TrLOX commited on
Commit
383f842
1 Parent(s): 81a5c85
all_results.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 20.0,
3
+ "init_mem_cpu_alloc_delta": 2547835,
4
+ "init_mem_cpu_peaked_delta": 38772,
5
+ "train_mem_cpu_alloc_delta": 1063085,
6
+ "train_mem_cpu_peaked_delta": 209273878,
7
+ "train_runtime": 264850.0877,
8
+ "train_samples": 438452,
9
+ "train_samples_per_second": 1.035
10
+ }
config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "gpt2-medium",
3
+ "activation_function": "gelu_new",
4
+ "architectures": [
5
+ "GPT2LMHeadModel"
6
+ ],
7
+ "attn_pdrop": 0.1,
8
+ "bos_token_id": 50256,
9
+ "embd_pdrop": 0.1,
10
+ "eos_token_id": 50256,
11
+ "gradient_checkpointing": false,
12
+ "initializer_range": 0.02,
13
+ "layer_norm_epsilon": 1e-05,
14
+ "model_type": "gpt2",
15
+ "n_ctx": 1024,
16
+ "n_embd": 1024,
17
+ "n_head": 16,
18
+ "n_inner": null,
19
+ "n_layer": 24,
20
+ "n_positions": 1024,
21
+ "n_special": 0,
22
+ "predict_special_tokens": true,
23
+ "resid_pdrop": 0.1,
24
+ "summary_activation": null,
25
+ "summary_first_dropout": 0.1,
26
+ "summary_proj_to_labels": true,
27
+ "summary_type": "cls_index",
28
+ "summary_use_proj": true,
29
+ "task_specific_params": {
30
+ "text-generation": {
31
+ "do_sample": true,
32
+ "max_length": 50
33
+ }
34
+ },
35
+ "transformers_version": "4.5.0.dev0",
36
+ "use_cache": true,
37
+ "vocab_size": 50257
38
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:977881c677372723a20824e85310f203b002b33d328453ab16c3af193d499da1
3
+ size 1650423212
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
1
+ {"bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "unk_token": "<|endoftext|>"}
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
1
+ {"unk_token": "<|endoftext|>", "bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "add_prefix_space": false, "model_max_length": 1024, "special_tokens_map_file": null, "name_or_path": "gpt2-medium"}
train_results.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 20.0,
3
+ "init_mem_cpu_alloc_delta": 2547835,
4
+ "init_mem_cpu_peaked_delta": 38772,
5
+ "train_mem_cpu_alloc_delta": 1063085,
6
+ "train_mem_cpu_peaked_delta": 209273878,
7
+ "train_runtime": 264850.0877,
8
+ "train_samples": 438452,
9
+ "train_samples_per_second": 1.035
10
+ }
trainer_state.json ADDED
@@ -0,0 +1,3311 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 20.0,
5
+ "global_step": 274040,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.04,
12
+ "learning_rate": 4.9908772441979274e-05,
13
+ "loss": 2.048,
14
+ "step": 500
15
+ },
16
+ {
17
+ "epoch": 0.07,
18
+ "learning_rate": 4.981754488395855e-05,
19
+ "loss": 2.048,
20
+ "step": 1000
21
+ },
22
+ {
23
+ "epoch": 0.11,
24
+ "learning_rate": 4.972631732593782e-05,
25
+ "loss": 2.048,
26
+ "step": 1500
27
+ },
28
+ {
29
+ "epoch": 0.15,
30
+ "learning_rate": 4.9635089767917094e-05,
31
+ "loss": 2.048,
32
+ "step": 2000
33
+ },
34
+ {
35
+ "epoch": 0.18,
36
+ "learning_rate": 4.954386220989637e-05,
37
+ "loss": 2.048,
38
+ "step": 2500
39
+ },
40
+ {
41
+ "epoch": 0.22,
42
+ "learning_rate": 4.945263465187564e-05,
43
+ "loss": 2.048,
44
+ "step": 3000
45
+ },
46
+ {
47
+ "epoch": 0.26,
48
+ "learning_rate": 4.9361407093854914e-05,
49
+ "loss": 2.048,
50
+ "step": 3500
51
+ },
52
+ {
53
+ "epoch": 0.29,
54
+ "learning_rate": 4.9270179535834185e-05,
55
+ "loss": 2.048,
56
+ "step": 4000
57
+ },
58
+ {
59
+ "epoch": 0.33,
60
+ "learning_rate": 4.9178951977813456e-05,
61
+ "loss": 2.048,
62
+ "step": 4500
63
+ },
64
+ {
65
+ "epoch": 0.36,
66
+ "learning_rate": 4.9087724419792734e-05,
67
+ "loss": 2.048,
68
+ "step": 5000
69
+ },
70
+ {
71
+ "epoch": 0.4,
72
+ "learning_rate": 4.8996496861772005e-05,
73
+ "loss": 2.048,
74
+ "step": 5500
75
+ },
76
+ {
77
+ "epoch": 0.44,
78
+ "learning_rate": 4.8905269303751276e-05,
79
+ "loss": 2.048,
80
+ "step": 6000
81
+ },
82
+ {
83
+ "epoch": 0.47,
84
+ "learning_rate": 4.8814041745730554e-05,
85
+ "loss": 2.048,
86
+ "step": 6500
87
+ },
88
+ {
89
+ "epoch": 0.51,
90
+ "learning_rate": 4.8722814187709825e-05,
91
+ "loss": 2.048,
92
+ "step": 7000
93
+ },
94
+ {
95
+ "epoch": 0.55,
96
+ "learning_rate": 4.8631586629689096e-05,
97
+ "loss": 2.048,
98
+ "step": 7500
99
+ },
100
+ {
101
+ "epoch": 0.58,
102
+ "learning_rate": 4.8540359071668374e-05,
103
+ "loss": 2.048,
104
+ "step": 8000
105
+ },
106
+ {
107
+ "epoch": 0.62,
108
+ "learning_rate": 4.8449131513647645e-05,
109
+ "loss": 2.048,
110
+ "step": 8500
111
+ },
112
+ {
113
+ "epoch": 0.66,
114
+ "learning_rate": 4.8357903955626916e-05,
115
+ "loss": 2.048,
116
+ "step": 9000
117
+ },
118
+ {
119
+ "epoch": 0.69,
120
+ "learning_rate": 4.8266676397606194e-05,
121
+ "loss": 2.048,
122
+ "step": 9500
123
+ },
124
+ {
125
+ "epoch": 0.73,
126
+ "learning_rate": 4.8175448839585465e-05,
127
+ "loss": 2.048,
128
+ "step": 10000
129
+ },
130
+ {
131
+ "epoch": 0.77,
132
+ "learning_rate": 4.808422128156474e-05,
133
+ "loss": 2.048,
134
+ "step": 10500
135
+ },
136
+ {
137
+ "epoch": 0.8,
138
+ "learning_rate": 4.7992993723544014e-05,
139
+ "loss": 2.048,
140
+ "step": 11000
141
+ },
142
+ {
143
+ "epoch": 0.84,
144
+ "learning_rate": 4.7901766165523285e-05,
145
+ "loss": 2.048,
146
+ "step": 11500
147
+ },
148
+ {
149
+ "epoch": 0.88,
150
+ "learning_rate": 4.7810538607502556e-05,
151
+ "loss": 2.048,
152
+ "step": 12000
153
+ },
154
+ {
155
+ "epoch": 0.91,
156
+ "learning_rate": 4.771931104948183e-05,
157
+ "loss": 2.048,
158
+ "step": 12500
159
+ },
160
+ {
161
+ "epoch": 0.95,
162
+ "learning_rate": 4.76280834914611e-05,
163
+ "loss": 2.048,
164
+ "step": 13000
165
+ },
166
+ {
167
+ "epoch": 0.99,
168
+ "learning_rate": 4.7536855933440376e-05,
169
+ "loss": 2.048,
170
+ "step": 13500
171
+ },
172
+ {
173
+ "epoch": 1.02,
174
+ "learning_rate": 4.744562837541965e-05,
175
+ "loss": 2.048,
176
+ "step": 14000
177
+ },
178
+ {
179
+ "epoch": 1.06,
180
+ "learning_rate": 4.735440081739892e-05,
181
+ "loss": 2.048,
182
+ "step": 14500
183
+ },
184
+ {
185
+ "epoch": 1.09,
186
+ "learning_rate": 4.7263173259378197e-05,
187
+ "loss": 2.048,
188
+ "step": 15000
189
+ },
190
+ {
191
+ "epoch": 1.13,
192
+ "learning_rate": 4.717194570135747e-05,
193
+ "loss": 2.048,
194
+ "step": 15500
195
+ },
196
+ {
197
+ "epoch": 1.17,
198
+ "learning_rate": 4.7080718143336746e-05,
199
+ "loss": 2.048,
200
+ "step": 16000
201
+ },
202
+ {
203
+ "epoch": 1.2,
204
+ "learning_rate": 4.698949058531602e-05,
205
+ "loss": 2.048,
206
+ "step": 16500
207
+ },
208
+ {
209
+ "epoch": 1.24,
210
+ "learning_rate": 4.689826302729529e-05,
211
+ "loss": 2.048,
212
+ "step": 17000
213
+ },
214
+ {
215
+ "epoch": 1.28,
216
+ "learning_rate": 4.6807035469274566e-05,
217
+ "loss": 2.048,
218
+ "step": 17500
219
+ },
220
+ {
221
+ "epoch": 1.31,
222
+ "learning_rate": 4.671580791125384e-05,
223
+ "loss": 2.048,
224
+ "step": 18000
225
+ },
226
+ {
227
+ "epoch": 1.35,
228
+ "learning_rate": 4.662458035323311e-05,
229
+ "loss": 2.048,
230
+ "step": 18500
231
+ },
232
+ {
233
+ "epoch": 1.39,
234
+ "learning_rate": 4.653335279521238e-05,
235
+ "loss": 2.048,
236
+ "step": 19000
237
+ },
238
+ {
239
+ "epoch": 1.42,
240
+ "learning_rate": 4.644212523719165e-05,
241
+ "loss": 2.048,
242
+ "step": 19500
243
+ },
244
+ {
245
+ "epoch": 1.46,
246
+ "learning_rate": 4.635089767917092e-05,
247
+ "loss": 2.048,
248
+ "step": 20000
249
+ },
250
+ {
251
+ "epoch": 1.5,
252
+ "learning_rate": 4.62596701211502e-05,
253
+ "loss": 2.048,
254
+ "step": 20500
255
+ },
256
+ {
257
+ "epoch": 1.53,
258
+ "learning_rate": 4.616844256312947e-05,
259
+ "loss": 2.048,
260
+ "step": 21000
261
+ },
262
+ {
263
+ "epoch": 1.57,
264
+ "learning_rate": 4.607721500510875e-05,
265
+ "loss": 2.048,
266
+ "step": 21500
267
+ },
268
+ {
269
+ "epoch": 1.61,
270
+ "learning_rate": 4.598598744708802e-05,
271
+ "loss": 2.048,
272
+ "step": 22000
273
+ },
274
+ {
275
+ "epoch": 1.64,
276
+ "learning_rate": 4.589475988906729e-05,
277
+ "loss": 2.048,
278
+ "step": 22500
279
+ },
280
+ {
281
+ "epoch": 1.68,
282
+ "learning_rate": 4.580353233104657e-05,
283
+ "loss": 2.048,
284
+ "step": 23000
285
+ },
286
+ {
287
+ "epoch": 1.72,
288
+ "learning_rate": 4.571230477302584e-05,
289
+ "loss": 2.048,
290
+ "step": 23500
291
+ },
292
+ {
293
+ "epoch": 1.75,
294
+ "learning_rate": 4.562107721500511e-05,
295
+ "loss": 2.048,
296
+ "step": 24000
297
+ },
298
+ {
299
+ "epoch": 1.79,
300
+ "learning_rate": 4.552984965698439e-05,
301
+ "loss": 2.048,
302
+ "step": 24500
303
+ },
304
+ {
305
+ "epoch": 1.82,
306
+ "learning_rate": 4.543862209896366e-05,
307
+ "loss": 2.048,
308
+ "step": 25000
309
+ },
310
+ {
311
+ "epoch": 1.86,
312
+ "learning_rate": 4.534739454094293e-05,
313
+ "loss": 2.048,
314
+ "step": 25500
315
+ },
316
+ {
317
+ "epoch": 1.9,
318
+ "learning_rate": 4.52561669829222e-05,
319
+ "loss": 2.048,
320
+ "step": 26000
321
+ },
322
+ {
323
+ "epoch": 1.93,
324
+ "learning_rate": 4.516493942490147e-05,
325
+ "loss": 2.048,
326
+ "step": 26500
327
+ },
328
+ {
329
+ "epoch": 1.97,
330
+ "learning_rate": 4.5073711866880744e-05,
331
+ "loss": 2.048,
332
+ "step": 27000
333
+ },
334
+ {
335
+ "epoch": 2.01,
336
+ "learning_rate": 4.498248430886002e-05,
337
+ "loss": 2.048,
338
+ "step": 27500
339
+ },
340
+ {
341
+ "epoch": 2.04,
342
+ "learning_rate": 4.489125675083929e-05,
343
