Nick H commited on
Commit ·
3b69445
1
Parent(s): c1f4cde
Initial commit
Browse files- added_tokens.json +3 -0
- config.json +41 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- special_tokens_map.json +15 -0
- spm.model +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +58 -0
- trainer_state.json +2585 -0
- training_args.bin +3 -0
added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[MASK]": 128000
|
| 3 |
+
}
|
config.json
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "results/checkpoint-25000",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"DebertaV2ForSequenceClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.1,
|
| 9 |
+
"hidden_size": 768,
|
| 10 |
+
"id2label": {
|
| 11 |
+
"0": "LABEL_0"
|
| 12 |
+
},
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 3072,
|
| 15 |
+
"label2id": {
|
| 16 |
+
"LABEL_0": 0
|
| 17 |
+
},
|
| 18 |
+
"layer_norm_eps": 1e-07,
|
| 19 |
+
"max_position_embeddings": 512,
|
| 20 |
+
"max_relative_positions": -1,
|
| 21 |
+
"model_type": "deberta-v2",
|
| 22 |
+
"norm_rel_ebd": "layer_norm",
|
| 23 |
+
"num_attention_heads": 12,
|
| 24 |
+
"num_hidden_layers": 12,
|
| 25 |
+
"pad_token_id": 0,
|
| 26 |
+
"pooler_dropout": 0,
|
| 27 |
+
"pooler_hidden_act": "gelu",
|
| 28 |
+
"pooler_hidden_size": 768,
|
| 29 |
+
"pos_att_type": [
|
| 30 |
+
"p2c",
|
| 31 |
+
"c2p"
|
| 32 |
+
],
|
| 33 |
+
"position_biased_input": true,
|
| 34 |
+
"position_buckets": 256,
|
| 35 |
+
"relative_attention": true,
|
| 36 |
+
"share_att_key": true,
|
| 37 |
+
"torch_dtype": "float32",
|
| 38 |
+
"transformers_version": "4.42.3",
|
| 39 |
+
"type_vocab_size": 0,
|
| 40 |
+
"vocab_size": 128100
|
| 41 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ed478963b7447a8f3e4bb2990487ccbc13a3b52dbb5b0ddd3a86b6c846c6ad24
|
| 3 |
+
size 739289172
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:40d6e18ae8fbc3711f3788dccb2405cadabf3dfbf24c412b2c10ed2ed6f6e3e1
|
| 3 |
+
size 1478698874
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0dc060a15caa2d6730d0bb07986669c84d6593d43c6d10f8bebbb5cd803f1c3c
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5aa07b3c00281fb18434b93ce2b2a8e51a3da90ed296ea2c862ed48d45752cbf
|
| 3 |
+
size 1064
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "[CLS]",
|
| 3 |
+
"cls_token": "[CLS]",
|
| 4 |
+
"eos_token": "[SEP]",
|
| 5 |
+
"mask_token": "[MASK]",
|
| 6 |
+
"pad_token": "[PAD]",
|
| 7 |
+
"sep_token": "[SEP]",
|
| 8 |
+
"unk_token": {
|
| 9 |
+
"content": "[UNK]",
|
| 10 |
+
"lstrip": false,
|
| 11 |
+
"normalized": true,
|
| 12 |
+
"rstrip": false,
|
| 13 |
+
"single_word": false
|
| 14 |
+
}
|
| 15 |
+
}
|
spm.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c679fbf93643d19aab7ee10c0b99e460bdbc02fedf34b92b05af343b4af586fd
|
| 3 |
+
size 2464616
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "[PAD]",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "[CLS]",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "[SEP]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "[UNK]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": true,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"128000": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"bos_token": "[CLS]",
|
| 45 |
+
"clean_up_tokenization_spaces": true,
|
| 46 |
+
"cls_token": "[CLS]",
|
| 47 |
+
"do_lower_case": false,
|
| 48 |
+
"eos_token": "[SEP]",
|
| 49 |
+
"mask_token": "[MASK]",
|
| 50 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 51 |
+
"pad_token": "[PAD]",
|
| 52 |
+
"sep_token": "[SEP]",
|
| 53 |
+
"sp_model_kwargs": {},
|
| 54 |
+
"split_by_punct": false,
|
| 55 |
+
"tokenizer_class": "DebertaV2Tokenizer",
|
| 56 |
+
"unk_token": "[UNK]",
|
| 57 |
+
"vocab_type": "spm"
|
| 58 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2585 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 40.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 160320,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.124750499001996,
|
| 13 |
+
"grad_norm": 13.176804542541504,
|
| 14 |
+
"learning_rate": 1.9937624750499e-06,
|
| 15 |
+
"loss": 0.2137,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.249500998003992,
|
| 20 |
+
"grad_norm": 52.68854904174805,
|
| 21 |
+
"learning_rate": 1.9875249500998005e-06,
|
| 22 |
+
"loss": 0.2463,
|
| 23 |
+
"step": 1000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.37425149700598803,
|
| 27 |
+
"grad_norm": 9.197150230407715,
|
| 28 |
+
"learning_rate": 1.9812874251497004e-06,
|
| 29 |
+
"loss": 0.2316,
|
| 30 |
+
"step": 1500
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.499001996007984,
|
| 34 |
+
"grad_norm": 23.94010353088379,
|
| 35 |
+
"learning_rate": 1.9750499001996007e-06,
|
| 36 |
+
"loss": 0.2095,
|
| 37 |
+
"step": 2000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.6237524950099801,
|
| 41 |
+
"grad_norm": 25.69223976135254,
|
| 42 |
+
"learning_rate": 1.968812375249501e-06,
|
| 43 |
+
"loss": 0.2102,
|
| 44 |
+
"step": 2500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.7485029940119761,
|
| 48 |
+
"grad_norm": 14.870789527893066,
|
| 49 |
+
"learning_rate": 1.9625748502994013e-06,
|
| 50 |
+
"loss": 0.2335,
|
| 51 |
+
"step": 3000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.873253493013972,
|
| 55 |
+
"grad_norm": 19.752464294433594,
|
| 56 |
+
"learning_rate": 1.9563373253493016e-06,
|
| 57 |
+
"loss": 0.2065,
|
| 58 |
+
"step": 3500
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.998003992015968,
|
| 62 |
+
"grad_norm": 7.356762409210205,
|
| 63 |
+
"learning_rate": 1.9500998003992014e-06,
|
| 64 |
+
"loss": 0.215,
|
| 65 |
+
"step": 4000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.0,
|
| 69 |
+
"eval_loss": 0.4418031871318817,
|
| 70 |
+
"eval_runtime": 50.9423,
|
| 71 |
+
"eval_samples_per_second": 62.934,
|
| 72 |
+
"eval_steps_per_second": 15.743,
|
| 73 |
+
"step": 4008
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 1.122754491017964,
|
| 77 |
+
"grad_norm": 1.049210786819458,
|
| 78 |
+
"learning_rate": 1.9438622754491017e-06,
|
| 79 |
+
"loss": 0.1786,
|
| 80 |
+
"step": 4500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 1.24750499001996,
|
| 84 |
+
"grad_norm": 33.95945358276367,
|
| 85 |
+
"learning_rate": 1.937624750499002e-06,
|
| 86 |
+
"loss": 0.2107,
|
| 87 |
+
"step": 5000
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 1.372255489021956,
|
| 91 |
+
"grad_norm": 3.9420273303985596,
|
| 92 |
+
"learning_rate": 1.931387225548902e-06,
|
| 93 |
+
"loss": 0.1877,
|
| 94 |
+
"step": 5500
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 1.4970059880239521,
|
| 98 |
+
"grad_norm": 15.459404945373535,
|
| 99 |
+
"learning_rate": 1.925149700598802e-06,
|
| 100 |
+
"loss": 0.1808,
|
| 101 |
+
"step": 6000
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 1.621756487025948,
|
| 105 |
+
"grad_norm": 0.35231631994247437,
|
| 106 |
+
"learning_rate": 1.9189121756487025e-06,
|
| 107 |
+
"loss": 0.1842,
|
| 108 |
+
"step": 6500
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 1.746506986027944,
|
| 112 |
+
"grad_norm": 22.17848014831543,
|
| 113 |
+
"learning_rate": 1.9126746506986028e-06,
|
| 114 |
+
"loss": 0.2165,
|
| 115 |
+
"step": 7000
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 1.8712574850299402,
|
| 119 |
+
"grad_norm": 31.21565055847168,
|
| 120 |
+
"learning_rate": 1.906437125748503e-06,
|
| 121 |
+
"loss": 0.1879,
|
| 122 |
+
"step": 7500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 1.996007984031936,
|
| 126 |
+
"grad_norm": 1.7563763856887817,
|
| 127 |
+
"learning_rate": 1.9001996007984032e-06,
|
| 128 |
+
"loss": 0.1913,
|
| 129 |
+
"step": 8000
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 2.0,
|
| 133 |
+
"eval_loss": 0.4956786632537842,
|
| 134 |
+
"eval_runtime": 48.4339,
|
| 135 |
+
"eval_samples_per_second": 66.193,
|
| 136 |
+
"eval_steps_per_second": 16.559,
|
| 137 |
+
"step": 8016
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.1207584830339323,
|
| 141 |
+
"grad_norm": 0.0910625234246254,
|
| 142 |
+
"learning_rate": 1.8939620758483032e-06,
|
| 143 |
+
"loss": 0.1712,
|
| 144 |
+
"step": 8500
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 2.245508982035928,
|
| 148 |
+
"grad_norm": 30.4615421295166,
|
| 149 |
+
"learning_rate": 1.8877245508982035e-06,
|
| 150 |
+
"loss": 0.1579,
|
| 151 |
+
"step": 9000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.370259481037924,
|
| 155 |
+
"grad_norm": 29.169662475585938,
|
| 156 |
+
"learning_rate": 1.8814870259481036e-06,
|
| 157 |
+
"loss": 0.1701,
|
| 158 |
+
"step": 9500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.49500998003992,
|
| 162 |
+
"grad_norm": 0.9950535893440247,
|
| 163 |
+
"learning_rate": 1.875249500998004e-06,
|
| 164 |
+
"loss": 0.1717,
|
| 165 |
+
"step": 10000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.6197604790419162,
|
| 169 |
+
"grad_norm": 0.30978772044181824,
|
| 170 |
+
"learning_rate": 1.8690119760479042e-06,
