Training in progress, epoch 7, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1575259780
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:00790e6eb3aaf7be06f42ed5ebbbf19f51da90eabde4485d44e5afbbaa258042
|
| 3 |
size 1575259780
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3150397656
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c7b07559a454669087b353898e147242640f3129e044989950556b889a6dbc02
|
| 3 |
size 3150397656
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7edf07803f417b0f28032144c1c24e7268187cdee6ec7471783e6c05da595f26
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:13b3ba23af7cfc783c09d1dceb8d6f5096a2c55f1fa03f3e194a3cfed394453e
|
| 3 |
size 1064
|
last-checkpoint/special_tokens_map.json
CHANGED
|
@@ -53,9 +53,27 @@
|
|
| 53 |
"gl_ES",
|
| 54 |
"sl_SI"
|
| 55 |
],
|
| 56 |
-
"bos_token":
|
| 57 |
-
|
| 58 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 59 |
"mask_token": {
|
| 60 |
"content": "<mask>",
|
| 61 |
"lstrip": true,
|
|
@@ -63,7 +81,25 @@
|
|
| 63 |
"rstrip": false,
|
| 64 |
"single_word": false
|
| 65 |
},
|
| 66 |
-
"pad_token":
|
| 67 |
-
|
| 68 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 69 |
}
|
|
|
|
| 53 |
"gl_ES",
|
| 54 |
"sl_SI"
|
| 55 |
],
|
| 56 |
+
"bos_token": {
|
| 57 |
+
"content": "<s>",
|
| 58 |
+
"lstrip": false,
|
| 59 |
+
"normalized": false,
|
| 60 |
+
"rstrip": false,
|
| 61 |
+
"single_word": false
|
| 62 |
+
},
|
| 63 |
+
"cls_token": {
|
| 64 |
+
"content": "<s>",
|
| 65 |
+
"lstrip": false,
|
| 66 |
+
"normalized": false,
|
| 67 |
+
"rstrip": false,
|
| 68 |
+
"single_word": false
|
| 69 |
+
},
|
| 70 |
+
"eos_token": {
|
| 71 |
+
"content": "</s>",
|
| 72 |
+
"lstrip": false,
|
| 73 |
+
"normalized": false,
|
| 74 |
+
"rstrip": false,
|
| 75 |
+
"single_word": false
|
| 76 |
+
},
|
| 77 |
"mask_token": {
|
| 78 |
"content": "<mask>",
|
| 79 |
"lstrip": true,
|
|
|
|
| 81 |
"rstrip": false,
|
| 82 |
"single_word": false
|
| 83 |
},
|
| 84 |
+
"pad_token": {
|
| 85 |
+
"content": "<pad>",
|
| 86 |
+
"lstrip": false,
|
| 87 |
+
"normalized": false,
|
| 88 |
+
"rstrip": false,
|
| 89 |
+
"single_word": false
|
| 90 |
+
},
|
| 91 |
+
"sep_token": {
|
| 92 |
+
"content": "</s>",
|
| 93 |
+
"lstrip": false,
|
| 94 |
+
"normalized": false,
|
| 95 |
+
"rstrip": false,
|
| 96 |
+
"single_word": false
|
| 97 |
+
},
|
| 98 |
+
"unk_token": {
|
| 99 |
+
"content": "<unk>",
|
| 100 |
+
"lstrip": false,
|
| 101 |
+
"normalized": false,
|
| 102 |
+
"rstrip": false,
|
| 103 |
+
"single_word": false
|
| 104 |
+
}
|
| 105 |
}
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint": "/kaggle/tmp/amr-tst-indo/AMRBART-id/fine-tune/../outputs/mbart-en-id-smaller-fted/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5865,6 +5865,2348 @@
|
|
| 5865 |
"eval_samples_per_second": 1.019,
|
| 5866 |
"eval_steps_per_second": 0.51,
|
| 5867 |
"step": 19347
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5868 |
}
|
| 5869 |
],
|
| 5870 |
"logging_steps": 20,
|
|
@@ -5884,7 +8226,7 @@
|
|
| 5884 |
"attributes": {}
|
| 5885 |
}
|
| 5886 |
},
|
| 5887 |
-
"total_flos": 1.
|
| 5888 |
"train_batch_size": 2,
|
| 5889 |
"trial_name": null,
|
| 5890 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 5.2201,
|
| 3 |
+
"best_model_checkpoint": "/kaggle/tmp/amr-tst-indo/AMRBART-id/fine-tune/../outputs/mbart-en-id-smaller-fted/checkpoint-27084",
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 27084,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5865 |
"eval_samples_per_second": 1.019,
|
| 5866 |
"eval_steps_per_second": 0.51,
|
| 5867 |
"step": 19347
|
| 5868 |
+
},
|
| 5869 |
+
{
|
| 5870 |
+
"epoch": 5.0038764698281435,
|
| 5871 |
+
"learning_rate": 1.6894852845844678e-07,
|
| 5872 |
+
"loss": 2.6522,
|
| 5873 |
+
"step": 19360
|
| 5874 |
+
},
|
| 5875 |
+
{
|
| 5876 |
+
"epoch": 5.009045096265667,
|
| 5877 |
+
"learning_rate": 1.6891611564890445e-07,
|
| 5878 |
+
"loss": 2.6008,
|
| 5879 |
+
"step": 19380
|
| 5880 |
+
},
|
| 5881 |
+
{
|
| 5882 |
+
"epoch": 5.014213722703191,
|
| 5883 |
+
"learning_rate": 1.688837028393621e-07,
|
| 5884 |
+
"loss": 2.6294,
|
| 5885 |
+
"step": 19400
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 5.019382349140716,
|
| 5889 |
+
"learning_rate": 1.6885129002981977e-07,
|
| 5890 |
+
"loss": 2.6519,
|
| 5891 |
+
"step": 19420
|
| 5892 |
+
},
|
| 5893 |
+
{
|
| 5894 |
+
"epoch": 5.02455097557824,
|
| 5895 |
+
"learning_rate": 1.6881887722027743e-07,
|
| 5896 |
+
"loss": 2.6537,
|
| 5897 |
+
"step": 19440
|
| 5898 |
+
},
|
| 5899 |
+
{
|
| 5900 |
+
"epoch": 5.029719602015764,
|
| 5901 |
+
"learning_rate": 1.6878646441073512e-07,
|
| 5902 |
+
"loss": 2.6256,
|
| 5903 |
+
"step": 19460
|
| 5904 |
+
},
|
| 5905 |
+
{
|
| 5906 |
+
"epoch": 5.034888228453289,
|
| 5907 |
+
"learning_rate": 1.6875405160119278e-07,
|
| 5908 |
+
"loss": 2.6721,
|
| 5909 |
+
"step": 19480
|
| 5910 |
+
},
|
| 5911 |
+
{
|
| 5912 |
+
"epoch": 5.040056854890813,
|
| 5913 |
+
"learning_rate": 1.6872163879165044e-07,
|
| 5914 |
+
"loss": 2.6894,
|
| 5915 |
+
"step": 19500
|
| 5916 |
+
},
|
| 5917 |
+
{
|
| 5918 |
+
"epoch": 5.045225481328337,
|
| 5919 |
+
"learning_rate": 1.6868922598210813e-07,
|
| 5920 |
+
"loss": 2.6702,
|
| 5921 |
+
"step": 19520
|
| 5922 |
+
},
|
| 5923 |
+
{
|
| 5924 |
+
"epoch": 5.050394107765861,
|
| 5925 |
+
"learning_rate": 1.686568131725658e-07,
|
| 5926 |
+
"loss": 2.7041,
|
| 5927 |
+
"step": 19540
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 5.055562734203385,
|
| 5931 |
+
"learning_rate": 1.6862440036302346e-07,
|
| 5932 |
+
"loss": 2.7243,
|
| 5933 |
+
"step": 19560
|
| 5934 |
+
},
|
| 5935 |
+
{
|
| 5936 |
+
"epoch": 5.06073136064091,
|
| 5937 |
+
"learning_rate": 1.6859198755348114e-07,
|
| 5938 |
+
"loss": 2.7082,
|
| 5939 |
+
"step": 19580
|
| 5940 |
+
},
|
| 5941 |
+
{
|
| 5942 |
+
"epoch": 5.065899987078434,
|
| 5943 |
+
"learning_rate": 1.685595747439388e-07,
|
| 5944 |
+
"loss": 2.6755,
|
| 5945 |
+
"step": 19600
|
| 5946 |
+
},
|
| 5947 |
+
{
|
| 5948 |
+
"epoch": 5.071068613515958,
|
| 5949 |
+
"learning_rate": 1.6852716193439647e-07,
|
| 5950 |
+
"loss": 2.6075,
|
| 5951 |
+
"step": 19620
|
| 5952 |
+
},
|
| 5953 |
+
{
|
| 5954 |
+
"epoch": 5.076237239953483,
|
| 5955 |
+
"learning_rate": 1.6849474912485413e-07,
|
| 5956 |
+
"loss": 2.6402,
|
| 5957 |
+
"step": 19640
|
| 5958 |
+
},
|
| 5959 |
+
{
|
| 5960 |
+
"epoch": 5.081405866391006,
|
| 5961 |
+
"learning_rate": 1.684623363153118e-07,
|
| 5962 |
+
"loss": 2.6928,
|
| 5963 |
+
"step": 19660
|
| 5964 |
+
},
|
| 5965 |
+
{
|
| 5966 |
+
"epoch": 5.086574492828531,
|
| 5967 |
+
"learning_rate": 1.6842992350576948e-07,
|
| 5968 |
+
"loss": 2.6689,
|
| 5969 |
+
"step": 19680
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 5.091743119266055,
|
| 5973 |
+
"learning_rate": 1.6839751069622714e-07,
|
| 5974 |
+
"loss": 2.6848,
|
| 5975 |
+
"step": 19700
|
| 5976 |
+
},
|
| 5977 |
+
{
|
| 5978 |
+
"epoch": 5.096911745703579,
|
| 5979 |
+
"learning_rate": 1.683650978866848e-07,
|
| 5980 |
+
"loss": 2.6315,
|
| 5981 |
+
"step": 19720
|
| 5982 |
+
},
|
| 5983 |
+
{
|
| 5984 |
+
"epoch": 5.102080372141104,
|
| 5985 |
+
"learning_rate": 1.683326850771425e-07,
|
| 5986 |
+
"loss": 2.6936,
|
| 5987 |
+
"step": 19740
|
| 5988 |
+
},
|
| 5989 |
+
{
|
| 5990 |
+
"epoch": 5.107248998578628,
|
| 5991 |
+
"learning_rate": 1.6830027226760015e-07,
|
| 5992 |
+
"loss": 2.6354,
|
| 5993 |
+
"step": 19760
|
| 5994 |
+
},
|
| 5995 |
+
{
|
| 5996 |
+
"epoch": 5.112417625016152,
|
| 5997 |
+
"learning_rate": 1.6826785945805782e-07,
|
| 5998 |
+
"loss": 2.6376,
|
| 5999 |
+
"step": 19780
|
| 6000 |
+
},
|
| 6001 |
+
{
|
| 6002 |
+
"epoch": 5.117586251453676,
|
| 6003 |
+
"learning_rate": 1.6823544664851548e-07,
|
| 6004 |
+
"loss": 2.7595,
|
| 6005 |
+
"step": 19800
|
| 6006 |
+
},
|
| 6007 |
+
{
|
| 6008 |
+
"epoch": 5.1227548778912,
|
| 6009 |
+
"learning_rate": 1.6820303383897317e-07,
|
| 6010 |
+
"loss": 2.5688,
|
| 6011 |
+
"step": 19820
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 5.127923504328725,
|
| 6015 |
+
"learning_rate": 1.6817062102943083e-07,
|
| 6016 |
+
"loss": 2.6504,
|
| 6017 |
+
"step": 19840
|
| 6018 |
+
},
|
| 6019 |
+
{
|
| 6020 |
+
"epoch": 5.133092130766249,
|
| 6021 |
+
"learning_rate": 1.681382082198885e-07,
|
| 6022 |
+
"loss": 2.701,
|
| 6023 |
+
"step": 19860
|
| 6024 |
+
},
|
| 6025 |
+
{
|
| 6026 |
+
"epoch": 5.138260757203773,
|
| 6027 |
+
"learning_rate": 1.6810579541034615e-07,
|
| 6028 |
+
"loss": 2.6691,
|
| 6029 |
+
"step": 19880
|
| 6030 |
+
},
|
| 6031 |
+
{
|
| 6032 |
+
"epoch": 5.143429383641298,
|
| 6033 |
+
"learning_rate": 1.6807338260080384e-07,
|
| 6034 |
+
"loss": 2.6396,
|
| 6035 |
+
"step": 19900
|
| 6036 |
+
},
|
| 6037 |
+
{
|
| 6038 |
+
"epoch": 5.148598010078821,
|
| 6039 |
+
"learning_rate": 1.680409697912615e-07,
|
| 6040 |
+
"loss": 2.6752,
|
| 6041 |
+
"step": 19920
|
| 6042 |
+
},
|
| 6043 |
+
{
|
| 6044 |
+
"epoch": 5.1537666365163455,
|
| 6045 |
+
"learning_rate": 1.6800855698171916e-07,
|
| 6046 |
+
"loss": 2.567,
|
| 6047 |
+
"step": 19940
|
| 6048 |
+
},
|
| 6049 |
+
{
|
| 6050 |
+
"epoch": 5.15893526295387,
|
| 6051 |
+
"learning_rate": 1.6797614417217685e-07,
|
| 6052 |
+
"loss": 2.6691,
|
| 6053 |
+
"step": 19960
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 5.164103889391394,
|
| 6057 |
+
"learning_rate": 1.6794373136263451e-07,
|
| 6058 |
+
"loss": 2.6172,
