,
|
| 1304 |
+
"loss": 0.053,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 7.181467181467181,
|
| 1309 |
+
"grad_norm": 0.48928484320640564,
|
| 1310 |
+
"learning_rate": 9.503511150370727e-05,
|
| 1311 |
+
"loss": 0.05,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 7.22007722007722,
|
| 1316 |
+
"grad_norm": 0.55223149061203,
|
| 1317 |
+
"learning_rate": 9.496303254224024e-05,
|
| 1318 |
+
"loss": 0.0584,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 7.258687258687258,
|
| 1323 |
+
"grad_norm": 0.46917882561683655,
|
| 1324 |
+
"learning_rate": 9.489046187225306e-05,
|
| 1325 |
+
"loss": 0.055,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 7.297297297297297,
|
| 1330 |
+
"grad_norm": 0.5167844295501709,
|
| 1331 |
+
"learning_rate": 9.481740028736692e-05,
|
| 1332 |
+
"loss": 0.0433,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 7.335907335907336,
|
| 1337 |
+
"grad_norm": 0.7333647012710571,
|
| 1338 |
+
"learning_rate": 9.474384858657164e-05,
|
| 1339 |
+
"loss": 0.0575,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 7.374517374517374,
|
| 1344 |
+
"grad_norm": 0.6839544177055359,
|
| 1345 |
+
"learning_rate": 9.466980757421679e-05,
|
| 1346 |
+
"loss": 0.0495,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 7.413127413127413,
|
| 1351 |
+
"grad_norm": 0.6659579873085022,
|
| 1352 |
+
"learning_rate": 9.459527806000305e-05,
|
| 1353 |
+
"loss": 0.0459,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 7.4517374517374515,
|
| 1358 |
+
"grad_norm": 0.44009920954704285,
|
| 1359 |
+
"learning_rate": 9.452026085897325e-05,
|
| 1360 |
+
"loss": 0.0509,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 7.4903474903474905,
|
| 1365 |
+
"grad_norm": 0.6139705181121826,
|
| 1366 |
+
"learning_rate": 9.444475679150348e-05,
|
| 1367 |
+
"loss": 0.05,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 7.528957528957529,
|
| 1372 |
+
"grad_norm": 0.39309248328208923,
|
| 1373 |
+
"learning_rate": 9.436876668329411e-05,
|
| 1374 |
+
"loss": 0.0425,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 7.5675675675675675,
|
| 1379 |
+
"grad_norm": 0.28272494673728943,
|
| 1380 |
+
"learning_rate": 9.429229136536079e-05,
|
| 1381 |
+
"loss": 0.0439,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 7.6061776061776065,
|
| 1386 |
+
"grad_norm": 0.4705607295036316,
|
| 1387 |
+
"learning_rate": 9.421533167402534e-05,
|
| 1388 |
+
"loss": 0.0507,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 7.644787644787645,
|
| 1393 |
+
"grad_norm": 0.39836612343788147,
|
| 1394 |
+
"learning_rate": 9.413788845090666e-05,
|
| 1395 |
+
"loss": 0.0537,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 7.683397683397684,
|
| 1400 |
+
"grad_norm": 0.437026709318161,
|
| 1401 |
+
"learning_rate": 9.405996254291136e-05,
|
| 1402 |
+
"loss": 0.0568,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 7.722007722007722,
|
| 1407 |
+
"grad_norm": 0.3932258188724518,
|
| 1408 |
+
"learning_rate": 9.398155480222474e-05,
|
| 1409 |
+
"loss": 0.0424,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 7.760617760617761,
|
| 1414 |
+
"grad_norm": 0.3578822612762451,
|
| 1415 |
+
"learning_rate": 9.390266608630128e-05,
|
| 1416 |
+
"loss": 0.0463,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 7.799227799227799,
|
| 1421 |
+
"grad_norm": 0.5391870141029358,
|
| 1422 |
+
"learning_rate": 9.38232972578553e-05,
|
| 1423 |
+
"loss": 0.0475,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 7.837837837837838,
|
| 1428 |
+
"grad_norm": 0.42705076932907104,
|
| 1429 |
+
"learning_rate": 9.374344918485164e-05,
|
| 1430 |
+
"loss": 0.0488,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 7.876447876447877,
|
| 1435 |
+
"grad_norm": 0.4739801287651062,
|
| 1436 |
+
"learning_rate": 9.366312274049602e-05,
|
| 1437 |
+
"loss": 0.0527,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 7.915057915057915,
|
| 1442 |
+
"grad_norm": 0.693805456161499,
|
| 1443 |
+
"learning_rate": 9.358231880322554e-05,
|
| 1444 |
+
"loss": 0.0453,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 7.953667953667954,
|
| 1449 |
+
"grad_norm": 0.34889355301856995,
|
| 1450 |
+
"learning_rate": 9.350103825669916e-05,
|
| 1451 |
+
"loss": 0.0431,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 7.992277992277993,
|
| 1456 |
+
"grad_norm": 0.4197182357311249,
|
| 1457 |
+
"learning_rate": 9.341928198978787e-05,
|
| 1458 |
+
"loss": 0.0454,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 8.03088803088803,
|
| 1463 |
+
"grad_norm": 0.4793173670768738,
|
| 1464 |
+
"learning_rate": 9.333705089656512e-05,
|
| 1465 |
+
"loss": 0.0451,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 8.069498069498069,
|
| 1470 |
+
"grad_norm": 0.41119933128356934,
|
| 1471 |
+
"learning_rate": 9.325434587629698e-05,
|
| 1472 |
+
"loss": 0.0417,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 8.108108108108109,
|
| 1477 |
+
"grad_norm": 0.5104618072509766,
|
| 1478 |
+
"learning_rate": 9.31711678334323e-05,
|
| 1479 |
+
"loss": 0.0486,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 8.146718146718147,
|
| 1484 |
+
"grad_norm": 0.3389429748058319,
|
| 1485 |
+
"learning_rate": 9.308751767759282e-05,
|
| 1486 |
+
"loss": 0.0453,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 8.185328185328185,
|
| 1491 |
+
"grad_norm": 0.40160703659057617,
|
| 1492 |
+
"learning_rate": 9.300339632356325e-05,
|
| 1493 |
+
"loss": 0.039,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 8.223938223938223,
|
| 1498 |
+
"grad_norm": 0.5155038833618164,
|
| 1499 |
+
"learning_rate": 9.291880469128124e-05,
|
| 1500 |
+
"loss": 0.0418,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 8.262548262548263,
|
| 1505 |
+
"grad_norm": 0.49120837450027466,
|
| 1506 |
+
"learning_rate": 9.283374370582732e-05,
|
| 1507 |
+
"loss": 0.0432,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 8.301158301158301,
|
| 1512 |
+
"grad_norm": 0.4529905915260315,
|