|
| 171 |
+
"loss": 0.1778,
|
| 172 |
+
"step": 10500
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 2.744510978043912,
|
| 176 |
+
"grad_norm": 1.4617693424224854,
|
| 177 |
+
"learning_rate": 1.8627744510978043e-06,
|
| 178 |
+
"loss": 0.1772,
|
| 179 |
+
"step": 11000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 2.8692614770459084,
|
| 183 |
+
"grad_norm": 13.257425308227539,
|
| 184 |
+
"learning_rate": 1.8565369261477044e-06,
|
| 185 |
+
"loss": 0.1697,
|
| 186 |
+
"step": 11500
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 2.9940119760479043,
|
| 190 |
+
"grad_norm": 11.522214889526367,
|
| 191 |
+
"learning_rate": 1.8502994011976047e-06,
|
| 192 |
+
"loss": 0.1651,
|
| 193 |
+
"step": 12000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 3.0,
|
| 197 |
+
"eval_loss": 0.4591982960700989,
|
| 198 |
+
"eval_runtime": 48.8032,
|
| 199 |
+
"eval_samples_per_second": 65.692,
|
| 200 |
+
"eval_steps_per_second": 16.433,
|
| 201 |
+
"step": 12024
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 3.1187624750499,
|
| 205 |
+
"grad_norm": 20.58974266052246,
|
| 206 |
+
"learning_rate": 1.844061876247505e-06,
|
| 207 |
+
"loss": 0.1607,
|
| 208 |
+
"step": 12500
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 3.243512974051896,
|
| 212 |
+
"grad_norm": 54.52241516113281,
|
| 213 |
+
"learning_rate": 1.8378243512974053e-06,
|
| 214 |
+
"loss": 0.1527,
|
| 215 |
+
"step": 13000
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 3.3682634730538923,
|
| 219 |
+
"grad_norm": 12.846843719482422,
|
| 220 |
+
"learning_rate": 1.8315868263473054e-06,
|
| 221 |
+
"loss": 0.1484,
|
| 222 |
+
"step": 13500
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 3.493013972055888,
|
| 226 |
+
"grad_norm": 0.6479830145835876,
|
| 227 |
+
"learning_rate": 1.8253493013972054e-06,
|
| 228 |
+
"loss": 0.1621,
|
| 229 |
+
"step": 14000
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 3.6177644710578845,
|
| 233 |
+
"grad_norm": 0.7256312370300293,
|
| 234 |
+
"learning_rate": 1.8191117764471057e-06,
|
| 235 |
+
"loss": 0.1428,
|
| 236 |
+
"step": 14500
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 3.7425149700598803,
|
| 240 |
+
"grad_norm": 12.274479866027832,
|
| 241 |
+
"learning_rate": 1.8128742514970058e-06,
|
| 242 |
+
"loss": 0.1433,
|
| 243 |
+
"step": 15000
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 3.867265469061876,
|
| 247 |
+
"grad_norm": 35.40715408325195,
|
| 248 |
+
"learning_rate": 1.8066367265469061e-06,
|
| 249 |
+
"loss": 0.161,
|
| 250 |
+
"step": 15500
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 3.992015968063872,
|
| 254 |
+
"grad_norm": 0.78450608253479,
|
| 255 |
+
"learning_rate": 1.8003992015968064e-06,
|
| 256 |
+
"loss": 0.171,
|
| 257 |
+
"step": 16000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 4.0,
|
| 261 |
+
"eval_loss": 0.4495457410812378,
|
| 262 |
+
"eval_runtime": 49.2624,
|
| 263 |
+
"eval_samples_per_second": 65.08,
|
| 264 |
+
"eval_steps_per_second": 16.28,
|
| 265 |
+
"step": 16032
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 4.116766467065868,
|
| 269 |
+
"grad_norm": 0.06905636936426163,
|
| 270 |
+
"learning_rate": 1.7941616766467065e-06,
|
| 271 |
+
"loss": 0.1475,
|
| 272 |
+
"step": 16500
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 4.241516966067865,
|
| 276 |
+
"grad_norm": 34.77931213378906,
|
| 277 |
+
"learning_rate": 1.7879241516966066e-06,
|
| 278 |
+
"loss": 0.1365,
|
| 279 |
+
"step": 17000
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 4.3662674650698605,
|
| 283 |
+
"grad_norm": 0.5809102058410645,
|
| 284 |
+
"learning_rate": 1.7816866267465069e-06,
|
| 285 |
+
"loss": 0.1366,
|
| 286 |
+
"step": 17500
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 4.491017964071856,
|
| 290 |
+
"grad_norm": 66.70156860351562,
|
| 291 |
+
"learning_rate": 1.775449101796407e-06,
|
| 292 |
+
"loss": 0.1526,
|
| 293 |
+
"step": 18000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 4.615768463073852,
|
| 297 |
+
"grad_norm": 29.423938751220703,
|
| 298 |
+
"learning_rate": 1.7692115768463075e-06,
|
| 299 |
+
"loss": 0.135,
|
| 300 |
+
"step": 18500
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 4.740518962075848,
|
| 304 |
+
"grad_norm": 0.48827868700027466,
|
| 305 |
+
"learning_rate": 1.7629740518962075e-06,
|
| 306 |
+
"loss": 0.1444,
|
| 307 |
+
"step": 19000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 4.865269461077844,
|
| 311 |
+
"grad_norm": 8.966581344604492,
|
| 312 |
+
"learning_rate": 1.7567365269461076e-06,
|
| 313 |
+
"loss": 0.1295,
|
| 314 |
+
"step": 19500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 4.99001996007984,
|
| 318 |
+
"grad_norm": 3.6332414150238037,
|
| 319 |
+
"learning_rate": 1.750499001996008e-06,
|
| 320 |
+
"loss": 0.1407,
|
| 321 |
+
"step": 20000
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 5.0,
|
| 325 |
+
"eval_loss": 0.5054113268852234,
|
| 326 |
+
"eval_runtime": 46.022,
|
| 327 |
+
"eval_samples_per_second": 69.662,
|
| 328 |
+
"eval_steps_per_second": 17.426,
|
| 329 |
+
"step": 20040
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 5.114770459081837,
|
| 333 |
+
"grad_norm": 26.99722671508789,
|
| 334 |
+
"learning_rate": 1.744261477045908e-06,
|
| 335 |
+
"loss": 0.1307,
|
| 336 |
+
"step": 20500
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 5.2395209580838324,
|
| 340 |
+
"grad_norm": 0.7371481657028198,
|
| 341 |
+
"learning_rate": 1.7380239520958083e-06,
|
| 342 |
+
"loss": 0.1153,
|
| 343 |
+
"step": 21000
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 5.364271457085828,
|
| 347 |
+
"grad_norm": 0.3232800364494324,
|
| 348 |
+
"learning_rate": 1.7317864271457086e-06,
|
| 349 |
+
"loss": 0.1154,
|
| 350 |
+
"step": 21500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 5.489021956087824,
|
| 354 |
+
"grad_norm": 1.8309438228607178,
|
| 355 |
+
"learning_rate": 1.7255489021956087e-06,
|
| 356 |
+
"loss": 0.1331,
|
| 357 |
+
"step": 22000
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 5.61377245508982,
|
| 361 |
+
"grad_norm": 0.4226222038269043,
|
| 362 |
+
"learning_rate": 1.719311377245509e-06,
|
| 363 |
+
"loss": 0.1206,
|
| 364 |
+
"step": 22500
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 5.738522954091817,
|
| 368 |
+
"grad_norm": 1.4337540864944458,
|
| 369 |
+
"learning_rate": 1.713073852295409e-06,
|
| 370 |
+
"loss": 0.13,
|
| 371 |
+
"step": 23000
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 5.863273453093813,
|
| 375 |
+
"grad_norm": 47.5312614440918,
|
| 376 |
+
"learning_rate": 1.7068363273453091e-06,
|
| 377 |
+
"loss": 0.1285,
|
| 378 |
+
"step": 23500
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 5.9880239520958085,
|
| 382 |
+
"grad_norm": 1.092816710472107,
|
| 383 |
+
"learning_rate": 1.7005988023952097e-06,
|
| 384 |
+
"loss": 0.1412,
|
| 385 |
+
"step": 24000
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 6.0,
|
| 389 |
+
"eval_loss": 0.4939550459384918,
|
| 390 |
+
"eval_runtime": 45.0298,
|
| 391 |
+
"eval_samples_per_second": 71.197,
|
| 392 |
+
"eval_steps_per_second": 17.81,
|
| 393 |
+
"step": 24048
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 6.112774451097804,
|
| 397 |
+
"grad_norm": 0.03936842083930969,
|
| 398 |
+
"learning_rate": 1.6943612774451097e-06,
|
| 399 |
+
"loss": 0.1134,
|
| 400 |
+
"step": 24500
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 6.2375249500998,
|
| 404 |
+
"grad_norm": 3.047616481781006,
|
| 405 |
+
"learning_rate": 1.6881237524950098e-06,
|
| 406 |
+
"loss": 0.1066,
|
| 407 |
+
"step": 25000
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 6.362275449101796,
|
| 411 |
+
"grad_norm": 16.7564754486084,
|
| 412 |
+
"learning_rate": 1.6818862275449101e-06,
|
| 413 |
+
"loss": 0.1615,
|
| 414 |
+
"step": 25500
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 6.487025948103792,
|
| 418 |
+
"grad_norm": 21.36778450012207,
|
| 419 |
+
"learning_rate": 1.6756487025948102e-06,
|
| 420 |
+
"loss": 0.1645,
|
| 421 |
+
"step": 26000
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 6.611776447105789,
|
| 425 |
+
"grad_norm": 78.45208740234375,
|
| 426 |
+
"learning_rate": 1.6694111776447105e-06,
|
| 427 |
+
"loss": 0.1675,
|
| 428 |
+
"step": 26500
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 6.736526946107785,
|
| 432 |
+
"grad_norm": 7.212148666381836,
|
| 433 |
+
"learning_rate": 1.6631736526946108e-06,
|
| 434 |
+
"loss": 0.146,
|
| 435 |
+
"step": 27000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 6.86127744510978,
|
| 439 |
+
"grad_norm": 9.503207206726074,
|
| 440 |
+
"learning_rate": 1.6569361277445109e-06,
|
| 441 |
+
"loss": 0.1606,
|
| 442 |
+
"step": 27500
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 6.986027944111776,
|
| 446 |
+
"grad_norm": 0.4464740753173828,
|
| 447 |
+
"learning_rate": 1.6506986027944112e-06,
|
| 448 |
+
"loss": 0.1429,
|
| 449 |
+