+ "loss": 2.048,
344
+ "step": 28000
345
+ },
346
+ {
347
+ "epoch": 2.08,
348
+ "learning_rate": 4.480002919281857e-05,
349
+ "loss": 2.048,
350
+ "step": 28500
351
+ },
352
+ {
353
+ "epoch": 2.12,
354
+ "learning_rate": 4.470880163479784e-05,
355
+ "loss": 2.048,
356
+ "step": 29000
357
+ },
358
+ {
359
+ "epoch": 2.15,
360
+ "learning_rate": 4.461757407677711e-05,
361
+ "loss": 2.048,
362
+ "step": 29500
363
+ },
364
+ {
365
+ "epoch": 2.19,
366
+ "learning_rate": 4.452634651875639e-05,
367
+ "loss": 2.048,
368
+ "step": 30000
369
+ },
370
+ {
371
+ "epoch": 2.23,
372
+ "learning_rate": 4.443511896073566e-05,
373
+ "loss": 2.048,
374
+ "step": 30500
375
+ },
376
+ {
377
+ "epoch": 2.26,
378
+ "learning_rate": 4.434389140271493e-05,
379
+ "loss": 2.048,
380
+ "step": 31000
381
+ },
382
+ {
383
+ "epoch": 2.3,
384
+ "learning_rate": 4.425266384469421e-05,
385
+ "loss": 2.048,
386
+ "step": 31500
387
+ },
388
+ {
389
+ "epoch": 2.34,
390
+ "learning_rate": 4.416143628667348e-05,
391
+ "loss": 2.048,
392
+ "step": 32000
393
+ },
394
+ {
395
+ "epoch": 2.37,
396
+ "learning_rate": 4.407020872865275e-05,
397
+ "loss": 2.048,
398
+ "step": 32500
399
+ },
400
+ {
401
+ "epoch": 2.41,
402
+ "learning_rate": 4.397898117063203e-05,
403
+ "loss": 2.048,
404
+ "step": 33000
405
+ },
406
+ {
407
+ "epoch": 2.44,
408
+ "learning_rate": 4.38877536126113e-05,
409
+ "loss": 2.048,
410
+ "step": 33500
411
+ },
412
+ {
413
+ "epoch": 2.48,
414
+ "learning_rate": 4.379652605459057e-05,
415
+ "loss": 2.048,
416
+ "step": 34000
417
+ },
418
+ {
419
+ "epoch": 2.52,
420
+ "learning_rate": 4.3705298496569844e-05,
421
+ "loss": 2.048,
422
+ "step": 34500
423
+ },
424
+ {
425
+ "epoch": 2.55,
426
+ "learning_rate": 4.3614070938549115e-05,
427
+ "loss": 2.048,
428
+ "step": 35000
429
+ },
430
+ {
431
+ "epoch": 2.59,
432
+ "learning_rate": 4.352284338052839e-05,
433
+ "loss": 2.048,
434
+ "step": 35500
435
+ },
436
+ {
437
+ "epoch": 2.63,
438
+ "learning_rate": 4.3431615822507664e-05,
439
+ "loss": 2.048,
440
+ "step": 36000
441
+ },
442
+ {
443
+ "epoch": 2.66,
444
+ "learning_rate": 4.3340388264486935e-05,
445
+ "loss": 2.048,
446
+ "step": 36500
447
+ },
448
+ {
449
+ "epoch": 2.7,
450
+ "learning_rate": 4.324916070646621e-05,
451
+ "loss": 2.048,
452
+ "step": 37000
453
+ },
454
+ {
455
+ "epoch": 2.74,
456
+ "learning_rate": 4.3157933148445484e-05,
457
+ "loss": 2.048,
458
+ "step": 37500
459
+ },
460
+ {
461
+ "epoch": 2.77,
462
+ "learning_rate": 4.3066705590424756e-05,
463
+ "loss": 2.048,
464
+ "step": 38000
465
+ },
466
+ {
467
+ "epoch": 2.81,
468
+ "learning_rate": 4.2975478032404033e-05,
469
+ "loss": 2.048,
470
+ "step": 38500
471
+ },
472
+ {
473
+ "epoch": 2.85,
474
+ "learning_rate": 4.2884250474383305e-05,
475
+ "loss": 2.048,
476
+ "step": 39000
477
+ },
478
+ {
479
+ "epoch": 2.88,
480
+ "learning_rate": 4.279302291636258e-05,
481
+ "loss": 2.048,
482
+ "step": 39500
483
+ },
484
+ {
485
+ "epoch": 2.92,
486
+ "learning_rate": 4.2701795358341853e-05,
487
+ "loss": 2.048,
488
+ "step": 40000
489
+ },
490
+ {
491
+ "epoch": 2.96,
492
+ "learning_rate": 4.2610567800321125e-05,
493
+ "loss": 2.048,
494
+ "step": 40500
495
+ },
496
+ {
497
+ "epoch": 2.99,
498
+ "learning_rate": 4.2519340242300396e-05,
499
+ "loss": 2.048,
500
+ "step": 41000
501
+ },
502
+ {
503
+ "epoch": 3.03,
504
+ "learning_rate": 4.242811268427967e-05,
505
+ "loss": 2.048,
506
+ "step": 41500
507
+ },
508
+ {
509
+ "epoch": 3.07,
510
+ "learning_rate": 4.233688512625894e-05,
511
+ "loss": 2.048,
512
+ "step": 42000
513
+ },
514
+ {
515
+ "epoch": 3.1,
516
+ "learning_rate": 4.2245657568238216e-05,
517
+ "loss": 2.048,
518
+ "step": 42500
519
+ },
520
+ {
521
+ "epoch": 3.14,
522
+ "learning_rate": 4.215443001021749e-05,
523
+ "loss": 2.048,
524
+ "step": 43000
525
+ },
526
+ {
527
+ "epoch": 3.17,
528
+ "learning_rate": 4.206320245219676e-05,
529
+ "loss": 2.048,
530
+ "step": 43500
531
+ },
532
+ {
533
+ "epoch": 3.21,
534
+ "learning_rate": 4.1971974894176036e-05,
535
+ "loss": 2.048,
536
+ "step": 44000
537
+ },
538
+ {
539
+ "epoch": 3.25,
540
+ "learning_rate": 4.188074733615531e-05,
541
+ "loss": 2.048,
542
+ "step": 44500
543
+ },
544
+ {
545
+ "epoch": 3.28,
546
+ "learning_rate": 4.178951977813458e-05,
547
+ "loss": 2.048,
548
+ "step": 45000
549
+ },
550
+ {
551
+ "epoch": 3.32,
552
+ "learning_rate": 4.1698292220113856e-05,
553
+ "loss": 2.048,
554
+ "step": 45500
555
+ },
556
+ {
557
+ "epoch": 3.36,
558
+ "learning_rate": 4.160706466209313e-05,
559
+ "loss": 2.048,
560
+ "step": 46000
561
+ },
562
+ {
563
+ "epoch": 3.39,
564
+ "learning_rate": 4.1515837104072405e-05,
565
+ "loss": 2.048,
566
+ "step": 46500
567
+ },
568
+ {
569
+ "epoch": 3.43,
570
+ "learning_rate": 4.1424609546051676e-05,
571
+ "loss": 2.048,
572
+ "step": 47000
573
+ },
574
+ {
575
+ "epoch": 3.47,
576
+ "learning_rate": 4.133338198803095e-05,
577
+ "loss": 2.048,
578
+ "step": 47500
579
+ },
580
+ {
581
+ "epoch": 3.5,
582
+ "learning_rate": 4.124215443001022e-05,
583
+ "loss": 2.048,
584
+ "step": 48000
585
+ },
586
+ {
587
+ "epoch": 3.54,
588
+ "learning_rate": 4.115092687198949e-05,
589
+ "loss": 2.048,
590
+ "step": 48500
591
+ },
592
+ {
593
+ "epoch": 3.58,
594
+ "learning_rate": 4.105969931396876e-05,
595
+ "loss": 2.048,
596
+ "step": 49000
597
+ },
598
+ {
599
+ "epoch": 3.61,
600
+ "learning_rate": 4.096847175594804e-05,
601
+ "loss": 2.048,
602
+ "step": 49500
603
+ },
604
+ {
605
+ "epoch": 3.65,
606
+ "learning_rate": 4.087724419792731e-05,
607
+ "loss": 2.048,
608
+ "step": 50000
609
+ },
610
+ {
611
+ "epoch": 3.69,
612
+ "learning_rate": 4.078601663990658e-05,
613
+ "loss": 2.048,
614
+ "step": 50500
615
+ },
616
+ {
617
+ "epoch": 3.72,
618
+ "learning_rate": 4.069478908188586e-05,
619
+ "loss": 2.048,
620
+ "step": 51000
621
+ },
622
+ {
623
+ "epoch": 3.76,
624
+ "learning_rate": 4.060356152386513e-05,
625
+ "loss": 2.048,
626
+ "step": 51500
627
+ },
628
+ {
629
+ "epoch": 3.8,
630
+ "learning_rate": 4.051233396584441e-05,
631
+ "loss": 2.048,
632
+ "step": 52000
633
+ },
634
+ {
635
+ "epoch": 3.83,
636
+ "learning_rate": 4.042110640782368e-05,
637
+ "loss": 2.048,
638
+ "step": 52500
639
+ },
640
+ {
641
+ "epoch": 3.87,
642
+ "learning_rate": 4.032987884980295e-05,
643
+ "loss": 2.048,
644
+ "step": 53000
645
+ },
646
+ {
647
+ "epoch": 3.9,
648
+ "learning_rate": 4.023865129178223e-05,
649
+ "loss": 2.048,
650
+ "step": 53500
651
+ },
652
+ {
653
+ "epoch": 3.94,
654
+ "learning_rate": 4.01474237337615e-05,
655
+ "loss": 2.048,
656
+ "step": 54000
657
+ },
658
+ {
659
+ "epoch": 3.98,
660
+ "learning_rate": 4.005619617574077e-05,
661
+ "loss": 2.048,
662
+ "step": 54500
663
+ },
664
+ {
665
+ "epoch": 4.01,
666
+ "learning_rate": 3.996496861772005e-05,
667
+ "loss": 2.048,
668
+ "step": 55000
669
+ },
670
+ {
671
+ "epoch": 4.05,
672
+ "learning_rate": 3.987374105969932e-05,
673
+ "loss": 2.048,
674
+ "step": 55500
675
+ },
676
+ {
677
+ "epoch": 4.09,
678
+ "learning_rate": 3.978251350167859e-05,
679
+ "loss": 2.048,
680
+ "step": 56000
681
+ },
682
+ {
683
+ "epoch": 4.12,
684
+ "learning_rate": 3.969128594365786e-05,
685
+ "loss": 2.048,
686
+ "step": 56500
687
+ },
688
+ {
689
+ "epoch": 4.16,
690
+ "learning_rate": 3.960005838563713e-05,
691
+ "loss": 2.048,
692
+ "step": 57000
693
+ },
694
+ {
695
+ "epoch": 4.2,
696
+ "learning_rate": 3.950883082761641e-05,
697
+ "loss": 2.048,
698
+ "step": 57500
699
+ },
700
+ {
701
+ "epoch": 4.23,
702
+ "learning_rate": 3.941760326959568e-05,
703
+ "loss": 2.048,
704
+ "step": 58000
705
+ },
706
+ {
707
+ "epoch": 4.27,
708
+ "learning_rate": 3.932637571157495e-05,
709
+ "loss": 2.048,
710
+ "step": 58500
711
+ },
712
+ {
713
+ "epoch": 4.31,
714
+ "learning_rate": 3.923514815355423e-05,
715
+ "loss": 2.048,
716
+ "step": 59000
717
+ },
718
+ {
719
+ "epoch": 4.34,
720
+ "learning_rate": 3.91439205955335e-05,
721
+ "loss": 2.048,
722
+ "step": 59500
723
+ },
724
+ {
725
+ "epoch": 4.38,
726
+ "learning_rate": 3.905269303751277e-05,
727
+ "loss": 2.048,
728
+ "step": 60000
729
+ },
730
+ {
731
+ "epoch": 4.42,
732
+ "learning_rate": 3.896146547949205e-05,
733
+ "loss": 2.048,
734
+ "step": 60500
735
+ },
736
+ {
737
+ "epoch": 4.45,
738
+ "learning_rate": 3.887023792147132e-05,
739
+ "loss": 2.048,
740
+ "step": 61000
741
+ },
742
+ {
743
+ "epoch": 4.49,
744
+ "learning_rate": 3.877901036345059e-05,
745
+ "loss": 2.048,
746
+ "step": 61500
747
+ },
748
+ {
749
+ "epoch": 4.52,
750
+ "learning_rate": 3.868778280542987e-05,
751
+ "loss": 2.048,
752
+ "step": 62000
753
+ },
754
+ {
755
+ "epoch": 4.56,
756
+ "learning_rate": 3.859655524740914e-05,
757
+ "loss": 2.048,
758
+ "step": 62500
759
+ },
760
+ {
761
+ "epoch": 4.6,
762
+ "learning_rate": 3.850532768938841e-05,
763
+ "loss": 2.048,
764
+ "step": 63000
765
+ },
766
+ {
767
+ "epoch": 4.63,
768
+ "learning_rate": 3.8414100131367684e-05,
769
+ "loss": 2.048,
770
+ "step": 63500
771
+ },
772
+ {
773
+ "epoch": 4.67,
774
+ "learning_rate": 3.8322872573346955e-05,
775
+ "loss": 2.048,
776
+ "step": 64000
777
+ },
778
+ {
779
+ "epoch": 4.71,
780
+ "learning_rate": 3.823164501532623e-05,
781
+ "loss": 2.048,
782
+ "step": 64500
783
+ },
784
+ {
785
+ "epoch": 4.74,
786
+ "learning_rate": 3.8140417457305504e-05,
787
+ "loss": 2.048,
788
+ "step": 65000
789
+ },
790
+ {
791