| 1513 |
+
"learning_rate": 9.274821429741482e-05,
|
| 1514 |
+
"loss": 0.056,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 8.339768339768339,
|
| 1519 |
+
"grad_norm": 0.3293595016002655,
|
| 1520 |
+
"learning_rate": 9.266221740137961e-05,
|
| 1521 |
+
"loss": 0.0492,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 8.378378378378379,
|
| 1526 |
+
"grad_norm": 0.4065215289592743,
|
| 1527 |
+
"learning_rate": 9.257575395817001e-05,
|
| 1528 |
+
"loss": 0.0547,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 8.416988416988417,
|
| 1533 |
+
"grad_norm": 0.4577319920063019,
|
| 1534 |
+
"learning_rate": 9.248882491333637e-05,
|
| 1535 |
+
"loss": 0.0425,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 8.455598455598455,
|
| 1540 |
+
"grad_norm": 0.5668477416038513,
|
| 1541 |
+
"learning_rate": 9.240143121752076e-05,
|
| 1542 |
+
"loss": 0.0517,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 8.494208494208495,
|
| 1547 |
+
"grad_norm": 0.5102901458740234,
|
| 1548 |
+
"learning_rate": 9.23135738264467e-05,
|
| 1549 |
+
"loss": 0.0364,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 8.532818532818533,
|
| 1554 |
+
"grad_norm": 0.4866316020488739,
|
| 1555 |
+
"learning_rate": 9.222525370090849e-05,
|
| 1556 |
+
"loss": 0.0436,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 8.571428571428571,
|
| 1561 |
+
"grad_norm": 0.32864415645599365,
|
| 1562 |
+
"learning_rate": 9.213647180676088e-05,
|
| 1563 |
+
"loss": 0.044,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 8.61003861003861,
|
| 1568 |
+
"grad_norm": 0.4218500852584839,
|
| 1569 |
+
"learning_rate": 9.204722911490846e-05,
|
| 1570 |
+
"loss": 0.0461,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 8.64864864864865,
|
| 1575 |
+
"grad_norm": 0.5844140648841858,
|
| 1576 |
+
"learning_rate": 9.1957526601295e-05,
|
| 1577 |
+
"loss": 0.047,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 8.687258687258687,
|
| 1582 |
+
"grad_norm": 0.37406060099601746,
|
| 1583 |
+
"learning_rate": 9.186736524689281e-05,
|
| 1584 |
+
"loss": 0.0539,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 8.725868725868725,
|
| 1589 |
+
"grad_norm": 0.4267013967037201,
|
| 1590 |
+
"learning_rate": 9.177674603769204e-05,
|
| 1591 |
+
"loss": 0.0529,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 8.764478764478765,
|
| 1596 |
+
"grad_norm": 0.46270227432250977,
|
| 1597 |
+
"learning_rate": 9.168566996468983e-05,
|
| 1598 |
+
"loss": 0.0481,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 8.803088803088803,
|
| 1603 |
+
"grad_norm": 0.5886604189872742,
|
| 1604 |
+
"learning_rate": 9.159413802387951e-05,
|
| 1605 |
+
"loss": 0.0458,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 8.841698841698841,
|
| 1610 |
+
"grad_norm": 0.3392360210418701,
|
| 1611 |
+
"learning_rate": 9.150215121623974e-05,
|
| 1612 |
+
"loss": 0.0396,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 8.880308880308881,
|
| 1617 |
+
"grad_norm": 0.5146806836128235,
|
| 1618 |
+
"learning_rate": 9.140971054772349e-05,
|
| 1619 |
+
"loss": 0.0432,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 8.91891891891892,
|
| 1624 |
+
"grad_norm": 0.30445799231529236,
|
| 1625 |
+
"learning_rate": 9.131681702924713e-05,
|
| 1626 |
+
"loss": 0.0475,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 8.957528957528957,
|
| 1631 |
+
"grad_norm": 0.5488026142120361,
|
| 1632 |
+
"learning_rate": 9.122347167667926e-05,
|
| 1633 |
+
"loss": 0.0459,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 8.996138996138995,
|
| 1638 |
+
"grad_norm": 0.44879812002182007,
|
| 1639 |
+
"learning_rate": 9.112967551082973e-05,
|
| 1640 |
+
"loss": 0.0456,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 9.034749034749035,
|
| 1645 |
+
"grad_norm": 0.413736492395401,
|
| 1646 |
+
"learning_rate": 9.103542955743835e-05,
|
| 1647 |
+
"loss": 0.0493,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 9.073359073359073,
|
| 1652 |
+
"grad_norm": 0.6810268759727478,
|
| 1653 |
+
"learning_rate": 9.094073484716381e-05,
|
| 1654 |
+
"loss": 0.0457,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 9.111969111969112,
|
| 1659 |
+
"grad_norm": 0.3536568582057953,
|
| 1660 |
+
"learning_rate": 9.084559241557226e-05,
|
| 1661 |
+
"loss": 0.0392,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 9.150579150579151,
|
| 1666 |
+
"grad_norm": 0.5432239174842834,
|
| 1667 |
+
"learning_rate": 9.075000330312608e-05,
|
| 1668 |
+
"loss": 0.0426,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 9.18918918918919,
|
| 1673 |
+
"grad_norm": 0.3981418013572693,
|
| 1674 |
+
"learning_rate": 9.065396855517253e-05,
|
| 1675 |
+
"loss": 0.0384,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 9.227799227799228,
|
| 1680 |
+
"grad_norm": 0.41987404227256775,
|
| 1681 |
+
"learning_rate": 9.055748922193219e-05,
|
| 1682 |
+
"loss": 0.0413,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 9.266409266409266,
|
| 1687 |
+
"grad_norm": 0.44830331206321716,
|
| 1688 |
+
"learning_rate": 9.046056635848761e-05,
|
| 1689 |
+
"loss": 0.0485,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 9.305019305019306,
|
| 1694 |
+
"grad_norm": 0.46031081676483154,
|
| 1695 |
+
"learning_rate": 9.036320102477169e-05,
|
| 1696 |
+
"loss": 0.0425,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 9.343629343629344,
|
| 1701 |
+
"grad_norm": 0.6913877129554749,
|
| 1702 |
+
"learning_rate": 9.02653942855561e-05,
|
| 1703 |
+
"loss": 0.041,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 9.382239382239382,
|
| 1708 |
+
"grad_norm": 0.4454318583011627,
|
| 1709 |
+
"learning_rate": 9.016714721043971e-05,
|
| 1710 |
+
"loss": 0.0455,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 9.420849420849422,
|
| 1715 |
+
"grad_norm": 0.4421475827693939,
|
| 1716 |
+
"learning_rate": 9.006846087383675e-05,
|
| 1717 |
+
"loss": 0.0412,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 9.45945945945946,
|
| 1722 |