|
| 6059 |
+
"step": 19980
|
| 6060 |
+
},
|
| 6061 |
+
{
|
| 6062 |
+
"epoch": 5.169272515828919,
|
| 6063 |
+
"learning_rate": 1.6791131855309218e-07,
|
| 6064 |
+
"loss": 2.6856,
|
| 6065 |
+
"step": 20000
|
| 6066 |
+
},
|
| 6067 |
+
{
|
| 6068 |
+
"epoch": 5.174441142266443,
|
| 6069 |
+
"learning_rate": 1.6787890574354984e-07,
|
| 6070 |
+
"loss": 2.6874,
|
| 6071 |
+
"step": 20020
|
| 6072 |
+
},
|
| 6073 |
+
{
|
| 6074 |
+
"epoch": 5.179609768703967,
|
| 6075 |
+
"learning_rate": 1.678464929340075e-07,
|
| 6076 |
+
"loss": 2.6737,
|
| 6077 |
+
"step": 20040
|
| 6078 |
+
},
|
| 6079 |
+
{
|
| 6080 |
+
"epoch": 5.184778395141491,
|
| 6081 |
+
"learning_rate": 1.678140801244652e-07,
|
| 6082 |
+
"loss": 2.6503,
|
| 6083 |
+
"step": 20060
|
| 6084 |
+
},
|
| 6085 |
+
{
|
| 6086 |
+
"epoch": 5.189947021579015,
|
| 6087 |
+
"learning_rate": 1.6778166731492285e-07,
|
| 6088 |
+
"loss": 2.6155,
|
| 6089 |
+
"step": 20080
|
| 6090 |
+
},
|
| 6091 |
+
{
|
| 6092 |
+
"epoch": 5.1951156480165395,
|
| 6093 |
+
"learning_rate": 1.677492545053805e-07,
|
| 6094 |
+
"loss": 2.7035,
|
| 6095 |
+
"step": 20100
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 5.200284274454064,
|
| 6099 |
+
"learning_rate": 1.677168416958382e-07,
|
| 6100 |
+
"loss": 2.6192,
|
| 6101 |
+
"step": 20120
|
| 6102 |
+
},
|
| 6103 |
+
{
|
| 6104 |
+
"epoch": 5.205452900891588,
|
| 6105 |
+
"learning_rate": 1.6768442888629586e-07,
|
| 6106 |
+
"loss": 2.5974,
|
| 6107 |
+
"step": 20140
|
| 6108 |
+
},
|
| 6109 |
+
{
|
| 6110 |
+
"epoch": 5.2106215273291125,
|
| 6111 |
+
"learning_rate": 1.6765201607675352e-07,
|
| 6112 |
+
"loss": 2.694,
|
| 6113 |
+
"step": 20160
|
| 6114 |
+
},
|
| 6115 |
+
{
|
| 6116 |
+
"epoch": 5.215790153766637,
|
| 6117 |
+
"learning_rate": 1.676196032672112e-07,
|
| 6118 |
+
"loss": 2.6391,
|
| 6119 |
+
"step": 20180
|
| 6120 |
+
},
|
| 6121 |
+
{
|
| 6122 |
+
"epoch": 5.22095878020416,
|
| 6123 |
+
"learning_rate": 1.6758719045766887e-07,
|
| 6124 |
+
"loss": 2.6428,
|
| 6125 |
+
"step": 20200
|
| 6126 |
+
},
|
| 6127 |
+
{
|
| 6128 |
+
"epoch": 5.226127406641685,
|
| 6129 |
+
"learning_rate": 1.6755477764812654e-07,
|
| 6130 |
+
"loss": 2.6796,
|
| 6131 |
+
"step": 20220
|
| 6132 |
+
},
|
| 6133 |
+
{
|
| 6134 |
+
"epoch": 5.231296033079209,
|
| 6135 |
+
"learning_rate": 1.675223648385842e-07,
|
| 6136 |
+
"loss": 2.6819,
|
| 6137 |
+
"step": 20240
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 5.2364646595167335,
|
| 6141 |
+
"learning_rate": 1.6748995202904186e-07,
|
| 6142 |
+
"loss": 2.6714,
|
| 6143 |
+
"step": 20260
|
| 6144 |
+
},
|
| 6145 |
+
{
|
| 6146 |
+
"epoch": 5.241633285954258,
|
| 6147 |
+
"learning_rate": 1.6745753921949955e-07,
|
| 6148 |
+
"loss": 2.6392,
|
| 6149 |
+
"step": 20280
|
| 6150 |
+
},
|
| 6151 |
+
{
|
| 6152 |
+
"epoch": 5.246801912391782,
|
| 6153 |
+
"learning_rate": 1.674251264099572e-07,
|
| 6154 |
+
"loss": 2.7057,
|
| 6155 |
+
"step": 20300
|
| 6156 |
+
},
|
| 6157 |
+
{
|
| 6158 |
+
"epoch": 5.2519705388293065,
|
| 6159 |
+
"learning_rate": 1.6739271360041487e-07,
|
| 6160 |
+
"loss": 2.6676,
|
| 6161 |
+
"step": 20320
|
| 6162 |
+
},
|
| 6163 |
+
{
|
| 6164 |
+
"epoch": 5.25713916526683,
|
| 6165 |
+
"learning_rate": 1.6736030079087256e-07,
|
| 6166 |
+
"loss": 2.7037,
|
| 6167 |
+
"step": 20340
|
| 6168 |
+
},
|
| 6169 |
+
{
|
| 6170 |
+
"epoch": 5.262307791704354,
|
| 6171 |
+
"learning_rate": 1.6732788798133022e-07,
|
| 6172 |
+
"loss": 2.7595,
|
| 6173 |
+
"step": 20360
|
| 6174 |
+
},
|
| 6175 |
+
{
|
| 6176 |
+
"epoch": 5.267476418141879,
|
| 6177 |
+
"learning_rate": 1.6729547517178788e-07,
|
| 6178 |
+
"loss": 2.6665,
|
| 6179 |
+
"step": 20380
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 5.272645044579403,
|
| 6183 |
+
"learning_rate": 1.6726306236224554e-07,
|
| 6184 |
+
"loss": 2.6281,
|
| 6185 |
+
"step": 20400
|
| 6186 |
+
},
|
| 6187 |
+
{
|
| 6188 |
+
"epoch": 5.277813671016927,
|
| 6189 |
+
"learning_rate": 1.6723064955270323e-07,
|
| 6190 |
+
"loss": 2.6047,
|
| 6191 |
+
"step": 20420
|
| 6192 |
+
},
|
| 6193 |
+
{
|
| 6194 |
+
"epoch": 5.282982297454452,
|
| 6195 |
+
"learning_rate": 1.671982367431609e-07,
|
| 6196 |
+
"loss": 2.6466,
|
| 6197 |
+
"step": 20440
|
| 6198 |
+
},
|
| 6199 |
+
{
|
| 6200 |
+
"epoch": 5.288150923891976,
|
| 6201 |
+
"learning_rate": 1.6716582393361856e-07,
|
| 6202 |
+
"loss": 2.6026,
|
| 6203 |
+
"step": 20460
|
| 6204 |
+
},
|
| 6205 |
+
{
|
| 6206 |
+
"epoch": 5.2933195503295,
|
| 6207 |
+
"learning_rate": 1.6713341112407622e-07,
|
| 6208 |
+
"loss": 2.6714,
|
| 6209 |
+
"step": 20480
|
| 6210 |
+
},
|
| 6211 |
+
{
|
| 6212 |
+
"epoch": 5.298488176767024,
|
| 6213 |
+
"learning_rate": 1.671009983145339e-07,
|
| 6214 |
+
"loss": 2.6351,
|
| 6215 |
+
"step": 20500
|
| 6216 |
+
},
|
| 6217 |
+
{
|
| 6218 |
+
"epoch": 5.303656803204548,
|
| 6219 |
+
"learning_rate": 1.6706858550499157e-07,
|
| 6220 |
+
"loss": 2.6293,
|
| 6221 |
+
"step": 20520
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 5.308825429642073,
|
| 6225 |
+
"learning_rate": 1.6703617269544923e-07,
|
| 6226 |
+
"loss": 2.6368,
|
| 6227 |
+
"step": 20540
|
| 6228 |
+
},
|
| 6229 |
+
{
|
| 6230 |
+
"epoch": 5.313994056079597,
|
| 6231 |
+
"learning_rate": 1.6700375988590692e-07,
|
| 6232 |
+
"loss": 2.6963,
|
| 6233 |
+
"step": 20560
|
| 6234 |
+
},
|
| 6235 |
+
{
|
| 6236 |
+
"epoch": 5.319162682517121,
|
| 6237 |
+
"learning_rate": 1.6697134707636458e-07,
|
| 6238 |
+
"loss": 2.6401,
|
| 6239 |
+
"step": 20580
|
| 6240 |
+
},
|
| 6241 |
+
{
|
| 6242 |
+
"epoch": 5.324331308954645,
|
| 6243 |
+
"learning_rate": 1.6693893426682224e-07,
|
| 6244 |
+
"loss": 2.669,
|
| 6245 |
+
"step": 20600
|
| 6246 |
+
},
|
| 6247 |
+
{
|
| 6248 |
+
"epoch": 5.329499935392169,
|
| 6249 |
+
"learning_rate": 1.669065214572799e-07,
|
| 6250 |
+
"loss": 2.6384,
|
| 6251 |
+
"step": 20620
|
| 6252 |
+
},
|
| 6253 |
+
{
|
| 6254 |
+
"epoch": 5.334668561829694,
|
| 6255 |
+
"learning_rate": 1.6687410864773757e-07,
|
| 6256 |
+
"loss": 2.7073,
|
| 6257 |
+
"step": 20640
|
| 6258 |
+
},
|
| 6259 |
+
{
|
| 6260 |
+
"epoch": 5.339837188267218,
|
| 6261 |
+
"learning_rate": 1.6684169583819526e-07,
|
| 6262 |
+
"loss": 2.6507,
|
| 6263 |
+
"step": 20660
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 5.345005814704742,
|
| 6267 |
+
"learning_rate": 1.6680928302865292e-07,
|
| 6268 |
+
"loss": 2.6378,
|
| 6269 |
+
"step": 20680
|
| 6270 |
+
},
|
| 6271 |
+
{
|
| 6272 |
+
"epoch": 5.350174441142267,
|
| 6273 |
+
"learning_rate": 1.6677687021911058e-07,
|
| 6274 |
+
"loss": 2.6756,
|
| 6275 |
+
"step": 20700
|
| 6276 |
+
},
|
| 6277 |
+
{
|
| 6278 |
+
"epoch": 5.355343067579791,
|
| 6279 |
+
"learning_rate": 1.6674445740956827e-07,
|
| 6280 |
+
"loss": 2.6667,
|
| 6281 |
+
"step": 20720
|
| 6282 |
+
},
|
| 6283 |
+
{
|
| 6284 |
+
"epoch": 5.3605116940173145,
|
| 6285 |
+
"learning_rate": 1.6671204460002593e-07,
|
| 6286 |
+
"loss": 2.6714,
|
| 6287 |
+
"step": 20740
|
| 6288 |
+
},
|
| 6289 |
+
{
|
| 6290 |
+
"epoch": 5.365680320454839,
|
| 6291 |
+
"learning_rate": 1.666796317904836e-07,
|
| 6292 |
+
"loss": 2.6117,
|
| 6293 |
+
"step": 20760
|
| 6294 |
+
},
|
| 6295 |
+
{
|
| 6296 |
+
"epoch": 5.370848946892363,
|
| 6297 |
+
"learning_rate": 1.6664721898094128e-07,
|
| 6298 |
+
"loss": 2.6513,
|
| 6299 |
+
"step": 20780
|
| 6300 |
+
},
|
| 6301 |
+
{
|
| 6302 |
+
"epoch": 5.376017573329888,
|
| 6303 |
+
"learning_rate": 1.6661480617139894e-07,
|
| 6304 |
+
"loss": 2.6997,
|
| 6305 |
+
"step": 20800
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 5.381186199767412,
|
| 6309 |
+
"learning_rate": 1.665823933618566e-07,
|
| 6310 |
+
"loss": 2.6395,
|
| 6311 |
+
"step": 20820
|
| 6312 |
+
},
|
| 6313 |
+
{
|
| 6314 |
+
"epoch": 5.386354826204936,
|
| 6315 |
+
"learning_rate": 1.6654998055231426e-07,
|
| 6316 |
+
"loss": 2.6615,
|
| 6317 |
+
"step": 20840
|
| 6318 |
+
},
|
| 6319 |
+
{
|
| 6320 |
+
"epoch": 5.39152345264246,
|
| 6321 |
+
"learning_rate": 1.6651756774277193e-07,
|
| 6322 |
+
"loss": 2.6915,
|
| 6323 |
+
"step": 20860
|
| 6324 |
+
},
|
| 6325 |
+
{
|
| 6326 |
+
"epoch": 5.396692079079984,
|
| 6327 |
+
"learning_rate": 1.6648515493322961e-07,
|
| 6328 |
+
"loss": 2.6248,
|
| 6329 |
+
"step": 20880
|
| 6330 |
+
},
|
| 6331 |
+
{
|
| 6332 |
+
"epoch": 5.4018607055175085,
|
| 6333 |
+
"learning_rate": 1.6645274212368728e-07,
|
| 6334 |
+
"loss": 2.7061,
|
| 6335 |
+
"step": 20900
|
| 6336 |
+
},
|
| 6337 |
+
{
|
| 6338 |
+
"epoch": 5.407029331955033,
|
| 6339 |
+
"learning_rate": 1.6642032931414494e-07,
|
| 6340 |
+
"loss": 2.642,
|
| 6341 |
+
"step": 20920
|
| 6342 |
+
},
|
| 6343 |
+
{
|
| 6344 |
+
"epoch": 5.412197958392557,
|
| 6345 |
+
"learning_rate": 1.6638791650460263e-07,
|
| 6346 |
+
"loss": 2.6377,
|
| 6347 |
+
"step": 20940
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 5.417366584830082,
|
| 6351 |
+
"learning_rate": 1.663555036950603e-07,
|
| 6352 |
+
"loss": 2.6452,
|
| 6353 |
+
"step": 20960
|
| 6354 |
+
},
|
| 6355 |
+
{
|
| 6356 |
+
"epoch": 5.422535211267606,
|
| 6357 |
+
"learning_rate": 1.6632309088551795e-07,
|
| 6358 |
+
"loss": 2.7206,
|
| 6359 |
+
"step": 20980
|
| 6360 |
+
},
|
| 6361 |
+
{
|
| 6362 |
+
"epoch": 5.427703837705129,
|
| 6363 |
+
"learning_rate": 1.662906780759756e-07,
|
| 6364 |
+
"loss": 2.6524,
|
| 6365 |
+
"step": 21000
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 5.432872464142654,
|
| 6369 |