+ "epoch": 4.78,
792
+ "learning_rate": 3.8049189899284775e-05,
793
+ "loss": 2.048,
794
+ "step": 65500
795
+ },
796
+ {
797
+ "epoch": 4.82,
798
+ "learning_rate": 3.795796234126405e-05,
799
+ "loss": 2.048,
800
+ "step": 66000
801
+ },
802
+ {
803
+ "epoch": 4.85,
804
+ "learning_rate": 3.7866734783243324e-05,
805
+ "loss": 2.048,
806
+ "step": 66500
807
+ },
808
+ {
809
+ "epoch": 4.89,
810
+ "learning_rate": 3.7775507225222595e-05,
811
+ "loss": 2.048,
812
+ "step": 67000
813
+ },
814
+ {
815
+ "epoch": 4.93,
816
+ "learning_rate": 3.768427966720187e-05,
817
+ "loss": 2.048,
818
+ "step": 67500
819
+ },
820
+ {
821
+ "epoch": 4.96,
822
+ "learning_rate": 3.7593052109181144e-05,
823
+ "loss": 2.048,
824
+ "step": 68000
825
+ },
826
+ {
827
+ "epoch": 5.0,
828
+ "learning_rate": 3.7501824551160415e-05,
829
+ "loss": 2.048,
830
+ "step": 68500
831
+ },
832
+ {
833
+ "epoch": 5.04,
834
+ "learning_rate": 3.741059699313969e-05,
835
+ "loss": 2.048,
836
+ "step": 69000
837
+ },
838
+ {
839
+ "epoch": 5.07,
840
+ "learning_rate": 3.7319369435118964e-05,
841
+ "loss": 2.048,
842
+ "step": 69500
843
+ },
844
+ {
845
+ "epoch": 5.11,
846
+ "learning_rate": 3.7228141877098235e-05,
847
+ "loss": 2.048,
848
+ "step": 70000
849
+ },
850
+ {
851
+ "epoch": 5.15,
852
+ "learning_rate": 3.7136914319077506e-05,
853
+ "loss": 2.048,
854
+ "step": 70500
855
+ },
856
+ {
857
+ "epoch": 5.18,
858
+ "learning_rate": 3.704568676105678e-05,
859
+ "loss": 2.048,
860
+ "step": 71000
861
+ },
862
+ {
863
+ "epoch": 5.22,
864
+ "learning_rate": 3.6954459203036055e-05,
865
+ "loss": 2.048,
866
+ "step": 71500
867
+ },
868
+ {
869
+ "epoch": 5.25,
870
+ "learning_rate": 3.6863231645015326e-05,
871
+ "loss": 2.048,
872
+ "step": 72000
873
+ },
874
+ {
875
+ "epoch": 5.29,
876
+ "learning_rate": 3.67720040869946e-05,
877
+ "loss": 2.048,
878
+ "step": 72500
879
+ },
880
+ {
881
+ "epoch": 5.33,
882
+ "learning_rate": 3.6680776528973875e-05,
883
+ "loss": 2.048,
884
+ "step": 73000
885
+ },
886
+ {
887
+ "epoch": 5.36,
888
+ "learning_rate": 3.6589548970953146e-05,
889
+ "loss": 2.048,
890
+ "step": 73500
891
+ },
892
+ {
893
+ "epoch": 5.4,
894
+ "learning_rate": 3.649832141293242e-05,
895
+ "loss": 2.048,
896
+ "step": 74000
897
+ },
898
+ {
899
+ "epoch": 5.44,
900
+ "learning_rate": 3.6407093854911695e-05,
901
+ "loss": 2.048,
902
+ "step": 74500
903
+ },
904
+ {
905
+ "epoch": 5.47,
906
+ "learning_rate": 3.6315866296890966e-05,
907
+ "loss": 2.048,
908
+ "step": 75000
909
+ },
910
+ {
911
+ "epoch": 5.51,
912
+ "learning_rate": 3.6224638738870244e-05,
913
+ "loss": 2.048,
914
+ "step": 75500
915
+ },
916
+ {
917
+ "epoch": 5.55,
918
+ "learning_rate": 3.6133411180849515e-05,
919
+ "loss": 2.048,
920
+ "step": 76000
921
+ },
922
+ {
923
+ "epoch": 5.58,
924
+ "learning_rate": 3.6042183622828787e-05,
925
+ "loss": 2.048,
926
+ "step": 76500
927
+ },
928
+ {
929
+ "epoch": 5.62,
930
+ "learning_rate": 3.5950956064808064e-05,
931
+ "loss": 2.048,
932
+ "step": 77000
933
+ },
934
+ {
935
+ "epoch": 5.66,
936
+ "learning_rate": 3.5859728506787336e-05,
937
+ "loss": 2.048,
938
+ "step": 77500
939
+ },
940
+ {
941
+ "epoch": 5.69,
942
+ "learning_rate": 3.576850094876661e-05,
943
+ "loss": 2.048,
944
+ "step": 78000
945
+ },
946
+ {
947
+ "epoch": 5.73,
948
+ "learning_rate": 3.567727339074588e-05,
949
+ "loss": 2.048,
950
+ "step": 78500
951
+ },
952
+ {
953
+ "epoch": 5.77,
954
+ "learning_rate": 3.558604583272515e-05,
955
+ "loss": 2.048,
956
+ "step": 79000
957
+ },
958
+ {
959
+ "epoch": 5.8,
960
+ "learning_rate": 3.549481827470442e-05,
961
+ "loss": 2.048,
962
+ "step": 79500
963
+ },
964
+ {
965
+ "epoch": 5.84,
966
+ "learning_rate": 3.54035907166837e-05,
967
+ "loss": 2.048,
968
+ "step": 80000
969
+ },
970
+ {
971
+ "epoch": 5.88,
972
+ "learning_rate": 3.531236315866297e-05,
973
+ "loss": 2.048,
974
+ "step": 80500
975
+ },
976
+ {
977
+ "epoch": 5.91,
978
+ "learning_rate": 3.522113560064224e-05,
979
+ "loss": 2.048,
980
+ "step": 81000
981
+ },
982
+ {
983
+ "epoch": 5.95,
984
+ "learning_rate": 3.512990804262152e-05,
985
+ "loss": 2.048,
986
+ "step": 81500
987
+ },
988
+ {
989
+ "epoch": 5.98,
990
+ "learning_rate": 3.503868048460079e-05,
991
+ "loss": 2.048,
992
+ "step": 82000
993
+ },
994
+ {
995
+ "epoch": 6.02,
996
+ "learning_rate": 3.494745292658007e-05,
997
+ "loss": 2.048,
998
+ "step": 82500
999
+ },
1000
+ {
1001
+ "epoch": 6.06,
1002
+ "learning_rate": 3.485622536855934e-05,
1003
+ "loss": 2.048,
1004
+ "step": 83000
1005
+ },
1006
+ {
1007
+ "epoch": 6.09,
1008
+ "learning_rate": 3.476499781053861e-05,
1009
+ "loss": 2.048,
1010
+ "step": 83500
1011
+ },
1012
+ {
1013
+ "epoch": 6.13,
1014
+ "learning_rate": 3.467377025251789e-05,
1015
+ "loss": 2.048,
1016
+ "step": 84000
1017
+ },
1018
+ {
1019
+ "epoch": 6.17,
1020
+ "learning_rate": 3.458254269449716e-05,
1021
+ "loss": 2.048,
1022
+ "step": 84500
1023
+ },
1024
+ {
1025
+ "epoch": 6.2,
1026
+ "learning_rate": 3.449131513647643e-05,
1027
+ "loss": 2.048,
1028
+ "step": 85000
1029
+ },
1030
+ {
1031
+ "epoch": 6.24,
1032
+ "learning_rate": 3.44000875784557e-05,
1033
+ "loss": 2.048,
1034
+ "step": 85500
1035
+ },
1036
+ {
1037
+ "epoch": 6.28,
1038
+ "learning_rate": 3.430886002043497e-05,
1039
+ "loss": 2.048,
1040
+ "step": 86000
1041
+ },
1042
+ {
1043
+ "epoch": 6.31,
1044
+ "learning_rate": 3.421763246241424e-05,
1045
+ "loss": 2.048,
1046
+ "step": 86500
1047
+ },
1048
+ {
1049
+ "epoch": 6.35,
1050
+ "learning_rate": 3.412640490439352e-05,
1051
+ "loss": 2.048,
1052
+ "step": 87000
1053
+ },
1054
+ {
1055
+ "epoch": 6.39,
1056
+ "learning_rate": 3.403517734637279e-05,
1057
+ "loss": 2.048,
1058
+ "step": 87500
1059
+ },
1060
+ {
1061
+ "epoch": 6.42,
1062
+ "learning_rate": 3.394394978835207e-05,
1063
+ "loss": 2.048,
1064
+ "step": 88000
1065
+ },
1066
+ {
1067
+ "epoch": 6.46,
1068
+ "learning_rate": 3.385272223033134e-05,
1069
+ "loss": 2.048,
1070
+ "step": 88500
1071
+ },
1072
+ {
1073
+ "epoch": 6.5,
1074
+ "learning_rate": 3.376149467231061e-05,
1075
+ "loss": 2.048,
1076
+ "step": 89000
1077
+ },
1078
+ {
1079
+ "epoch": 6.53,
1080
+ "learning_rate": 3.367026711428989e-05,
1081
+ "loss": 2.048,
1082
+ "step": 89500
1083
+ },
1084
+ {
1085
+ "epoch": 6.57,
1086
+ "learning_rate": 3.357903955626916e-05,
1087
+ "loss": 2.048,
1088
+ "step": 90000
1089
+ },
1090
+ {
1091
+ "epoch": 6.6,
1092
+ "learning_rate": 3.348781199824843e-05,
1093
+ "loss": 2.048,
1094
+ "step": 90500
1095
+ },
1096
+ {
1097
+ "epoch": 6.64,
1098
+ "learning_rate": 3.339658444022771e-05,
1099
+ "loss": 2.048,
1100
+ "step": 91000
1101
+ },
1102
+ {
1103
+ "epoch": 6.68,
1104
+ "learning_rate": 3.330535688220698e-05,
1105
+ "loss": 2.048,
1106
+ "step": 91500
1107
+ },
1108
+ {
1109
+ "epoch": 6.71,
1110
+ "learning_rate": 3.321412932418625e-05,
1111
+ "loss": 2.048,
1112
+ "step": 92000
1113
+ },
1114
+ {
1115
+ "epoch": 6.75,
1116
+ "learning_rate": 3.312290176616552e-05,
1117
+ "loss": 2.048,
1118
+ "step": 92500
1119
+ },
1120
+ {
1121
+ "epoch": 6.79,
1122
+ "learning_rate": 3.3031674208144794e-05,
1123
+ "loss": 2.048,
1124
+ "step": 93000
1125
+ },
1126
+ {
1127
+ "epoch": 6.82,
1128
+ "learning_rate": 3.294044665012407e-05,
1129
+ "loss": 2.048,
1130
+ "step": 93500
1131
+ },
1132
+ {
1133
+ "epoch": 6.86,
1134
+ "learning_rate": 3.284921909210334e-05,
1135
+ "loss": 2.048,
1136
+ "step": 94000
1137
+ },
1138
+ {
1139
+ "epoch": 6.9,
1140
+ "learning_rate": 3.2757991534082614e-05,
1141
+ "loss": 2.048,
1142
+ "step": 94500
1143
+ },
1144
+ {
1145
+ "epoch": 6.93,
1146
+ "learning_rate": 3.266676397606189e-05,
1147
+ "loss": 2.048,
1148
+ "step": 95000
1149
+ },
1150
+ {
1151
+ "epoch": 6.97,
1152
+ "learning_rate": 3.257553641804116e-05,
1153
+ "loss": 2.048,
1154
+ "step": 95500
1155
+ },
1156
+ {
1157
+ "epoch": 7.01,
1158
+ "learning_rate": 3.2484308860020434e-05,
1159
+ "loss": 2.048,
1160
+ "step": 96000
1161
+ },
1162
+ {
1163
+ "epoch": 7.04,
1164
+ "learning_rate": 3.239308130199971e-05,
1165
+ "loss": 2.048,
1166
+ "step": 96500
1167
+ },
1168
+ {
1169
+ "epoch": 7.08,
1170
+ "learning_rate": 3.230185374397898e-05,
1171
+ "loss": 2.048,
1172
+ "step": 97000
1173
+ },
1174
+ {
1175
+ "epoch": 7.12,
1176
+ "learning_rate": 3.2210626185958254e-05,
1177
+ "loss": 2.048,
1178
+ "step": 97500
1179
+ },
1180
+ {
1181
+ "epoch": 7.15,
1182
+ "learning_rate": 3.211939862793753e-05,
1183
+ "loss": 2.048,
1184
+ "step": 98000
1185
+ },
1186
+ {
1187
+ "epoch": 7.19,
1188
+ "learning_rate": 3.20281710699168e-05,
1189
+ "loss": 2.048,
1190
+ "step": 98500
1191
+ },
1192
+ {
1193
+ "epoch": 7.23,
1194
+ "learning_rate": 3.1936943511896074e-05,
1195
+ "loss": 2.048,
1196
+ "step": 99000
1197
+ },
1198
+ {
1199
+ "epoch": 7.26,
1200
+ "learning_rate": 3.184571595387535e-05,
1201
+ "loss": 2.048,
1202
+ "step": 99500
1203
+ },
1204
+ {
1205
+ "epoch": 7.3,
1206
+ "learning_rate": 3.175448839585462e-05,
1207
+ "loss": 2.048,
1208
+ "step": 100000
1209
+ },
1210
+ {
1211
+ "epoch": 7.33,
1212
+ "learning_rate": 3.1663260837833894e-05,
1213
+ "loss": 2.048,
1214
+ "step": 100500
1215
+ },
1216
+ {
1217
+ "epoch": 7.37,
1218
+ "learning_rate": 3.1572033279813166e-05,
1219
+ "loss": 2.048,
1220
+ "step": 101000
1221
+ },
1222
+ {
1223
+ "epoch": 7.41,
1224
+ "learning_rate": 3.148080572179244e-05,
1225