+
"grad_norm": 0.3897337019443512,
|
| 1723 |
+
"learning_rate": 8.996933635496523e-05,
|
| 1724 |
+
"loss": 0.0385,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 9.498069498069498,
|
| 1729 |
+
"grad_norm": 0.49044185876846313,
|
| 1730 |
+
"learning_rate": 8.986977473783498e-05,
|
| 1731 |
+
"loss": 0.0408,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 9.536679536679536,
|
| 1736 |
+
"grad_norm": 0.4835743308067322,
|
| 1737 |
+
"learning_rate": 8.97697771112359e-05,
|
| 1738 |
+
"loss": 0.0406,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 9.575289575289576,
|
| 1743 |
+
"grad_norm": 0.5740953683853149,
|
| 1744 |
+
"learning_rate": 8.966934456872602e-05,
|
| 1745 |
+
"loss": 0.0375,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 9.613899613899614,
|
| 1750 |
+
"grad_norm": 0.44186949729919434,
|
| 1751 |
+
"learning_rate": 8.95684782086195e-05,
|
| 1752 |
+
"loss": 0.0424,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 9.652509652509652,
|
| 1757 |
+
"grad_norm": 0.351376473903656,
|
| 1758 |
+
"learning_rate": 8.946717913397476e-05,
|
| 1759 |
+
"loss": 0.0394,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 9.691119691119692,
|
| 1764 |
+
"grad_norm": 0.4355716109275818,
|
| 1765 |
+
"learning_rate": 8.93654484525822e-05,
|
| 1766 |
+
"loss": 0.0426,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 9.72972972972973,
|
| 1771 |
+
"grad_norm": 0.43863701820373535,
|
| 1772 |
+
"learning_rate": 8.926328727695226e-05,
|
| 1773 |
+
"loss": 0.0423,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 9.768339768339768,
|
| 1778 |
+
"grad_norm": 0.4783848524093628,
|
| 1779 |
+
"learning_rate": 8.916069672430319e-05,
|
| 1780 |
+
"loss": 0.031,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 9.806949806949808,
|
| 1785 |
+
"grad_norm": 0.4009969234466553,
|
| 1786 |
+
"learning_rate": 8.905767791654884e-05,
|
| 1787 |
+
"loss": 0.0454,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 9.845559845559846,
|
| 1792 |
+
"grad_norm": 0.4583117365837097,
|
| 1793 |
+
"learning_rate": 8.895423198028638e-05,
|
| 1794 |
+
"loss": 0.0376,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 9.884169884169884,
|
| 1799 |
+
"grad_norm": 0.35600435733795166,
|
| 1800 |
+
"learning_rate": 8.885036004678402e-05,
|
| 1801 |
+
"loss": 0.049,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 9.922779922779922,
|
| 1806 |
+
"grad_norm": 0.3151102364063263,
|
| 1807 |
+
"learning_rate": 8.874606325196857e-05,
|
| 1808 |
+
"loss": 0.0397,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 9.961389961389962,
|
| 1813 |
+
"grad_norm": 0.459255188703537,
|
| 1814 |
+
"learning_rate": 8.864134273641304e-05,
|
| 1815 |
+
"loss": 0.0436,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 10.0,
|
| 1820 |
+
"grad_norm": 0.38972389698028564,
|
| 1821 |
+
"learning_rate": 8.853619964532427e-05,
|
| 1822 |
+
"loss": 0.0357,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 10.038610038610038,
|
| 1827 |
+
"grad_norm": 0.33490484952926636,
|
| 1828 |
+
"learning_rate": 8.843063512853019e-05,
|
| 1829 |
+
"loss": 0.0426,
|
| 1830 |
+
"step": 2600
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 10.077220077220078,
|
| 1834 |
+
"grad_norm": 0.348417192697525,
|
| 1835 |
+
"learning_rate": 8.832465034046749e-05,
|
| 1836 |
+
"loss": 0.0432,
|
| 1837 |
+
"step": 2610
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 10.115830115830116,
|
| 1841 |
+
"grad_norm": 0.4167700409889221,
|
| 1842 |
+
"learning_rate": 8.821824644016882e-05,
|
| 1843 |
+
"loss": 0.0386,
|
| 1844 |
+
"step": 2620
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 10.154440154440154,
|
| 1848 |
+
"grad_norm": 0.3185581564903259,
|
| 1849 |
+
"learning_rate": 8.811142459125019e-05,
|
| 1850 |
+
"loss": 0.0369,
|
| 1851 |
+
"step": 2630
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 10.193050193050192,
|
| 1855 |
+
"grad_norm": 0.306863933801651,
|
| 1856 |
+
"learning_rate": 8.800418596189822e-05,
|
| 1857 |
+
"loss": 0.0423,
|
| 1858 |
+
"step": 2640
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 10.231660231660232,
|
| 1862 |
+
"grad_norm": 0.3122132122516632,
|
| 1863 |
+
"learning_rate": 8.789653172485737e-05,
|
| 1864 |
+
"loss": 0.0316,
|
| 1865 |
+
"step": 2650
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 10.27027027027027,
|
| 1869 |
+
"grad_norm": 0.326272577047348,
|
| 1870 |
+
"learning_rate": 8.778846305741715e-05,
|
| 1871 |
+
"loss": 0.045,
|
| 1872 |
+
"step": 2660
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 10.308880308880308,
|
| 1876 |
+
"grad_norm": 0.4160739779472351,
|
| 1877 |
+
"learning_rate": 8.767998114139918e-05,
|
| 1878 |
+
"loss": 0.0429,
|
| 1879 |
+
"step": 2670
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 10.347490347490348,
|
| 1883 |
+
"grad_norm": 0.2969798147678375,
|
| 1884 |
+
"learning_rate": 8.757108716314429e-05,
|
| 1885 |
+
"loss": 0.04,
|
| 1886 |
+
"step": 2680
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 10.386100386100386,
|
| 1890 |
+
"grad_norm": 0.4636528789997101,
|
| 1891 |
+
"learning_rate": 8.746178231349962e-05,
|
| 1892 |
+
"loss": 0.0467,
|
| 1893 |
+
"step": 2690
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 10.424710424710424,
|
| 1897 |
+
"grad_norm": 0.4343163073062897,
|
| 1898 |
+
"learning_rate": 8.735206778780549e-05,
|
| 1899 |
+
"loss": 0.0432,
|
| 1900 |
+
"step": 2700
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 10.463320463320464,
|
| 1904 |
+
"grad_norm": 0.39633092284202576,
|
| 1905 |
+
"learning_rate": 8.724194478588234e-05,
|
| 1906 |
+
"loss": 0.0428,
|
| 1907 |
+
"step": 2710
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 10.501930501930502,
|
| 1911 |
+
"grad_norm": 0.53948575258255,
|
| 1912 |
+
"learning_rate": 8.713141451201772e-05,
|
| 1913 |
+
"loss": 0.0457,
|
| 1914 |
+
"step": 2720
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 10.54054054054054,
|
| 1918 |
+