+
"learning_rate": 1.662582652664333e-07,
|
| 6370 |
+
"loss": 2.6398,
|
| 6371 |
+
"step": 21020
|
| 6372 |
+
},
|
| 6373 |
+
{
|
| 6374 |
+
"epoch": 5.438041090580178,
|
| 6375 |
+
"learning_rate": 1.6622585245689096e-07,
|
| 6376 |
+
"loss": 2.6486,
|
| 6377 |
+
"step": 21040
|
| 6378 |
+
},
|
| 6379 |
+
{
|
| 6380 |
+
"epoch": 5.4432097170177025,
|
| 6381 |
+
"learning_rate": 1.6619343964734862e-07,
|
| 6382 |
+
"loss": 2.6466,
|
| 6383 |
+
"step": 21060
|
| 6384 |
+
},
|
| 6385 |
+
{
|
| 6386 |
+
"epoch": 5.448378343455227,
|
| 6387 |
+
"learning_rate": 1.6616102683780629e-07,
|
| 6388 |
+
"loss": 2.6818,
|
| 6389 |
+
"step": 21080
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 5.453546969892751,
|
| 6393 |
+
"learning_rate": 1.6612861402826397e-07,
|
| 6394 |
+
"loss": 2.6826,
|
| 6395 |
+
"step": 21100
|
| 6396 |
+
},
|
| 6397 |
+
{
|
| 6398 |
+
"epoch": 5.458715596330276,
|
| 6399 |
+
"learning_rate": 1.6609620121872164e-07,
|
| 6400 |
+
"loss": 2.665,
|
| 6401 |
+
"step": 21120
|
| 6402 |
+
},
|
| 6403 |
+
{
|
| 6404 |
+
"epoch": 5.463884222767799,
|
| 6405 |
+
"learning_rate": 1.660637884091793e-07,
|
| 6406 |
+
"loss": 2.6018,
|
| 6407 |
+
"step": 21140
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 5.469052849205323,
|
| 6411 |
+
"learning_rate": 1.66031375599637e-07,
|
| 6412 |
+
"loss": 2.6867,
|
| 6413 |
+
"step": 21160
|
| 6414 |
+
},
|
| 6415 |
+
{
|
| 6416 |
+
"epoch": 5.474221475642848,
|
| 6417 |
+
"learning_rate": 1.6599896279009465e-07,
|
| 6418 |
+
"loss": 2.6972,
|
| 6419 |
+
"step": 21180
|
| 6420 |
+
},
|
| 6421 |
+
{
|
| 6422 |
+
"epoch": 5.479390102080372,
|
| 6423 |
+
"learning_rate": 1.659665499805523e-07,
|
| 6424 |
+
"loss": 2.6775,
|
| 6425 |
+
"step": 21200
|
| 6426 |
+
},
|
| 6427 |
+
{
|
| 6428 |
+
"epoch": 5.4845587285178965,
|
| 6429 |
+
"learning_rate": 1.6593413717100997e-07,
|
| 6430 |
+
"loss": 2.7239,
|
| 6431 |
+
"step": 21220
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 5.489727354955421,
|
| 6435 |
+
"learning_rate": 1.6590172436146763e-07,
|
| 6436 |
+
"loss": 2.6349,
|
| 6437 |
+
"step": 21240
|
| 6438 |
+
},
|
| 6439 |
+
{
|
| 6440 |
+
"epoch": 5.494895981392945,
|
| 6441 |
+
"learning_rate": 1.6586931155192532e-07,
|
| 6442 |
+
"loss": 2.6607,
|
| 6443 |
+
"step": 21260
|
| 6444 |
+
},
|
| 6445 |
+
{
|
| 6446 |
+
"epoch": 5.500064607830469,
|
| 6447 |
+
"learning_rate": 1.6583689874238298e-07,
|
| 6448 |
+
"loss": 2.6806,
|
| 6449 |
+
"step": 21280
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 5.505233234267993,
|
| 6453 |
+
"learning_rate": 1.6580448593284065e-07,
|
| 6454 |
+
"loss": 2.7212,
|
| 6455 |
+
"step": 21300
|
| 6456 |
+
},
|
| 6457 |
+
{
|
| 6458 |
+
"epoch": 5.510401860705517,
|
| 6459 |
+
"learning_rate": 1.6577207312329833e-07,
|
| 6460 |
+
"loss": 2.6997,
|
| 6461 |
+
"step": 21320
|
| 6462 |
+
},
|
| 6463 |
+
{
|
| 6464 |
+
"epoch": 5.515570487143042,
|
| 6465 |
+
"learning_rate": 1.65739660313756e-07,
|
| 6466 |
+
"loss": 2.6277,
|
| 6467 |
+
"step": 21340
|
| 6468 |
+
},
|
| 6469 |
+
{
|
| 6470 |
+
"epoch": 5.520739113580566,
|
| 6471 |
+
"learning_rate": 1.6570724750421366e-07,
|
| 6472 |
+
"loss": 2.6953,
|
| 6473 |
+
"step": 21360
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 5.5259077400180905,
|
| 6477 |
+
"learning_rate": 1.6567483469467135e-07,
|
| 6478 |
+
"loss": 2.7586,
|
| 6479 |
+
"step": 21380
|
| 6480 |
+
},
|
| 6481 |
+
{
|
| 6482 |
+
"epoch": 5.531076366455615,
|
| 6483 |
+
"learning_rate": 1.65642421885129e-07,
|
| 6484 |
+
"loss": 2.5887,
|
| 6485 |
+
"step": 21400
|
| 6486 |
+
},
|
| 6487 |
+
{
|
| 6488 |
+
"epoch": 5.536244992893138,
|
| 6489 |
+
"learning_rate": 1.6561000907558667e-07,
|
| 6490 |
+
"loss": 2.6364,
|
| 6491 |
+
"step": 21420
|
| 6492 |
+
},
|
| 6493 |
+
{
|
| 6494 |
+
"epoch": 5.541413619330663,
|
| 6495 |
+
"learning_rate": 1.6557759626604433e-07,
|
| 6496 |
+
"loss": 2.5951,
|
| 6497 |
+
"step": 21440
|
| 6498 |
+
},
|
| 6499 |
+
{
|
| 6500 |
+
"epoch": 5.546582245768187,
|
| 6501 |
+
"learning_rate": 1.65545183456502e-07,
|
| 6502 |
+
"loss": 2.6595,
|
| 6503 |
+
"step": 21460
|
| 6504 |
+
},
|
| 6505 |
+
{
|
| 6506 |
+
"epoch": 5.551750872205711,
|
| 6507 |
+
"learning_rate": 1.6551277064695968e-07,
|
| 6508 |
+
"loss": 2.602,
|
| 6509 |
+
"step": 21480
|
| 6510 |
+
},
|
| 6511 |
+
{
|
| 6512 |
+
"epoch": 5.556919498643236,
|
| 6513 |
+
"learning_rate": 1.6548035783741734e-07,
|
| 6514 |
+
"loss": 2.5996,
|
| 6515 |
+
"step": 21500
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 5.56208812508076,
|
| 6519 |
+
"learning_rate": 1.65447945027875e-07,
|
| 6520 |
+
"loss": 2.6529,
|
| 6521 |
+
"step": 21520
|
| 6522 |
+
},
|
| 6523 |
+
{
|
| 6524 |
+
"epoch": 5.5672567515182845,
|
| 6525 |
+
"learning_rate": 1.654155322183327e-07,
|
| 6526 |
+
"loss": 2.6917,
|
| 6527 |
+
"step": 21540
|
| 6528 |
+
},
|
| 6529 |
+
{
|
| 6530 |
+
"epoch": 5.572425377955808,
|
| 6531 |
+
"learning_rate": 1.6538311940879036e-07,
|
| 6532 |
+
"loss": 2.6509,
|
| 6533 |
+
"step": 21560
|
| 6534 |
+
},
|
| 6535 |
+
{
|
| 6536 |
+
"epoch": 5.577594004393332,
|
| 6537 |
+
"learning_rate": 1.6535070659924802e-07,
|
| 6538 |
+
"loss": 2.6524,
|
| 6539 |
+
"step": 21580
|
| 6540 |
+
},
|
| 6541 |
+
{
|
| 6542 |
+
"epoch": 5.582762630830857,
|
| 6543 |
+
"learning_rate": 1.6531829378970568e-07,
|
| 6544 |
+
"loss": 2.6894,
|
| 6545 |
+
"step": 21600
|
| 6546 |
+
},
|
| 6547 |
+
{
|
| 6548 |
+
"epoch": 5.587931257268381,
|
| 6549 |
+
"learning_rate": 1.6528588098016334e-07,
|
| 6550 |
+
"loss": 2.6451,
|
| 6551 |
+
"step": 21620
|
| 6552 |
+
},
|
| 6553 |
+
{
|
| 6554 |
+
"epoch": 5.593099883705905,
|
| 6555 |
+
"learning_rate": 1.6525346817062103e-07,
|
| 6556 |
+
"loss": 2.6636,
|
| 6557 |
+
"step": 21640
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 5.59826851014343,
|
| 6561 |
+
"learning_rate": 1.652210553610787e-07,
|
| 6562 |
+
"loss": 2.6696,
|
| 6563 |
+
"step": 21660
|
| 6564 |
+
},
|
| 6565 |
+
{
|
| 6566 |
+
"epoch": 5.603437136580954,
|
| 6567 |
+
"learning_rate": 1.6518864255153635e-07,
|
| 6568 |
+
"loss": 2.6577,
|
| 6569 |
+
"step": 21680
|
| 6570 |
+
},
|
| 6571 |
+
{
|
| 6572 |
+
"epoch": 5.608605763018478,
|
| 6573 |
+
"learning_rate": 1.6515622974199404e-07,
|
| 6574 |
+
"loss": 2.5968,
|
| 6575 |
+
"step": 21700
|
| 6576 |
+
},
|
| 6577 |
+
{
|
| 6578 |
+
"epoch": 5.613774389456002,
|
| 6579 |
+
"learning_rate": 1.651238169324517e-07,
|
| 6580 |
+
"loss": 2.6492,
|
| 6581 |
+
"step": 21720
|
| 6582 |
+
},
|
| 6583 |
+
{
|
| 6584 |
+
"epoch": 5.618943015893526,
|
| 6585 |
+
"learning_rate": 1.6509140412290937e-07,
|
| 6586 |
+
"loss": 2.6357,
|
| 6587 |
+
"step": 21740
|
| 6588 |
+
},
|
| 6589 |
+
{
|
| 6590 |
+
"epoch": 5.624111642331051,
|
| 6591 |
+
"learning_rate": 1.6505899131336705e-07,
|
| 6592 |
+
"loss": 2.6401,
|
| 6593 |
+
"step": 21760
|
| 6594 |
+
},
|
| 6595 |
+
{
|
| 6596 |
+
"epoch": 5.629280268768575,
|
| 6597 |
+
"learning_rate": 1.650265785038247e-07,
|
| 6598 |
+
"loss": 2.5984,
|
| 6599 |
+
"step": 21780
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 5.6344488952060985,
|
| 6603 |
+
"learning_rate": 1.6499416569428238e-07,
|
| 6604 |
+
"loss": 2.6678,
|
| 6605 |
+
"step": 21800
|
| 6606 |
+
},
|
| 6607 |
+
{
|
| 6608 |
+
"epoch": 5.639617521643623,
|
| 6609 |
+
"learning_rate": 1.6496175288474004e-07,
|
| 6610 |
+
"loss": 2.6161,
|
| 6611 |
+
"step": 21820
|
| 6612 |
+
},
|
| 6613 |
+
{
|
| 6614 |
+
"epoch": 5.644786148081147,
|
| 6615 |
+
"learning_rate": 1.649293400751977e-07,
|
| 6616 |
+
"loss": 2.6262,
|
| 6617 |
+
"step": 21840
|
| 6618 |
+
},
|
| 6619 |
+
{
|
| 6620 |
+
"epoch": 5.649954774518672,
|
| 6621 |
+
"learning_rate": 1.648969272656554e-07,
|
| 6622 |
+
"loss": 2.6514,
|
| 6623 |
+
"step": 21860
|
| 6624 |
+
},
|
| 6625 |
+
{
|
| 6626 |
+
"epoch": 5.655123400956196,
|
| 6627 |
+
"learning_rate": 1.6486451445611305e-07,
|
| 6628 |
+
"loss": 2.6629,
|
| 6629 |
+
"step": 21880
|
| 6630 |
+
},
|
| 6631 |
+
{
|
| 6632 |
+
"epoch": 5.66029202739372,
|
| 6633 |
+
"learning_rate": 1.6483210164657071e-07,
|
| 6634 |
+
"loss": 2.6764,
|
| 6635 |
+
"step": 21900
|
| 6636 |
+
},
|
| 6637 |
+
{
|
| 6638 |
+
"epoch": 5.665460653831245,
|
| 6639 |
+
"learning_rate": 1.647996888370284e-07,
|
| 6640 |
+
"loss": 2.6414,
|
| 6641 |
+
"step": 21920
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 5.670629280268768,
|
| 6645 |
+
"learning_rate": 1.6476727602748604e-07,
|
| 6646 |
+
"loss": 2.5379,
|
| 6647 |
+
"step": 21940
|
| 6648 |
+
},
|
| 6649 |
+
{
|
| 6650 |
+
"epoch": 5.6757979067062925,
|
| 6651 |
+
"learning_rate": 1.6473486321794373e-07,
|
| 6652 |
+
"loss": 2.6744,
|
| 6653 |
+
"step": 21960
|
| 6654 |
+
},
|
| 6655 |
+
{
|
| 6656 |
+
"epoch": 5.680966533143817,
|
| 6657 |
+
"learning_rate": 1.647024504084014e-07,
|
| 6658 |
+
"loss": 2.7254,
|
| 6659 |
+
"step": 21980
|
| 6660 |
+
},
|
| 6661 |
+
{
|
| 6662 |
+
"epoch": 5.686135159581341,
|
| 6663 |
+
"learning_rate": 1.6467003759885905e-07,
|
| 6664 |
+
"loss": 2.6408,
|
| 6665 |
+
"step": 22000
|
| 6666 |
+
},
|
| 6667 |
+
{
|
| 6668 |
+
"epoch": 5.6913037860188656,
|
| 6669 |
+
"learning_rate": 1.6463762478931674e-07,
|
| 6670 |
+
"loss": 2.6751,
|
| 6671 |
+
"step": 22020
|
| 6672 |
+
},
|
| 6673 |
+
{
|
| 6674 |
+
"epoch": 5.69647241245639,
|
| 6675 |
+
"learning_rate": 1.646052119797744e-07,
|
| 6676 |
+
"loss": 2.6391,
|
| 6677 |
+
"step": 22040
|
| 6678 |