+ "loss": 2.048,
1226
+ "step": 101500
1227
+ },
1228
+ {
1229
+ "epoch": 7.44,
1230
+ "learning_rate": 3.1389578163771715e-05,
1231
+ "loss": 2.048,
1232
+ "step": 102000
1233
+ },
1234
+ {
1235
+ "epoch": 7.48,
1236
+ "learning_rate": 3.1298350605750986e-05,
1237
+ "loss": 2.048,
1238
+ "step": 102500
1239
+ },
1240
+ {
1241
+ "epoch": 7.52,
1242
+ "learning_rate": 3.120712304773026e-05,
1243
+ "loss": 2.048,
1244
+ "step": 103000
1245
+ },
1246
+ {
1247
+ "epoch": 7.55,
1248
+ "learning_rate": 3.1115895489709535e-05,
1249
+ "loss": 2.048,
1250
+ "step": 103500
1251
+ },
1252
+ {
1253
+ "epoch": 7.59,
1254
+ "learning_rate": 3.1024667931688806e-05,
1255
+ "loss": 2.048,
1256
+ "step": 104000
1257
+ },
1258
+ {
1259
+ "epoch": 7.63,
1260
+ "learning_rate": 3.093344037366808e-05,
1261
+ "loss": 2.048,
1262
+ "step": 104500
1263
+ },
1264
+ {
1265
+ "epoch": 7.66,
1266
+ "learning_rate": 3.0842212815647355e-05,
1267
+ "loss": 2.048,
1268
+ "step": 105000
1269
+ },
1270
+ {
1271
+ "epoch": 7.7,
1272
+ "learning_rate": 3.0750985257626626e-05,
1273
+ "loss": 2.048,
1274
+ "step": 105500
1275
+ },
1276
+ {
1277
+ "epoch": 7.74,
1278
+ "learning_rate": 3.0659757699605904e-05,
1279
+ "loss": 2.048,
1280
+ "step": 106000
1281
+ },
1282
+ {
1283
+ "epoch": 7.77,
1284
+ "learning_rate": 3.0568530141585175e-05,
1285
+ "loss": 2.048,
1286
+ "step": 106500
1287
+ },
1288
+ {
1289
+ "epoch": 7.81,
1290
+ "learning_rate": 3.0477302583564443e-05,
1291
+ "loss": 2.048,
1292
+ "step": 107000
1293
+ },
1294
+ {
1295
+ "epoch": 7.85,
1296
+ "learning_rate": 3.038607502554372e-05,
1297
+ "loss": 2.048,
1298
+ "step": 107500
1299
+ },
1300
+ {
1301
+ "epoch": 7.88,
1302
+ "learning_rate": 3.029484746752299e-05,
1303
+ "loss": 2.048,
1304
+ "step": 108000
1305
+ },
1306
+ {
1307
+ "epoch": 7.92,
1308
+ "learning_rate": 3.0203619909502263e-05,
1309
+ "loss": 2.048,
1310
+ "step": 108500
1311
+ },
1312
+ {
1313
+ "epoch": 7.96,
1314
+ "learning_rate": 3.0112392351481537e-05,
1315
+ "loss": 2.048,
1316
+ "step": 109000
1317
+ },
1318
+ {
1319
+ "epoch": 7.99,
1320
+ "learning_rate": 3.0021164793460808e-05,
1321
+ "loss": 2.048,
1322
+ "step": 109500
1323
+ },
1324
+ {
1325
+ "epoch": 8.03,
1326
+ "learning_rate": 2.992993723544008e-05,
1327
+ "loss": 2.048,
1328
+ "step": 110000
1329
+ },
1330
+ {
1331
+ "epoch": 8.06,
1332
+ "learning_rate": 2.9838709677419357e-05,
1333
+ "loss": 2.048,
1334
+ "step": 110500
1335
+ },
1336
+ {
1337
+ "epoch": 8.1,
1338
+ "learning_rate": 2.974748211939863e-05,
1339
+ "loss": 2.048,
1340
+ "step": 111000
1341
+ },
1342
+ {
1343
+ "epoch": 8.14,
1344
+ "learning_rate": 2.96562545613779e-05,
1345
+ "loss": 2.048,
1346
+ "step": 111500
1347
+ },
1348
+ {
1349
+ "epoch": 8.17,
1350
+ "learning_rate": 2.9565027003357177e-05,
1351
+ "loss": 2.048,
1352
+ "step": 112000
1353
+ },
1354
+ {
1355
+ "epoch": 8.21,
1356
+ "learning_rate": 2.947379944533645e-05,
1357
+ "loss": 2.048,
1358
+ "step": 112500
1359
+ },
1360
+ {
1361
+ "epoch": 8.25,
1362
+ "learning_rate": 2.9382571887315723e-05,
1363
+ "loss": 2.048,
1364
+ "step": 113000
1365
+ },
1366
+ {
1367
+ "epoch": 8.28,
1368
+ "learning_rate": 2.9291344329294994e-05,
1369
+ "loss": 2.048,
1370
+ "step": 113500
1371
+ },
1372
+ {
1373
+ "epoch": 8.32,
1374
+ "learning_rate": 2.9200116771274265e-05,
1375
+ "loss": 2.048,
1376
+ "step": 114000
1377
+ },
1378
+ {
1379
+ "epoch": 8.36,
1380
+ "learning_rate": 2.9108889213253543e-05,
1381
+ "loss": 2.048,
1382
+ "step": 114500
1383
+ },
1384
+ {
1385
+ "epoch": 8.39,
1386
+ "learning_rate": 2.9017661655232814e-05,
1387
+ "loss": 2.048,
1388
+ "step": 115000
1389
+ },
1390
+ {
1391
+ "epoch": 8.43,
1392
+ "learning_rate": 2.8926434097212085e-05,
1393
+ "loss": 2.048,
1394
+ "step": 115500
1395
+ },
1396
+ {
1397
+ "epoch": 8.47,
1398
+ "learning_rate": 2.8835206539191363e-05,
1399
+ "loss": 2.048,
1400
+ "step": 116000
1401
+ },
1402
+ {
1403
+ "epoch": 8.5,
1404
+ "learning_rate": 2.8743978981170634e-05,
1405
+ "loss": 2.048,
1406
+ "step": 116500
1407
+ },
1408
+ {
1409
+ "epoch": 8.54,
1410
+ "learning_rate": 2.8652751423149905e-05,
1411
+ "loss": 2.048,
1412
+ "step": 117000
1413
+ },
1414
+ {
1415
+ "epoch": 8.58,
1416
+ "learning_rate": 2.856152386512918e-05,
1417
+ "loss": 2.048,
1418
+ "step": 117500
1419
+ },
1420
+ {
1421
+ "epoch": 8.61,
1422
+ "learning_rate": 2.847029630710845e-05,
1423
+ "loss": 2.048,
1424
+ "step": 118000
1425
+ },
1426
+ {
1427
+ "epoch": 8.65,
1428
+ "learning_rate": 2.837906874908773e-05,
1429
+ "loss": 2.048,
1430
+ "step": 118500
1431
+ },
1432
+ {
1433
+ "epoch": 8.68,
1434
+ "learning_rate": 2.8287841191067e-05,
1435
+ "loss": 2.048,
1436
+ "step": 119000
1437
+ },
1438
+ {
1439
+ "epoch": 8.72,
1440
+ "learning_rate": 2.819661363304627e-05,
1441
+ "loss": 2.048,
1442
+ "step": 119500
1443
+ },
1444
+ {
1445
+ "epoch": 8.76,
1446
+ "learning_rate": 2.8105386075025546e-05,
1447
+ "loss": 2.048,
1448
+ "step": 120000
1449
+ },
1450
+ {
1451
+ "epoch": 8.79,
1452
+ "learning_rate": 2.8014158517004817e-05,
1453
+ "loss": 2.048,
1454
+ "step": 120500
1455
+ },
1456
+ {
1457
+ "epoch": 8.83,
1458
+ "learning_rate": 2.7922930958984088e-05,
1459
+ "loss": 2.048,
1460
+ "step": 121000
1461
+ },
1462
+ {
1463
+ "epoch": 8.87,
1464
+ "learning_rate": 2.7831703400963366e-05,
1465
+ "loss": 2.048,
1466
+ "step": 121500
1467
+ },
1468
+ {
1469
+ "epoch": 8.9,
1470
+ "learning_rate": 2.7740475842942637e-05,
1471
+ "loss": 2.048,
1472
+ "step": 122000
1473
+ },
1474
+ {
1475
+ "epoch": 8.94,
1476
+ "learning_rate": 2.7649248284921908e-05,
1477
+ "loss": 2.048,
1478
+ "step": 122500
1479
+ },
1480
+ {
1481
+ "epoch": 8.98,
1482
+ "learning_rate": 2.7558020726901186e-05,
1483
+ "loss": 2.048,
1484
+ "step": 123000
1485
+ },
1486
+ {
1487
+ "epoch": 9.01,
1488
+ "learning_rate": 2.7466793168880457e-05,
1489
+ "loss": 2.048,
1490
+ "step": 123500
1491
+ },
1492
+ {
1493
+ "epoch": 9.05,
1494
+ "learning_rate": 2.737556561085973e-05,
1495
+ "loss": 2.048,
1496
+ "step": 124000
1497
+ },
1498
+ {
1499
+ "epoch": 9.09,
1500
+ "learning_rate": 2.7284338052839002e-05,
1501
+ "loss": 2.048,
1502
+ "step": 124500
1503
+ },
1504
+ {
1505
+ "epoch": 9.12,
1506
+ "learning_rate": 2.7193110494818274e-05,
1507
+ "loss": 2.048,
1508
+ "step": 125000
1509
+ },
1510
+ {
1511
+ "epoch": 9.16,
1512
+ "learning_rate": 2.710188293679755e-05,
1513
+ "loss": 2.048,
1514
+ "step": 125500
1515
+ },
1516
+ {
1517
+ "epoch": 9.2,
1518
+ "learning_rate": 2.7010655378776823e-05,
1519
+ "loss": 2.048,
1520
+ "step": 126000
1521
+ },
1522
+ {
1523
+ "epoch": 9.23,
1524
+ "learning_rate": 2.6919427820756094e-05,
1525
+ "loss": 2.048,
1526
+ "step": 126500
1527
+ },
1528
+ {
1529
+ "epoch": 9.27,
1530
+ "learning_rate": 2.682820026273537e-05,
1531
+ "loss": 2.048,
1532
+ "step": 127000
1533
+ },
1534
+ {
1535
+ "epoch": 9.31,
1536
+ "learning_rate": 2.6736972704714643e-05,
1537
+ "loss": 2.048,
1538
+ "step": 127500
1539
+ },
1540
+ {
1541
+ "epoch": 9.34,
1542
+ "learning_rate": 2.6645745146693914e-05,
1543
+ "loss": 2.048,
1544
+ "step": 128000
1545
+ },
1546
+ {
1547
+ "epoch": 9.38,
1548
+ "learning_rate": 2.6554517588673188e-05,
1549
+ "loss": 2.048,
1550
+ "step": 128500
1551
+ },
1552
+ {
1553
+ "epoch": 9.41,
1554
+ "learning_rate": 2.646329003065246e-05,
1555
+ "loss": 2.048,
1556
+ "step": 129000
1557
+ },
1558
+ {
1559
+ "epoch": 9.45,
1560
+ "learning_rate": 2.637206247263173e-05,
1561
+ "loss": 2.048,
1562
+ "step": 129500
1563
+ },
1564
+ {
1565
+ "epoch": 9.49,
1566
+ "learning_rate": 2.628083491461101e-05,
1567
+ "loss": 2.048,
1568
+ "step": 130000
1569
+ },
1570
+ {
1571
+ "epoch": 9.52,
1572
+ "learning_rate": 2.618960735659028e-05,
1573
+ "loss": 2.048,
1574
+ "step": 130500
1575
+ },
1576
+ {
1577
+ "epoch": 9.56,
1578
+ "learning_rate": 2.6098379798569554e-05,
1579
+ "loss": 2.048,
1580
+ "step": 131000
1581
+ },
1582
+ {
1583
+ "epoch": 9.6,
1584
+ "learning_rate": 2.6007152240548825e-05,
1585
+ "loss": 2.048,
1586
+ "step": 131500
1587
+ },
1588
+ {
1589
+ "epoch": 9.63,
1590
+ "learning_rate": 2.5915924682528096e-05,
1591
+ "loss": 2.048,
1592
+ "step": 132000
1593
+ },
1594
+ {
1595
+ "epoch": 9.67,
1596
+ "learning_rate": 2.5824697124507374e-05,
1597
+ "loss": 2.048,
1598
+ "step": 132500
1599
+ },
1600
+ {
1601
+ "epoch": 9.71,
1602
+ "learning_rate": 2.5733469566486645e-05,
1603
+ "loss": 2.048,
1604
+ "step": 133000
1605
+ },
1606
+ {
1607
+ "epoch": 9.74,
1608
+ "learning_rate": 2.5642242008465916e-05,
1609
+ "loss": 2.048,
1610
+ "step": 133500
1611
+ },
1612
+ {
1613
+ "epoch": 9.78,
1614
+ "learning_rate": 2.5551014450445194e-05,
1615
+ "loss": 2.048,
1616
+ "step": 134000
1617
+ },
1618
+ {
1619
+ "epoch": 9.82,
1620
+ "learning_rate": 2.5459786892424465e-05,
1621
+ "loss": 2.048,
1622
+ "step": 134500
1623
+ },
1624
+ {
1625
+ "epoch": 9.85,
1626
+ "learning_rate": 2.5368559334403736e-05,
1627
+ "loss": 2.048,
1628
+ "step": 135000
1629
+ },
1630
+ {
1631
+ "epoch": 9.89,
1632
+ "learning_rate": 2.527733177638301e-05,
1633
+ "loss": 2.048,
1634
+ "step": 135500
1635
+ },
1636
+ {
1637
+ "epoch": 9.93,
1638
+ "learning_rate": 2.5186104218362282e-05,
1639
+ "loss": 2.048,
1640
+ "step": 136000
1641
+ },
1642
+ {
1643
+ "epoch": 9.96,
1644
+ "learning_rate": 2.509487666034156e-05,
1645
+ "loss": 2.048,
1646
+ "step": 136500