"grad_norm": 0.5596122741699219,
|
| 1919 |
+
"learning_rate": 8.702047817495295e-05,
|
| 1920 |
+
"loss": 0.0465,
|
| 1921 |
+
"step": 2730
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 10.579150579150578,
|
| 1925 |
+
"grad_norm": 0.41859936714172363,
|
| 1926 |
+
"learning_rate": 8.69091369878701e-05,
|
| 1927 |
+
"loss": 0.0421,
|
| 1928 |
+
"step": 2740
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 10.617760617760618,
|
| 1932 |
+
"grad_norm": 0.43876025080680847,
|
| 1933 |
+
"learning_rate": 8.679739216837849e-05,
|
| 1934 |
+
"loss": 0.0375,
|
| 1935 |
+
"step": 2750
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 10.656370656370656,
|
| 1939 |
+
"grad_norm": 0.3008074462413788,
|
| 1940 |
+
"learning_rate": 8.66852449385016e-05,
|
| 1941 |
+
"loss": 0.0399,
|
| 1942 |
+
"step": 2760
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 10.694980694980694,
|
| 1946 |
+
"grad_norm": 0.45133543014526367,
|
| 1947 |
+
"learning_rate": 8.657269652466356e-05,
|
| 1948 |
+
"loss": 0.0402,
|
| 1949 |
+
"step": 2770
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 10.733590733590734,
|
| 1953 |
+
"grad_norm": 0.4804655909538269,
|
| 1954 |
+
"learning_rate": 8.645974815767577e-05,
|
| 1955 |
+
"loss": 0.0439,
|
| 1956 |
+
"step": 2780
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 10.772200772200772,
|
| 1960 |
+
"grad_norm": 0.39003893733024597,
|
| 1961 |
+
"learning_rate": 8.634640107272351e-05,
|
| 1962 |
+
"loss": 0.0422,
|
| 1963 |
+
"step": 2790
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 10.81081081081081,
|
| 1967 |
+
"grad_norm": 0.4365558624267578,
|
| 1968 |
+
"learning_rate": 8.623265650935234e-05,
|
| 1969 |
+
"loss": 0.0354,
|
| 1970 |
+
"step": 2800
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 10.849420849420849,
|
| 1974 |
+
"grad_norm": 0.3692835867404938,
|
| 1975 |
+
"learning_rate": 8.611851571145456e-05,
|
| 1976 |
+
"loss": 0.0391,
|
| 1977 |
+
"step": 2810
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 10.888030888030888,
|
| 1981 |
+
"grad_norm": 0.404334157705307,
|
| 1982 |
+
"learning_rate": 8.600397992725566e-05,
|
| 1983 |
+
"loss": 0.0428,
|
| 1984 |
+
"step": 2820
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 10.926640926640927,
|
| 1988 |
+
"grad_norm": 0.8677656054496765,
|
| 1989 |
+
"learning_rate": 8.588905040930061e-05,
|
| 1990 |
+
"loss": 0.0449,
|
| 1991 |
+
"step": 2830
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 10.965250965250965,
|
| 1995 |
+
"grad_norm": 0.5743116736412048,
|
| 1996 |
+
"learning_rate": 8.577372841444022e-05,
|
| 1997 |
+
"loss": 0.0416,
|
| 1998 |
+
"step": 2840
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 11.003861003861005,
|
| 2002 |
+
"grad_norm": 0.38388240337371826,
|
| 2003 |
+
"learning_rate": 8.565801520381736e-05,
|
| 2004 |
+
"loss": 0.0465,
|
| 2005 |
+
"step": 2850
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 11.042471042471043,
|
| 2009 |
+
"grad_norm": 0.3360496461391449,
|
| 2010 |
+
"learning_rate": 8.554191204285313e-05,
|
| 2011 |
+
"loss": 0.0441,
|
| 2012 |
+
"step": 2860
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 11.08108108108108,
|
| 2016 |
+
"grad_norm": 0.5388212203979492,
|
| 2017 |
+
"learning_rate": 8.542542020123315e-05,
|
| 2018 |
+
"loss": 0.0473,
|
| 2019 |
+
"step": 2870
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 11.11969111969112,
|
| 2023 |
+
"grad_norm": 0.4236580431461334,
|
| 2024 |
+
"learning_rate": 8.530854095289347e-05,
|
| 2025 |
+
"loss": 0.0412,
|
| 2026 |
+
"step": 2880
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 11.158301158301159,
|
| 2030 |
+
"grad_norm": 0.23769927024841309,
|
| 2031 |
+
"learning_rate": 8.519127557600688e-05,
|
| 2032 |
+
"loss": 0.0348,
|
| 2033 |
+
"step": 2890
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 11.196911196911197,
|
| 2037 |
+
"grad_norm": 0.4001792073249817,
|
| 2038 |
+
"learning_rate": 8.507362535296871e-05,
|
| 2039 |
+
"loss": 0.041,
|
| 2040 |
+
"step": 2900
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 11.235521235521235,
|
| 2044 |
+
"grad_norm": 0.5066102147102356,
|
| 2045 |
+
"learning_rate": 8.495559157038299e-05,
|
| 2046 |
+
"loss": 0.0456,
|
| 2047 |
+
"step": 2910
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 11.274131274131275,
|
| 2051 |
+
"grad_norm": 0.396445095539093,
|
| 2052 |
+
"learning_rate": 8.483717551904823e-05,
|
| 2053 |
+
"loss": 0.0505,
|
| 2054 |
+
"step": 2920
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 11.312741312741313,
|
| 2058 |
+
"grad_norm": 0.3678114116191864,
|
| 2059 |
+
"learning_rate": 8.47183784939434e-05,
|
| 2060 |
+
"loss": 0.0392,
|
| 2061 |
+
"step": 2930
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 11.35135135135135,
|
| 2065 |
+
"grad_norm": 0.5877802968025208,
|
| 2066 |
+
"learning_rate": 8.459920179421374e-05,
|
| 2067 |
+
"loss": 0.0397,
|
| 2068 |
+
"step": 2940
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 11.38996138996139,
|
| 2072 |
+
"grad_norm": 0.364266961812973,
|
| 2073 |
+
"learning_rate": 8.447964672315656e-05,
|
| 2074 |
+
"loss": 0.043,
|
| 2075 |
+
"step": 2950
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 11.428571428571429,
|
| 2079 |
+
"grad_norm": 0.5190859436988831,
|
| 2080 |
+
"learning_rate": 8.435971458820692e-05,
|
| 2081 |
+
"loss": 0.0424,
|
| 2082 |
+
"step": 2960
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 11.467181467181467,
|
| 2086 |
+
"grad_norm": 0.3852812647819519,
|
| 2087 |
+
"learning_rate": 8.423940670092345e-05,
|
| 2088 |
+
"loss": 0.0399,
|
| 2089 |
+
"step": 2970
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 11.505791505791505,
|
| 2093 |
+
"grad_norm": 0.606198787689209,
|
| 2094 |
+
"learning_rate": 8.411872437697394e-05,
|
| 2095 |
+
"loss": 0.0411,
|
| 2096 |
+
"step": 2980
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 11.544401544401545,
|
| 2100 |
+
"grad_norm": 0.475053995847702,
|
| 2101 |