+
},
|
| 6679 |
+
{
|
| 6680 |
+
"epoch": 5.701641038893914,
|
| 6681 |
+
"learning_rate": 1.6457279917023206e-07,
|
| 6682 |
+
"loss": 2.625,
|
| 6683 |
+
"step": 22060
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 5.706809665331438,
|
| 6687 |
+
"learning_rate": 1.6454038636068975e-07,
|
| 6688 |
+
"loss": 2.607,
|
| 6689 |
+
"step": 22080
|
| 6690 |
+
},
|
| 6691 |
+
{
|
| 6692 |
+
"epoch": 5.711978291768962,
|
| 6693 |
+
"learning_rate": 1.6450797355114739e-07,
|
| 6694 |
+
"loss": 2.6629,
|
| 6695 |
+
"step": 22100
|
| 6696 |
+
},
|
| 6697 |
+
{
|
| 6698 |
+
"epoch": 5.7171469182064865,
|
| 6699 |
+
"learning_rate": 1.6447556074160507e-07,
|
| 6700 |
+
"loss": 2.6358,
|
| 6701 |
+
"step": 22120
|
| 6702 |
+
},
|
| 6703 |
+
{
|
| 6704 |
+
"epoch": 5.722315544644011,
|
| 6705 |
+
"learning_rate": 1.6444314793206274e-07,
|
| 6706 |
+
"loss": 2.6962,
|
| 6707 |
+
"step": 22140
|
| 6708 |
+
},
|
| 6709 |
+
{
|
| 6710 |
+
"epoch": 5.727484171081535,
|
| 6711 |
+
"learning_rate": 1.644107351225204e-07,
|
| 6712 |
+
"loss": 2.6403,
|
| 6713 |
+
"step": 22160
|
| 6714 |
+
},
|
| 6715 |
+
{
|
| 6716 |
+
"epoch": 5.7326527975190595,
|
| 6717 |
+
"learning_rate": 1.6437832231297809e-07,
|
| 6718 |
+
"loss": 2.641,
|
| 6719 |
+
"step": 22180
|
| 6720 |
+
},
|
| 6721 |
+
{
|
| 6722 |
+
"epoch": 5.737821423956584,
|
| 6723 |
+
"learning_rate": 1.6434590950343575e-07,
|
| 6724 |
+
"loss": 2.6213,
|
| 6725 |
+
"step": 22200
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 5.742990050394107,
|
| 6729 |
+
"learning_rate": 1.643134966938934e-07,
|
| 6730 |
+
"loss": 2.6508,
|
| 6731 |
+
"step": 22220
|
| 6732 |
+
},
|
| 6733 |
+
{
|
| 6734 |
+
"epoch": 5.748158676831632,
|
| 6735 |
+
"learning_rate": 1.642810838843511e-07,
|
| 6736 |
+
"loss": 2.6528,
|
| 6737 |
+
"step": 22240
|
| 6738 |
+
},
|
| 6739 |
+
{
|
| 6740 |
+
"epoch": 5.753327303269156,
|
| 6741 |
+
"learning_rate": 1.6424867107480873e-07,
|
| 6742 |
+
"loss": 2.5921,
|
| 6743 |
+
"step": 22260
|
| 6744 |
+
},
|
| 6745 |
+
{
|
| 6746 |
+
"epoch": 5.7584959297066804,
|
| 6747 |
+
"learning_rate": 1.6421625826526642e-07,
|
| 6748 |
+
"loss": 2.6379,
|
| 6749 |
+
"step": 22280
|
| 6750 |
+
},
|
| 6751 |
+
{
|
| 6752 |
+
"epoch": 5.763664556144205,
|
| 6753 |
+
"learning_rate": 1.641838454557241e-07,
|
| 6754 |
+
"loss": 2.6838,
|
| 6755 |
+
"step": 22300
|
| 6756 |
+
},
|
| 6757 |
+
{
|
| 6758 |
+
"epoch": 5.768833182581729,
|
| 6759 |
+
"learning_rate": 1.6415143264618175e-07,
|
| 6760 |
+
"loss": 2.6511,
|
| 6761 |
+
"step": 22320
|
| 6762 |
+
},
|
| 6763 |
+
{
|
| 6764 |
+
"epoch": 5.7740018090192535,
|
| 6765 |
+
"learning_rate": 1.6411901983663943e-07,
|
| 6766 |
+
"loss": 2.6547,
|
| 6767 |
+
"step": 22340
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 5.779170435456777,
|
| 6771 |
+
"learning_rate": 1.640866070270971e-07,
|
| 6772 |
+
"loss": 2.5962,
|
| 6773 |
+
"step": 22360
|
| 6774 |
+
},
|
| 6775 |
+
{
|
| 6776 |
+
"epoch": 5.784339061894301,
|
| 6777 |
+
"learning_rate": 1.6405419421755476e-07,
|
| 6778 |
+
"loss": 2.6074,
|
| 6779 |
+
"step": 22380
|
| 6780 |
+
},
|
| 6781 |
+
{
|
| 6782 |
+
"epoch": 5.789507688331826,
|
| 6783 |
+
"learning_rate": 1.6402178140801245e-07,
|
| 6784 |
+
"loss": 2.5927,
|
| 6785 |
+
"step": 22400
|
| 6786 |
+
},
|
| 6787 |
+
{
|
| 6788 |
+
"epoch": 5.79467631476935,
|
| 6789 |
+
"learning_rate": 1.6398936859847008e-07,
|
| 6790 |
+
"loss": 2.6697,
|
| 6791 |
+
"step": 22420
|
| 6792 |
+
},
|
| 6793 |
+
{
|
| 6794 |
+
"epoch": 5.799844941206874,
|
| 6795 |
+
"learning_rate": 1.6395695578892777e-07,
|
| 6796 |
+
"loss": 2.6267,
|
| 6797 |
+
"step": 22440
|
| 6798 |
+
},
|
| 6799 |
+
{
|
| 6800 |
+
"epoch": 5.805013567644399,
|
| 6801 |
+
"learning_rate": 1.6392454297938546e-07,
|
| 6802 |
+
"loss": 2.6833,
|
| 6803 |
+
"step": 22460
|
| 6804 |
+
},
|
| 6805 |
+
{
|
| 6806 |
+
"epoch": 5.810182194081923,
|
| 6807 |
+
"learning_rate": 1.638921301698431e-07,
|
| 6808 |
+
"loss": 2.6647,
|
| 6809 |
+
"step": 22480
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 5.815350820519447,
|
| 6813 |
+
"learning_rate": 1.6385971736030078e-07,
|
| 6814 |
+
"loss": 2.6619,
|
| 6815 |
+
"step": 22500
|
| 6816 |
+
},
|
| 6817 |
+
{
|
| 6818 |
+
"epoch": 5.820519446956971,
|
| 6819 |
+
"learning_rate": 1.6382730455075847e-07,
|
| 6820 |
+
"loss": 2.6341,
|
| 6821 |
+
"step": 22520
|
| 6822 |
+
},
|
| 6823 |
+
{
|
| 6824 |
+
"epoch": 5.825688073394495,
|
| 6825 |
+
"learning_rate": 1.637948917412161e-07,
|
| 6826 |
+
"loss": 2.604,
|
| 6827 |
+
"step": 22540
|
| 6828 |
+
},
|
| 6829 |
+
{
|
| 6830 |
+
"epoch": 5.83085669983202,
|
| 6831 |
+
"learning_rate": 1.637624789316738e-07,
|
| 6832 |
+
"loss": 2.6344,
|
| 6833 |
+
"step": 22560
|
| 6834 |
+
},
|
| 6835 |
+
{
|
| 6836 |
+
"epoch": 5.836025326269544,
|
| 6837 |
+
"learning_rate": 1.6373006612213146e-07,
|
| 6838 |
+
"loss": 2.6453,
|
| 6839 |
+
"step": 22580
|
| 6840 |
+
},
|
| 6841 |
+
{
|
| 6842 |
+
"epoch": 5.841193952707068,
|
| 6843 |
+
"learning_rate": 1.6369765331258912e-07,
|
| 6844 |
+
"loss": 2.6433,
|
| 6845 |
+
"step": 22600
|
| 6846 |
+
},
|
| 6847 |
+
{
|
| 6848 |
+
"epoch": 5.846362579144593,
|
| 6849 |
+
"learning_rate": 1.636652405030468e-07,
|
| 6850 |
+
"loss": 2.5673,
|
| 6851 |
+
"step": 22620
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 5.851531205582116,
|
| 6855 |
+
"learning_rate": 1.6363282769350444e-07,
|
| 6856 |
+
"loss": 2.6455,
|
| 6857 |
+
"step": 22640
|
| 6858 |
+
},
|
| 6859 |
+
{
|
| 6860 |
+
"epoch": 5.856699832019641,
|
| 6861 |
+
"learning_rate": 1.6360041488396213e-07,
|
| 6862 |
+
"loss": 2.6617,
|
| 6863 |
+
"step": 22660
|
| 6864 |
+
},
|
| 6865 |
+
{
|
| 6866 |
+
"epoch": 5.861868458457165,
|
| 6867 |
+
"learning_rate": 1.6356800207441982e-07,
|
| 6868 |
+
"loss": 2.6395,
|
| 6869 |
+
"step": 22680
|
| 6870 |
+
},
|
| 6871 |
+
{
|
| 6872 |
+
"epoch": 5.867037084894689,
|
| 6873 |
+
"learning_rate": 1.6353558926487745e-07,
|
| 6874 |
+
"loss": 2.6358,
|
| 6875 |
+
"step": 22700
|
| 6876 |
+
},
|
| 6877 |
+
{
|
| 6878 |
+
"epoch": 5.872205711332214,
|
| 6879 |
+
"learning_rate": 1.6350317645533514e-07,
|
| 6880 |
+
"loss": 2.618,
|
| 6881 |
+
"step": 22720
|
| 6882 |
+
},
|
| 6883 |
+
{
|
| 6884 |
+
"epoch": 5.877374337769738,
|
| 6885 |
+
"learning_rate": 1.634707636457928e-07,
|
| 6886 |
+
"loss": 2.6219,
|
| 6887 |
+
"step": 22740
|
| 6888 |
+
},
|
| 6889 |
+
{
|
| 6890 |
+
"epoch": 5.882542964207262,
|
| 6891 |
+
"learning_rate": 1.6343835083625047e-07,
|
| 6892 |
+
"loss": 2.6028,
|
| 6893 |
+
"step": 22760
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 5.887711590644786,
|
| 6897 |
+
"learning_rate": 1.6340593802670815e-07,
|
| 6898 |
+
"loss": 2.599,
|
| 6899 |
+
"step": 22780
|
| 6900 |
+
},
|
| 6901 |
+
{
|
| 6902 |
+
"epoch": 5.89288021708231,
|
| 6903 |
+
"learning_rate": 1.633735252171658e-07,
|
| 6904 |
+
"loss": 2.6062,
|
| 6905 |
+
"step": 22800
|
| 6906 |
+
},
|
| 6907 |
+
{
|
| 6908 |
+
"epoch": 5.898048843519835,
|
| 6909 |
+
"learning_rate": 1.6334111240762348e-07,
|
| 6910 |
+
"loss": 2.5854,
|
| 6911 |
+
"step": 22820
|
| 6912 |
+
},
|
| 6913 |
+
{
|
| 6914 |
+
"epoch": 5.903217469957359,
|
| 6915 |
+
"learning_rate": 1.6330869959808117e-07,
|
| 6916 |
+
"loss": 2.5416,
|
| 6917 |
+
"step": 22840
|
| 6918 |
+
},
|
| 6919 |
+
{
|
| 6920 |
+
"epoch": 5.908386096394883,
|
| 6921 |
+
"learning_rate": 1.632762867885388e-07,
|
| 6922 |
+
"loss": 2.6731,
|
| 6923 |
+
"step": 22860
|
| 6924 |
+
},
|
| 6925 |
+
{
|
| 6926 |
+
"epoch": 5.913554722832407,
|
| 6927 |
+
"learning_rate": 1.632438739789965e-07,
|
| 6928 |
+
"loss": 2.6271,
|
| 6929 |
+
"step": 22880
|
| 6930 |
+
},
|
| 6931 |
+
{
|
| 6932 |
+
"epoch": 5.918723349269931,
|
| 6933 |
+
"learning_rate": 1.6321146116945418e-07,
|
| 6934 |
+
"loss": 2.5896,
|
| 6935 |
+
"step": 22900
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 5.9238919757074555,
|
| 6939 |
+
"learning_rate": 1.6317904835991181e-07,
|
| 6940 |
+
"loss": 2.6794,
|
| 6941 |
+
"step": 22920
|
| 6942 |
+
},
|
| 6943 |
+
{
|
| 6944 |
+
"epoch": 5.92906060214498,
|
| 6945 |
+
"learning_rate": 1.631466355503695e-07,
|
| 6946 |
+
"loss": 2.6051,
|
| 6947 |
+
"step": 22940
|
| 6948 |
+
},
|
| 6949 |
+
{
|
| 6950 |
+
"epoch": 5.934229228582504,
|
| 6951 |
+
"learning_rate": 1.6311422274082716e-07,
|
| 6952 |
+
"loss": 2.6901,
|
| 6953 |
+
"step": 22960
|
| 6954 |
+
},
|
| 6955 |
+
{
|
| 6956 |
+
"epoch": 5.939397855020029,
|
| 6957 |
+
"learning_rate": 1.6308180993128483e-07,
|
| 6958 |
+
"loss": 2.6288,
|
| 6959 |
+
"step": 22980
|
| 6960 |
+
},
|
| 6961 |
+
{
|
| 6962 |
+
"epoch": 5.944566481457553,
|
| 6963 |
+
"learning_rate": 1.6304939712174251e-07,
|
| 6964 |
+
"loss": 2.6806,
|
| 6965 |
+
"step": 23000
|
| 6966 |
+
},
|
| 6967 |
+
{
|
| 6968 |
+
"epoch": 5.949735107895076,
|
| 6969 |
+
"learning_rate": 1.6301698431220015e-07,
|
| 6970 |
+
"loss": 2.6003,
|
| 6971 |
+
"step": 23020
|
| 6972 |
+
},
|
| 6973 |
+
{
|
| 6974 |
+
"epoch": 5.954903734332601,
|
| 6975 |
+
"learning_rate": 1.6298457150265784e-07,
|
| 6976 |
+
"loss": 2.6595,
|
| 6977 |
+
"step": 23040
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 5.960072360770125,
|
| 6981 |
+
"learning_rate": 1.6295215869311553e-07,
|
| 6982 |
+
"loss": 2.6563,
|
| 6983 |
+
"step": 23060
|
| 6984 |
+
},
|
| 6985 |
+
{
|
| 6986 |
+
"epoch": 5.9652409872076495,
|
| 6987 |
+
"learning_rate": 1.6291974588357316e-07,