1647
+ },
1648
+ {
1649
+ "epoch": 10.0,
1650
+ "learning_rate": 2.500364910232083e-05,
1651
+ "loss": 2.048,
1652
+ "step": 137000
1653
+ },
1654
+ {
1655
+ "epoch": 10.04,
1656
+ "learning_rate": 2.4912421544300105e-05,
1657
+ "loss": 2.048,
1658
+ "step": 137500
1659
+ },
1660
+ {
1661
+ "epoch": 10.07,
1662
+ "learning_rate": 2.4821193986279376e-05,
1663
+ "loss": 2.048,
1664
+ "step": 138000
1665
+ },
1666
+ {
1667
+ "epoch": 10.11,
1668
+ "learning_rate": 2.472996642825865e-05,
1669
+ "loss": 2.048,
1670
+ "step": 138500
1671
+ },
1672
+ {
1673
+ "epoch": 10.14,
1674
+ "learning_rate": 2.4638738870237922e-05,
1675
+ "loss": 2.048,
1676
+ "step": 139000
1677
+ },
1678
+ {
1679
+ "epoch": 10.18,
1680
+ "learning_rate": 2.4547511312217193e-05,
1681
+ "loss": 2.048,
1682
+ "step": 139500
1683
+ },
1684
+ {
1685
+ "epoch": 10.22,
1686
+ "learning_rate": 2.4456283754196468e-05,
1687
+ "loss": 2.048,
1688
+ "step": 140000
1689
+ },
1690
+ {
1691
+ "epoch": 10.25,
1692
+ "learning_rate": 2.4365056196175742e-05,
1693
+ "loss": 2.048,
1694
+ "step": 140500
1695
+ },
1696
+ {
1697
+ "epoch": 10.29,
1698
+ "learning_rate": 2.4273828638155017e-05,
1699
+ "loss": 2.048,
1700
+ "step": 141000
1701
+ },
1702
+ {
1703
+ "epoch": 10.33,
1704
+ "learning_rate": 2.4182601080134288e-05,
1705
+ "loss": 2.048,
1706
+ "step": 141500
1707
+ },
1708
+ {
1709
+ "epoch": 10.36,
1710
+ "learning_rate": 2.4091373522113562e-05,
1711
+ "loss": 2.048,
1712
+ "step": 142000
1713
+ },
1714
+ {
1715
+ "epoch": 10.4,
1716
+ "learning_rate": 2.4000145964092833e-05,
1717
+ "loss": 2.048,
1718
+ "step": 142500
1719
+ },
1720
+ {
1721
+ "epoch": 10.44,
1722
+ "learning_rate": 2.3908918406072104e-05,
1723
+ "loss": 2.048,
1724
+ "step": 143000
1725
+ },
1726
+ {
1727
+ "epoch": 10.47,
1728
+ "learning_rate": 2.381769084805138e-05,
1729
+ "loss": 2.048,
1730
+ "step": 143500
1731
+ },
1732
+ {
1733
+ "epoch": 10.51,
1734
+ "learning_rate": 2.3726463290030653e-05,
1735
+ "loss": 2.048,
1736
+ "step": 144000
1737
+ },
1738
+ {
1739
+ "epoch": 10.55,
1740
+ "learning_rate": 2.3635235732009928e-05,
1741
+ "loss": 2.048,
1742
+ "step": 144500
1743
+ },
1744
+ {
1745
+ "epoch": 10.58,
1746
+ "learning_rate": 2.35440081739892e-05,
1747
+ "loss": 2.048,
1748
+ "step": 145000
1749
+ },
1750
+ {
1751
+ "epoch": 10.62,
1752
+ "learning_rate": 2.3452780615968474e-05,
1753
+ "loss": 2.048,
1754
+ "step": 145500
1755
+ },
1756
+ {
1757
+ "epoch": 10.66,
1758
+ "learning_rate": 2.3361553057947748e-05,
1759
+ "loss": 2.048,
1760
+ "step": 146000
1761
+ },
1762
+ {
1763
+ "epoch": 10.69,
1764
+ "learning_rate": 2.327032549992702e-05,
1765
+ "loss": 2.048,
1766
+ "step": 146500
1767
+ },
1768
+ {
1769
+ "epoch": 10.73,
1770
+ "learning_rate": 2.317909794190629e-05,
1771
+ "loss": 2.048,
1772
+ "step": 147000
1773
+ },
1774
+ {
1775
+ "epoch": 10.76,
1776
+ "learning_rate": 2.3087870383885565e-05,
1777
+ "loss": 2.048,
1778
+ "step": 147500
1779
+ },
1780
+ {
1781
+ "epoch": 10.8,
1782
+ "learning_rate": 2.299664282586484e-05,
1783
+ "loss": 2.048,
1784
+ "step": 148000
1785
+ },
1786
+ {
1787
+ "epoch": 10.84,
1788
+ "learning_rate": 2.290541526784411e-05,
1789
+ "loss": 2.048,
1790
+ "step": 148500
1791
+ },
1792
+ {
1793
+ "epoch": 10.87,
1794
+ "learning_rate": 2.2814187709823385e-05,
1795
+ "loss": 2.048,
1796
+ "step": 149000
1797
+ },
1798
+ {
1799
+ "epoch": 10.91,
1800
+ "learning_rate": 2.272296015180266e-05,
1801
+ "loss": 2.048,
1802
+ "step": 149500
1803
+ },
1804
+ {
1805
+ "epoch": 10.95,
1806
+ "learning_rate": 2.263173259378193e-05,
1807
+ "loss": 2.048,
1808
+ "step": 150000
1809
+ },
1810
+ {
1811
+ "epoch": 10.98,
1812
+ "learning_rate": 2.25405050357612e-05,
1813
+ "loss": 2.048,
1814
+ "step": 150500
1815
+ },
1816
+ {
1817
+ "epoch": 11.02,
1818
+ "learning_rate": 2.2449277477740476e-05,
1819
+ "loss": 2.048,
1820
+ "step": 151000
1821
+ },
1822
+ {
1823
+ "epoch": 11.06,
1824
+ "learning_rate": 2.235804991971975e-05,
1825
+ "loss": 2.048,
1826
+ "step": 151500
1827
+ },
1828
+ {
1829
+ "epoch": 11.09,
1830
+ "learning_rate": 2.226682236169902e-05,
1831
+ "loss": 2.048,
1832
+ "step": 152000
1833
+ },
1834
+ {
1835
+ "epoch": 11.13,
1836
+ "learning_rate": 2.2175594803678296e-05,
1837
+ "loss": 2.048,
1838
+ "step": 152500
1839
+ },
1840
+ {
1841
+ "epoch": 11.17,
1842
+ "learning_rate": 2.208436724565757e-05,
1843
+ "loss": 2.048,
1844
+ "step": 153000
1845
+ },
1846
+ {
1847
+ "epoch": 11.2,
1848
+ "learning_rate": 2.1993139687636842e-05,
1849
+ "loss": 2.048,
1850
+ "step": 153500
1851
+ },
1852
+ {
1853
+ "epoch": 11.24,
1854
+ "learning_rate": 2.1901912129616113e-05,
1855
+ "loss": 2.048,
1856
+ "step": 154000
1857
+ },
1858
+ {
1859
+ "epoch": 11.28,
1860
+ "learning_rate": 2.1810684571595387e-05,
1861
+ "loss": 2.048,
1862
+ "step": 154500
1863
+ },
1864
+ {
1865
+ "epoch": 11.31,
1866
+ "learning_rate": 2.1719457013574662e-05,
1867
+ "loss": 2.048,
1868
+ "step": 155000
1869
+ },
1870
+ {
1871
+ "epoch": 11.35,
1872
+ "learning_rate": 2.1628229455553936e-05,
1873
+ "loss": 2.048,
1874
+ "step": 155500
1875
+ },
1876
+ {
1877
+ "epoch": 11.39,
1878
+ "learning_rate": 2.1537001897533207e-05,
1879
+ "loss": 2.048,
1880
+ "step": 156000
1881
+ },
1882
+ {
1883
+ "epoch": 11.42,
1884
+ "learning_rate": 2.1445774339512482e-05,
1885
+ "loss": 2.048,
1886
+ "step": 156500
1887
+ },
1888
+ {
1889
+ "epoch": 11.46,
1890
+ "learning_rate": 2.1354546781491756e-05,
1891
+ "loss": 2.048,
1892
+ "step": 157000
1893
+ },
1894
+ {
1895
+ "epoch": 11.49,
1896
+ "learning_rate": 2.1263319223471028e-05,
1897
+ "loss": 2.048,
1898
+ "step": 157500
1899
+ },
1900
+ {
1901
+ "epoch": 11.53,
1902
+ "learning_rate": 2.11720916654503e-05,
1903
+ "loss": 2.048,
1904
+ "step": 158000
1905
+ },
1906
+ {
1907
+ "epoch": 11.57,
1908
+ "learning_rate": 2.1080864107429573e-05,
1909
+ "loss": 2.048,
1910
+ "step": 158500
1911
+ },
1912
+ {
1913
+ "epoch": 11.6,
1914
+ "learning_rate": 2.0989636549408848e-05,
1915
+ "loss": 2.048,
1916
+ "step": 159000
1917
+ },
1918
+ {
1919
+ "epoch": 11.64,
1920
+ "learning_rate": 2.089840899138812e-05,
1921
+ "loss": 2.048,
1922
+ "step": 159500
1923
+ },
1924
+ {
1925
+ "epoch": 11.68,
1926
+ "learning_rate": 2.0807181433367393e-05,
1927
+ "loss": 2.048,
1928
+ "step": 160000
1929
+ },
1930
+ {
1931
+ "epoch": 11.71,
1932
+ "learning_rate": 2.0715953875346668e-05,
1933
+ "loss": 2.048,
1934
+ "step": 160500
1935
+ },
1936
+ {
1937
+ "epoch": 11.75,
1938
+ "learning_rate": 2.062472631732594e-05,
1939
+ "loss": 2.048,
1940
+ "step": 161000
1941
+ },
1942
+ {
1943
+ "epoch": 11.79,
1944
+ "learning_rate": 2.053349875930521e-05,
1945
+ "loss": 2.048,
1946
+ "step": 161500
1947
+ },
1948
+ {
1949
+ "epoch": 11.82,
1950
+ "learning_rate": 2.0442271201284484e-05,
1951
+ "loss": 2.048,
1952
+ "step": 162000
1953
+ },
1954
+ {
1955
+ "epoch": 11.86,
1956
+ "learning_rate": 2.035104364326376e-05,
1957
+ "loss": 2.048,
1958
+ "step": 162500
1959
+ },
1960
+ {
1961
+ "epoch": 11.9,
1962
+ "learning_rate": 2.025981608524303e-05,
1963
+ "loss": 2.048,
1964
+ "step": 163000
1965
+ },
1966
+ {
1967
+ "epoch": 11.93,
1968
+ "learning_rate": 2.0168588527222305e-05,
1969
+ "loss": 2.048,
1970
+ "step": 163500
1971
+ },
1972
+ {
1973
+ "epoch": 11.97,
1974
+ "learning_rate": 2.007736096920158e-05,
1975
+ "loss": 2.048,
1976
+ "step": 164000
1977
+ },
1978
+ {
1979
+ "epoch": 12.01,
1980
+ "learning_rate": 1.998613341118085e-05,
1981
+ "loss": 2.048,
1982
+ "step": 164500
1983
+ },
1984
+ {
1985
+ "epoch": 12.04,
1986
+ "learning_rate": 1.989490585316012e-05,
1987
+ "loss": 2.048,
1988
+ "step": 165000
1989
+ },
1990
+ {
1991
+ "epoch": 12.08,
1992
+ "learning_rate": 1.9803678295139396e-05,
1993
+ "loss": 2.048,
1994
+ "step": 165500
1995
+ },
1996
+ {
1997
+ "epoch": 12.12,
1998
+ "learning_rate": 1.971245073711867e-05,
1999
+ "loss": 2.048,
2000
+ "step": 166000
2001
+ },
2002
+ {
2003
+ "epoch": 12.15,
2004
+ "learning_rate": 1.962122317909794e-05,
2005
+ "loss": 2.048,
2006
+ "step": 166500
2007
+ },
2008
+ {
2009
+ "epoch": 12.19,
2010
+ "learning_rate": 1.9529995621077216e-05,
2011
+ "loss": 2.048,
2012
+ "step": 167000
2013
+ },
2014
+ {
2015
+ "epoch": 12.22,
2016
+ "learning_rate": 1.943876806305649e-05,
2017
+ "loss": 2.048,
2018
+ "step": 167500
2019
+ },
2020
+ {
2021
+ "epoch": 12.26,
2022
+ "learning_rate": 1.9347540505035765e-05,
2023
+ "loss": 2.048,
2024
+ "step": 168000
2025
+ },
2026
+ {
2027
+ "epoch": 12.3,
2028
+ "learning_rate": 1.9256312947015036e-05,
2029
+ "loss": 2.048,
2030
+ "step": 168500
2031
+ },
2032
+ {
2033
+ "epoch": 12.33,
2034
+ "learning_rate": 1.9165085388994307e-05,
2035
+ "loss": 2.048,
2036
+ "step": 169000
2037
+ },
2038
+ {
2039
+ "epoch": 12.37,
2040
+ "learning_rate": 1.907385783097358e-05,
2041
+ "loss": 2.048,
2042
+ "step": 169500
2043
+ },
2044
+ {
2045
+ "epoch": 12.41,
2046
+ "learning_rate": 1.8982630272952853e-05,
2047
+ "loss": 2.048,
2048
+ "step": 170000
2049
+ },
2050
+ {
2051
+ "epoch": 12.44,
2052
+ "learning_rate": 1.8891402714932127e-05,
2053
+ "loss": 2.048,
2054
+ "step": 170500
2055
+ },
2056
+ {
2057
+ "epoch": 12.48,
2058
+ "learning_rate": 1.88001751569114e-05,
2059
+ "loss": 2.048,
2060
+ "step": 171000
2061
+ },
2062
+ {
2063
+ "epoch": 12.52,
2064
+ "learning_rate": 1.8708947598890676e-05,
2065
+ "loss": 2.048,