+
"learning_rate": 8.399766893612096e-05,
|
| 2102 |
+
"loss": 0.0381,
|
| 2103 |
+
"step": 2990
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 11.583011583011583,
|
| 2107 |
+
"grad_norm": 0.4046951234340668,
|
| 2108 |
+
"learning_rate": 8.38762417022074e-05,
|
| 2109 |
+
"loss": 0.04,
|
| 2110 |
+
"step": 3000
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 11.621621621621621,
|
| 2114 |
+
"grad_norm": 0.33849042654037476,
|
| 2115 |
+
"learning_rate": 8.375444400314204e-05,
|
| 2116 |
+
"loss": 0.0408,
|
| 2117 |
+
"step": 3010
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 11.660231660231661,
|
| 2121 |
+
"grad_norm": 0.3920499086380005,
|
| 2122 |
+
"learning_rate": 8.3632277170885e-05,
|
| 2123 |
+
"loss": 0.0389,
|
| 2124 |
+
"step": 3020
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 11.698841698841699,
|
| 2128 |
+
"grad_norm": 0.5179415345191956,
|
| 2129 |
+
"learning_rate": 8.350974254143318e-05,
|
| 2130 |
+
"loss": 0.0397,
|
| 2131 |
+
"step": 3030
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 11.737451737451737,
|
| 2135 |
+
"grad_norm": 0.4171680510044098,
|
| 2136 |
+
"learning_rate": 8.338684145480566e-05,
|
| 2137 |
+
"loss": 0.0384,
|
| 2138 |
+
"step": 3040
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 11.776061776061777,
|
| 2142 |
+
"grad_norm": 0.45747077465057373,
|
| 2143 |
+
"learning_rate": 8.326357525502904e-05,
|
| 2144 |
+
"loss": 0.0347,
|
| 2145 |
+
"step": 3050
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 11.814671814671815,
|
| 2149 |
+
"grad_norm": 0.2903911769390106,
|
| 2150 |
+
"learning_rate": 8.313994529012273e-05,
|
| 2151 |
+
"loss": 0.0394,
|
| 2152 |
+
"step": 3060
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 11.853281853281853,
|
| 2156 |
+
"grad_norm": 0.5158092379570007,
|
| 2157 |
+
"learning_rate": 8.301595291208422e-05,
|
| 2158 |
+
"loss": 0.0375,
|
| 2159 |
+
"step": 3070
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 11.891891891891891,
|
| 2163 |
+
"grad_norm": 0.35226407647132874,
|
| 2164 |
+
"learning_rate": 8.289159947687427e-05,
|
| 2165 |
+
"loss": 0.0386,
|
| 2166 |
+
"step": 3080
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 11.930501930501931,
|
| 2170 |
+
"grad_norm": 0.3436617851257324,
|
| 2171 |
+
"learning_rate": 8.276688634440216e-05,
|
| 2172 |
+
"loss": 0.0329,
|
| 2173 |
+
"step": 3090
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 11.96911196911197,
|
| 2177 |
+
"grad_norm": 0.3412259519100189,
|
| 2178 |
+
"learning_rate": 8.26418148785107e-05,
|
| 2179 |
+
"loss": 0.0361,
|
| 2180 |
+
"step": 3100
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 12.007722007722007,
|
| 2184 |
+
"grad_norm": 0.24303047358989716,
|
| 2185 |
+
"learning_rate": 8.251638644696141e-05,
|
| 2186 |
+
"loss": 0.0381,
|
| 2187 |
+
"step": 3110
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 12.046332046332047,
|
| 2191 |
+
"grad_norm": 0.46407920122146606,
|
| 2192 |
+
"learning_rate": 8.23906024214195e-05,
|
| 2193 |
+
"loss": 0.047,
|
| 2194 |
+
"step": 3120
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 12.084942084942085,
|
| 2198 |
+
"grad_norm": 0.3811705708503723,
|
| 2199 |
+
"learning_rate": 8.226446417743897e-05,
|
| 2200 |
+
"loss": 0.0362,
|
| 2201 |
+
"step": 3130
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 12.123552123552123,
|
| 2205 |
+
"grad_norm": 0.45096784830093384,
|
| 2206 |
+
"learning_rate": 8.213797309444742e-05,
|
| 2207 |
+
"loss": 0.0387,
|
| 2208 |
+
"step": 3140
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 12.162162162162161,
|
| 2212 |
+
"grad_norm": 0.3665260970592499,
|
| 2213 |
+
"learning_rate": 8.201113055573105e-05,
|
| 2214 |
+
"loss": 0.0387,
|
| 2215 |
+
"step": 3150
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 12.200772200772201,
|
| 2219 |
+
"grad_norm": 0.35533803701400757,
|
| 2220 |
+
"learning_rate": 8.188393794841958e-05,
|
| 2221 |
+
"loss": 0.0381,
|
| 2222 |
+
"step": 3160
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 12.23938223938224,
|
| 2226 |
+
"grad_norm": 0.3760382831096649,
|
| 2227 |
+
"learning_rate": 8.175639666347094e-05,
|
| 2228 |
+
"loss": 0.0381,
|
| 2229 |
+
"step": 3170
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 12.277992277992277,
|
| 2233 |
+
"grad_norm": 0.4703519940376282,
|
| 2234 |
+
"learning_rate": 8.162850809565623e-05,
|
| 2235 |
+
"loss": 0.0389,
|
| 2236 |
+
"step": 3180
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 12.316602316602317,
|
| 2240 |
+
"grad_norm": 0.3851741850376129,
|
| 2241 |
+
"learning_rate": 8.150027364354431e-05,
|
| 2242 |
+
"loss": 0.0344,
|
| 2243 |
+
"step": 3190
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 12.355212355212355,
|
| 2247 |
+
"grad_norm": 0.34900107979774475,
|
| 2248 |
+
"learning_rate": 8.137169470948662e-05,
|
| 2249 |
+
"loss": 0.0416,
|
| 2250 |
+
"step": 3200
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 12.393822393822393,
|
| 2254 |
+
"grad_norm": 0.498517781496048,
|
| 2255 |
+
"learning_rate": 8.124277269960179e-05,
|
| 2256 |
+
"loss": 0.0389,
|
| 2257 |
+
"step": 3210
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 12.432432432432432,
|
| 2261 |
+
"grad_norm": 0.4294493496417999,
|
| 2262 |
+
"learning_rate": 8.111350902376023e-05,
|
| 2263 |
+
"loss": 0.0374,
|
| 2264 |
+
"step": 3220
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 12.471042471042471,
|
| 2268 |
+
"grad_norm": 0.551791250705719,
|
| 2269 |
+
"learning_rate": 8.098390509556883e-05,
|
| 2270 |
+
"loss": 0.0357,
|
| 2271 |
+
"step": 3230
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 12.50965250965251,
|
| 2275 |
+
"grad_norm": 0.43380701541900635,
|
| 2276 |
+
"learning_rate": 8.085396233235536e-05,
|
| 2277 |
+
"loss": 0.0385,
|
| 2278 |
+
"step": 3240
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 12.548262548262548,
|
| 2282 |
+
"grad_norm": 0.3815891146659851,
|
| 2283 |
+
"learning_rate": 8.072368215515306e-05,