"step": 28000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 7.0,
|
| 453 |
+
"eval_loss": 0.4717544615268707,
|
| 454 |
+
"eval_runtime": 47.386,
|
| 455 |
+
"eval_samples_per_second": 67.657,
|
| 456 |
+
"eval_steps_per_second": 16.925,
|
| 457 |
+
"step": 28056
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 7.110778443113772,
|
| 461 |
+
"grad_norm": 0.42686018347740173,
|
| 462 |
+
"learning_rate": 1.6444610778443113e-06,
|
| 463 |
+
"loss": 0.1207,
|
| 464 |
+
"step": 28500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 7.235528942115769,
|
| 468 |
+
"grad_norm": 24.92848014831543,
|
| 469 |
+
"learning_rate": 1.6382235528942113e-06,
|
| 470 |
+
"loss": 0.1351,
|
| 471 |
+
"step": 29000
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 7.360279441117765,
|
| 475 |
+
"grad_norm": 7.397327423095703,
|
| 476 |
+
"learning_rate": 1.6319860279441118e-06,
|
| 477 |
+
"loss": 0.1543,
|
| 478 |
+
"step": 29500
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 7.485029940119761,
|
| 482 |
+
"grad_norm": 0.43539106845855713,
|
| 483 |
+
"learning_rate": 1.625748502994012e-06,
|
| 484 |
+
"loss": 0.1494,
|
| 485 |
+
"step": 30000
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 7.6097804391217565,
|
| 489 |
+
"grad_norm": 14.456055641174316,
|
| 490 |
+
"learning_rate": 1.619510978043912e-06,
|
| 491 |
+
"loss": 0.1419,
|
| 492 |
+
"step": 30500
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 7.734530938123752,
|
| 496 |
+
"grad_norm": 9.563997268676758,
|
| 497 |
+
"learning_rate": 1.6132734530938123e-06,
|
| 498 |
+
"loss": 0.1357,
|
| 499 |
+
"step": 31000
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 7.859281437125748,
|
| 503 |
+
"grad_norm": 1.7568217515945435,
|
| 504 |
+
"learning_rate": 1.6070359281437124e-06,
|
| 505 |
+
"loss": 0.1369,
|
| 506 |
+
"step": 31500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 7.984031936127744,
|
| 510 |
+
"grad_norm": 2.780186653137207,
|
| 511 |
+
"learning_rate": 1.600798403193613e-06,
|
| 512 |
+
"loss": 0.1451,
|
| 513 |
+
"step": 32000
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 8.0,
|
| 517 |
+
"eval_loss": 0.45947006344795227,
|
| 518 |
+
"eval_runtime": 44.2941,
|
| 519 |
+
"eval_samples_per_second": 72.38,
|
| 520 |
+
"eval_steps_per_second": 18.106,
|
| 521 |
+
"step": 32064
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 8.10878243512974,
|
| 525 |
+
"grad_norm": 10.451217651367188,
|
| 526 |
+
"learning_rate": 1.594560878243513e-06,
|
| 527 |
+
"loss": 0.1136,
|
| 528 |
+
"step": 32500
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 8.233532934131736,
|
| 532 |
+
"grad_norm": 0.18200552463531494,
|
| 533 |
+
"learning_rate": 1.588323353293413e-06,
|
| 534 |
+
"loss": 0.1259,
|
| 535 |
+
"step": 33000
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 8.358283433133732,
|
| 539 |
+
"grad_norm": 1.9428528547286987,
|
| 540 |
+
"learning_rate": 1.5820858283433134e-06,
|
| 541 |
+
"loss": 0.1279,
|
| 542 |
+
"step": 33500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 8.48303393213573,
|
| 546 |
+
"grad_norm": 1.7016535997390747,
|
| 547 |
+
"learning_rate": 1.5758483033932135e-06,
|
| 548 |
+
"loss": 0.1231,
|
| 549 |
+
"step": 34000
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 8.607784431137725,
|
| 553 |
+
"grad_norm": 0.7158037424087524,
|
| 554 |
+
"learning_rate": 1.5696107784431135e-06,
|
| 555 |
+
"loss": 0.1446,
|
| 556 |
+
"step": 34500
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 8.732534930139721,
|
| 560 |
+
"grad_norm": 0.4712078273296356,
|
| 561 |
+
"learning_rate": 1.563373253493014e-06,
|
| 562 |
+
"loss": 0.1344,
|
| 563 |
+
"step": 35000
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 8.857285429141717,
|
| 567 |
+
"grad_norm": 24.5105037689209,
|
| 568 |
+
"learning_rate": 1.5571357285429141e-06,
|
| 569 |
+
"loss": 0.1331,
|
| 570 |
+
"step": 35500
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 8.982035928143713,
|
| 574 |
+
"grad_norm": 27.750621795654297,
|
| 575 |
+
"learning_rate": 1.5508982035928142e-06,
|
| 576 |
+
"loss": 0.1296,
|
| 577 |
+
"step": 36000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 9.0,
|
| 581 |
+
"eval_loss": 0.47351646423339844,
|
| 582 |
+
"eval_runtime": 41.3902,
|
| 583 |
+
"eval_samples_per_second": 77.458,
|
| 584 |
+
"eval_steps_per_second": 19.377,
|
| 585 |
+
"step": 36072
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 9.106786427145709,
|
| 589 |
+
"grad_norm": 28.095134735107422,
|
| 590 |
+
"learning_rate": 1.5446606786427145e-06,
|
| 591 |
+
"loss": 0.119,
|
| 592 |
+
"step": 36500
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 9.231536926147704,
|
| 596 |
+
"grad_norm": 0.07204411178827286,
|
| 597 |
+
"learning_rate": 1.5384231536926146e-06,
|
| 598 |
+
"loss": 0.1098,
|
| 599 |
+
"step": 37000
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 9.3562874251497,
|
| 603 |
+
"grad_norm": 0.2767297327518463,
|
| 604 |
+
"learning_rate": 1.532185628742515e-06,
|
| 605 |
+
"loss": 0.1224,
|
| 606 |
+
"step": 37500
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 9.481037924151696,
|
| 610 |
+
"grad_norm": 0.14060889184474945,
|
| 611 |
+
"learning_rate": 1.5259481037924152e-06,
|
| 612 |
+
"loss": 0.1247,
|
| 613 |
+
"step": 38000
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 9.605788423153692,
|
| 617 |
+
"grad_norm": 32.673011779785156,
|
| 618 |
+
"learning_rate": 1.5197105788423153e-06,
|
| 619 |
+
"loss": 0.122,
|
| 620 |
+
"step": 38500
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 9.730538922155688,
|
| 624 |
+
"grad_norm": 0.21247480809688568,
|
| 625 |
+
"learning_rate": 1.5134730538922156e-06,
|
| 626 |
+
"loss": 0.1233,
|
| 627 |
+
"step": 39000
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 9.855289421157684,
|
| 631 |
+
"grad_norm": 0.4861377775669098,
|
| 632 |
+
"learning_rate": 1.5072355289421156e-06,
|
| 633 |
+
"loss": 0.1286,
|
| 634 |
+
"step": 39500
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 9.980039920159681,
|
| 638 |
+
"grad_norm": 11.489697456359863,
|
| 639 |
+
"learning_rate": 1.5009980039920157e-06,
|
| 640 |
+
"loss": 0.1203,
|
| 641 |
+
"step": 40000
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 10.0,
|
| 645 |
+
"eval_loss": 0.44174808263778687,
|
| 646 |
+
"eval_runtime": 40.4714,
|
| 647 |
+
"eval_samples_per_second": 79.216,
|
| 648 |
+
"eval_steps_per_second": 19.816,
|
| 649 |
+
"step": 40080
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 10.104790419161677,
|
| 653 |
+
"grad_norm": 0.06284382939338684,
|
| 654 |
+
"learning_rate": 1.4947604790419162e-06,
|
| 655 |
+
"loss": 0.1176,
|
| 656 |
+
"step": 40500
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 10.229540918163673,
|
| 660 |
+
"grad_norm": 0.8282334804534912,
|
| 661 |
+
"learning_rate": 1.4885229540918163e-06,
|
| 662 |
+
"loss": 0.1133,
|
| 663 |
+
"step": 41000
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 10.354291417165669,
|
| 667 |
+
"grad_norm": 0.675163984298706,
|
| 668 |
+
"learning_rate": 1.4822854291417164e-06,
|
| 669 |
+
"loss": 0.0977,
|
| 670 |
+
"step": 41500
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 10.479041916167665,
|
| 674 |
+
"grad_norm": 6.970102310180664,
|
| 675 |
+
"learning_rate": 1.4760479041916167e-06,
|
| 676 |
+
"loss": 0.1113,
|
| 677 |
+
"step": 42000
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 10.60379241516966,
|
| 681 |
+
"grad_norm": 8.85517406463623,
|
| 682 |
+
"learning_rate": 1.4698103792415168e-06,
|
| 683 |
+
"loss": 0.1164,
|
| 684 |
+
"step": 42500
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 10.728542914171657,
|
| 688 |
+
"grad_norm": 0.9282238483428955,
|
| 689 |
+
"learning_rate": 1.4635728542914173e-06,
|
| 690 |
+
"loss": 0.1167,
|
| 691 |
+
"step": 43000
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 10.853293413173652,
|
| 695 |
+
"grad_norm": 9.984148979187012,
|
| 696 |
+
"learning_rate": 1.4573353293413174e-06,
|
| 697 |
+
"loss": 0.1261,
|
| 698 |
+
"step": 43500
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 10.978043912175648,
|
| 702 |
+
"grad_norm": 0.20773719251155853,
|
| 703 |
+
"learning_rate": 1.4510978043912175e-06,
|
| 704 |
+
"loss": 0.1132,
|
| 705 |
+
"step": 44000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 11.0,
|
| 709 |
+
"eval_loss": 0.49900639057159424,
|
| 710 |
+
"eval_runtime": 43.241,
|
| 711 |
+
"eval_samples_per_second": 74.143,
|
| 712 |
+
"eval_steps_per_second": 18.547,
|
| 713 |
+
"step": 44088
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 11.102794411177644,
|
| 717 |
+
"grad_norm": 12.603593826293945,
|
| 718 |
+
"learning_rate": 1.4448602794411178e-06,
|
| 719 |
+
"loss": 0.1061,
|
| 720 |
+
"step": 44500
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 11.22754491017964,
|
| 724 |
+
"grad_norm": 51.32432174682617,
|
| 725 |
+
"learning_rate": 1.4386227544910178e-06,
|
| 726 |
+
"loss": 0.1079,
|
| 727 |
+