|
| 6988 |
+
"loss": 2.6499,
|
| 6989 |
+
"step": 23080
|
| 6990 |
+
},
|
| 6991 |
+
{
|
| 6992 |
+
"epoch": 5.970409613645174,
|
| 6993 |
+
"learning_rate": 1.6288733307403085e-07,
|
| 6994 |
+
"loss": 2.6836,
|
| 6995 |
+
"step": 23100
|
| 6996 |
+
},
|
| 6997 |
+
{
|
| 6998 |
+
"epoch": 5.975578240082698,
|
| 6999 |
+
"learning_rate": 1.6285492026448854e-07,
|
| 7000 |
+
"loss": 2.6193,
|
| 7001 |
+
"step": 23120
|
| 7002 |
+
},
|
| 7003 |
+
{
|
| 7004 |
+
"epoch": 5.980746866520223,
|
| 7005 |
+
"learning_rate": 1.6282250745494617e-07,
|
| 7006 |
+
"loss": 2.6293,
|
| 7007 |
+
"step": 23140
|
| 7008 |
+
},
|
| 7009 |
+
{
|
| 7010 |
+
"epoch": 5.985915492957746,
|
| 7011 |
+
"learning_rate": 1.6279009464540386e-07,
|
| 7012 |
+
"loss": 2.7275,
|
| 7013 |
+
"step": 23160
|
| 7014 |
+
},
|
| 7015 |
+
{
|
| 7016 |
+
"epoch": 5.99108411939527,
|
| 7017 |
+
"learning_rate": 1.6275768183586152e-07,
|
| 7018 |
+
"loss": 2.6021,
|
| 7019 |
+
"step": 23180
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 5.996252745832795,
|
| 7023 |
+
"learning_rate": 1.6272526902631919e-07,
|
| 7024 |
+
"loss": 2.6331,
|
| 7025 |
+
"step": 23200
|
| 7026 |
+
},
|
| 7027 |
+
{
|
| 7028 |
+
"epoch": 5.999870784339062,
|
| 7029 |
+
"eval_bleu": 1.6991,
|
| 7030 |
+
"eval_gen_len": 66.9245,
|
| 7031 |
+
"eval_loss": 2.596095085144043,
|
| 7032 |
+
"eval_runtime": 1347.5897,
|
| 7033 |
+
"eval_samples_per_second": 1.278,
|
| 7034 |
+
"eval_steps_per_second": 0.639,
|
| 7035 |
+
"step": 23214
|
| 7036 |
+
},
|
| 7037 |
+
{
|
| 7038 |
+
"epoch": 6.001421372270319,
|
| 7039 |
+
"learning_rate": 1.6269285621677687e-07,
|
| 7040 |
+
"loss": 2.6423,
|
| 7041 |
+
"step": 23220
|
| 7042 |
+
},
|
| 7043 |
+
{
|
| 7044 |
+
"epoch": 6.0065899987078435,
|
| 7045 |
+
"learning_rate": 1.626604434072345e-07,
|
| 7046 |
+
"loss": 2.667,
|
| 7047 |
+
"step": 23240
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 6.011758625145368,
|
| 7051 |
+
"learning_rate": 1.626280305976922e-07,
|
| 7052 |
+
"loss": 2.6741,
|
| 7053 |
+
"step": 23260
|
| 7054 |
+
},
|
| 7055 |
+
{
|
| 7056 |
+
"epoch": 6.016927251582892,
|
| 7057 |
+
"learning_rate": 1.6259561778814989e-07,
|
| 7058 |
+
"loss": 2.6269,
|
| 7059 |
+
"step": 23280
|
| 7060 |
+
},
|
| 7061 |
+
{
|
| 7062 |
+
"epoch": 6.022095878020416,
|
| 7063 |
+
"learning_rate": 1.6256320497860752e-07,
|
| 7064 |
+
"loss": 2.6817,
|
| 7065 |
+
"step": 23300
|
| 7066 |
+
},
|
| 7067 |
+
{
|
| 7068 |
+
"epoch": 6.02726450445794,
|
| 7069 |
+
"learning_rate": 1.625307921690652e-07,
|
| 7070 |
+
"loss": 2.6804,
|
| 7071 |
+
"step": 23320
|
| 7072 |
+
},
|
| 7073 |
+
{
|
| 7074 |
+
"epoch": 6.032433130895464,
|
| 7075 |
+
"learning_rate": 1.6249837935952287e-07,
|
| 7076 |
+
"loss": 2.6283,
|
| 7077 |
+
"step": 23340
|
| 7078 |
+
},
|
| 7079 |
+
{
|
| 7080 |
+
"epoch": 6.037601757332989,
|
| 7081 |
+
"learning_rate": 1.6246596654998053e-07,
|
| 7082 |
+
"loss": 2.6344,
|
| 7083 |
+
"step": 23360
|
| 7084 |
+
},
|
| 7085 |
+
{
|
| 7086 |
+
"epoch": 6.042770383770513,
|
| 7087 |
+
"learning_rate": 1.6243355374043822e-07,
|
| 7088 |
+
"loss": 2.6453,
|
| 7089 |
+
"step": 23380
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 6.0479390102080375,
|
| 7093 |
+
"learning_rate": 1.6240114093089586e-07,
|
| 7094 |
+
"loss": 2.5926,
|
| 7095 |
+
"step": 23400
|
| 7096 |
+
},
|
| 7097 |
+
{
|
| 7098 |
+
"epoch": 6.053107636645562,
|
| 7099 |
+
"learning_rate": 1.6236872812135355e-07,
|
| 7100 |
+
"loss": 2.6709,
|
| 7101 |
+
"step": 23420
|
| 7102 |
+
},
|
| 7103 |
+
{
|
| 7104 |
+
"epoch": 6.058276263083085,
|
| 7105 |
+
"learning_rate": 1.6233631531181123e-07,
|
| 7106 |
+
"loss": 2.6228,
|
| 7107 |
+
"step": 23440
|
| 7108 |
+
},
|
| 7109 |
+
{
|
| 7110 |
+
"epoch": 6.06344488952061,
|
| 7111 |
+
"learning_rate": 1.6230390250226887e-07,
|
| 7112 |
+
"loss": 2.6006,
|
| 7113 |
+
"step": 23460
|
| 7114 |
+
},
|
| 7115 |
+
{
|
| 7116 |
+
"epoch": 6.068613515958134,
|
| 7117 |
+
"learning_rate": 1.6227148969272656e-07,
|
| 7118 |
+
"loss": 2.6179,
|
| 7119 |
+
"step": 23480
|
| 7120 |
+
},
|
| 7121 |
+
{
|
| 7122 |
+
"epoch": 6.073782142395658,
|
| 7123 |
+
"learning_rate": 1.6223907688318425e-07,
|
| 7124 |
+
"loss": 2.63,
|
| 7125 |
+
"step": 23500
|
| 7126 |
+
},
|
| 7127 |
+
{
|
| 7128 |
+
"epoch": 6.078950768833183,
|
| 7129 |
+
"learning_rate": 1.6220666407364188e-07,
|
| 7130 |
+
"loss": 2.653,
|
| 7131 |
+
"step": 23520
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 6.084119395270707,
|
| 7135 |
+
"learning_rate": 1.6217425126409957e-07,
|
| 7136 |
+
"loss": 2.6129,
|
| 7137 |
+
"step": 23540
|
| 7138 |
+
},
|
| 7139 |
+
{
|
| 7140 |
+
"epoch": 6.0892880217082315,
|
| 7141 |
+
"learning_rate": 1.6214183845455723e-07,
|
| 7142 |
+
"loss": 2.6008,
|
| 7143 |
+
"step": 23560
|
| 7144 |
+
},
|
| 7145 |
+
{
|
| 7146 |
+
"epoch": 6.094456648145755,
|
| 7147 |
+
"learning_rate": 1.621094256450149e-07,
|
| 7148 |
+
"loss": 2.6594,
|
| 7149 |
+
"step": 23580
|
| 7150 |
+
},
|
| 7151 |
+
{
|
| 7152 |
+
"epoch": 6.099625274583279,
|
| 7153 |
+
"learning_rate": 1.6207701283547258e-07,
|
| 7154 |
+
"loss": 2.617,
|
| 7155 |
+
"step": 23600
|
| 7156 |
+
},
|
| 7157 |
+
{
|
| 7158 |
+
"epoch": 6.104793901020804,
|
| 7159 |
+
"learning_rate": 1.6204460002593022e-07,
|
| 7160 |
+
"loss": 2.6392,
|
| 7161 |
+
"step": 23620
|
| 7162 |
+
},
|
| 7163 |
+
{
|
| 7164 |
+
"epoch": 6.109962527458328,
|
| 7165 |
+
"learning_rate": 1.620121872163879e-07,
|
| 7166 |
+
"loss": 2.6315,
|
| 7167 |
+
"step": 23640
|
| 7168 |
+
},
|
| 7169 |
+
{
|
| 7170 |
+
"epoch": 6.115131153895852,
|
| 7171 |
+
"learning_rate": 1.619797744068456e-07,
|
| 7172 |
+
"loss": 2.6131,
|
| 7173 |
+
"step": 23660
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 6.120299780333377,
|
| 7177 |
+
"learning_rate": 1.6194736159730323e-07,
|
| 7178 |
+
"loss": 2.6512,
|
| 7179 |
+
"step": 23680
|
| 7180 |
+
},
|
| 7181 |
+
{
|
| 7182 |
+
"epoch": 6.1254684067709,
|
| 7183 |
+
"learning_rate": 1.6191494878776092e-07,
|
| 7184 |
+
"loss": 2.6196,
|
| 7185 |
+
"step": 23700
|
| 7186 |
+
},
|
| 7187 |
+
{
|
| 7188 |
+
"epoch": 6.130637033208425,
|
| 7189 |
+
"learning_rate": 1.618825359782186e-07,
|
| 7190 |
+
"loss": 2.6234,
|
| 7191 |
+
"step": 23720
|
| 7192 |
+
},
|
| 7193 |
+
{
|
| 7194 |
+
"epoch": 6.135805659645949,
|
| 7195 |
+
"learning_rate": 1.6185012316867624e-07,
|
| 7196 |
+
"loss": 2.5817,
|
| 7197 |
+
"step": 23740
|
| 7198 |
+
},
|
| 7199 |
+
{
|
| 7200 |
+
"epoch": 6.140974286083473,
|
| 7201 |
+
"learning_rate": 1.6181771035913393e-07,
|
| 7202 |
+
"loss": 2.5873,
|
| 7203 |
+
"step": 23760
|
| 7204 |
+
},
|
| 7205 |
+
{
|
| 7206 |
+
"epoch": 6.146142912520998,
|
| 7207 |
+
"learning_rate": 1.617852975495916e-07,
|
| 7208 |
+
"loss": 2.5957,
|
| 7209 |
+
"step": 23780
|
| 7210 |
+
},
|
| 7211 |
+
{
|
| 7212 |
+
"epoch": 6.151311538958522,
|
| 7213 |
+
"learning_rate": 1.6175288474004925e-07,
|
| 7214 |
+
"loss": 2.613,
|
| 7215 |
+
"step": 23800
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 6.156480165396046,
|
| 7219 |
+
"learning_rate": 1.6172047193050694e-07,
|
| 7220 |
+
"loss": 2.5577,
|
| 7221 |
+
"step": 23820
|
| 7222 |
+
},
|
| 7223 |
+
{
|
| 7224 |
+
"epoch": 6.16164879183357,
|
| 7225 |
+
"learning_rate": 1.6168805912096458e-07,
|
| 7226 |
+
"loss": 2.6101,
|
| 7227 |
+
"step": 23840
|
| 7228 |
+
},
|
| 7229 |
+
{
|
| 7230 |
+
"epoch": 6.166817418271094,
|
| 7231 |
+
"learning_rate": 1.6165564631142227e-07,
|
| 7232 |
+
"loss": 2.5553,
|
| 7233 |
+
"step": 23860
|
| 7234 |
+
},
|
| 7235 |
+
{
|
| 7236 |
+
"epoch": 6.171986044708619,
|
| 7237 |
+
"learning_rate": 1.6162323350187995e-07,
|
| 7238 |
+
"loss": 2.6326,
|
| 7239 |
+
"step": 23880
|
| 7240 |
+
},
|
| 7241 |
+
{
|
| 7242 |
+
"epoch": 6.177154671146143,
|
| 7243 |
+
"learning_rate": 1.615908206923376e-07,
|
| 7244 |
+
"loss": 2.5922,
|
| 7245 |
+
"step": 23900
|
| 7246 |
+
},
|
| 7247 |
+
{
|
| 7248 |
+
"epoch": 6.182323297583667,
|
| 7249 |
+
"learning_rate": 1.6155840788279528e-07,
|
| 7250 |
+
"loss": 2.5913,
|
| 7251 |
+
"step": 23920
|
| 7252 |
+
},
|
| 7253 |
+
{
|
| 7254 |
+
"epoch": 6.187491924021192,
|
| 7255 |
+
"learning_rate": 1.6152599507325294e-07,
|
| 7256 |
+
"loss": 2.6378,
|
| 7257 |
+
"step": 23940
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 6.192660550458716,
|
| 7261 |
+
"learning_rate": 1.614935822637106e-07,
|
| 7262 |
+
"loss": 2.5969,
|
| 7263 |
+
"step": 23960
|
| 7264 |
+
},
|
| 7265 |
+
{
|
| 7266 |
+
"epoch": 6.1978291768962395,
|
| 7267 |
+
"learning_rate": 1.614611694541683e-07,
|
| 7268 |
+
"loss": 2.5971,
|
| 7269 |
+
"step": 23980
|
| 7270 |
+
},
|
| 7271 |
+
{
|
| 7272 |
+
"epoch": 6.202997803333764,
|
| 7273 |
+
"learning_rate": 1.6142875664462593e-07,
|
| 7274 |
+
"loss": 2.616,
|
| 7275 |
+
"step": 24000
|
| 7276 |
+
},
|
| 7277 |
+
{
|
| 7278 |
+
"epoch": 6.208166429771288,
|
| 7279 |
+
"learning_rate": 1.6139634383508361e-07,
|
| 7280 |
+
"loss": 2.6352,
|
| 7281 |
+
"step": 24020
|
| 7282 |
+
},
|
| 7283 |
+
{
|
| 7284 |
+
"epoch": 6.2133350562088125,
|
| 7285 |
+
"learning_rate": 1.613639310255413e-07,
|
| 7286 |
+
"loss": 2.6371,
|
| 7287 |
+
"step": 24040
|
| 7288 |
+
},
|
| 7289 |
+
{
|
| 7290 |
+
"epoch": 6.218503682646337,
|
| 7291 |
+
"learning_rate": 1.6133151821599894e-07,
|
| 7292 |
+
"loss": 2.5946,
|
| 7293 |
+
"step": 24060
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 6.223672309083861,