|
| 2284 |
+
"loss": 0.0324,
|
| 2285 |
+
"step": 3250
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 12.586872586872587,
|
| 2289 |
+
"grad_norm": 0.3850783109664917,
|
| 2290 |
+
"learning_rate": 8.059306598868506e-05,
|
| 2291 |
+
"loss": 0.0371,
|
| 2292 |
+
"step": 3260
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 12.625482625482626,
|
| 2296 |
+
"grad_norm": 0.3633018136024475,
|
| 2297 |
+
"learning_rate": 8.046211526134888e-05,
|
| 2298 |
+
"loss": 0.0428,
|
| 2299 |
+
"step": 3270
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 12.664092664092664,
|
| 2303 |
+
"grad_norm": 0.2936514914035797,
|
| 2304 |
+
"learning_rate": 8.033083140520065e-05,
|
| 2305 |
+
"loss": 0.0357,
|
| 2306 |
+
"step": 3280
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 12.702702702702704,
|
| 2310 |
+
"grad_norm": 0.47930997610092163,
|
| 2311 |
+
"learning_rate": 8.019921585593962e-05,
|
| 2312 |
+
"loss": 0.0358,
|
| 2313 |
+
"step": 3290
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 12.741312741312742,
|
| 2317 |
+
"grad_norm": 0.5421620607376099,
|
| 2318 |
+
"learning_rate": 8.006727005289232e-05,
|
| 2319 |
+
"loss": 0.035,
|
| 2320 |
+
"step": 3300
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 12.77992277992278,
|
| 2324 |
+
"grad_norm": 0.4249776303768158,
|
| 2325 |
+
"learning_rate": 7.993499543899692e-05,
|
| 2326 |
+
"loss": 0.0327,
|
| 2327 |
+
"step": 3310
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 12.818532818532818,
|
| 2331 |
+
"grad_norm": 0.3001653254032135,
|
| 2332 |
+
"learning_rate": 7.980239346078742e-05,
|
| 2333 |
+
"loss": 0.0355,
|
| 2334 |
+
"step": 3320
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 12.857142857142858,
|
| 2338 |
+
"grad_norm": 0.40948551893234253,
|
| 2339 |
+
"learning_rate": 7.966946556837778e-05,
|
| 2340 |
+
"loss": 0.0348,
|
| 2341 |
+
"step": 3330
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 12.895752895752896,
|
| 2345 |
+
"grad_norm": 0.24407273530960083,
|
| 2346 |
+
"learning_rate": 7.953621321544616e-05,
|
| 2347 |
+
"loss": 0.0394,
|
| 2348 |
+
"step": 3340
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 12.934362934362934,
|
| 2352 |
+
"grad_norm": 0.2720007002353668,
|
| 2353 |
+
"learning_rate": 7.940263785921896e-05,
|
| 2354 |
+
"loss": 0.0347,
|
| 2355 |
+
"step": 3350
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 12.972972972972974,
|
| 2359 |
+
"grad_norm": 0.40845194458961487,
|
| 2360 |
+
"learning_rate": 7.926874096045482e-05,
|
| 2361 |
+
"loss": 0.0408,
|
| 2362 |
+
"step": 3360
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 13.011583011583012,
|
| 2366 |
+
"grad_norm": 0.3387102484703064,
|
| 2367 |
+
"learning_rate": 7.913452398342881e-05,
|
| 2368 |
+
"loss": 0.0352,
|
| 2369 |
+
"step": 3370
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 13.05019305019305,
|
| 2373 |
+
"grad_norm": 0.3375230133533478,
|
| 2374 |
+
"learning_rate": 7.89999883959163e-05,
|
| 2375 |
+
"loss": 0.0432,
|
| 2376 |
+
"step": 3380
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 13.088803088803088,
|
| 2380 |
+
"grad_norm": 0.4011669158935547,
|
| 2381 |
+
"learning_rate": 7.886513566917687e-05,
|
| 2382 |
+
"loss": 0.041,
|
| 2383 |
+
"step": 3390
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 13.127413127413128,
|
| 2387 |
+
"grad_norm": 0.32482975721359253,
|
| 2388 |
+
"learning_rate": 7.872996727793838e-05,
|
| 2389 |
+
"loss": 0.0367,
|
| 2390 |
+
"step": 3400
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 13.166023166023166,
|
| 2394 |
+
"grad_norm": 0.3560713827610016,
|
| 2395 |
+
"learning_rate": 7.859448470038069e-05,
|
| 2396 |
+
"loss": 0.0345,
|
| 2397 |
+
"step": 3410
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 13.204633204633204,
|
| 2401 |
+
"grad_norm": 0.35355344414711,
|
| 2402 |
+
"learning_rate": 7.845868941811956e-05,
|
| 2403 |
+
"loss": 0.0405,
|
| 2404 |
+
"step": 3420
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 13.243243243243244,
|
| 2408 |
+
"grad_norm": 0.3198929727077484,
|
| 2409 |
+
"learning_rate": 7.832258291619043e-05,
|
| 2410 |
+
"loss": 0.0321,
|
| 2411 |
+
"step": 3430
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 13.281853281853282,
|
| 2415 |
+
"grad_norm": 0.42199987173080444,
|
| 2416 |
+
"learning_rate": 7.81861666830322e-05,
|
| 2417 |
+
"loss": 0.0319,
|
| 2418 |
+
"step": 3440
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 13.32046332046332,
|
| 2422 |
+
"grad_norm": 0.5426796078681946,
|
| 2423 |
+
"learning_rate": 7.804944221047097e-05,
|
| 2424 |
+
"loss": 0.0395,
|
| 2425 |
+
"step": 3450
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 13.35907335907336,
|
| 2429 |
+
"grad_norm": 0.3558543622493744,
|
| 2430 |
+
"learning_rate": 7.791241099370364e-05,
|
| 2431 |
+
"loss": 0.0357,
|
| 2432 |
+
"step": 3460
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 13.397683397683398,
|
| 2436 |
+
"grad_norm": 0.42568162083625793,
|
| 2437 |
+
"learning_rate": 7.777507453128163e-05,
|
| 2438 |
+
"loss": 0.0411,
|
| 2439 |
+
"step": 3470
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 13.436293436293436,
|
| 2443 |
+
"grad_norm": 0.3781692683696747,
|
| 2444 |
+
"learning_rate": 7.763743432509451e-05,
|
| 2445 |
+
"loss": 0.0371,
|
| 2446 |
+
"step": 3480
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 13.474903474903474,
|
| 2450 |
+
"grad_norm": 0.41086286306381226,
|
| 2451 |
+
"learning_rate": 7.749949188035353e-05,
|
| 2452 |
+
"loss": 0.04,
|
| 2453 |
+
"step": 3490
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 13.513513513513514,
|
| 2457 |
+
"grad_norm": 0.41704869270324707,
|
| 2458 |
+
"learning_rate": 7.736124870557516e-05,
|
| 2459 |
+
"loss": 0.0452,
|
| 2460 |
+
"step": 3500
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 13.552123552123552,
|
| 2464 |
+
"grad_norm": 0.3421342074871063,
|
| 2465 |
+
"learning_rate": 7.722270631256459e-05,
|
| 2466 |
+