"step": 45000
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 11.352295409181636,
|
| 731 |
+
"grad_norm": 10.22624397277832,
|
| 732 |
+
"learning_rate": 1.432385229540918e-06,
|
| 733 |
+
"loss": 0.1166,
|
| 734 |
+
"step": 45500
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 11.477045908183634,
|
| 738 |
+
"grad_norm": 11.041003227233887,
|
| 739 |
+
"learning_rate": 1.4261477045908184e-06,
|
| 740 |
+
"loss": 0.105,
|
| 741 |
+
"step": 46000
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 11.60179640718563,
|
| 745 |
+
"grad_norm": 35.79409408569336,
|
| 746 |
+
"learning_rate": 1.4199101796407185e-06,
|
| 747 |
+
"loss": 0.1124,
|
| 748 |
+
"step": 46500
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 11.726546906187625,
|
| 752 |
+
"grad_norm": 0.18676696717739105,
|
| 753 |
+
"learning_rate": 1.4136726546906188e-06,
|
| 754 |
+
"loss": 0.0928,
|
| 755 |
+
"step": 47000
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 11.851297405189621,
|
| 759 |
+
"grad_norm": 1.4925884008407593,
|
| 760 |
+
"learning_rate": 1.4074351297405189e-06,
|
| 761 |
+
"loss": 0.1098,
|
| 762 |
+
"step": 47500
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 11.976047904191617,
|
| 766 |
+
"grad_norm": 0.32953181862831116,
|
| 767 |
+
"learning_rate": 1.401197604790419e-06,
|
| 768 |
+
"loss": 0.1117,
|
| 769 |
+
"step": 48000
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 12.0,
|
| 773 |
+
"eval_loss": 0.4872562289237976,
|
| 774 |
+
"eval_runtime": 41.6872,
|
| 775 |
+
"eval_samples_per_second": 76.906,
|
| 776 |
+
"eval_steps_per_second": 19.239,
|
| 777 |
+
"step": 48096
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 12.100798403193613,
|
| 781 |
+
"grad_norm": 0.027937307953834534,
|
| 782 |
+
"learning_rate": 1.3949600798403195e-06,
|
| 783 |
+
"loss": 0.0992,
|
| 784 |
+
"step": 48500
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 12.225548902195609,
|
| 788 |
+
"grad_norm": 0.29068148136138916,
|
| 789 |
+
"learning_rate": 1.3887225548902196e-06,
|
| 790 |
+
"loss": 0.0921,
|
| 791 |
+
"step": 49000
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 12.350299401197605,
|
| 795 |
+
"grad_norm": 0.127395898103714,
|
| 796 |
+
"learning_rate": 1.3824850299401197e-06,
|
| 797 |
+
"loss": 0.0933,
|
| 798 |
+
"step": 49500
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 12.4750499001996,
|
| 802 |
+
"grad_norm": 0.09435238689184189,
|
| 803 |
+
"learning_rate": 1.37624750499002e-06,
|
| 804 |
+
"loss": 0.116,
|
| 805 |
+
"step": 50000
|
| 806 |
+
},
|
| 807 |
+
{
|
| 808 |
+
"epoch": 12.599800399201596,
|
| 809 |
+
"grad_norm": 39.19729232788086,
|
| 810 |
+
"learning_rate": 1.37000998003992e-06,
|
| 811 |
+
"loss": 0.1052,
|
| 812 |
+
"step": 50500
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 12.724550898203592,
|
| 816 |
+
"grad_norm": 0.28930047154426575,
|
| 817 |
+
"learning_rate": 1.3637724550898201e-06,
|
| 818 |
+
"loss": 0.1038,
|
| 819 |
+
"step": 51000
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 12.849301397205588,
|
| 823 |
+
"grad_norm": 0.15510033071041107,
|
| 824 |
+
"learning_rate": 1.3575349301397206e-06,
|
| 825 |
+
"loss": 0.0983,
|
| 826 |
+
"step": 51500
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 12.974051896207584,
|
| 830 |
+
"grad_norm": 81.58076477050781,
|
| 831 |
+
"learning_rate": 1.3512974051896207e-06,
|
| 832 |
+
"loss": 0.1117,
|
| 833 |
+
"step": 52000
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 13.0,
|
| 837 |
+
"eval_loss": 0.45387548208236694,
|
| 838 |
+
"eval_runtime": 43.4012,
|
| 839 |
+
"eval_samples_per_second": 73.869,
|
| 840 |
+
"eval_steps_per_second": 18.479,
|
| 841 |
+
"step": 52104
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 13.098802395209582,
|
| 845 |
+
"grad_norm": 4.060844421386719,
|
| 846 |
+
"learning_rate": 1.345059880239521e-06,
|
| 847 |
+
"loss": 0.0983,
|
| 848 |
+
"step": 52500
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 13.223552894211577,
|
| 852 |
+
"grad_norm": 33.315853118896484,
|
| 853 |
+
"learning_rate": 1.338822355289421e-06,
|
| 854 |
+
"loss": 0.0941,
|
| 855 |
+
"step": 53000
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 13.348303393213573,
|
| 859 |
+
"grad_norm": 0.1183587834239006,
|
| 860 |
+
"learning_rate": 1.3325848303393212e-06,
|
| 861 |
+
"loss": 0.0973,
|
| 862 |
+
"step": 53500
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 13.47305389221557,
|
| 866 |
+
"grad_norm": 40.30908966064453,
|
| 867 |
+
"learning_rate": 1.3263473053892215e-06,
|
| 868 |
+
"loss": 0.0871,
|
| 869 |
+
"step": 54000
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 13.597804391217565,
|
| 873 |
+
"grad_norm": 0.619777262210846,
|
| 874 |
+
"learning_rate": 1.3201097804391218e-06,
|
| 875 |
+
"loss": 0.1001,
|
| 876 |
+
"step": 54500
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 13.72255489021956,
|
| 880 |
+
"grad_norm": 0.2705942392349243,
|
| 881 |
+
"learning_rate": 1.3138722554890218e-06,
|
| 882 |
+
"loss": 0.0983,
|
| 883 |
+
"step": 55000
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 13.847305389221557,
|
| 887 |
+
"grad_norm": 6.151524066925049,
|
| 888 |
+
"learning_rate": 1.3076347305389221e-06,
|
| 889 |
+
"loss": 0.0793,
|
| 890 |
+
"step": 55500
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 13.972055888223553,
|
| 894 |
+
"grad_norm": 2.340573787689209,
|
| 895 |
+
"learning_rate": 1.3013972055888222e-06,
|
| 896 |
+
"loss": 0.099,
|
| 897 |
+
"step": 56000
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 14.0,
|
| 901 |
+
"eval_loss": 0.47363531589508057,
|
| 902 |
+
"eval_runtime": 42.3279,
|
| 903 |
+
"eval_samples_per_second": 75.742,
|
| 904 |
+
"eval_steps_per_second": 18.947,
|
| 905 |
+
"step": 56112
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 14.096806387225548,
|
| 909 |
+
"grad_norm": 2.052589178085327,
|
| 910 |
+
"learning_rate": 1.2951596806387225e-06,
|
| 911 |
+
"loss": 0.0875,
|
| 912 |
+
"step": 56500
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 14.221556886227544,
|
| 916 |
+
"grad_norm": 1.2925941944122314,
|
| 917 |
+
"learning_rate": 1.2889221556886228e-06,
|
| 918 |
+
"loss": 0.0812,
|
| 919 |
+
"step": 57000
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 14.34630738522954,
|
| 923 |
+
"grad_norm": 0.062304213643074036,
|
| 924 |
+
"learning_rate": 1.282684630738523e-06,
|
| 925 |
+
"loss": 0.1017,
|
| 926 |
+
"step": 57500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 14.471057884231538,
|
| 930 |
+
"grad_norm": 0.1741693764925003,
|
| 931 |
+
"learning_rate": 1.2764471057884232e-06,
|
| 932 |
+
"loss": 0.0836,
|
| 933 |
+
"step": 58000
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 14.595808383233534,
|
| 937 |
+
"grad_norm": 0.6444254517555237,
|
| 938 |
+
"learning_rate": 1.2702095808383233e-06,
|
| 939 |
+
"loss": 0.0804,
|
| 940 |
+
"step": 58500
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 14.72055888223553,
|
| 944 |
+
"grad_norm": 2.0034759044647217,
|
| 945 |
+
"learning_rate": 1.2639720558882234e-06,
|
| 946 |
+
"loss": 0.0953,
|
| 947 |
+
"step": 59000
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 14.845309381237525,
|
| 951 |
+
"grad_norm": 52.82548522949219,
|
| 952 |
+
"learning_rate": 1.2577345309381237e-06,
|
| 953 |
+
"loss": 0.0996,
|
| 954 |
+
"step": 59500
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 14.970059880239521,
|
| 958 |
+
"grad_norm": 6.955111503601074,
|
| 959 |
+
"learning_rate": 1.251497005988024e-06,
|
| 960 |
+
"loss": 0.0857,
|
| 961 |
+
"step": 60000
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 15.0,
|
| 965 |
+
"eval_loss": 0.45942702889442444,
|
| 966 |
+
"eval_runtime": 43.409,
|
| 967 |
+
"eval_samples_per_second": 73.856,
|
| 968 |
+
"eval_steps_per_second": 18.475,
|
| 969 |
+
"step": 60120
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 15.094810379241517,
|
| 973 |
+
"grad_norm": 3.2324092388153076,
|
| 974 |
+
"learning_rate": 1.245259481037924e-06,
|
| 975 |
+
"loss": 0.0849,
|
| 976 |
+
"step": 60500
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 15.219560878243513,
|
| 980 |
+
"grad_norm": 61.83153533935547,
|
| 981 |
+
"learning_rate": 1.2390219560878243e-06,
|
| 982 |
+
"loss": 0.0798,
|
| 983 |
+
"step": 61000
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 15.344311377245509,
|
| 987 |
+
"grad_norm": 0.015876924619078636,
|
| 988 |
+
"learning_rate": 1.2327844311377244e-06,
|
| 989 |
+
"loss": 0.0785,
|
| 990 |
+
"step": 61500
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 15.469061876247505,
|
| 994 |
+
"grad_norm": 3.0025134086608887,
|
| 995 |
+
"learning_rate": 1.2265469061876247e-06,
|
| 996 |
+
"loss": 0.0881,
|
| 997 |
+
"step": 62000
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 15.5938123752495,
|
| 1001 |
+
"grad_norm": 12.912367820739746,
|
| 1002 |
+
"learning_rate": 1.220309381237525e-06,
|
| 1003 |
+
"loss": 0.0802,
|
| 1004 |
+
"step": 62500