|
| 7297 |
+
"learning_rate": 1.6129910540645663e-07,
|
| 7298 |
+
"loss": 2.6379,
|
| 7299 |
+
"step": 24080
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 6.228840935521385,
|
| 7303 |
+
"learning_rate": 1.6126669259691431e-07,
|
| 7304 |
+
"loss": 2.6046,
|
| 7305 |
+
"step": 24100
|
| 7306 |
+
},
|
| 7307 |
+
{
|
| 7308 |
+
"epoch": 6.234009561958909,
|
| 7309 |
+
"learning_rate": 1.6123427978737195e-07,
|
| 7310 |
+
"loss": 2.653,
|
| 7311 |
+
"step": 24120
|
| 7312 |
+
},
|
| 7313 |
+
{
|
| 7314 |
+
"epoch": 6.2391781883964335,
|
| 7315 |
+
"learning_rate": 1.6120186697782964e-07,
|
| 7316 |
+
"loss": 2.6409,
|
| 7317 |
+
"step": 24140
|
| 7318 |
+
},
|
| 7319 |
+
{
|
| 7320 |
+
"epoch": 6.244346814833958,
|
| 7321 |
+
"learning_rate": 1.611694541682873e-07,
|
| 7322 |
+
"loss": 2.6077,
|
| 7323 |
+
"step": 24160
|
| 7324 |
+
},
|
| 7325 |
+
{
|
| 7326 |
+
"epoch": 6.249515441271482,
|
| 7327 |
+
"learning_rate": 1.6113704135874496e-07,
|
| 7328 |
+
"loss": 2.5993,
|
| 7329 |
+
"step": 24180
|
| 7330 |
+
},
|
| 7331 |
+
{
|
| 7332 |
+
"epoch": 6.2546840677090065,
|
| 7333 |
+
"learning_rate": 1.6110462854920265e-07,
|
| 7334 |
+
"loss": 2.6326,
|
| 7335 |
+
"step": 24200
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 6.259852694146531,
|
| 7339 |
+
"learning_rate": 1.6107221573966029e-07,
|
| 7340 |
+
"loss": 2.612,
|
| 7341 |
+
"step": 24220
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 6.265021320584054,
|
| 7345 |
+
"learning_rate": 1.6103980293011797e-07,
|
| 7346 |
+
"loss": 2.662,
|
| 7347 |
+
"step": 24240
|
| 7348 |
+
},
|
| 7349 |
+
{
|
| 7350 |
+
"epoch": 6.270189947021579,
|
| 7351 |
+
"learning_rate": 1.6100739012057566e-07,
|
| 7352 |
+
"loss": 2.6377,
|
| 7353 |
+
"step": 24260
|
| 7354 |
+
},
|
| 7355 |
+
{
|
| 7356 |
+
"epoch": 6.275358573459103,
|
| 7357 |
+
"learning_rate": 1.609749773110333e-07,
|
| 7358 |
+
"loss": 2.6113,
|
| 7359 |
+
"step": 24280
|
| 7360 |
+
},
|
| 7361 |
+
{
|
| 7362 |
+
"epoch": 6.280527199896627,
|
| 7363 |
+
"learning_rate": 1.6094256450149099e-07,
|
| 7364 |
+
"loss": 2.6094,
|
| 7365 |
+
"step": 24300
|
| 7366 |
+
},
|
| 7367 |
+
{
|
| 7368 |
+
"epoch": 6.285695826334152,
|
| 7369 |
+
"learning_rate": 1.6091015169194867e-07,
|
| 7370 |
+
"loss": 2.6185,
|
| 7371 |
+
"step": 24320
|
| 7372 |
+
},
|
| 7373 |
+
{
|
| 7374 |
+
"epoch": 6.290864452771676,
|
| 7375 |
+
"learning_rate": 1.608777388824063e-07,
|
| 7376 |
+
"loss": 2.6523,
|
| 7377 |
+
"step": 24340
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 6.2960330792092005,
|
| 7381 |
+
"learning_rate": 1.60845326072864e-07,
|
| 7382 |
+
"loss": 2.6458,
|
| 7383 |
+
"step": 24360
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 6.301201705646724,
|
| 7387 |
+
"learning_rate": 1.6081291326332166e-07,
|
| 7388 |
+
"loss": 2.5826,
|
| 7389 |
+
"step": 24380
|
| 7390 |
+
},
|
| 7391 |
+
{
|
| 7392 |
+
"epoch": 6.306370332084248,
|
| 7393 |
+
"learning_rate": 1.6078050045377932e-07,
|
| 7394 |
+
"loss": 2.6395,
|
| 7395 |
+
"step": 24400
|
| 7396 |
+
},
|
| 7397 |
+
{
|
| 7398 |
+
"epoch": 6.311538958521773,
|
| 7399 |
+
"learning_rate": 1.60748087644237e-07,
|
| 7400 |
+
"loss": 2.6305,
|
| 7401 |
+
"step": 24420
|
| 7402 |
+
},
|
| 7403 |
+
{
|
| 7404 |
+
"epoch": 6.316707584959297,
|
| 7405 |
+
"learning_rate": 1.6071567483469465e-07,
|
| 7406 |
+
"loss": 2.6493,
|
| 7407 |
+
"step": 24440
|
| 7408 |
+
},
|
| 7409 |
+
{
|
| 7410 |
+
"epoch": 6.321876211396821,
|
| 7411 |
+
"learning_rate": 1.6068326202515233e-07,
|
| 7412 |
+
"loss": 2.6198,
|
| 7413 |
+
"step": 24460
|
| 7414 |
+
},
|
| 7415 |
+
{
|
| 7416 |
+
"epoch": 6.327044837834346,
|
| 7417 |
+
"learning_rate": 1.6065084921561002e-07,
|
| 7418 |
+
"loss": 2.5705,
|
| 7419 |
+
"step": 24480
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 6.33221346427187,
|
| 7423 |
+
"learning_rate": 1.6061843640606766e-07,
|
| 7424 |
+
"loss": 2.5797,
|
| 7425 |
+
"step": 24500
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 6.337382090709394,
|
| 7429 |
+
"learning_rate": 1.6058602359652535e-07,
|
| 7430 |
+
"loss": 2.6152,
|
| 7431 |
+
"step": 24520
|
| 7432 |
+
},
|
| 7433 |
+
{
|
| 7434 |
+
"epoch": 6.342550717146918,
|
| 7435 |
+
"learning_rate": 1.60553610786983e-07,
|
| 7436 |
+
"loss": 2.6075,
|
| 7437 |
+
"step": 24540
|
| 7438 |
+
},
|
| 7439 |
+
{
|
| 7440 |
+
"epoch": 6.347719343584442,
|
| 7441 |
+
"learning_rate": 1.6052119797744067e-07,
|
| 7442 |
+
"loss": 2.6446,
|
| 7443 |
+
"step": 24560
|
| 7444 |
+
},
|
| 7445 |
+
{
|
| 7446 |
+
"epoch": 6.352887970021967,
|
| 7447 |
+
"learning_rate": 1.6048878516789836e-07,
|
| 7448 |
+
"loss": 2.6204,
|
| 7449 |
+
"step": 24580
|
| 7450 |
+
},
|
| 7451 |
+
{
|
| 7452 |
+
"epoch": 6.358056596459491,
|
| 7453 |
+
"learning_rate": 1.60456372358356e-07,
|
| 7454 |
+
"loss": 2.6079,
|
| 7455 |
+
"step": 24600
|
| 7456 |
+
},
|
| 7457 |
+
{
|
| 7458 |
+
"epoch": 6.363225222897015,
|
| 7459 |
+
"learning_rate": 1.6042395954881368e-07,
|
| 7460 |
+
"loss": 2.59,
|
| 7461 |
+
"step": 24620
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 6.36839384933454,
|
| 7465 |
+
"learning_rate": 1.6039154673927137e-07,
|
| 7466 |
+
"loss": 2.6417,
|
| 7467 |
+
"step": 24640
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 6.373562475772063,
|
| 7471 |
+
"learning_rate": 1.60359133929729e-07,
|
| 7472 |
+
"loss": 2.6426,
|
| 7473 |
+
"step": 24660
|
| 7474 |
+
},
|
| 7475 |
+
{
|
| 7476 |
+
"epoch": 6.378731102209588,
|
| 7477 |
+
"learning_rate": 1.603267211201867e-07,
|
| 7478 |
+
"loss": 2.6004,
|
| 7479 |
+
"step": 24680
|
| 7480 |
+
},
|
| 7481 |
+
{
|
| 7482 |
+
"epoch": 6.383899728647112,
|
| 7483 |
+
"learning_rate": 1.6029430831064438e-07,
|
| 7484 |
+
"loss": 2.6422,
|
| 7485 |
+
"step": 24700
|
| 7486 |
+
},
|
| 7487 |
+
{
|
| 7488 |
+
"epoch": 6.389068355084636,
|
| 7489 |
+
"learning_rate": 1.6026189550110202e-07,
|
| 7490 |
+
"loss": 2.595,
|
| 7491 |
+
"step": 24720
|
| 7492 |
+
},
|
| 7493 |
+
{
|
| 7494 |
+
"epoch": 6.394236981522161,
|
| 7495 |
+
"learning_rate": 1.602294826915597e-07,
|
| 7496 |
+
"loss": 2.6091,
|
| 7497 |
+
"step": 24740
|
| 7498 |
+
},
|
| 7499 |
+
{
|
| 7500 |
+
"epoch": 6.399405607959685,
|
| 7501 |
+
"learning_rate": 1.6019706988201737e-07,
|
| 7502 |
+
"loss": 2.5978,
|
| 7503 |
+
"step": 24760
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 6.404574234397209,
|
| 7507 |
+
"learning_rate": 1.6016465707247503e-07,
|
| 7508 |
+
"loss": 2.5525,
|
| 7509 |
+
"step": 24780
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 6.409742860834733,
|
| 7513 |
+
"learning_rate": 1.6013224426293272e-07,
|
| 7514 |
+
"loss": 2.6011,
|
| 7515 |
+
"step": 24800
|
| 7516 |
+
},
|
| 7517 |
+
{
|
| 7518 |
+
"epoch": 6.414911487272257,
|
| 7519 |
+
"learning_rate": 1.6009983145339035e-07,
|
| 7520 |
+
"loss": 2.5727,
|
| 7521 |
+
"step": 24820
|
| 7522 |
+
},
|
| 7523 |
+
{
|
| 7524 |
+
"epoch": 6.420080113709782,
|
| 7525 |
+
"learning_rate": 1.6006741864384804e-07,
|
| 7526 |
+
"loss": 2.6214,
|
| 7527 |
+
"step": 24840
|
| 7528 |
+
},
|
| 7529 |
+
{
|
| 7530 |
+
"epoch": 6.425248740147306,
|
| 7531 |
+
"learning_rate": 1.6003500583430573e-07,
|
| 7532 |
+
"loss": 2.6492,
|
| 7533 |
+
"step": 24860
|
| 7534 |
+
},
|
| 7535 |
+
{
|
| 7536 |
+
"epoch": 6.43041736658483,
|
| 7537 |
+
"learning_rate": 1.6000259302476336e-07,
|
| 7538 |
+
"loss": 2.628,
|
| 7539 |
+
"step": 24880
|
| 7540 |
+
},
|
| 7541 |
+
{
|
| 7542 |
+
"epoch": 6.435585993022355,
|
| 7543 |
+
"learning_rate": 1.5997018021522105e-07,
|
| 7544 |
+
"loss": 2.6282,
|
| 7545 |
+
"step": 24900
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 6.440754619459878,
|
| 7549 |
+
"learning_rate": 1.5993776740567874e-07,
|
| 7550 |
+
"loss": 2.6483,
|
| 7551 |
+
"step": 24920
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 6.4459232458974025,
|
| 7555 |
+
"learning_rate": 1.5990535459613638e-07,
|
| 7556 |
+
"loss": 2.6026,
|
| 7557 |
+
"step": 24940
|
| 7558 |
+
},
|
| 7559 |
+
{
|
| 7560 |
+
"epoch": 6.451091872334927,
|
| 7561 |
+
"learning_rate": 1.5987294178659407e-07,
|
| 7562 |
+
"loss": 2.6324,
|
| 7563 |
+
"step": 24960
|
| 7564 |
+
},
|
| 7565 |
+
{
|
| 7566 |
+
"epoch": 6.456260498772451,
|
| 7567 |
+
"learning_rate": 1.5984052897705173e-07,
|
| 7568 |
+
"loss": 2.5601,
|
| 7569 |
+
"step": 24980
|
| 7570 |
+
},
|
| 7571 |
+
{
|
| 7572 |
+
"epoch": 6.461429125209976,
|
| 7573 |
+
"learning_rate": 1.598081161675094e-07,
|
| 7574 |
+
"loss": 2.6437,
|
| 7575 |
+
"step": 25000
|
| 7576 |
+
},
|
| 7577 |
+
{
|
| 7578 |
+
"epoch": 6.4665977516475,
|
| 7579 |
+
"learning_rate": 1.5977570335796708e-07,
|
| 7580 |
+
"loss": 2.6326,
|
| 7581 |
+
"step": 25020
|
| 7582 |
+
},
|
| 7583 |
+
{
|
| 7584 |
+
"epoch": 6.471766378085024,
|
| 7585 |
+
"learning_rate": 1.597432905484247e-07,
|
| 7586 |
+
"loss": 2.6664,
|
| 7587 |
+
"step": 25040
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 6.476935004522548,
|
| 7591 |
+
"learning_rate": 1.597108777388824e-07,
|
| 7592 |
+
"loss": 2.6057,
|
| 7593 |
+
"step": 25060
|
| 7594 |
+
},
|
| 7595 |
+
{
|
| 7596 |
+
"epoch": 6.482103630960072,
|
| 7597 |
+
"learning_rate": 1.596784649293401e-07,
|
| 7598 |
+
"loss": 2.6007,
|
| 7599 |
+
"step": 25080
|
| 7600 |
+
},
|
| 7601 |
+
{
|
| 7602 |
+
"epoch": 6.4872722573975965,
|
| 7603 |
+
"learning_rate": 1.5964605211979772e-07,
|
| 7604 |
+
"loss": 2.5935,
|
| 7605 |
+
"step": 25100