"loss": 0.0338,
|
| 2467 |
+
"step": 3510
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 13.59073359073359,
|
| 2471 |
+
"grad_norm": 0.4111107885837555,
|
| 2472 |
+
"learning_rate": 7.708386621639925e-05,
|
| 2473 |
+
"loss": 0.0414,
|
| 2474 |
+
"step": 3520
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 13.62934362934363,
|
| 2478 |
+
"grad_norm": 0.4858056604862213,
|
| 2479 |
+
"learning_rate": 7.694472993541219e-05,
|
| 2480 |
+
"loss": 0.0358,
|
| 2481 |
+
"step": 3530
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 13.667953667953668,
|
| 2485 |
+
"grad_norm": 0.5249147415161133,
|
| 2486 |
+
"learning_rate": 7.680529899117547e-05,
|
| 2487 |
+
"loss": 0.035,
|
| 2488 |
+
"step": 3540
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 13.706563706563706,
|
| 2492 |
+
"grad_norm": 0.5796403884887695,
|
| 2493 |
+
"learning_rate": 7.666557490848358e-05,
|
| 2494 |
+
"loss": 0.0406,
|
| 2495 |
+
"step": 3550
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 13.745173745173744,
|
| 2499 |
+
"grad_norm": 0.524755597114563,
|
| 2500 |
+
"learning_rate": 7.65255592153367e-05,
|
| 2501 |
+
"loss": 0.0397,
|
| 2502 |
+
"step": 3560
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 13.783783783783784,
|
| 2506 |
+
"grad_norm": 0.4135282337665558,
|
| 2507 |
+
"learning_rate": 7.638525344292402e-05,
|
| 2508 |
+
"loss": 0.039,
|
| 2509 |
+
"step": 3570
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 13.822393822393822,
|
| 2513 |
+
"grad_norm": 0.3764033019542694,
|
| 2514 |
+
"learning_rate": 7.624465912560697e-05,
|
| 2515 |
+
"loss": 0.0371,
|
| 2516 |
+
"step": 3580
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 13.86100386100386,
|
| 2520 |
+
"grad_norm": 0.501422107219696,
|
| 2521 |
+
"learning_rate": 7.610377780090249e-05,
|
| 2522 |
+
"loss": 0.0424,
|
| 2523 |
+
"step": 3590
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 13.8996138996139,
|
| 2527 |
+
"grad_norm": 0.3695226013660431,
|
| 2528 |
+
"learning_rate": 7.596261100946618e-05,
|
| 2529 |
+
"loss": 0.0267,
|
| 2530 |
+
"step": 3600
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 13.938223938223938,
|
| 2534 |
+
"grad_norm": 0.3322920501232147,
|
| 2535 |
+
"learning_rate": 7.582116029507542e-05,
|
| 2536 |
+
"loss": 0.0319,
|
| 2537 |
+
"step": 3610
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 13.976833976833976,
|
| 2541 |
+
"grad_norm": 0.4884520173072815,
|
| 2542 |
+
"learning_rate": 7.56794272046126e-05,
|
| 2543 |
+
"loss": 0.03,
|
| 2544 |
+
"step": 3620
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 14.015444015444016,
|
| 2548 |
+
"grad_norm": 0.4386620819568634,
|
| 2549 |
+
"learning_rate": 7.55374132880481e-05,
|
| 2550 |
+
"loss": 0.0364,
|
| 2551 |
+
"step": 3630
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 14.054054054054054,
|
| 2555 |
+
"grad_norm": 0.3752287030220032,
|
| 2556 |
+
"learning_rate": 7.539512009842333e-05,
|
| 2557 |
+
"loss": 0.0377,
|
| 2558 |
+
"step": 3640
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 14.092664092664092,
|
| 2562 |
+
"grad_norm": 0.31382790207862854,
|
| 2563 |
+
"learning_rate": 7.525254919183382e-05,
|
| 2564 |
+
"loss": 0.036,
|
| 2565 |
+
"step": 3650
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 14.13127413127413,
|
| 2569 |
+
"grad_norm": 0.3197014629840851,
|
| 2570 |
+
"learning_rate": 7.510970212741215e-05,
|
| 2571 |
+
"loss": 0.0354,
|
| 2572 |
+
"step": 3660
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 14.16988416988417,
|
| 2576 |
+
"grad_norm": 0.33206719160079956,
|
| 2577 |
+
"learning_rate": 7.496658046731096e-05,
|
| 2578 |
+
"loss": 0.0306,
|
| 2579 |
+
"step": 3670
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 14.208494208494209,
|
| 2583 |
+
"grad_norm": 0.37065961956977844,
|
| 2584 |
+
"learning_rate": 7.482318577668578e-05,
|
| 2585 |
+
"loss": 0.039,
|
| 2586 |
+
"step": 3680
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 14.247104247104247,
|
| 2590 |
+
"grad_norm": 0.5282608270645142,
|
| 2591 |
+
"learning_rate": 7.467951962367796e-05,
|
| 2592 |
+
"loss": 0.0428,
|
| 2593 |
+
"step": 3690
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 14.285714285714286,
|
| 2597 |
+
"grad_norm": 0.4893680512905121,
|
| 2598 |
+
"learning_rate": 7.453558357939755e-05,
|
| 2599 |
+
"loss": 0.0436,
|
| 2600 |
+
"step": 3700
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 14.324324324324325,
|
| 2604 |
+
"grad_norm": 0.3082546591758728,
|
| 2605 |
+
"learning_rate": 7.439137921790606e-05,
|
| 2606 |
+
"loss": 0.0326,
|
| 2607 |
+
"step": 3710
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 14.362934362934363,
|
| 2611 |
+
"grad_norm": 0.35221901535987854,
|
| 2612 |
+
"learning_rate": 7.42469081161993e-05,
|
| 2613 |
+
"loss": 0.0406,
|
| 2614 |
+
"step": 3720
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 14.4015444015444,
|
| 2618 |
+
"grad_norm": 0.40901821851730347,
|
| 2619 |
+
"learning_rate": 7.410217185419006e-05,
|
| 2620 |
+
"loss": 0.0299,
|
| 2621 |
+
"step": 3730
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 14.44015444015444,
|
| 2625 |
+
"grad_norm": 0.3284573554992676,
|
| 2626 |
+
"learning_rate": 7.395717201469095e-05,
|
| 2627 |
+
"loss": 0.0323,
|
| 2628 |
+
"step": 3740
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 14.478764478764479,
|
| 2632 |
+
"grad_norm": 0.34251612424850464,
|
| 2633 |
+
"learning_rate": 7.381191018339696e-05,
|
| 2634 |
+
"loss": 0.0428,
|
| 2635 |
+
"step": 3750
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 14.517374517374517,
|
| 2639 |
+
"grad_norm": 0.4092288315296173,
|
| 2640 |
+
"learning_rate": 7.36663879488682e-05,
|
| 2641 |
+
"loss": 0.0327,
|
| 2642 |
+
"step": 3760
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 14.555984555984557,
|
| 2646 |
+
"grad_norm": 0.31500324606895447,
|
| 2647 |
+
"learning_rate": 7.352060690251254e-05,
|
| 2648 |
+
"loss": 0.033,
|
| 2649 |
+
"step": 3770