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 15.718562874251496,
|
| 1008 |
+
"grad_norm": 0.3600245714187622,
|
| 1009 |
+
"learning_rate": 1.214071856287425e-06,
|
| 1010 |
+
"loss": 0.0849,
|
| 1011 |
+
"step": 63000
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 15.843313373253492,
|
| 1015 |
+
"grad_norm": 0.21024100482463837,
|
| 1016 |
+
"learning_rate": 1.2078343313373254e-06,
|
| 1017 |
+
"loss": 0.078,
|
| 1018 |
+
"step": 63500
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 15.968063872255488,
|
| 1022 |
+
"grad_norm": 9.392132759094238,
|
| 1023 |
+
"learning_rate": 1.2015968063872255e-06,
|
| 1024 |
+
"loss": 0.0865,
|
| 1025 |
+
"step": 64000
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 16.0,
|
| 1029 |
+
"eval_loss": 0.48642057180404663,
|
| 1030 |
+
"eval_runtime": 46.7976,
|
| 1031 |
+
"eval_samples_per_second": 68.508,
|
| 1032 |
+
"eval_steps_per_second": 17.138,
|
| 1033 |
+
"step": 64128
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 16.092814371257486,
|
| 1037 |
+
"grad_norm": 0.5227041244506836,
|
| 1038 |
+
"learning_rate": 1.1953592814371256e-06,
|
| 1039 |
+
"loss": 0.0722,
|
| 1040 |
+
"step": 64500
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 16.21756487025948,
|
| 1044 |
+
"grad_norm": 25.282564163208008,
|
| 1045 |
+
"learning_rate": 1.1891217564870259e-06,
|
| 1046 |
+
"loss": 0.0981,
|
| 1047 |
+
"step": 65000
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 16.342315369261478,
|
| 1051 |
+
"grad_norm": 0.6670591235160828,
|
| 1052 |
+
"learning_rate": 1.1828842315369261e-06,
|
| 1053 |
+
"loss": 0.0787,
|
| 1054 |
+
"step": 65500
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 16.46706586826347,
|
| 1058 |
+
"grad_norm": 22.668352127075195,
|
| 1059 |
+
"learning_rate": 1.1766467065868262e-06,
|
| 1060 |
+
"loss": 0.0764,
|
| 1061 |
+
"step": 66000
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 16.59181636726547,
|
| 1065 |
+
"grad_norm": 0.22597374022006989,
|
| 1066 |
+
"learning_rate": 1.1704091816367265e-06,
|
| 1067 |
+
"loss": 0.078,
|
| 1068 |
+
"step": 66500
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 16.716566866267463,
|
| 1072 |
+
"grad_norm": 21.123409271240234,
|
| 1073 |
+
"learning_rate": 1.1641716566866266e-06,
|
| 1074 |
+
"loss": 0.0766,
|
| 1075 |
+
"step": 67000
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 16.84131736526946,
|
| 1079 |
+
"grad_norm": 0.04259370267391205,
|
| 1080 |
+
"learning_rate": 1.157934131736527e-06,
|
| 1081 |
+
"loss": 0.0765,
|
| 1082 |
+
"step": 67500
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 16.96606786427146,
|
| 1086 |
+
"grad_norm": 0.021560240536928177,
|
| 1087 |
+
"learning_rate": 1.1516966067864272e-06,
|
| 1088 |
+
"loss": 0.0785,
|
| 1089 |
+
"step": 68000
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 17.0,
|
| 1093 |
+
"eval_loss": 0.4793809652328491,
|
| 1094 |
+
"eval_runtime": 45.0906,
|
| 1095 |
+
"eval_samples_per_second": 71.101,
|
| 1096 |
+
"eval_steps_per_second": 17.786,
|
| 1097 |
+
"step": 68136
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 17.090818363273453,
|
| 1101 |
+
"grad_norm": 9.094868659973145,
|
| 1102 |
+
"learning_rate": 1.1454590818363273e-06,
|
| 1103 |
+
"loss": 0.0647,
|
| 1104 |
+
"step": 68500
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 17.21556886227545,
|
| 1108 |
+
"grad_norm": 0.195833221077919,
|
| 1109 |
+
"learning_rate": 1.1392215568862276e-06,
|
| 1110 |
+
"loss": 0.0698,
|
| 1111 |
+
"step": 69000
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 17.340319361277444,
|
| 1115 |
+
"grad_norm": 0.18507197499275208,
|
| 1116 |
+
"learning_rate": 1.1329840319361277e-06,
|
| 1117 |
+
"loss": 0.0712,
|
| 1118 |
+
"step": 69500
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 17.465069860279442,
|
| 1122 |
+
"grad_norm": 0.9911601543426514,
|
| 1123 |
+
"learning_rate": 1.1267465069860278e-06,
|
| 1124 |
+
"loss": 0.0752,
|
| 1125 |
+
"step": 70000
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 17.589820359281436,
|
| 1129 |
+
"grad_norm": 1.9703953266143799,
|
| 1130 |
+
"learning_rate": 1.120508982035928e-06,
|
| 1131 |
+
"loss": 0.0675,
|
| 1132 |
+
"step": 70500
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 17.714570858283434,
|
| 1136 |
+
"grad_norm": 41.10940933227539,
|
| 1137 |
+
"learning_rate": 1.1142714570858283e-06,
|
| 1138 |
+
"loss": 0.0705,
|
| 1139 |
+
"step": 71000
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 17.839321357285428,
|
| 1143 |
+
"grad_norm": 15.87336254119873,
|
| 1144 |
+
"learning_rate": 1.1080339321357286e-06,
|
| 1145 |
+
"loss": 0.0763,
|
| 1146 |
+
"step": 71500
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 17.964071856287426,
|
| 1150 |
+
"grad_norm": 0.060888275504112244,
|
| 1151 |
+
"learning_rate": 1.1017964071856287e-06,
|
| 1152 |
+
"loss": 0.0784,
|
| 1153 |
+
"step": 72000
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 18.0,
|
| 1157 |
+
"eval_loss": 0.4715409278869629,
|
| 1158 |
+
"eval_runtime": 44.1035,
|
| 1159 |
+
"eval_samples_per_second": 72.693,
|
| 1160 |
+
"eval_steps_per_second": 18.184,
|
| 1161 |
+
"step": 72144
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 18.08882235528942,
|
| 1165 |
+
"grad_norm": 2.47182035446167,
|
| 1166 |
+
"learning_rate": 1.0955588822355288e-06,
|
| 1167 |
+
"loss": 0.0747,
|
| 1168 |
+
"step": 72500
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 18.213572854291417,
|
| 1172 |
+
"grad_norm": 40.5880126953125,
|
| 1173 |
+
"learning_rate": 1.089321357285429e-06,
|
| 1174 |
+
"loss": 0.0678,
|
| 1175 |
+
"step": 73000
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 18.338323353293415,
|
| 1179 |
+
"grad_norm": 0.4340246915817261,
|
| 1180 |
+
"learning_rate": 1.0830838323353294e-06,
|
| 1181 |
+
"loss": 0.0713,
|
| 1182 |
+
"step": 73500
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 18.46307385229541,
|
| 1186 |
+
"grad_norm": 4.4763312339782715,
|
| 1187 |
+
"learning_rate": 1.0768463073852295e-06,
|
| 1188 |
+
"loss": 0.065,
|
| 1189 |
+
"step": 74000
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 18.587824351297407,
|
| 1193 |
+
"grad_norm": 0.1397508829832077,
|
| 1194 |
+
"learning_rate": 1.0706087824351298e-06,
|
| 1195 |
+
"loss": 0.0727,
|
| 1196 |
+
"step": 74500
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 18.7125748502994,
|
| 1200 |
+
"grad_norm": 7.134496212005615,
|
| 1201 |
+
"learning_rate": 1.0643712574850299e-06,
|
| 1202 |
+
"loss": 0.0605,
|
| 1203 |
+
"step": 75000
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 18.8373253493014,
|
| 1207 |
+
"grad_norm": 0.05227530747652054,
|
| 1208 |
+
"learning_rate": 1.05813373253493e-06,
|
| 1209 |
+
"loss": 0.0764,
|
| 1210 |
+
"step": 75500
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 18.962075848303392,
|
| 1214 |
+
"grad_norm": 17.22441864013672,
|
| 1215 |
+
"learning_rate": 1.0518962075848302e-06,
|
| 1216 |
+
"loss": 0.0696,
|
| 1217 |
+
"step": 76000
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 19.0,
|
| 1221 |
+
"eval_loss": 0.4802711308002472,
|
| 1222 |
+
"eval_runtime": 45.7109,
|
| 1223 |
+
"eval_samples_per_second": 70.136,
|
| 1224 |
+
"eval_steps_per_second": 17.545,
|
| 1225 |
+
"step": 76152
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 19.08682634730539,
|
| 1229 |
+
"grad_norm": 0.02889215387403965,
|
| 1230 |
+
"learning_rate": 1.0456586826347305e-06,
|
| 1231 |
+
"loss": 0.0625,
|
| 1232 |
+
"step": 76500
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 19.211576846307384,
|
| 1236 |
+
"grad_norm": 128.0497283935547,
|
| 1237 |
+
"learning_rate": 1.0394211576846308e-06,
|
| 1238 |
+
"loss": 0.0548,
|
| 1239 |
+
"step": 77000
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 19.336327345309382,
|
| 1243 |
+
"grad_norm": 0.22108981013298035,
|
| 1244 |
+
"learning_rate": 1.033183632734531e-06,
|
| 1245 |
+
"loss": 0.0695,
|
| 1246 |
+
"step": 77500
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 19.461077844311376,
|
| 1250 |
+
"grad_norm": 55.13557815551758,
|
| 1251 |
+
"learning_rate": 1.026946107784431e-06,
|
| 1252 |
+
"loss": 0.0679,
|
| 1253 |
+
"step": 78000
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 19.585828343313374,
|
| 1257 |
+
"grad_norm": 3.5990562438964844,
|
| 1258 |
+
"learning_rate": 1.0207085828343313e-06,
|
| 1259 |
+
"loss": 0.0697,
|
| 1260 |
+
"step": 78500
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 19.710578842315368,
|
| 1264 |
+
"grad_norm": 3.9640650749206543,
|
| 1265 |
+
"learning_rate": 1.0144710578842316e-06,
|
| 1266 |
+
"loss": 0.0699,
|
| 1267 |
+
"step": 79000
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 19.835329341317365,
|
| 1271 |
+
"grad_norm": 0.3529013395309448,
|
| 1272 |
+
"learning_rate": 1.0082335329341317e-06,
|
| 1273 |
+
"loss": 0.0676,
|
| 1274 |
+
"step": 79500
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 19.960079840319363,