|
| 7606 |
+
},
|
| 7607 |
+
{
|
| 7608 |
+
"epoch": 6.492440883835121,
|
| 7609 |
+
"learning_rate": 1.5961363931025541e-07,
|
| 7610 |
+
"loss": 2.5344,
|
| 7611 |
+
"step": 25120
|
| 7612 |
+
},
|
| 7613 |
+
{
|
| 7614 |
+
"epoch": 6.497609510272645,
|
| 7615 |
+
"learning_rate": 1.5958122650071308e-07,
|
| 7616 |
+
"loss": 2.6249,
|
| 7617 |
+
"step": 25140
|
| 7618 |
+
},
|
| 7619 |
+
{
|
| 7620 |
+
"epoch": 6.50277813671017,
|
| 7621 |
+
"learning_rate": 1.5954881369117074e-07,
|
| 7622 |
+
"loss": 2.5882,
|
| 7623 |
+
"step": 25160
|
| 7624 |
+
},
|
| 7625 |
+
{
|
| 7626 |
+
"epoch": 6.507946763147693,
|
| 7627 |
+
"learning_rate": 1.5951640088162843e-07,
|
| 7628 |
+
"loss": 2.6219,
|
| 7629 |
+
"step": 25180
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 6.513115389585217,
|
| 7633 |
+
"learning_rate": 1.5948398807208606e-07,
|
| 7634 |
+
"loss": 2.5838,
|
| 7635 |
+
"step": 25200
|
| 7636 |
+
},
|
| 7637 |
+
{
|
| 7638 |
+
"epoch": 6.518284016022742,
|
| 7639 |
+
"learning_rate": 1.5945157526254375e-07,
|
| 7640 |
+
"loss": 2.5943,
|
| 7641 |
+
"step": 25220
|
| 7642 |
+
},
|
| 7643 |
+
{
|
| 7644 |
+
"epoch": 6.523452642460266,
|
| 7645 |
+
"learning_rate": 1.5941916245300144e-07,
|
| 7646 |
+
"loss": 2.6468,
|
| 7647 |
+
"step": 25240
|
| 7648 |
+
},
|
| 7649 |
+
{
|
| 7650 |
+
"epoch": 6.5286212688977905,
|
| 7651 |
+
"learning_rate": 1.5938674964345907e-07,
|
| 7652 |
+
"loss": 2.6726,
|
| 7653 |
+
"step": 25260
|
| 7654 |
+
},
|
| 7655 |
+
{
|
| 7656 |
+
"epoch": 6.533789895335315,
|
| 7657 |
+
"learning_rate": 1.5935433683391676e-07,
|
| 7658 |
+
"loss": 2.5732,
|
| 7659 |
+
"step": 25280
|
| 7660 |
+
},
|
| 7661 |
+
{
|
| 7662 |
+
"epoch": 6.538958521772839,
|
| 7663 |
+
"learning_rate": 1.5932192402437442e-07,
|
| 7664 |
+
"loss": 2.5739,
|
| 7665 |
+
"step": 25300
|
| 7666 |
+
},
|
| 7667 |
+
{
|
| 7668 |
+
"epoch": 6.544127148210363,
|
| 7669 |
+
"learning_rate": 1.5928951121483208e-07,
|
| 7670 |
+
"loss": 2.5914,
|
| 7671 |
+
"step": 25320
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 6.549295774647887,
|
| 7675 |
+
"learning_rate": 1.5925709840528977e-07,
|
| 7676 |
+
"loss": 2.6142,
|
| 7677 |
+
"step": 25340
|
| 7678 |
+
},
|
| 7679 |
+
{
|
| 7680 |
+
"epoch": 6.554464401085411,
|
| 7681 |
+
"learning_rate": 1.5922468559574744e-07,
|
| 7682 |
+
"loss": 2.6145,
|
| 7683 |
+
"step": 25360
|
| 7684 |
+
},
|
| 7685 |
+
{
|
| 7686 |
+
"epoch": 6.559633027522936,
|
| 7687 |
+
"learning_rate": 1.591922727862051e-07,
|
| 7688 |
+
"loss": 2.6449,
|
| 7689 |
+
"step": 25380
|
| 7690 |
+
},
|
| 7691 |
+
{
|
| 7692 |
+
"epoch": 6.56480165396046,
|
| 7693 |
+
"learning_rate": 1.5915985997666279e-07,
|
| 7694 |
+
"loss": 2.586,
|
| 7695 |
+
"step": 25400
|
| 7696 |
+
},
|
| 7697 |
+
{
|
| 7698 |
+
"epoch": 6.5699702803979845,
|
| 7699 |
+
"learning_rate": 1.5912744716712042e-07,
|
| 7700 |
+
"loss": 2.6136,
|
| 7701 |
+
"step": 25420
|
| 7702 |
+
},
|
| 7703 |
+
{
|
| 7704 |
+
"epoch": 6.575138906835509,
|
| 7705 |
+
"learning_rate": 1.590950343575781e-07,
|
| 7706 |
+
"loss": 2.6556,
|
| 7707 |
+
"step": 25440
|
| 7708 |
+
},
|
| 7709 |
+
{
|
| 7710 |
+
"epoch": 6.580307533273032,
|
| 7711 |
+
"learning_rate": 1.5906262154803577e-07,
|
| 7712 |
+
"loss": 2.6557,
|
| 7713 |
+
"step": 25460
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 6.585476159710557,
|
| 7717 |
+
"learning_rate": 1.5903020873849343e-07,
|
| 7718 |
+
"loss": 2.5518,
|
| 7719 |
+
"step": 25480
|
| 7720 |
+
},
|
| 7721 |
+
{
|
| 7722 |
+
"epoch": 6.590644786148081,
|
| 7723 |
+
"learning_rate": 1.5899779592895112e-07,
|
| 7724 |
+
"loss": 2.6054,
|
| 7725 |
+
"step": 25500
|
| 7726 |
+
},
|
| 7727 |
+
{
|
| 7728 |
+
"epoch": 6.595813412585605,
|
| 7729 |
+
"learning_rate": 1.5896538311940878e-07,
|
| 7730 |
+
"loss": 2.5766,
|
| 7731 |
+
"step": 25520
|
| 7732 |
+
},
|
| 7733 |
+
{
|
| 7734 |
+
"epoch": 6.60098203902313,
|
| 7735 |
+
"learning_rate": 1.5893297030986644e-07,
|
| 7736 |
+
"loss": 2.5573,
|
| 7737 |
+
"step": 25540
|
| 7738 |
+
},
|
| 7739 |
+
{
|
| 7740 |
+
"epoch": 6.606150665460654,
|
| 7741 |
+
"learning_rate": 1.5890055750032413e-07,
|
| 7742 |
+
"loss": 2.6429,
|
| 7743 |
+
"step": 25560
|
| 7744 |
+
},
|
| 7745 |
+
{
|
| 7746 |
+
"epoch": 6.6113192918981785,
|
| 7747 |
+
"learning_rate": 1.588681446907818e-07,
|
| 7748 |
+
"loss": 2.6795,
|
| 7749 |
+
"step": 25580
|
| 7750 |
+
},
|
| 7751 |
+
{
|
| 7752 |
+
"epoch": 6.616487918335702,
|
| 7753 |
+
"learning_rate": 1.5883573188123946e-07,
|
| 7754 |
+
"loss": 2.6573,
|
| 7755 |
+
"step": 25600
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 6.621656544773226,
|
| 7759 |
+
"learning_rate": 1.5880331907169712e-07,
|
| 7760 |
+
"loss": 2.5762,
|
| 7761 |
+
"step": 25620
|
| 7762 |
+
},
|
| 7763 |
+
{
|
| 7764 |
+
"epoch": 6.626825171210751,
|
| 7765 |
+
"learning_rate": 1.5877090626215478e-07,
|
| 7766 |
+
"loss": 2.6336,
|
| 7767 |
+
"step": 25640
|
| 7768 |
+
},
|
| 7769 |
+
{
|
| 7770 |
+
"epoch": 6.631993797648275,
|
| 7771 |
+
"learning_rate": 1.5873849345261247e-07,
|
| 7772 |
+
"loss": 2.5999,
|
| 7773 |
+
"step": 25660
|
| 7774 |
+
},
|
| 7775 |
+
{
|
| 7776 |
+
"epoch": 6.637162424085799,
|
| 7777 |
+
"learning_rate": 1.5870608064307013e-07,
|
| 7778 |
+
"loss": 2.6625,
|
| 7779 |
+
"step": 25680
|
| 7780 |
+
},
|
| 7781 |
+
{
|
| 7782 |
+
"epoch": 6.642331050523324,
|
| 7783 |
+
"learning_rate": 1.586736678335278e-07,
|
| 7784 |
+
"loss": 2.5277,
|
| 7785 |
+
"step": 25700
|
| 7786 |
+
},
|
| 7787 |
+
{
|
| 7788 |
+
"epoch": 6.647499676960848,
|
| 7789 |
+
"learning_rate": 1.5864125502398548e-07,
|
| 7790 |
+
"loss": 2.5688,
|
| 7791 |
+
"step": 25720
|
| 7792 |
+
},
|
| 7793 |
+
{
|
| 7794 |
+
"epoch": 6.652668303398372,
|
| 7795 |
+
"learning_rate": 1.5860884221444314e-07,
|
| 7796 |
+
"loss": 2.5813,
|
| 7797 |
+
"step": 25740
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 6.657836929835896,
|
| 7801 |
+
"learning_rate": 1.585764294049008e-07,
|
| 7802 |
+
"loss": 2.622,
|
| 7803 |
+
"step": 25760
|
| 7804 |
+
},
|
| 7805 |
+
{
|
| 7806 |
+
"epoch": 6.66300555627342,
|
| 7807 |
+
"learning_rate": 1.5854401659535847e-07,
|
| 7808 |
+
"loss": 2.5956,
|
| 7809 |
+
"step": 25780
|
| 7810 |
+
},
|
| 7811 |
+
{
|
| 7812 |
+
"epoch": 6.668174182710945,
|
| 7813 |
+
"learning_rate": 1.5851160378581613e-07,
|
| 7814 |
+
"loss": 2.5834,
|
| 7815 |
+
"step": 25800
|
| 7816 |
+
},
|
| 7817 |
+
{
|
| 7818 |
+
"epoch": 6.673342809148469,
|
| 7819 |
+
"learning_rate": 1.5847919097627382e-07,
|
| 7820 |
+
"loss": 2.5737,
|
| 7821 |
+
"step": 25820
|
| 7822 |
+
},
|
| 7823 |
+
{
|
| 7824 |
+
"epoch": 6.678511435585993,
|
| 7825 |
+
"learning_rate": 1.5844677816673148e-07,
|
| 7826 |
+
"loss": 2.5815,
|
| 7827 |
+
"step": 25840
|
| 7828 |
+
},
|
| 7829 |
+
{
|
| 7830 |
+
"epoch": 6.683680062023518,
|
| 7831 |
+
"learning_rate": 1.5841436535718914e-07,
|
| 7832 |
+
"loss": 2.5868,
|
| 7833 |
+
"step": 25860
|
| 7834 |
+
},
|
| 7835 |
+
{
|
| 7836 |
+
"epoch": 6.688848688461041,
|
| 7837 |
+
"learning_rate": 1.5838195254764683e-07,
|
| 7838 |
+
"loss": 2.6376,
|
| 7839 |
+
"step": 25880
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 6.6940173148985656,
|
| 7843 |
+
"learning_rate": 1.583495397381045e-07,
|
| 7844 |
+
"loss": 2.5857,
|
| 7845 |
+
"step": 25900
|
| 7846 |
+
},
|
| 7847 |
+
{
|
| 7848 |
+
"epoch": 6.69918594133609,
|
| 7849 |
+
"learning_rate": 1.5831712692856215e-07,
|
| 7850 |
+
"loss": 2.6078,
|
| 7851 |
+
"step": 25920
|
| 7852 |
+
},
|
| 7853 |
+
{
|
| 7854 |
+
"epoch": 6.704354567773614,
|
| 7855 |
+
"learning_rate": 1.5828471411901981e-07,
|
| 7856 |
+
"loss": 2.6262,
|
| 7857 |
+
"step": 25940
|
| 7858 |
+
},
|
| 7859 |
+
{
|
| 7860 |
+
"epoch": 6.709523194211139,
|
| 7861 |
+
"learning_rate": 1.582523013094775e-07,
|
| 7862 |
+
"loss": 2.5676,
|
| 7863 |
+
"step": 25960
|
| 7864 |
+
},
|
| 7865 |
+
{
|
| 7866 |
+
"epoch": 6.714691820648663,
|
| 7867 |
+
"learning_rate": 1.5821988849993516e-07,
|
| 7868 |
+
"loss": 2.6036,
|
| 7869 |
+
"step": 25980
|
| 7870 |
+
},
|
| 7871 |
+
{
|
| 7872 |
+
"epoch": 6.7198604470861865,
|
| 7873 |
+
"learning_rate": 1.5818747569039283e-07,
|
| 7874 |
+
"loss": 2.5575,
|
| 7875 |
+
"step": 26000
|
| 7876 |
+
},
|
| 7877 |
+
{
|
| 7878 |
+
"epoch": 6.725029073523711,
|
| 7879 |
+
"learning_rate": 1.581550628808505e-07,
|
| 7880 |
+
"loss": 2.5708,
|
| 7881 |
+
"step": 26020
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 6.730197699961235,
|
| 7885 |
+
"learning_rate": 1.5812265007130818e-07,
|
| 7886 |
+
"loss": 2.6489,
|
| 7887 |
+
"step": 26040
|
| 7888 |
+
},
|
| 7889 |
+
{
|
| 7890 |
+
"epoch": 6.7353663263987595,
|
| 7891 |
+
"learning_rate": 1.5809023726176584e-07,
|
| 7892 |
+
"loss": 2.6338,
|
| 7893 |
+
"step": 26060
|
| 7894 |
+
},
|
| 7895 |
+
{
|
| 7896 |
+
"epoch": 6.740534952836284,
|
| 7897 |
+
"learning_rate": 1.580578244522235e-07,
|
| 7898 |
+
"loss": 2.6506,
|
| 7899 |
+
"step": 26080
|
| 7900 |
+
},
|
| 7901 |
+
{
|
| 7902 |
+
"epoch": 6.745703579273808,
|
| 7903 |
+
"learning_rate": 1.5802541164268116e-07,
|
| 7904 |
+
"loss": 2.541,
|
| 7905 |
+
"step": 26100
|
| 7906 |
+
},
|
| 7907 |
+
{
|
| 7908 |
+
"epoch": 6.750872205711332,
|
| 7909 |
+
"learning_rate": 1.5799299883313885e-07,
|
| 7910 |
+
"loss": 2.5367,
|
| 7911 |
+
"step": 26120
|
| 7912 |
+
},
|
| 7913 |
+
{
|
| 7914 |
+
"epoch": 6.756040832148856,
|
| 7915 |
+