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 14.594594594594595,
|
| 2653 |
+
"grad_norm": 0.4034515917301178,
|
| 2654 |
+
"learning_rate": 7.337456863856811e-05,
|
| 2655 |
+
"loss": 0.0407,
|
| 2656 |
+
"step": 3780
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 14.633204633204633,
|
| 2660 |
+
"grad_norm": 0.2690200209617615,
|
| 2661 |
+
"learning_rate": 7.3228274754086e-05,
|
| 2662 |
+
"loss": 0.0329,
|
| 2663 |
+
"step": 3790
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 14.671814671814673,
|
| 2667 |
+
"grad_norm": 0.32889077067375183,
|
| 2668 |
+
"learning_rate": 7.308172684891267e-05,
|
| 2669 |
+
"loss": 0.0344,
|
| 2670 |
+
"step": 3800
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 14.71042471042471,
|
| 2674 |
+
"grad_norm": 0.35578179359436035,
|
| 2675 |
+
"learning_rate": 7.293492652567255e-05,
|
| 2676 |
+
"loss": 0.0343,
|
| 2677 |
+
"step": 3810
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 14.749034749034749,
|
| 2681 |
+
"grad_norm": 0.3409169912338257,
|
| 2682 |
+
"learning_rate": 7.278787538975043e-05,
|
| 2683 |
+
"loss": 0.0327,
|
| 2684 |
+
"step": 3820
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 14.787644787644787,
|
| 2688 |
+
"grad_norm": 0.34640923142433167,
|
| 2689 |
+
"learning_rate": 7.2640575049274e-05,
|
| 2690 |
+
"loss": 0.033,
|
| 2691 |
+
"step": 3830
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 14.826254826254827,
|
| 2695 |
+
"grad_norm": 0.3829593360424042,
|
| 2696 |
+
"learning_rate": 7.249302711509616e-05,
|
| 2697 |
+
"loss": 0.0346,
|
| 2698 |
+
"step": 3840
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 14.864864864864865,
|
| 2702 |
+
"grad_norm": 0.3409728407859802,
|
| 2703 |
+
"learning_rate": 7.23452332007775e-05,
|
| 2704 |
+
"loss": 0.0335,
|
| 2705 |
+
"step": 3850
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 14.903474903474903,
|
| 2709 |
+
"grad_norm": 0.4900124669075012,
|
| 2710 |
+
"learning_rate": 7.219719492256858e-05,
|
| 2711 |
+
"loss": 0.0346,
|
| 2712 |
+
"step": 3860
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 14.942084942084943,
|
| 2716 |
+
"grad_norm": 0.23285947740077972,
|
| 2717 |
+
"learning_rate": 7.20489138993923e-05,
|
| 2718 |
+
"loss": 0.0352,
|
| 2719 |
+
"step": 3870
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 14.980694980694981,
|
| 2723 |
+
"grad_norm": 0.5416049957275391,
|
| 2724 |
+
"learning_rate": 7.190039175282614e-05,
|
| 2725 |
+
"loss": 0.0362,
|
| 2726 |
+
"step": 3880
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 15.019305019305019,
|
| 2730 |
+
"grad_norm": 0.18925291299819946,
|
| 2731 |
+
"learning_rate": 7.175163010708455e-05,
|
| 2732 |
+
"loss": 0.0321,
|
| 2733 |
+
"step": 3890
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 15.057915057915057,
|
| 2737 |
+
"grad_norm": 0.3688291013240814,
|
| 2738 |
+
"learning_rate": 7.1602630589001e-05,
|
| 2739 |
+
"loss": 0.0286,
|
| 2740 |
+
"step": 3900
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 15.096525096525097,
|
| 2744 |
+
"grad_norm": 0.26431047916412354,
|
| 2745 |
+
"learning_rate": 7.14533948280104e-05,
|
| 2746 |
+
"loss": 0.0326,
|
| 2747 |
+
"step": 3910
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 15.135135135135135,
|
| 2751 |
+
"grad_norm": 0.1818448156118393,
|
| 2752 |
+
"learning_rate": 7.130392445613109e-05,
|
| 2753 |
+
"loss": 0.0329,
|
| 2754 |
+
"step": 3920
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 15.173745173745173,
|
| 2758 |
+
"grad_norm": 0.49927157163619995,
|
| 2759 |
+
"learning_rate": 7.115422110794711e-05,
|
| 2760 |
+
"loss": 0.0319,
|
| 2761 |
+
"step": 3930
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 15.212355212355213,
|
| 2765 |
+
"grad_norm": 0.45622697472572327,
|
| 2766 |
+
"learning_rate": 7.100428642059033e-05,
|
| 2767 |
+
"loss": 0.0341,
|
| 2768 |
+
"step": 3940
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 15.250965250965251,
|
| 2772 |
+
"grad_norm": 0.579872727394104,
|
| 2773 |
+
"learning_rate": 7.08541220337224e-05,
|
| 2774 |
+
"loss": 0.0358,
|
| 2775 |
+
"step": 3950
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 15.28957528957529,
|
| 2779 |
+
"grad_norm": 0.4018189311027527,
|
| 2780 |
+
"learning_rate": 7.070372958951706e-05,
|
| 2781 |
+
"loss": 0.0362,
|
| 2782 |
+
"step": 3960
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 15.328185328185327,
|
| 2786 |
+
"grad_norm": 0.5537483096122742,
|
| 2787 |
+
"learning_rate": 7.055311073264194e-05,
|
| 2788 |
+
"loss": 0.0333,
|
| 2789 |
+
"step": 3970
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 15.366795366795367,
|
| 2793 |
+
"grad_norm": 0.45532456040382385,
|
| 2794 |
+
"learning_rate": 7.040226711024077e-05,
|
| 2795 |
+
"loss": 0.0355,
|
| 2796 |
+
"step": 3980
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 15.405405405405405,
|
| 2800 |
+
"grad_norm": 0.22638791799545288,
|
| 2801 |
+
"learning_rate": 7.02512003719152e-05,
|
| 2802 |
+
"loss": 0.03,
|
| 2803 |
+
"step": 3990
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 15.444015444015443,
|
| 2807 |
+
"grad_norm": 0.28028252720832825,
|
| 2808 |
+
"learning_rate": 7.00999121697069e-05,
|
| 2809 |
+
"loss": 0.0308,
|
| 2810 |
+
"step": 4000
|
| 2811 |
+
}
|
| 2812 |
+
],
|
| 2813 |
+
"logging_steps": 10,
|
| 2814 |
+
"max_steps": 10000,
|
| 2815 |
+
"num_input_tokens_seen": 0,
|
| 2816 |
+
"num_train_epochs": 39,
|
| 2817 |
+
"save_steps": 1000,
|
| 2818 |
+
"stateful_callbacks": {
|
| 2819 |
+
"TrainerControl": {
|
| 2820 |
+
"args": {
|
| 2821 |
+
"should_epoch_stop": false,
|
| 2822 |
+
"should_evaluate": false,
|
| 2823 |
+
"should_log": false,
|
| 2824 |
+
"should_save": true,
|
| 2825 |
+
"should_training_stop": false
|
| 2826 |
+
},
|
| 2827 |
+
"attributes": {}
|
| 2828 |
+
}
|
| 2829 |
+
},
|
| 2830 |
+
"total_flos": 0.0,
|
| 2831 |
+
"train_batch_size": 32,
|
| 2832 |
+
"trial_name": null,
|
| 2833 |
+
"trial_params": null
|
| 2834 |
+
}
|