|
| 1278 |
+
"grad_norm": 1.3875175714492798,
|
| 1279 |
+
"learning_rate": 1.001996007984032e-06,
|
| 1280 |
+
"loss": 0.0683,
|
| 1281 |
+
"step": 80000
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 20.0,
|
| 1285 |
+
"eval_loss": 0.5128437280654907,
|
| 1286 |
+
"eval_runtime": 46.0282,
|
| 1287 |
+
"eval_samples_per_second": 69.653,
|
| 1288 |
+
"eval_steps_per_second": 17.424,
|
| 1289 |
+
"step": 80160
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 20.084830339321357,
|
| 1293 |
+
"grad_norm": 6.171479225158691,
|
| 1294 |
+
"learning_rate": 9.95758483033932e-07,
|
| 1295 |
+
"loss": 0.0698,
|
| 1296 |
+
"step": 80500
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 20.209580838323355,
|
| 1300 |
+
"grad_norm": 0.012239497154951096,
|
| 1301 |
+
"learning_rate": 9.895209580838323e-07,
|
| 1302 |
+
"loss": 0.0532,
|
| 1303 |
+
"step": 81000
|
| 1304 |
+
},
|
| 1305 |
+
{
|
| 1306 |
+
"epoch": 20.33433133732535,
|
| 1307 |
+
"grad_norm": 7.920960426330566,
|
| 1308 |
+
"learning_rate": 9.832834331337324e-07,
|
| 1309 |
+
"loss": 0.0609,
|
| 1310 |
+
"step": 81500
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 20.459081836327346,
|
| 1314 |
+
"grad_norm": 59.41933822631836,
|
| 1315 |
+
"learning_rate": 9.770459081836327e-07,
|
| 1316 |
+
"loss": 0.0653,
|
| 1317 |
+
"step": 82000
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 20.58383233532934,
|
| 1321 |
+
"grad_norm": 0.10031065344810486,
|
| 1322 |
+
"learning_rate": 9.708083832335328e-07,
|
| 1323 |
+
"loss": 0.0497,
|
| 1324 |
+
"step": 82500
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 20.708582834331338,
|
| 1328 |
+
"grad_norm": 5.42900276184082,
|
| 1329 |
+
"learning_rate": 9.645708582834331e-07,
|
| 1330 |
+
"loss": 0.061,
|
| 1331 |
+
"step": 83000
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 20.833333333333332,
|
| 1335 |
+
"grad_norm": 20.380285263061523,
|
| 1336 |
+
"learning_rate": 9.583333333333334e-07,
|
| 1337 |
+
"loss": 0.0717,
|
| 1338 |
+
"step": 83500
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 20.95808383233533,
|
| 1342 |
+
"grad_norm": 0.10651753097772598,
|
| 1343 |
+
"learning_rate": 9.520958083832335e-07,
|
| 1344 |
+
"loss": 0.0638,
|
| 1345 |
+
"step": 84000
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 21.0,
|
| 1349 |
+
"eval_loss": 0.4833807945251465,
|
| 1350 |
+
"eval_runtime": 46.5592,
|
| 1351 |
+
"eval_samples_per_second": 68.859,
|
| 1352 |
+
"eval_steps_per_second": 17.225,
|
| 1353 |
+
"step": 84168
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 21.082834331337324,
|
| 1357 |
+
"grad_norm": 0.3842374086380005,
|
| 1358 |
+
"learning_rate": 9.458582834331337e-07,
|
| 1359 |
+
"loss": 0.0603,
|
| 1360 |
+
"step": 84500
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 21.20758483033932,
|
| 1364 |
+
"grad_norm": 51.563140869140625,
|
| 1365 |
+
"learning_rate": 9.396207584830339e-07,
|
| 1366 |
+
"loss": 0.06,
|
| 1367 |
+
"step": 85000
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 21.33233532934132,
|
| 1371 |
+
"grad_norm": 0.037806153297424316,
|
| 1372 |
+
"learning_rate": 9.333832335329342e-07,
|
| 1373 |
+
"loss": 0.0612,
|
| 1374 |
+
"step": 85500
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 21.457085828343313,
|
| 1378 |
+
"grad_norm": 0.11586946994066238,
|
| 1379 |
+
"learning_rate": 9.271457085828342e-07,
|
| 1380 |
+
"loss": 0.0664,
|
| 1381 |
+
"step": 86000
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 21.58183632734531,
|
| 1385 |
+
"grad_norm": 0.34262338280677795,
|
| 1386 |
+
"learning_rate": 9.209081836327344e-07,
|
| 1387 |
+
"loss": 0.0602,
|
| 1388 |
+
"step": 86500
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 21.706586826347305,
|
| 1392 |
+
"grad_norm": 0.11894870549440384,
|
| 1393 |
+
"learning_rate": 9.146706586826347e-07,
|
| 1394 |
+
"loss": 0.0522,
|
| 1395 |
+
"step": 87000
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 21.831337325349303,
|
| 1399 |
+
"grad_norm": 0.1180167868733406,
|
| 1400 |
+
"learning_rate": 9.084331337325349e-07,
|
| 1401 |
+
"loss": 0.0616,
|
| 1402 |
+
"step": 87500
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 21.956087824351297,
|
| 1406 |
+
"grad_norm": 0.09437087923288345,
|
| 1407 |
+
"learning_rate": 9.02195608782435e-07,
|
| 1408 |
+
"loss": 0.0607,
|
| 1409 |
+
"step": 88000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 22.0,
|
| 1413 |
+
"eval_loss": 0.4958905279636383,
|
| 1414 |
+
"eval_runtime": 44.4581,
|
| 1415 |
+
"eval_samples_per_second": 72.113,
|
| 1416 |
+
"eval_steps_per_second": 18.039,
|
| 1417 |
+
"step": 88176
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 22.080838323353294,
|
| 1421 |
+
"grad_norm": 0.5892271399497986,
|
| 1422 |
+
"learning_rate": 8.959580838323353e-07,
|
| 1423 |
+
"loss": 0.058,
|
| 1424 |
+
"step": 88500
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 22.20558882235529,
|
| 1428 |
+
"grad_norm": 1.0569002628326416,
|
| 1429 |
+
"learning_rate": 8.897205588822355e-07,
|
| 1430 |
+
"loss": 0.0559,
|
| 1431 |
+
"step": 89000
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 22.330339321357286,
|
| 1435 |
+
"grad_norm": 50.68812561035156,
|
| 1436 |
+
"learning_rate": 8.834830339321357e-07,
|
| 1437 |
+
"loss": 0.05,
|
| 1438 |
+
"step": 89500
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 22.45508982035928,
|
| 1442 |
+
"grad_norm": 0.08090469241142273,
|
| 1443 |
+
"learning_rate": 8.772455089820359e-07,
|
| 1444 |
+
"loss": 0.0595,
|
| 1445 |
+
"step": 90000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 22.579840319361278,
|
| 1449 |
+
"grad_norm": 14.62991714477539,
|
| 1450 |
+
"learning_rate": 8.710079840319361e-07,
|
| 1451 |
+
"loss": 0.059,
|
| 1452 |
+
"step": 90500
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 22.704590818363272,
|
| 1456 |
+
"grad_norm": 0.2893312871456146,
|
| 1457 |
+
"learning_rate": 8.647704590818364e-07,
|
| 1458 |
+
"loss": 0.0518,
|
| 1459 |
+
"step": 91000
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 22.82934131736527,
|
| 1463 |
+
"grad_norm": 22.239938735961914,
|
| 1464 |
+
"learning_rate": 8.585329341317364e-07,
|
| 1465 |
+
"loss": 0.0493,
|
| 1466 |
+
"step": 91500
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 22.954091816367267,
|
| 1470 |
+
"grad_norm": 0.09933929890394211,
|
| 1471 |
+
"learning_rate": 8.522954091816366e-07,
|
| 1472 |
+
"loss": 0.0536,
|
| 1473 |
+
"step": 92000
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 23.0,
|
| 1477 |
+
"eval_loss": 0.48672357201576233,
|
| 1478 |
+
"eval_runtime": 44.6856,
|
| 1479 |
+
"eval_samples_per_second": 71.746,
|
| 1480 |
+
"eval_steps_per_second": 17.948,
|
| 1481 |
+
"step": 92184
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 23.07884231536926,
|
| 1485 |
+
"grad_norm": 0.821902871131897,
|
| 1486 |
+
"learning_rate": 8.460578842315369e-07,
|
| 1487 |
+
"loss": 0.0553,
|
| 1488 |
+
"step": 92500
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 23.20359281437126,
|
| 1492 |
+
"grad_norm": 0.2537296414375305,
|
| 1493 |
+
"learning_rate": 8.398203592814371e-07,
|
| 1494 |
+
"loss": 0.046,
|
| 1495 |
+
"step": 93000
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 23.328343313373253,
|
| 1499 |
+
"grad_norm": 0.198989599943161,
|
| 1500 |
+
"learning_rate": 8.335828343313372e-07,
|
| 1501 |
+
"loss": 0.0496,
|
| 1502 |
+
"step": 93500
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 23.45309381237525,
|
| 1506 |
+
"grad_norm": 14.523540496826172,
|
| 1507 |
+
"learning_rate": 8.273453093812375e-07,
|
| 1508 |
+
"loss": 0.0465,
|
| 1509 |
+
"step": 94000
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 23.577844311377245,
|
| 1513 |
+
"grad_norm": 0.3473449945449829,
|
| 1514 |
+
"learning_rate": 8.211077844311377e-07,
|
| 1515 |
+
"loss": 0.048,
|
| 1516 |
+
"step": 94500
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 23.702594810379242,
|
| 1520 |
+
"grad_norm": 4.4253129959106445,
|
| 1521 |
+
"learning_rate": 8.14870259481038e-07,
|
| 1522 |
+
"loss": 0.0489,
|
| 1523 |
+
"step": 95000
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 23.827345309381236,
|
| 1527 |
+
"grad_norm": 159.51025390625,
|
| 1528 |
+
"learning_rate": 8.086327345309381e-07,
|
| 1529 |
+
"loss": 0.0552,
|
| 1530 |
+
"step": 95500
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 23.952095808383234,
|
| 1534 |
+
"grad_norm": 0.31450316309928894,
|
| 1535 |
+
"learning_rate": 8.023952095808383e-07,
|
| 1536 |
+
"loss": 0.0537,
|
| 1537 |
+
"step": 96000
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 24.0,
|
| 1541 |
+
"eval_loss": 0.5026536583900452,
|
| 1542 |
+
"eval_runtime": 46.0362,
|
| 1543 |
+
"eval_samples_per_second": 69.641,
|
| 1544 |
+
"eval_steps_per_second": 17.421,
|
| 1545 |
+
"step": 96192
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 24.076846307385228,
|
| 1549 |
+
"grad_norm": 1.8670942783355713,
|
| 1550 |
+