2066
+ "step": 171500
2067
+ },
2068
+ {
2069
+ "epoch": 12.55,
2070
+ "learning_rate": 1.8617720040869947e-05,
2071
+ "loss": 2.048,
2072
+ "step": 172000
2073
+ },
2074
+ {
2075
+ "epoch": 12.59,
2076
+ "learning_rate": 1.852649248284922e-05,
2077
+ "loss": 2.048,
2078
+ "step": 172500
2079
+ },
2080
+ {
2081
+ "epoch": 12.63,
2082
+ "learning_rate": 1.8435264924828493e-05,
2083
+ "loss": 2.048,
2084
+ "step": 173000
2085
+ },
2086
+ {
2087
+ "epoch": 12.66,
2088
+ "learning_rate": 1.8344037366807767e-05,
2089
+ "loss": 2.048,
2090
+ "step": 173500
2091
+ },
2092
+ {
2093
+ "epoch": 12.7,
2094
+ "learning_rate": 1.825280980878704e-05,
2095
+ "loss": 2.048,
2096
+ "step": 174000
2097
+ },
2098
+ {
2099
+ "epoch": 12.74,
2100
+ "learning_rate": 1.8161582250766313e-05,
2101
+ "loss": 2.048,
2102
+ "step": 174500
2103
+ },
2104
+ {
2105
+ "epoch": 12.77,
2106
+ "learning_rate": 1.8070354692745587e-05,
2107
+ "loss": 2.048,
2108
+ "step": 175000
2109
+ },
2110
+ {
2111
+ "epoch": 12.81,
2112
+ "learning_rate": 1.797912713472486e-05,
2113
+ "loss": 2.048,
2114
+ "step": 175500
2115
+ },
2116
+ {
2117
+ "epoch": 12.84,
2118
+ "learning_rate": 1.788789957670413e-05,
2119
+ "loss": 2.048,
2120
+ "step": 176000
2121
+ },
2122
+ {
2123
+ "epoch": 12.88,
2124
+ "learning_rate": 1.7796672018683404e-05,
2125
+ "loss": 2.048,
2126
+ "step": 176500
2127
+ },
2128
+ {
2129
+ "epoch": 12.92,
2130
+ "learning_rate": 1.770544446066268e-05,
2131
+ "loss": 2.048,
2132
+ "step": 177000
2133
+ },
2134
+ {
2135
+ "epoch": 12.95,
2136
+ "learning_rate": 1.761421690264195e-05,
2137
+ "loss": 2.048,
2138
+ "step": 177500
2139
+ },
2140
+ {
2141
+ "epoch": 12.99,
2142
+ "learning_rate": 1.7522989344621224e-05,
2143
+ "loss": 2.048,
2144
+ "step": 178000
2145
+ },
2146
+ {
2147
+ "epoch": 13.03,
2148
+ "learning_rate": 1.74317617866005e-05,
2149
+ "loss": 2.048,
2150
+ "step": 178500
2151
+ },
2152
+ {
2153
+ "epoch": 13.06,
2154
+ "learning_rate": 1.734053422857977e-05,
2155
+ "loss": 2.048,
2156
+ "step": 179000
2157
+ },
2158
+ {
2159
+ "epoch": 13.1,
2160
+ "learning_rate": 1.7249306670559044e-05,
2161
+ "loss": 2.048,
2162
+ "step": 179500
2163
+ },
2164
+ {
2165
+ "epoch": 13.14,
2166
+ "learning_rate": 1.7158079112538315e-05,
2167
+ "loss": 2.048,
2168
+ "step": 180000
2169
+ },
2170
+ {
2171
+ "epoch": 13.17,
2172
+ "learning_rate": 1.706685155451759e-05,
2173
+ "loss": 2.048,
2174
+ "step": 180500
2175
+ },
2176
+ {
2177
+ "epoch": 13.21,
2178
+ "learning_rate": 1.697562399649686e-05,
2179
+ "loss": 2.048,
2180
+ "step": 181000
2181
+ },
2182
+ {
2183
+ "epoch": 13.25,
2184
+ "learning_rate": 1.6884396438476136e-05,
2185
+ "loss": 2.048,
2186
+ "step": 181500
2187
+ },
2188
+ {
2189
+ "epoch": 13.28,
2190
+ "learning_rate": 1.679316888045541e-05,
2191
+ "loss": 2.048,
2192
+ "step": 182000
2193
+ },
2194
+ {
2195
+ "epoch": 13.32,
2196
+ "learning_rate": 1.6701941322434684e-05,
2197
+ "loss": 2.048,
2198
+ "step": 182500
2199
+ },
2200
+ {
2201
+ "epoch": 13.36,
2202
+ "learning_rate": 1.6610713764413956e-05,
2203
+ "loss": 2.048,
2204
+ "step": 183000
2205
+ },
2206
+ {
2207
+ "epoch": 13.39,
2208
+ "learning_rate": 1.6519486206393227e-05,
2209
+ "loss": 2.048,
2210
+ "step": 183500
2211
+ },
2212
+ {
2213
+ "epoch": 13.43,
2214
+ "learning_rate": 1.64282586483725e-05,
2215
+ "loss": 2.048,
2216
+ "step": 184000
2217
+ },
2218
+ {
2219
+ "epoch": 13.47,
2220
+ "learning_rate": 1.6337031090351772e-05,
2221
+ "loss": 2.048,
2222
+ "step": 184500
2223
+ },
2224
+ {
2225
+ "epoch": 13.5,
2226
+ "learning_rate": 1.6245803532331047e-05,
2227
+ "loss": 2.048,
2228
+ "step": 185000
2229
+ },
2230
+ {
2231
+ "epoch": 13.54,
2232
+ "learning_rate": 1.615457597431032e-05,
2233
+ "loss": 2.048,
2234
+ "step": 185500
2235
+ },
2236
+ {
2237
+ "epoch": 13.57,
2238
+ "learning_rate": 1.6063348416289596e-05,
2239
+ "loss": 2.048,
2240
+ "step": 186000
2241
+ },
2242
+ {
2243
+ "epoch": 13.61,
2244
+ "learning_rate": 1.5972120858268867e-05,
2245
+ "loss": 2.048,
2246
+ "step": 186500
2247
+ },
2248
+ {
2249
+ "epoch": 13.65,
2250
+ "learning_rate": 1.588089330024814e-05,
2251
+ "loss": 2.048,
2252
+ "step": 187000
2253
+ },
2254
+ {
2255
+ "epoch": 13.68,
2256
+ "learning_rate": 1.5789665742227412e-05,
2257
+ "loss": 2.048,
2258
+ "step": 187500
2259
+ },
2260
+ {
2261
+ "epoch": 13.72,
2262
+ "learning_rate": 1.5698438184206684e-05,
2263
+ "loss": 2.048,
2264
+ "step": 188000
2265
+ },
2266
+ {
2267
+ "epoch": 13.76,
2268
+ "learning_rate": 1.5607210626185958e-05,
2269
+ "loss": 2.048,
2270
+ "step": 188500
2271
+ },
2272
+ {
2273
+ "epoch": 13.79,
2274
+ "learning_rate": 1.5515983068165233e-05,
2275
+ "loss": 2.048,
2276
+ "step": 189000
2277
+ },
2278
+ {
2279
+ "epoch": 13.83,
2280
+ "learning_rate": 1.5424755510144507e-05,
2281
+ "loss": 2.048,
2282
+ "step": 189500
2283
+ },
2284
+ {
2285
+ "epoch": 13.87,
2286
+ "learning_rate": 1.5333527952123778e-05,
2287
+ "loss": 2.048,
2288
+ "step": 190000
2289
+ },
2290
+ {
2291
+ "epoch": 13.9,
2292
+ "learning_rate": 1.5242300394103051e-05,
2293
+ "loss": 2.048,
2294
+ "step": 190500
2295
+ },
2296
+ {
2297
+ "epoch": 13.94,
2298
+ "learning_rate": 1.5151072836082325e-05,
2299
+ "loss": 2.048,
2300
+ "step": 191000
2301
+ },
2302
+ {
2303
+ "epoch": 13.98,
2304
+ "learning_rate": 1.5059845278061598e-05,
2305
+ "loss": 2.048,
2306
+ "step": 191500
2307
+ },
2308
+ {
2309
+ "epoch": 14.01,
2310
+ "learning_rate": 1.496861772004087e-05,
2311
+ "loss": 2.048,
2312
+ "step": 192000
2313
+ },
2314
+ {
2315
+ "epoch": 14.05,
2316
+ "learning_rate": 1.4877390162020144e-05,
2317
+ "loss": 2.048,
2318
+ "step": 192500
2319
+ },
2320
+ {
2321
+ "epoch": 14.09,
2322
+ "learning_rate": 1.4786162603999418e-05,
2323
+ "loss": 2.048,
2324
+ "step": 193000
2325
+ },
2326
+ {
2327
+ "epoch": 14.12,
2328
+ "learning_rate": 1.469493504597869e-05,
2329
+ "loss": 2.048,
2330
+ "step": 193500
2331
+ },
2332
+ {
2333
+ "epoch": 14.16,
2334
+ "learning_rate": 1.4603707487957962e-05,
2335
+ "loss": 2.048,
2336
+ "step": 194000
2337
+ },
2338
+ {
2339
+ "epoch": 14.2,
2340
+ "learning_rate": 1.4512479929937237e-05,
2341
+ "loss": 2.048,
2342
+ "step": 194500
2343
+ },
2344
+ {
2345
+ "epoch": 14.23,
2346
+ "learning_rate": 1.4421252371916511e-05,
2347
+ "loss": 2.048,
2348
+ "step": 195000
2349
+ },
2350
+ {
2351
+ "epoch": 14.27,
2352
+ "learning_rate": 1.4330024813895782e-05,
2353
+ "loss": 2.048,
2354
+ "step": 195500
2355
+ },
2356
+ {
2357
+ "epoch": 14.3,
2358
+ "learning_rate": 1.4238797255875055e-05,
2359
+ "loss": 2.048,
2360
+ "step": 196000
2361
+ },
2362
+ {
2363
+ "epoch": 14.34,
2364
+ "learning_rate": 1.414756969785433e-05,
2365
+ "loss": 2.048,
2366
+ "step": 196500
2367
+ },
2368
+ {
2369
+ "epoch": 14.38,
2370
+ "learning_rate": 1.40563421398336e-05,
2371
+ "loss": 2.048,
2372
+ "step": 197000
2373
+ },
2374
+ {
2375
+ "epoch": 14.41,
2376
+ "learning_rate": 1.3965114581812874e-05,
2377
+ "loss": 2.048,
2378
+ "step": 197500
2379
+ },
2380
+ {
2381
+ "epoch": 14.45,
2382
+ "learning_rate": 1.3873887023792148e-05,
2383
+ "loss": 2.048,
2384
+ "step": 198000
2385
+ },
2386
+ {
2387
+ "epoch": 14.49,
2388
+ "learning_rate": 1.3782659465771423e-05,
2389
+ "loss": 2.048,
2390
+ "step": 198500
2391
+ },
2392
+ {
2393
+ "epoch": 14.52,
2394
+ "learning_rate": 1.3691431907750694e-05,
2395
+ "loss": 2.048,
2396
+ "step": 199000
2397
+ },
2398
+ {
2399
+ "epoch": 14.56,
2400
+ "learning_rate": 1.3600204349729966e-05,
2401
+ "loss": 2.048,
2402
+ "step": 199500
2403
+ },
2404
+ {
2405
+ "epoch": 14.6,
2406
+ "learning_rate": 1.3508976791709241e-05,
2407
+ "loss": 2.048,
2408
+ "step": 200000
2409
+ },
2410
+ {
2411
+ "epoch": 14.63,
2412
+ "learning_rate": 1.3417749233688512e-05,
2413
+ "loss": 2.048,
2414
+ "step": 200500
2415
+ },
2416
+ {
2417
+ "epoch": 14.67,
2418
+ "learning_rate": 1.3326521675667787e-05,
2419
+ "loss": 2.048,
2420
+ "step": 201000
2421
+ },
2422
+ {
2423
+ "epoch": 14.71,
2424
+ "learning_rate": 1.323529411764706e-05,
2425
+ "loss": 2.048,
2426
+ "step": 201500
2427
+ },
2428
+ {
2429
+ "epoch": 14.74,
2430
+ "learning_rate": 1.3144066559626334e-05,
2431
+ "loss": 2.048,
2432
+ "step": 202000
2433
+ },
2434
+ {
2435
+ "epoch": 14.78,
2436
+ "learning_rate": 1.3052839001605605e-05,
2437
+ "loss": 2.048,
2438
+ "step": 202500
2439
+ },
2440
+ {
2441
+ "epoch": 14.82,
2442
+ "learning_rate": 1.2961611443584878e-05,
2443
+ "loss": 2.048,
2444
+ "step": 203000
2445
+ },
2446
+ {
2447
+ "epoch": 14.85,
2448
+ "learning_rate": 1.2870383885564152e-05,
2449
+ "loss": 2.048,
2450
+ "step": 203500
2451
+ },
2452
+ {
2453
+ "epoch": 14.89,
2454
+ "learning_rate": 1.2779156327543427e-05,
2455
+ "loss": 2.048,
2456
+ "step": 204000
2457
+ },
2458
+ {
2459
+ "epoch": 14.92,
2460
+ "learning_rate": 1.2687928769522698e-05,
2461
+ "loss": 2.048,
2462
+ "step": 204500
2463
+ },
2464
+ {
2465
+ "epoch": 14.96,
2466
+ "learning_rate": 1.259670121150197e-05,
2467
+ "loss": 2.048,
2468
+ "step": 205000
2469
+ },
2470
+ {
2471
+ "epoch": 15.0,
2472
+ "learning_rate": 1.2505473653481245e-05,
2473
+ "loss": 2.048,
2474
+ "step": 205500
2475
+ },
2476
+ {
2477
+ "epoch": 15.03,
2478
+ "learning_rate": 1.2414246095460518e-05,
2479
+ "loss": 2.048,
2480
+ "step": 206000
2481
+ },
2482
+ {
2483
+ "epoch": 15.07,
2484
+ "learning_rate": 1.232301853743979e-05,