"learning_rate": 1.579605860235965e-07,
|
| 7916 |
+
"loss": 2.5891,
|
| 7917 |
+
"step": 26140
|
| 7918 |
+
},
|
| 7919 |
+
{
|
| 7920 |
+
"epoch": 6.7612094585863804,
|
| 7921 |
+
"learning_rate": 1.5792817321405417e-07,
|
| 7922 |
+
"loss": 2.6226,
|
| 7923 |
+
"step": 26160
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 6.766378085023905,
|
| 7927 |
+
"learning_rate": 1.5789576040451186e-07,
|
| 7928 |
+
"loss": 2.6313,
|
| 7929 |
+
"step": 26180
|
| 7930 |
+
},
|
| 7931 |
+
{
|
| 7932 |
+
"epoch": 6.771546711461429,
|
| 7933 |
+
"learning_rate": 1.5786334759496952e-07,
|
| 7934 |
+
"loss": 2.5891,
|
| 7935 |
+
"step": 26200
|
| 7936 |
+
},
|
| 7937 |
+
{
|
| 7938 |
+
"epoch": 6.7767153378989535,
|
| 7939 |
+
"learning_rate": 1.5783093478542719e-07,
|
| 7940 |
+
"loss": 2.5806,
|
| 7941 |
+
"step": 26220
|
| 7942 |
+
},
|
| 7943 |
+
{
|
| 7944 |
+
"epoch": 6.781883964336478,
|
| 7945 |
+
"learning_rate": 1.5779852197588485e-07,
|
| 7946 |
+
"loss": 2.5518,
|
| 7947 |
+
"step": 26240
|
| 7948 |
+
},
|
| 7949 |
+
{
|
| 7950 |
+
"epoch": 6.787052590774001,
|
| 7951 |
+
"learning_rate": 1.5776610916634254e-07,
|
| 7952 |
+
"loss": 2.6104,
|
| 7953 |
+
"step": 26260
|
| 7954 |
+
},
|
| 7955 |
+
{
|
| 7956 |
+
"epoch": 6.792221217211526,
|
| 7957 |
+
"learning_rate": 1.577336963568002e-07,
|
| 7958 |
+
"loss": 2.6052,
|
| 7959 |
+
"step": 26280
|
| 7960 |
+
},
|
| 7961 |
+
{
|
| 7962 |
+
"epoch": 6.79738984364905,
|
| 7963 |
+
"learning_rate": 1.5770128354725786e-07,
|
| 7964 |
+
"loss": 2.5845,
|
| 7965 |
+
"step": 26300
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 6.802558470086574,
|
| 7969 |
+
"learning_rate": 1.5766887073771552e-07,
|
| 7970 |
+
"loss": 2.6096,
|
| 7971 |
+
"step": 26320
|
| 7972 |
+
},
|
| 7973 |
+
{
|
| 7974 |
+
"epoch": 6.807727096524099,
|
| 7975 |
+
"learning_rate": 1.576364579281732e-07,
|
| 7976 |
+
"loss": 2.5916,
|
| 7977 |
+
"step": 26340
|
| 7978 |
+
},
|
| 7979 |
+
{
|
| 7980 |
+
"epoch": 6.812895722961623,
|
| 7981 |
+
"learning_rate": 1.5760404511863087e-07,
|
| 7982 |
+
"loss": 2.584,
|
| 7983 |
+
"step": 26360
|
| 7984 |
+
},
|
| 7985 |
+
{
|
| 7986 |
+
"epoch": 6.8180643493991475,
|
| 7987 |
+
"learning_rate": 1.5757163230908853e-07,
|
| 7988 |
+
"loss": 2.6194,
|
| 7989 |
+
"step": 26380
|
| 7990 |
+
},
|
| 7991 |
+
{
|
| 7992 |
+
"epoch": 6.823232975836671,
|
| 7993 |
+
"learning_rate": 1.575392194995462e-07,
|
| 7994 |
+
"loss": 2.6581,
|
| 7995 |
+
"step": 26400
|
| 7996 |
+
},
|
| 7997 |
+
{
|
| 7998 |
+
"epoch": 6.828401602274195,
|
| 7999 |
+
"learning_rate": 1.5750680669000388e-07,
|
| 8000 |
+
"loss": 2.6636,
|
| 8001 |
+
"step": 26420
|
| 8002 |
+
},
|
| 8003 |
+
{
|
| 8004 |
+
"epoch": 6.83357022871172,
|
| 8005 |
+
"learning_rate": 1.5747439388046155e-07,
|
| 8006 |
+
"loss": 2.6135,
|
| 8007 |
+
"step": 26440
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 6.838738855149244,
|
| 8011 |
+
"learning_rate": 1.574419810709192e-07,
|
| 8012 |
+
"loss": 2.609,
|
| 8013 |
+
"step": 26460
|
| 8014 |
+
},
|
| 8015 |
+
{
|
| 8016 |
+
"epoch": 6.843907481586768,
|
| 8017 |
+
"learning_rate": 1.574095682613769e-07,
|
| 8018 |
+
"loss": 2.6168,
|
| 8019 |
+
"step": 26480
|
| 8020 |
+
},
|
| 8021 |
+
{
|
| 8022 |
+
"epoch": 6.849076108024293,
|
| 8023 |
+
"learning_rate": 1.5737715545183456e-07,
|
| 8024 |
+
"loss": 2.5408,
|
| 8025 |
+
"step": 26500
|
| 8026 |
+
},
|
| 8027 |
+
{
|
| 8028 |
+
"epoch": 6.854244734461817,
|
| 8029 |
+
"learning_rate": 1.5734474264229222e-07,
|
| 8030 |
+
"loss": 2.6252,
|
| 8031 |
+
"step": 26520
|
| 8032 |
+
},
|
| 8033 |
+
{
|
| 8034 |
+
"epoch": 6.859413360899341,
|
| 8035 |
+
"learning_rate": 1.5731232983274988e-07,
|
| 8036 |
+
"loss": 2.5935,
|
| 8037 |
+
"step": 26540
|
| 8038 |
+
},
|
| 8039 |
+
{
|
| 8040 |
+
"epoch": 6.864581987336865,
|
| 8041 |
+
"learning_rate": 1.5727991702320757e-07,
|
| 8042 |
+
"loss": 2.5767,
|
| 8043 |
+
"step": 26560
|
| 8044 |
+
},
|
| 8045 |
+
{
|
| 8046 |
+
"epoch": 6.869750613774389,
|
| 8047 |
+
"learning_rate": 1.5724750421366523e-07,
|
| 8048 |
+
"loss": 2.6351,
|
| 8049 |
+
"step": 26580
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 6.874919240211914,
|
| 8053 |
+
"learning_rate": 1.572150914041229e-07,
|
| 8054 |
+
"loss": 2.5793,
|
| 8055 |
+
"step": 26600
|
| 8056 |
+
},
|
| 8057 |
+
{
|
| 8058 |
+
"epoch": 6.880087866649438,
|
| 8059 |
+
"learning_rate": 1.5718267859458056e-07,
|
| 8060 |
+
"loss": 2.6026,
|
| 8061 |
+
"step": 26620
|
| 8062 |
+
},
|
| 8063 |
+
{
|
| 8064 |
+
"epoch": 6.885256493086962,
|
| 8065 |
+
"learning_rate": 1.5715026578503824e-07,
|
| 8066 |
+
"loss": 2.5796,
|
| 8067 |
+
"step": 26640
|
| 8068 |
+
},
|
| 8069 |
+
{
|
| 8070 |
+
"epoch": 6.890425119524487,
|
| 8071 |
+
"learning_rate": 1.571178529754959e-07,
|
| 8072 |
+
"loss": 2.6488,
|
| 8073 |
+
"step": 26660
|
| 8074 |
+
},
|
| 8075 |
+
{
|
| 8076 |
+
"epoch": 6.89559374596201,
|
| 8077 |
+
"learning_rate": 1.5708544016595357e-07,
|
| 8078 |
+
"loss": 2.6461,
|
| 8079 |
+
"step": 26680
|
| 8080 |
+
},
|
| 8081 |
+
{
|
| 8082 |
+
"epoch": 6.900762372399535,
|
| 8083 |
+
"learning_rate": 1.5705302735641123e-07,
|
| 8084 |
+
"loss": 2.5738,
|
| 8085 |
+
"step": 26700
|
| 8086 |
+
},
|
| 8087 |
+
{
|
| 8088 |
+
"epoch": 6.905930998837059,
|
| 8089 |
+
"learning_rate": 1.5702061454686892e-07,
|
| 8090 |
+
"loss": 2.5407,
|
| 8091 |
+
"step": 26720
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 6.911099625274583,
|
| 8095 |
+
"learning_rate": 1.5698820173732658e-07,
|
| 8096 |
+
"loss": 2.6304,
|
| 8097 |
+
"step": 26740
|
| 8098 |
+
},
|
| 8099 |
+
{
|
| 8100 |
+
"epoch": 6.916268251712108,
|
| 8101 |
+
"learning_rate": 1.5695578892778424e-07,
|
| 8102 |
+
"loss": 2.576,
|
| 8103 |
+
"step": 26760
|
| 8104 |
+
},
|
| 8105 |
+
{
|
| 8106 |
+
"epoch": 6.921436878149632,
|
| 8107 |
+
"learning_rate": 1.5692337611824193e-07,
|
| 8108 |
+
"loss": 2.5784,
|
| 8109 |
+
"step": 26780
|
| 8110 |
+
},
|
| 8111 |
+
{
|
| 8112 |
+
"epoch": 6.926605504587156,
|
| 8113 |
+
"learning_rate": 1.568909633086996e-07,
|
| 8114 |
+
"loss": 2.6124,
|
| 8115 |
+
"step": 26800
|
| 8116 |
+
},
|
| 8117 |
+
{
|
| 8118 |
+
"epoch": 6.93177413102468,
|
| 8119 |
+
"learning_rate": 1.5685855049915725e-07,
|
| 8120 |
+
"loss": 2.5992,
|
| 8121 |
+
"step": 26820
|
| 8122 |
+
},
|
| 8123 |
+
{
|
| 8124 |
+
"epoch": 6.936942757462204,
|
| 8125 |
+
"learning_rate": 1.5682613768961492e-07,
|
| 8126 |
+
"loss": 2.5961,
|
| 8127 |
+
"step": 26840
|
| 8128 |
+
},
|
| 8129 |
+
{
|
| 8130 |
+
"epoch": 6.942111383899729,
|
| 8131 |
+
"learning_rate": 1.567937248800726e-07,
|
| 8132 |
+
"loss": 2.5989,
|
| 8133 |
+
"step": 26860
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 6.947280010337253,
|
| 8137 |
+
"learning_rate": 1.5676131207053027e-07,
|
| 8138 |
+
"loss": 2.6514,
|
| 8139 |
+
"step": 26880
|
| 8140 |
+
},
|
| 8141 |
+
{
|
| 8142 |
+
"epoch": 6.952448636774777,
|
| 8143 |
+
"learning_rate": 1.5672889926098793e-07,
|
| 8144 |
+
"loss": 2.5921,
|
| 8145 |
+
"step": 26900
|
| 8146 |
+
},
|
| 8147 |
+
{
|
| 8148 |
+
"epoch": 6.957617263212302,
|
| 8149 |
+
"learning_rate": 1.566964864514456e-07,
|
| 8150 |
+
"loss": 2.5907,
|
| 8151 |
+
"step": 26920
|
| 8152 |
+
},
|
| 8153 |
+
{
|
| 8154 |
+
"epoch": 6.962785889649826,
|
| 8155 |
+
"learning_rate": 1.5666407364190328e-07,
|
| 8156 |
+
"loss": 2.5183,
|
| 8157 |
+
"step": 26940
|
| 8158 |
+
},
|
| 8159 |
+
{
|
| 8160 |
+
"epoch": 6.9679545160873495,
|
| 8161 |
+
"learning_rate": 1.5663166083236094e-07,
|
| 8162 |
+
"loss": 2.6151,
|
| 8163 |
+
"step": 26960
|
| 8164 |
+
},
|
| 8165 |
+
{
|
| 8166 |
+
"epoch": 6.973123142524874,
|
| 8167 |
+
"learning_rate": 1.565992480228186e-07,
|
| 8168 |
+
"loss": 2.5454,
|
| 8169 |
+
"step": 26980
|
| 8170 |
+
},
|
| 8171 |
+
{
|
| 8172 |
+
"epoch": 6.978291768962398,
|
| 8173 |
+
"learning_rate": 1.5656683521327626e-07,
|
| 8174 |
+
"loss": 2.5349,
|
| 8175 |
+
"step": 27000
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 6.983460395399923,
|
| 8179 |
+
"learning_rate": 1.5653442240373395e-07,
|
| 8180 |
+
"loss": 2.6422,
|
| 8181 |
+
"step": 27020
|
| 8182 |
+
},
|
| 8183 |
+
{
|
| 8184 |
+
"epoch": 6.988629021837447,
|
| 8185 |
+
"learning_rate": 1.5650200959419161e-07,
|
| 8186 |
+
"loss": 2.6273,
|
| 8187 |
+
"step": 27040
|
| 8188 |
+
},
|
| 8189 |
+
{
|
| 8190 |
+
"epoch": 6.993797648274971,
|
| 8191 |
+
"learning_rate": 1.5646959678464928e-07,
|
| 8192 |
+
"loss": 2.5851,
|
| 8193 |
+
"step": 27060
|
| 8194 |
+
},
|
| 8195 |
+
{
|
| 8196 |
+
"epoch": 6.998966274712495,
|
| 8197 |
+
"learning_rate": 1.5643718397510696e-07,
|
| 8198 |
+
"loss": 2.5716,
|
| 8199 |
+
"step": 27080
|
| 8200 |
+
},
|
| 8201 |
+
{
|
| 8202 |
+
"epoch": 7.0,
|
| 8203 |
+
"eval_bleu": 5.2201,
|
| 8204 |
+
"eval_gen_len": 46.1405,
|
| 8205 |
+
"eval_loss": 2.561117172241211,
|
| 8206 |
+
"eval_runtime": 958.6568,
|
| 8207 |
+
"eval_samples_per_second": 1.796,
|
| 8208 |
+
"eval_steps_per_second": 0.898,
|
| 8209 |
+
"step": 27084
|
| 8210 |
}
|
| 8211 |
],
|
| 8212 |
"logging_steps": 20,
|
|
|
|
| 8226 |
"attributes": {}
|
| 8227 |
}
|
| 8228 |
},
|
| 8229 |
+
"total_flos": 1.7781178479766733e+17,
|
| 8230 |
"train_batch_size": 2,
|
| 8231 |
"trial_name": null,
|
| 8232 |
"trial_params": null
|
last-checkpoint/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d11a3b4036b5ce40442d47a4051217581bc97fdd9805ef76304488e63693998
|
| 3 |
+
size 5752
|