"learning_rate": 7.961576846307386e-07,
|
| 1551 |
+
"loss": 0.0556,
|
| 1552 |
+
"step": 96500
|
| 1553 |
+
},
|
| 1554 |
+
{
|
| 1555 |
+
"epoch": 24.201596806387226,
|
| 1556 |
+
"grad_norm": 0.4119631052017212,
|
| 1557 |
+
"learning_rate": 7.899201596806386e-07,
|
| 1558 |
+
"loss": 0.0427,
|
| 1559 |
+
"step": 97000
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 24.32634730538922,
|
| 1563 |
+
"grad_norm": 4.47167444229126,
|
| 1564 |
+
"learning_rate": 7.836826347305388e-07,
|
| 1565 |
+
"loss": 0.0579,
|
| 1566 |
+
"step": 97500
|
| 1567 |
+
},
|
| 1568 |
+
{
|
| 1569 |
+
"epoch": 24.451097804391217,
|
| 1570 |
+
"grad_norm": 0.940743625164032,
|
| 1571 |
+
"learning_rate": 7.774451097804391e-07,
|
| 1572 |
+
"loss": 0.0462,
|
| 1573 |
+
"step": 98000
|
| 1574 |
+
},
|
| 1575 |
+
{
|
| 1576 |
+
"epoch": 24.575848303393215,
|
| 1577 |
+
"grad_norm": 4.091241359710693,
|
| 1578 |
+
"learning_rate": 7.712075848303393e-07,
|
| 1579 |
+
"loss": 0.0524,
|
| 1580 |
+
"step": 98500
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 24.70059880239521,
|
| 1584 |
+
"grad_norm": 11.099757194519043,
|
| 1585 |
+
"learning_rate": 7.649700598802394e-07,
|
| 1586 |
+
"loss": 0.0549,
|
| 1587 |
+
"step": 99000
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 24.825349301397207,
|
| 1591 |
+
"grad_norm": 2.001067876815796,
|
| 1592 |
+
"learning_rate": 7.587325349301397e-07,
|
| 1593 |
+
"loss": 0.0485,
|
| 1594 |
+
"step": 99500
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 24.9500998003992,
|
| 1598 |
+
"grad_norm": 0.15496690571308136,
|
| 1599 |
+
"learning_rate": 7.524950099800399e-07,
|
| 1600 |
+
"loss": 0.0537,
|
| 1601 |
+
"step": 100000
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 25.0,
|
| 1605 |
+
"eval_loss": 0.48970088362693787,
|
| 1606 |
+
"eval_runtime": 48.0502,
|
| 1607 |
+
"eval_samples_per_second": 66.722,
|
| 1608 |
+
"eval_steps_per_second": 16.691,
|
| 1609 |
+
"step": 100200
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 25.0748502994012,
|
| 1613 |
+
"grad_norm": 5.718461513519287,
|
| 1614 |
+
"learning_rate": 7.462574850299402e-07,
|
| 1615 |
+
"loss": 0.0471,
|
| 1616 |
+
"step": 100500
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 25.199600798403193,
|
| 1620 |
+
"grad_norm": 53.097293853759766,
|
| 1621 |
+
"learning_rate": 7.400199600798403e-07,
|
| 1622 |
+
"loss": 0.0467,
|
| 1623 |
+
"step": 101000
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 25.32435129740519,
|
| 1627 |
+
"grad_norm": 70.51046752929688,
|
| 1628 |
+
"learning_rate": 7.337824351297404e-07,
|
| 1629 |
+
"loss": 0.0464,
|
| 1630 |
+
"step": 101500
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 25.449101796407184,
|
| 1634 |
+
"grad_norm": 6.485039234161377,
|
| 1635 |
+
"learning_rate": 7.275449101796407e-07,
|
| 1636 |
+
"loss": 0.0501,
|
| 1637 |
+
"step": 102000
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 25.573852295409182,
|
| 1641 |
+
"grad_norm": 0.2076825648546219,
|
| 1642 |
+
"learning_rate": 7.213073852295409e-07,
|
| 1643 |
+
"loss": 0.05,
|
| 1644 |
+
"step": 102500
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 25.698602794411176,
|
| 1648 |
+
"grad_norm": 40.60255432128906,
|
| 1649 |
+
"learning_rate": 7.15069860279441e-07,
|
| 1650 |
+
"loss": 0.0374,
|
| 1651 |
+
"step": 103000
|
| 1652 |
+
},
|
| 1653 |
+
{
|
| 1654 |
+
"epoch": 25.823353293413174,
|
| 1655 |
+
"grad_norm": 1.1958940029144287,
|
| 1656 |
+
"learning_rate": 7.088323353293413e-07,
|
| 1657 |
+
"loss": 0.0533,
|
| 1658 |
+
"step": 103500
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 25.948103792415168,
|
| 1662 |
+
"grad_norm": 11.201072692871094,
|
| 1663 |
+
"learning_rate": 7.025948103792415e-07,
|
| 1664 |
+
"loss": 0.0388,
|
| 1665 |
+
"step": 104000
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 26.0,
|
| 1669 |
+
"eval_loss": 0.48730549216270447,
|
| 1670 |
+
"eval_runtime": 48.7336,
|
| 1671 |
+
"eval_samples_per_second": 65.786,
|
| 1672 |
+
"eval_steps_per_second": 16.457,
|
| 1673 |
+
"step": 104208
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 26.072854291417165,
|
| 1677 |
+
"grad_norm": 0.08899884670972824,
|
| 1678 |
+
"learning_rate": 6.963572854291417e-07,
|
| 1679 |
+
"loss": 0.0482,
|
| 1680 |
+
"step": 104500
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 26.197604790419163,
|
| 1684 |
+
"grad_norm": 0.08736108243465424,
|
| 1685 |
+
"learning_rate": 6.901197604790419e-07,
|
| 1686 |
+
"loss": 0.042,
|
| 1687 |
+
"step": 105000
|
| 1688 |
+
},
|
| 1689 |
+
{
|
| 1690 |
+
"epoch": 26.322355289421157,
|
| 1691 |
+
"grad_norm": 0.050059039145708084,
|
| 1692 |
+
"learning_rate": 6.838822355289421e-07,
|
| 1693 |
+
"loss": 0.0443,
|
| 1694 |
+
"step": 105500
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 26.447105788423155,
|
| 1698 |
+
"grad_norm": 0.3098917603492737,
|
| 1699 |
+
"learning_rate": 6.776447105788423e-07,
|
| 1700 |
+
"loss": 0.0431,
|
| 1701 |
+
"step": 106000
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 26.57185628742515,
|
| 1705 |
+
"grad_norm": 0.601845920085907,
|
| 1706 |
+
"learning_rate": 6.714071856287425e-07,
|
| 1707 |
+
"loss": 0.0474,
|
| 1708 |
+
"step": 106500
|
| 1709 |
+
},
|
| 1710 |
+
{
|
| 1711 |
+
"epoch": 26.696606786427147,
|
| 1712 |
+
"grad_norm": 43.90340805053711,
|
| 1713 |
+
"learning_rate": 6.651696606786426e-07,
|
| 1714 |
+
"loss": 0.0546,
|
| 1715 |
+
"step": 107000
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 26.82135728542914,
|
| 1719 |
+
"grad_norm": 0.1658441424369812,
|
| 1720 |
+
"learning_rate": 6.589321357285429e-07,
|
| 1721 |
+
"loss": 0.0463,
|
| 1722 |
+
"step": 107500
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 26.94610778443114,
|
| 1726 |
+
"grad_norm": 0.7097954154014587,
|
| 1727 |
+
"learning_rate": 6.526946107784431e-07,
|
| 1728 |
+
"loss": 0.0413,
|
| 1729 |
+
"step": 108000
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 27.0,
|
| 1733 |
+
"eval_loss": 0.49195966124534607,
|
| 1734 |
+
"eval_runtime": 48.5815,
|
| 1735 |
+
"eval_samples_per_second": 65.992,
|
| 1736 |
+
"eval_steps_per_second": 16.508,
|
| 1737 |
+
"step": 108216
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 27.070858283433132,
|
| 1741 |
+
"grad_norm": 0.12945351004600525,
|
| 1742 |
+
"learning_rate": 6.464570858283432e-07,
|
| 1743 |
+
"loss": 0.0514,
|
| 1744 |
+
"step": 108500
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 27.19560878243513,
|
| 1748 |
+
"grad_norm": 0.09241262078285217,
|
| 1749 |
+
"learning_rate": 6.402195608782435e-07,
|
| 1750 |
+
"loss": 0.0454,
|
| 1751 |
+
"step": 109000
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 27.320359281437124,
|
| 1755 |
+
"grad_norm": 0.07145562022924423,
|
| 1756 |
+
"learning_rate": 6.339820359281437e-07,
|
| 1757 |
+
"loss": 0.0381,
|
| 1758 |
+
"step": 109500
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 27.44510978043912,
|
| 1762 |
+
"grad_norm": 0.003607134334743023,
|
| 1763 |
+
"learning_rate": 6.277445109780439e-07,
|
| 1764 |
+
"loss": 0.0476,
|
| 1765 |
+
"step": 110000
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 27.56986027944112,
|
| 1769 |
+
"grad_norm": 10.220846176147461,
|
| 1770 |
+
"learning_rate": 6.215069860279441e-07,
|
| 1771 |
+
"loss": 0.0441,
|
| 1772 |
+
"step": 110500
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 27.694610778443113,
|
| 1776 |
+
"grad_norm": 0.18386581540107727,
|
| 1777 |
+
"learning_rate": 6.152694610778443e-07,
|
| 1778 |
+
"loss": 0.0461,
|
| 1779 |
+
"step": 111000
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 27.81936127744511,
|
| 1783 |
+
"grad_norm": 0.26254481077194214,
|
| 1784 |
+
"learning_rate": 6.090319361277445e-07,
|
| 1785 |
+
"loss": 0.0367,
|
| 1786 |
+
"step": 111500
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 27.944111776447105,
|
| 1790 |
+
"grad_norm": 68.7042007446289,
|
| 1791 |
+
"learning_rate": 6.027944111776448e-07,
|
| 1792 |
+
"loss": 0.0471,
|
| 1793 |
+
"step": 112000
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 28.0,
|
| 1797 |
+
"eval_loss": 0.4870954751968384,
|
| 1798 |
+
"eval_runtime": 45.0714,
|
| 1799 |
+
"eval_samples_per_second": 71.132,
|
| 1800 |
+
"eval_steps_per_second": 17.794,
|
| 1801 |
+
"step": 112224
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 28.068862275449103,
|
| 1805 |
+
"grad_norm": 0.0271464716643095,
|
| 1806 |
+
"learning_rate": 5.965568862275448e-07,
|
| 1807 |
+
"loss": 0.0433,
|
| 1808 |
+
"step": 112500
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 28.193612774451097,
|
| 1812 |
+
"grad_norm": 0.0086235161870718,
|
| 1813 |
+
"learning_rate": 5.903193612774451e-07,
|
| 1814 |
+
"loss": 0.0475,
|
| 1815 |
+
"step": 113000
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 28.318363273453095,
|
| 1819 |
+
"grad_norm": 0.11506126821041107,
|
| 1820 |
+
"learning_rate": 5.840818363273453e-07,
|
| 1821 |
+
"loss": 0.0353,
|
| 1822 |
+