2485
+ "loss": 2.048,
2486
+ "step": 206500
2487
+ },
2488
+ {
2489
+ "epoch": 15.11,
2490
+ "learning_rate": 1.2231790979419064e-05,
2491
+ "loss": 2.048,
2492
+ "step": 207000
2493
+ },
2494
+ {
2495
+ "epoch": 15.14,
2496
+ "learning_rate": 1.2140563421398336e-05,
2497
+ "loss": 2.048,
2498
+ "step": 207500
2499
+ },
2500
+ {
2501
+ "epoch": 15.18,
2502
+ "learning_rate": 1.204933586337761e-05,
2503
+ "loss": 2.048,
2504
+ "step": 208000
2505
+ },
2506
+ {
2507
+ "epoch": 15.22,
2508
+ "learning_rate": 1.1958108305356882e-05,
2509
+ "loss": 2.048,
2510
+ "step": 208500
2511
+ },
2512
+ {
2513
+ "epoch": 15.25,
2514
+ "learning_rate": 1.1866880747336156e-05,
2515
+ "loss": 2.048,
2516
+ "step": 209000
2517
+ },
2518
+ {
2519
+ "epoch": 15.29,
2520
+ "learning_rate": 1.177565318931543e-05,
2521
+ "loss": 2.048,
2522
+ "step": 209500
2523
+ },
2524
+ {
2525
+ "epoch": 15.33,
2526
+ "learning_rate": 1.1684425631294702e-05,
2527
+ "loss": 2.048,
2528
+ "step": 210000
2529
+ },
2530
+ {
2531
+ "epoch": 15.36,
2532
+ "learning_rate": 1.1593198073273975e-05,
2533
+ "loss": 2.048,
2534
+ "step": 210500
2535
+ },
2536
+ {
2537
+ "epoch": 15.4,
2538
+ "learning_rate": 1.1501970515253248e-05,
2539
+ "loss": 2.048,
2540
+ "step": 211000
2541
+ },
2542
+ {
2543
+ "epoch": 15.44,
2544
+ "learning_rate": 1.1410742957232522e-05,
2545
+ "loss": 2.048,
2546
+ "step": 211500
2547
+ },
2548
+ {
2549
+ "epoch": 15.47,
2550
+ "learning_rate": 1.1319515399211795e-05,
2551
+ "loss": 2.048,
2552
+ "step": 212000
2553
+ },
2554
+ {
2555
+ "epoch": 15.51,
2556
+ "learning_rate": 1.1228287841191068e-05,
2557
+ "loss": 2.048,
2558
+ "step": 212500
2559
+ },
2560
+ {
2561
+ "epoch": 15.55,
2562
+ "learning_rate": 1.113706028317034e-05,
2563
+ "loss": 2.048,
2564
+ "step": 213000
2565
+ },
2566
+ {
2567
+ "epoch": 15.58,
2568
+ "learning_rate": 1.1045832725149615e-05,
2569
+ "loss": 2.048,
2570
+ "step": 213500
2571
+ },
2572
+ {
2573
+ "epoch": 15.62,
2574
+ "learning_rate": 1.0954605167128886e-05,
2575
+ "loss": 2.048,
2576
+ "step": 214000
2577
+ },
2578
+ {
2579
+ "epoch": 15.65,
2580
+ "learning_rate": 1.0863377609108159e-05,
2581
+ "loss": 2.048,
2582
+ "step": 214500
2583
+ },
2584
+ {
2585
+ "epoch": 15.69,
2586
+ "learning_rate": 1.0772150051087433e-05,
2587
+ "loss": 2.048,
2588
+ "step": 215000
2589
+ },
2590
+ {
2591
+ "epoch": 15.73,
2592
+ "learning_rate": 1.0680922493066706e-05,
2593
+ "loss": 2.048,
2594
+ "step": 215500
2595
+ },
2596
+ {
2597
+ "epoch": 15.76,
2598
+ "learning_rate": 1.0589694935045979e-05,
2599
+ "loss": 2.048,
2600
+ "step": 216000
2601
+ },
2602
+ {
2603
+ "epoch": 15.8,
2604
+ "learning_rate": 1.0498467377025252e-05,
2605
+ "loss": 2.048,
2606
+ "step": 216500
2607
+ },
2608
+ {
2609
+ "epoch": 15.84,
2610
+ "learning_rate": 1.0407239819004526e-05,
2611
+ "loss": 2.048,
2612
+ "step": 217000
2613
+ },
2614
+ {
2615
+ "epoch": 15.87,
2616
+ "learning_rate": 1.0316012260983799e-05,
2617
+ "loss": 2.048,
2618
+ "step": 217500
2619
+ },
2620
+ {
2621
+ "epoch": 15.91,
2622
+ "learning_rate": 1.0224784702963072e-05,
2623
+ "loss": 2.048,
2624
+ "step": 218000
2625
+ },
2626
+ {
2627
+ "epoch": 15.95,
2628
+ "learning_rate": 1.0133557144942345e-05,
2629
+ "loss": 2.048,
2630
+ "step": 218500
2631
+ },
2632
+ {
2633
+ "epoch": 15.98,
2634
+ "learning_rate": 1.0042329586921618e-05,
2635
+ "loss": 2.048,
2636
+ "step": 219000
2637
+ },
2638
+ {
2639
+ "epoch": 16.02,
2640
+ "learning_rate": 9.95110202890089e-06,
2641
+ "loss": 2.048,
2642
+ "step": 219500
2643
+ },
2644
+ {
2645
+ "epoch": 16.06,
2646
+ "learning_rate": 9.859874470880163e-06,
2647
+ "loss": 2.048,
2648
+ "step": 220000
2649
+ },
2650
+ {
2651
+ "epoch": 16.09,
2652
+ "learning_rate": 9.768646912859438e-06,
2653
+ "loss": 2.048,
2654
+ "step": 220500
2655
+ },
2656
+ {
2657
+ "epoch": 16.13,
2658
+ "learning_rate": 9.67741935483871e-06,
2659
+ "loss": 2.048,
2660
+ "step": 221000
2661
+ },
2662
+ {
2663
+ "epoch": 16.17,
2664
+ "learning_rate": 9.586191796817983e-06,
2665
+ "loss": 2.048,
2666
+ "step": 221500
2667
+ },
2668
+ {
2669
+ "epoch": 16.2,
2670
+ "learning_rate": 9.494964238797256e-06,
2671
+ "loss": 2.048,
2672
+ "step": 222000
2673
+ },
2674
+ {
2675
+ "epoch": 16.24,
2676
+ "learning_rate": 9.40373668077653e-06,
2677
+ "loss": 2.048,
2678
+ "step": 222500
2679
+ },
2680
+ {
2681
+ "epoch": 16.27,
2682
+ "learning_rate": 9.312509122755803e-06,
2683
+ "loss": 2.048,
2684
+ "step": 223000
2685
+ },
2686
+ {
2687
+ "epoch": 16.31,
2688
+ "learning_rate": 9.221281564735074e-06,
2689
+ "loss": 2.048,
2690
+ "step": 223500
2691
+ },
2692
+ {
2693
+ "epoch": 16.35,
2694
+ "learning_rate": 9.130054006714349e-06,
2695
+ "loss": 2.048,
2696
+ "step": 224000
2697
+ },
2698
+ {
2699
+ "epoch": 16.38,
2700
+ "learning_rate": 9.038826448693622e-06,
2701
+ "loss": 2.048,
2702
+ "step": 224500
2703
+ },
2704
+ {
2705
+ "epoch": 16.42,
2706
+ "learning_rate": 8.947598890672895e-06,
2707
+ "loss": 2.048,
2708
+ "step": 225000
2709
+ },
2710
+ {
2711
+ "epoch": 16.46,
2712
+ "learning_rate": 8.856371332652167e-06,
2713
+ "loss": 2.048,
2714
+ "step": 225500
2715
+ },
2716
+ {
2717
+ "epoch": 16.49,
2718
+ "learning_rate": 8.765143774631442e-06,
2719
+ "loss": 2.048,
2720
+ "step": 226000
2721
+ },
2722
+ {
2723
+ "epoch": 16.53,
2724
+ "learning_rate": 8.673916216610715e-06,
2725
+ "loss": 2.048,
2726
+ "step": 226500
2727
+ },
2728
+ {
2729
+ "epoch": 16.57,
2730
+ "learning_rate": 8.582688658589987e-06,
2731
+ "loss": 2.048,
2732
+ "step": 227000
2733
+ },
2734
+ {
2735
+ "epoch": 16.6,
2736
+ "learning_rate": 8.49146110056926e-06,
2737
+ "loss": 2.048,
2738
+ "step": 227500
2739
+ },
2740
+ {
2741
+ "epoch": 16.64,
2742
+ "learning_rate": 8.400233542548533e-06,
2743
+ "loss": 2.048,
2744
+ "step": 228000
2745
+ },
2746
+ {
2747
+ "epoch": 16.68,
2748
+ "learning_rate": 8.309005984527807e-06,
2749
+ "loss": 2.048,
2750
+ "step": 228500
2751
+ },
2752
+ {
2753
+ "epoch": 16.71,
2754
+ "learning_rate": 8.217778426507079e-06,
2755
+ "loss": 2.048,
2756
+ "step": 229000
2757
+ },
2758
+ {
2759
+ "epoch": 16.75,
2760
+ "learning_rate": 8.126550868486353e-06,
2761
+ "loss": 2.048,
2762
+ "step": 229500
2763
+ },
2764
+ {
2765
+ "epoch": 16.79,
2766
+ "learning_rate": 8.035323310465626e-06,
2767
+ "loss": 2.048,
2768
+ "step": 230000
2769
+ },
2770
+ {
2771
+ "epoch": 16.82,
2772
+ "learning_rate": 7.944095752444899e-06,
2773
+ "loss": 2.048,
2774
+ "step": 230500
2775
+ },
2776
+ {
2777
+ "epoch": 16.86,
2778
+ "learning_rate": 7.852868194424171e-06,
2779
+ "loss": 2.048,
2780
+ "step": 231000
2781
+ },
2782
+ {
2783
+ "epoch": 16.9,
2784
+ "learning_rate": 7.761640636403446e-06,
2785
+ "loss": 2.048,
2786
+ "step": 231500
2787
+ },
2788
+ {
2789
+ "epoch": 16.93,
2790
+ "learning_rate": 7.670413078382719e-06,
2791
+ "loss": 2.048,
2792
+ "step": 232000
2793
+ },
2794
+ {
2795
+ "epoch": 16.97,
2796
+ "learning_rate": 7.579185520361991e-06,
2797
+ "loss": 2.048,
2798
+ "step": 232500
2799
+ },
2800
+ {
2801
+ "epoch": 17.0,
2802
+ "learning_rate": 7.487957962341264e-06,
2803
+ "loss": 2.048,
2804
+ "step": 233000
2805
+ },
2806
+ {
2807
+ "epoch": 17.04,
2808
+ "learning_rate": 7.396730404320537e-06,
2809
+ "loss": 2.048,
2810
+ "step": 233500
2811
+ },
2812
+ {
2813
+ "epoch": 17.08,
2814
+ "learning_rate": 7.305502846299811e-06,
2815
+ "loss": 2.048,
2816
+ "step": 234000
2817
+ },
2818
+ {
2819
+ "epoch": 17.11,
2820
+ "learning_rate": 7.214275288279084e-06,
2821
+ "loss": 2.048,
2822
+ "step": 234500
2823
+ },
2824
+ {
2825
+ "epoch": 17.15,
2826
+ "learning_rate": 7.123047730258357e-06,
2827
+ "loss": 2.048,
2828
+ "step": 235000
2829
+ },
2830
+ {
2831
+ "epoch": 17.19,
2832
+ "learning_rate": 7.03182017223763e-06,
2833
+ "loss": 2.048,
2834
+ "step": 235500
2835
+ },
2836
+ {
2837
+ "epoch": 17.22,
2838
+ "learning_rate": 6.940592614216904e-06,
2839
+ "loss": 2.048,
2840
+ "step": 236000
2841
+ },
2842
+ {
2843
+ "epoch": 17.26,
2844
+ "learning_rate": 6.8493650561961765e-06,
2845
+ "loss": 2.048,
2846
+ "step": 236500
2847
+ },
2848
+ {
2849
+ "epoch": 17.3,
2850
+ "learning_rate": 6.7581374981754485e-06,
2851
+ "loss": 2.048,
2852
+ "step": 237000
2853
+ },
2854
+ {
2855
+ "epoch": 17.33,
2856
+ "learning_rate": 6.666909940154722e-06,
2857
+ "loss": 2.048,
2858
+ "step": 237500
2859
+ },
2860
+ {
2861
+ "epoch": 17.37,
2862
+ "learning_rate": 6.575682382133995e-06,
2863
+ "loss": 2.048,
2864
+ "step": 238000
2865
+ },
2866
+ {
2867
+ "epoch": 17.41,
2868
+ "learning_rate": 6.484454824113269e-06,
2869
+ "loss": 2.048,
2870
+ "step": 238500
2871
+ },
2872
+ {
2873
+ "epoch": 17.44,
2874
+ "learning_rate": 6.393227266092541e-06,
2875
+ "loss": 2.048,
2876
+ "step": 239000
2877
+ },
2878
+ {
2879
+ "epoch": 17.48,
2880
+ "learning_rate": 6.301999708071815e-06,
2881
+ "loss": 2.048,
2882
+ "step": 239500
2883
+ },
2884
+ {
2885
+ "epoch": 17.52,
2886
+ "learning_rate": 6.210772150051088e-06,
2887
+ "loss": 2.048,
2888
+ "step": 240000
2889
+ },
2890
+ {
2891
+ "epoch": 17.55,
2892
+ "learning_rate": 6.119544592030361e-06,
2893
+ "loss": 2.048,
2894
+ "step": 240500
2895
+ },
2896
+ {
2897
+ "epoch": 17.59,
2898
+ "learning_rate": 6.028317034009634e-06,
2899
+ "loss": 2.048,
2900
+ "step": 241000
2901
+ },
2902
+ {
2903
+ "epoch": 17.63,
2904
+ "learning_rate": 5.937089475988907e-06,
2905
+ "loss": 2.048,
2906
+ "step": 241500
2907
+ },
2908
+ {
2909
+ "epoch": 17.66,
2910
+ "learning_rate": 5.845861917968181e-06,
2911
+ "loss": 2.048,
2912
+ "step": 242000
2913
+ },
2914
+ {
2915
+ "epoch": 17.7,
2916
+ "learning_rate": 5.7546343599474535e-06,
2917
+ "loss": 2.048,
2918
+ "step": 242500
2919
+ },
2920
+ {
2921
+ "epoch": 17.73,
2922
+ "learning_rate": 5.663406801926726e-06,
2923
+ "loss": 2.048,
2924
+ "step": 243000
2925
+ },
2926
+ {
2927
+ "epoch": 17.77,
2928
+ "learning_rate": 5.572179243905999e-06,
2929
+ "loss": 2.048,
2930
+ "step": 243500
2931
+ },
2932
+ {
2933
+ "epoch": 17.81,
2934
+ "learning_rate": 5.480951685885272e-06,
2935
+ "loss": 2.048,
2936
+ "step": 244000
2937
+ },
2938
+ {
2939
+ "epoch": 17.84,
2940
+ "learning_rate": 5.3897241278645456e-06,
2941
+ "loss": 2.048,
2942
+ "step": 244500
2943
+ },
2944
+ {
2945
+ "epoch": 17.88,
2946
+ "learning_rate": 5.298496569843818e-06,
2947
+ "loss": 2.048,
2948
+ "step": 245000
2949
+ },
2950
+ {
2951
+ "epoch": 17.92,
2952
+ "learning_rate": 5.207269011823092e-06,
2953
+ "loss": 2.048,
2954
+ "step": 245500
2955
+ },
2956
+ {
2957
+ "epoch": 17.95,
2958
+ "learning_rate": 5.116041453802365e-06,
2959
+ "loss": 2.048,
2960
+ "step": 246000
2961
+ },
2962
+ {
2963
+ "epoch": 17.99,
2964
+ "learning_rate": 5.0248138957816385e-06,
2965
+ "loss": 2.048,
2966
+ "step": 246500
2967
+ },
2968
+ {
2969
+ "epoch": 18.03,
2970
+ "learning_rate": 4.933586337760911e-06,
2971
+ "loss": 2.048,
2972
+ "step": 247000
2973
+ },
2974
+ {
2975
+ "epoch": 18.06,
2976
+ "learning_rate": 4.842358779740184e-06,
2977
+ "loss": 2.048,
2978
+ "step": 247500
2979
+ },
2980
+ {
2981
+ "epoch": 18.1,
2982
+ "learning_rate": 4.751131221719457e-06,
2983
+ "loss": 2.048,
2984
+ "step": 248000
2985
+ },
2986
+ {
2987
+ "epoch": 18.14,
2988
+ "learning_rate": 4.6599036636987305e-06,
2989
+ "loss": 2.048,
2990
+ "step": 248500
2991
+ },
2992
+ {
2993
+ "epoch": 18.17,
2994
+ "learning_rate": 4.568676105678003e-06,
2995
+ "loss": 2.048,
2996
+ "step": 249000
2997
+ },
2998
+ {
2999
+ "epoch": 18.21,
3000
+ "learning_rate": 4.477448547657277e-06,
3001
+ "loss": 2.048,
3002
+ "step": 249500
3003
+ },
3004
+ {
3005
+ "epoch": 18.25,
3006
+ "learning_rate": 4.38622098963655e-06,
3007
+ "loss": 2.048,
3008
+ "step": 250000
3009
+ },
3010
+ {
3011
+ "epoch": 18.28,
3012
+ "learning_rate": 4.2949934316158226e-06,
3013
+ "loss": 2.048,
3014
+ "step": 250500
3015
+ },
3016
+ {
3017
+ "epoch": 18.32,
3018
+ "learning_rate": 4.203765873595096e-06,
3019
+ "loss": 2.048,
3020
+ "step": 251000
3021
+ },
3022
+ {
3023
+ "epoch": 18.35,
3024
+ "learning_rate": 4.112538315574369e-06,
3025
+ "loss": 2.048,
3026
+ "step": 251500
3027
+ },
3028
+ {
3029
+ "epoch": 18.39,
3030
+ "learning_rate": 4.021310757553642e-06,
3031
+ "loss": 2.048,
3032
+ "step": 252000
3033
+ },
3034
+ {
3035
+ "epoch": 18.43,
3036
+ "learning_rate": 3.930083199532915e-06,
3037
+ "loss": 2.048,
3038
+ "step": 252500
3039
+ },
3040
+ {
3041
+ "epoch": 18.46,
3042
+ "learning_rate": 3.838855641512188e-06,
3043
+ "loss": 2.048,
3044
+ "step": 253000
3045
+ },
3046
+ {
3047
+ "epoch": 18.5,
3048
+ "learning_rate": 3.747628083491461e-06,
3049
+ "loss": 2.048,
3050
+ "step": 253500
3051
+ },
3052
+ {
3053
+ "epoch": 18.54,
3054
+ "learning_rate": 3.6564005254707343e-06,
3055
+ "loss": 2.048,
3056
+ "step": 254000
3057
+ },
3058
+ {
3059
+ "epoch": 18.57,
3060
+ "learning_rate": 3.5651729674500075e-06,
3061
+ "loss": 2.048,
3062
+ "step": 254500
3063
+ },
3064
+ {
3065
+ "epoch": 18.61,
3066
+ "learning_rate": 3.4739454094292807e-06,
3067
+ "loss": 2.048,
3068
+ "step": 255000
3069
+ },
3070
+ {
3071
+ "epoch": 18.65,
3072
+ "learning_rate": 3.382717851408554e-06,
3073
+ "loss": 2.048,
3074
+ "step": 255500
3075
+ },
3076
+ {
3077
+ "epoch": 18.68,
3078
+ "learning_rate": 3.291490293387827e-06,
3079
+ "loss": 2.048,
3080
+ "step": 256000
3081
+ },
3082
+ {
3083
+ "epoch": 18.72,
3084
+ "learning_rate": 3.2002627353670995e-06,
3085
+ "loss": 2.048,
3086
+ "step": 256500
3087
+ },
3088
+ {
3089
+ "epoch": 18.76,
3090
+ "learning_rate": 3.109035177346373e-06,
3091
+ "loss": 2.048,
3092
+ "step": 257000
3093
+ },
3094
+ {
3095
+ "epoch": 18.79,
3096
+ "learning_rate": 3.017807619325646e-06,
3097
+ "loss": 2.048,
3098
+ "step": 257500
3099
+ },
3100
+ {
3101
+ "epoch": 18.83,
3102
+ "learning_rate": 2.926580061304919e-06,
3103
+ "loss": 2.048,
3104
+ "step": 258000
3105
+ },
3106
+ {
3107
+ "epoch": 18.87,
3108
+ "learning_rate": 2.8353525032841924e-06,
3109
+ "loss": 2.048,
3110
+ "step": 258500
3111
+ },
3112
+ {
3113
+ "epoch": 18.9,
3114
+ "learning_rate": 2.7441249452634652e-06,
3115
+ "loss": 2.048,
3116
+ "step": 259000
3117
+ },
3118
+ {
3119
+ "epoch": 18.94,
3120
+ "learning_rate": 2.652897387242738e-06,
3121
+ "loss": 2.048,
3122
+ "step": 259500
3123
+ },
3124
+ {
3125
+ "epoch": 18.98,
3126
+ "learning_rate": 2.5616698292220113e-06,
3127
+ "loss": 2.048,
3128
+ "step": 260000
3129
+ },
3130
+ {
3131
+ "epoch": 19.01,
3132
+ "learning_rate": 2.4704422712012845e-06,
3133
+ "loss": 2.048,
3134
+ "step": 260500
3135
+ },
3136
+ {
3137
+ "epoch": 19.05,
3138
+ "learning_rate": 2.3792147131805577e-06,
3139
+ "loss": 2.048,
3140
+ "step": 261000
3141
+ },
3142
+ {
3143
+ "epoch": 19.08,
3144
+ "learning_rate": 2.287987155159831e-06,
3145
+ "loss": 2.048,
3146
+ "step": 261500
3147
+ },
3148
+ {
3149
+ "epoch": 19.12,
3150
+ "learning_rate": 2.1967595971391037e-06,
3151
+ "loss": 2.048,
3152
+ "step": 262000
3153
+ },
3154
+ {
3155
+ "epoch": 19.16,
3156
+ "learning_rate": 2.105532039118377e-06,
3157
+ "loss": 2.048,
3158
+ "step": 262500
3159
+ },
3160
+ {
3161
+ "epoch": 19.19,
3162
+ "learning_rate": 2.01430448109765e-06,
3163
+ "loss": 2.048,
3164
+ "step": 263000
3165
+ },
3166
+ {
3167
+ "epoch": 19.23,
3168
+ "learning_rate": 1.9230769230769234e-06,
3169
+ "loss": 2.048,
3170
+ "step": 263500
3171
+ },
3172
+ {
3173
+ "epoch": 19.27,
3174
+ "learning_rate": 1.8318493650561962e-06,
3175
+ "loss": 2.048,
3176
+ "step": 264000
3177
+ },
3178
+ {
3179
+ "epoch": 19.3,
3180
+ "learning_rate": 1.7406218070354692e-06,
3181
+ "loss": 2.048,
3182
+ "step": 264500
3183
+ },
3184
+ {
3185
+ "epoch": 19.34,
3186
+ "learning_rate": 1.6493942490147424e-06,
3187
+ "loss": 2.048,
3188
+ "step": 265000
3189
+ },
3190
+ {
3191
+ "epoch": 19.38,
3192
+ "learning_rate": 1.5581666909940157e-06,
3193
+ "loss": 2.048,
3194
+ "step": 265500
3195
+ },
3196
+ {
3197
+ "epoch": 19.41,
3198
+ "learning_rate": 1.4669391329732887e-06,
3199
+ "loss": 2.048,
3200
+ "step": 266000
3201
+ },
3202
+ {
3203
+ "epoch": 19.45,
3204
+ "learning_rate": 1.3757115749525619e-06,
3205
+ "loss": 2.048,
3206
+ "step": 266500
3207
+ },
3208
+ {
3209
+ "epoch": 19.49,
3210
+ "learning_rate": 1.2844840169318347e-06,
3211
+ "loss": 2.048,
3212
+ "step": 267000
3213
+ },
3214
+ {
3215
+ "epoch": 19.52,
3216
+ "learning_rate": 1.193256458911108e-06,
3217
+ "loss": 2.048,
3218
+ "step": 267500
3219
+ },
3220
+ {
3221
+ "epoch": 19.56,
3222
+ "learning_rate": 1.102028900890381e-06,
3223
+ "loss": 2.048,
3224
+ "step": 268000
3225
+ },
3226
+ {
3227
+ "epoch": 19.6,
3228
+ "learning_rate": 1.0108013428696541e-06,
3229
+ "loss": 2.048,
3230
+ "step": 268500
3231
+ },
3232
+ {
3233
+ "epoch": 19.63,
3234
+ "learning_rate": 9.195737848489272e-07,
3235
+ "loss": 2.048,
3236
+ "step": 269000
3237
+ },
3238
+ {
3239
+ "epoch": 19.67,
3240
+ "learning_rate": 8.283462268282003e-07,
3241
+ "loss": 2.048,
3242
+ "step": 269500
3243
+ },
3244
+ {
3245
+ "epoch": 19.71,
3246
+ "learning_rate": 7.371186688074734e-07,
3247
+ "loss": 2.048,
3248
+ "step": 270000
3249
+ },
3250
+ {
3251
+ "epoch": 19.74,
3252
+ "learning_rate": 6.458911107867465e-07,
3253
+ "loss": 2.048,
3254
+ "step": 270500
3255
+ },
3256
+ {
3257
+ "epoch": 19.78,
3258
+ "learning_rate": 5.546635527660195e-07,
3259
+ "loss": 2.048,
3260
+ "step": 271000
3261
+ },
3262
+ {
3263
+ "epoch": 19.81,
3264
+ "learning_rate": 4.6343599474529264e-07,
3265
+ "loss": 2.048,
3266
+ "step": 271500
3267
+ },
3268
+ {
3269
+ "epoch": 19.85,
3270
+ "learning_rate": 3.7220843672456576e-07,
3271
+ "loss": 2.048,
3272
+ "step": 272000
3273
+ },
3274
+ {
3275
+ "epoch": 19.89,
3276
+ "learning_rate": 2.809808787038389e-07,
3277
+ "loss": 2.048,
3278
+ "step": 272500
3279
+ },
3280
+ {
3281
+ "epoch": 19.92,
3282
+ "learning_rate": 1.8975332068311197e-07,
3283
+ "loss": 2.048,
3284
+ "step": 273000
3285
+ },
3286
+ {
3287
+ "epoch": 19.96,
3288
+ "learning_rate": 9.852576266238505e-08,
3289
+ "loss": 2.048,
3290
+ "step": 273500
3291
+ },
3292
+ {
3293
+ "epoch": 20.0,
3294
+ "learning_rate": 7.298204641658152e-09,
3295
+ "loss": 2.048,
3296
+ "step": 274000
3297
+ },
3298
+ {
3299
+ "epoch": 20.0,
3300
+ "step": 274040,
3301
+ "total_flos": 2.7362600759486054e+18,
3302
+ "train_runtime": 264850.0877,
3303
+ "train_samples_per_second": 1.035
3304
+ }
3305
+ ],
3306
+ "max_steps": 274040,
3307
+ "num_train_epochs": 20,
3308
+ "total_flos": 2.7362600759486054e+18,
3309
+ "trial_name": null,
3310
+ "trial_params": null
3311
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e32613937d90041db08ada1be6c8bf90bc879e0f1a6f8429de214969c93a132c
3
+ size 2287
vocab.json ADDED
The diff for this file is too large to render. See raw diff