"step": 113500
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 28.44311377245509,
|
| 1826 |
+
"grad_norm": 10.355070114135742,
|
| 1827 |
+
"learning_rate": 5.778443113772454e-07,
|
| 1828 |
+
"loss": 0.0416,
|
| 1829 |
+
"step": 114000
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 28.567864271457086,
|
| 1833 |
+
"grad_norm": 0.2200528234243393,
|
| 1834 |
+
"learning_rate": 5.716067864271457e-07,
|
| 1835 |
+
"loss": 0.0325,
|
| 1836 |
+
"step": 114500
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 28.69261477045908,
|
| 1840 |
+
"grad_norm": 0.05802537873387337,
|
| 1841 |
+
"learning_rate": 5.653692614770459e-07,
|
| 1842 |
+
"loss": 0.0468,
|
| 1843 |
+
"step": 115000
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 28.817365269461078,
|
| 1847 |
+
"grad_norm": 0.10829133540391922,
|
| 1848 |
+
"learning_rate": 5.591317365269461e-07,
|
| 1849 |
+
"loss": 0.042,
|
| 1850 |
+
"step": 115500
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 28.942115768463076,
|
| 1854 |
+
"grad_norm": 0.162460595369339,
|
| 1855 |
+
"learning_rate": 5.528942115768463e-07,
|
| 1856 |
+
"loss": 0.049,
|
| 1857 |
+
"step": 116000
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 29.0,
|
| 1861 |
+
"eval_loss": 0.4795687198638916,
|
| 1862 |
+
"eval_runtime": 45.1647,
|
| 1863 |
+
"eval_samples_per_second": 70.985,
|
| 1864 |
+
"eval_steps_per_second": 17.757,
|
| 1865 |
+
"step": 116232
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 29.06686626746507,
|
| 1869 |
+
"grad_norm": 134.6587677001953,
|
| 1870 |
+
"learning_rate": 5.466566866267465e-07,
|
| 1871 |
+
"loss": 0.0416,
|
| 1872 |
+
"step": 116500
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 29.191616766467067,
|
| 1876 |
+
"grad_norm": 0.09312257915735245,
|
| 1877 |
+
"learning_rate": 5.404191616766467e-07,
|
| 1878 |
+
"loss": 0.0287,
|
| 1879 |
+
"step": 117000
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 29.31636726546906,
|
| 1883 |
+
"grad_norm": 0.3530866503715515,
|
| 1884 |
+
"learning_rate": 5.341816367265469e-07,
|
| 1885 |
+
"loss": 0.0384,
|
| 1886 |
+
"step": 117500
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 29.44111776447106,
|
| 1890 |
+
"grad_norm": 0.033993642777204514,
|
| 1891 |
+
"learning_rate": 5.27944111776447e-07,
|
| 1892 |
+
"loss": 0.043,
|
| 1893 |
+
"step": 118000
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 29.565868263473053,
|
| 1897 |
+
"grad_norm": 0.3124711513519287,
|
| 1898 |
+
"learning_rate": 5.217065868263473e-07,
|
| 1899 |
+
"loss": 0.04,
|
| 1900 |
+
"step": 118500
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 29.69061876247505,
|
| 1904 |
+
"grad_norm": 10.49288272857666,
|
| 1905 |
+
"learning_rate": 5.154690618762475e-07,
|
| 1906 |
+
"loss": 0.0463,
|
| 1907 |
+
"step": 119000
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 29.815369261477045,
|
| 1911 |
+
"grad_norm": 0.024224599823355675,
|
| 1912 |
+
"learning_rate": 5.092315369261477e-07,
|
| 1913 |
+
"loss": 0.0411,
|
| 1914 |
+
"step": 119500
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 29.940119760479043,
|
| 1918 |
+
"grad_norm": 3.9215731620788574,
|
| 1919 |
+
"learning_rate": 5.029940119760479e-07,
|
| 1920 |
+
"loss": 0.0408,
|
| 1921 |
+
"step": 120000
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 30.0,
|
| 1925 |
+
"eval_loss": 0.492553174495697,
|
| 1926 |
+
"eval_runtime": 46.2042,
|
| 1927 |
+
"eval_samples_per_second": 69.388,
|
| 1928 |
+
"eval_steps_per_second": 17.358,
|
| 1929 |
+
"step": 120240
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 30.064870259481037,
|
| 1933 |
+
"grad_norm": 0.021667474880814552,
|
| 1934 |
+
"learning_rate": 4.967564870259481e-07,
|
| 1935 |
+
"loss": 0.0374,
|
| 1936 |
+
"step": 120500
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"epoch": 30.189620758483034,
|
| 1940 |
+
"grad_norm": 0.5888983011245728,
|
| 1941 |
+
"learning_rate": 4.905189620758483e-07,
|
| 1942 |
+
"loss": 0.0463,
|
| 1943 |
+
"step": 121000
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 30.31437125748503,
|
| 1947 |
+
"grad_norm": 0.09637131541967392,
|
| 1948 |
+
"learning_rate": 4.842814371257485e-07,
|
| 1949 |
+
"loss": 0.033,
|
| 1950 |
+
"step": 121500
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 30.439121756487026,
|
| 1954 |
+
"grad_norm": 0.23179832100868225,
|
| 1955 |
+
"learning_rate": 4.780439121756487e-07,
|
| 1956 |
+
"loss": 0.0402,
|
| 1957 |
+
"step": 122000
|
| 1958 |
+
},
|
| 1959 |
+
{
|
| 1960 |
+
"epoch": 30.563872255489024,
|
| 1961 |
+
"grad_norm": 0.14170564711093903,
|
| 1962 |
+
"learning_rate": 4.718063872255489e-07,
|
| 1963 |
+
"loss": 0.0395,
|
| 1964 |
+
"step": 122500
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 30.688622754491018,
|
| 1968 |
+
"grad_norm": 0.006093321368098259,
|
| 1969 |
+
"learning_rate": 4.6556886227544903e-07,
|
| 1970 |
+
"loss": 0.0356,
|
| 1971 |
+
"step": 123000
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 30.813373253493015,
|
| 1975 |
+
"grad_norm": 0.1018219068646431,
|
| 1976 |
+
"learning_rate": 4.593313373253493e-07,
|
| 1977 |
+
"loss": 0.0419,
|
| 1978 |
+
"step": 123500
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 30.93812375249501,
|
| 1982 |
+
"grad_norm": 2.9131383895874023,
|
| 1983 |
+
"learning_rate": 4.5309381237524947e-07,
|
| 1984 |
+
"loss": 0.0378,
|
| 1985 |
+
"step": 124000
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 31.0,
|
| 1989 |
+
"eval_loss": 0.5052226781845093,
|
| 1990 |
+
"eval_runtime": 43.1611,
|
| 1991 |
+
"eval_samples_per_second": 74.28,
|
| 1992 |
+
"eval_steps_per_second": 18.582,
|
| 1993 |
+
"step": 124248
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 31.062874251497007,
|
| 1997 |
+
"grad_norm": 11.588695526123047,
|
| 1998 |
+
"learning_rate": 4.468562874251497e-07,
|
| 1999 |
+
"loss": 0.0346,
|
| 2000 |
+
"step": 124500
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 31.187624750499,
|
| 2004 |
+
"grad_norm": 0.2488149255514145,
|
| 2005 |
+
"learning_rate": 4.4061876247504985e-07,
|
| 2006 |
+
"loss": 0.0351,
|
| 2007 |
+
"step": 125000
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 31.312375249501,
|
| 2011 |
+
"grad_norm": 12.691544532775879,
|
| 2012 |
+
"learning_rate": 4.343812375249501e-07,
|
| 2013 |
+
"loss": 0.0323,
|
| 2014 |
+
"step": 125500
|
| 2015 |
+
},
|
| 2016 |
+
{
|
| 2017 |
+
"epoch": 31.437125748502993,
|
| 2018 |
+
"grad_norm": 0.004168800078332424,
|
| 2019 |
+
"learning_rate": 4.281437125748503e-07,
|
| 2020 |
+
"loss": 0.033,
|
| 2021 |
+
"step": 126000
|
| 2022 |
+
},
|
| 2023 |
+
{
|
| 2024 |
+
"epoch": 31.56187624750499,
|
| 2025 |
+
"grad_norm": 0.042690277099609375,
|
| 2026 |
+
"learning_rate": 4.219061876247505e-07,
|
| 2027 |
+
"loss": 0.039,
|
| 2028 |
+
"step": 126500
|
| 2029 |
+
},
|
| 2030 |
+
{
|
| 2031 |
+
"epoch": 31.686626746506985,
|
| 2032 |
+
"grad_norm": 1.1096973419189453,
|
| 2033 |
+
"learning_rate": 4.1566866267465066e-07,
|
| 2034 |
+
"loss": 0.0349,
|
| 2035 |
+
"step": 127000
|
| 2036 |
+
},
|
| 2037 |
+
{
|
| 2038 |
+
"epoch": 31.811377245508982,
|
| 2039 |
+
"grad_norm": 0.2642970085144043,
|
| 2040 |
+
"learning_rate": 4.094311377245509e-07,
|
| 2041 |
+
"loss": 0.0338,
|
| 2042 |
+
"step": 127500
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 31.936127744510976,
|
| 2046 |
+
"grad_norm": 0.21338249742984772,
|
| 2047 |
+
"learning_rate": 4.031936127744511e-07,
|
| 2048 |
+
"loss": 0.0349,
|
| 2049 |
+
"step": 128000
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 32.0,
|
| 2053 |
+
"eval_loss": 0.4927305281162262,
|
| 2054 |
+
"eval_runtime": 43.7641,
|
| 2055 |
+
"eval_samples_per_second": 73.256,
|
| 2056 |
+
"eval_steps_per_second": 18.326,
|
| 2057 |
+
"step": 128256
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 32.060878243512974,
|
| 2061 |
+
"grad_norm": 0.1497274786233902,
|
| 2062 |
+
"learning_rate": 3.969560878243513e-07,
|
| 2063 |
+
"loss": 0.0403,
|
| 2064 |
+
"step": 128500
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 32.18562874251497,
|
| 2068 |
+
"grad_norm": 0.5848351120948792,
|
| 2069 |
+
"learning_rate": 3.9071856287425147e-07,
|
| 2070 |
+
"loss": 0.037,
|
| 2071 |
+
"step": 129000
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 32.31037924151697,
|
| 2075 |
+
"grad_norm": 0.11372077465057373,
|
| 2076 |
+
"learning_rate": 3.8448103792415166e-07,
|
| 2077 |
+
"loss": 0.0383,
|
| 2078 |
+
"step": 129500
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 32.43512974051896,
|
| 2082 |
+
"grad_norm": 0.1047956719994545,
|
| 2083 |
+
"learning_rate": 3.782435129740519e-07,
|
| 2084 |
+
"loss": 0.0315,
|
| 2085 |
+
"step": 130000
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 32.55988023952096,
|
| 2089 |
+
"grad_norm": 0.2975727617740631,
|
| 2090 |
+
"learning_rate": 3.7200598802395204e-07,
|
| 2091 |
+
"loss": 0.0264,
|
| 2092 |
+
"step": 130500
|
| 2093 |
+
},
|
| 2094 |
+
{
|
| 2095 |
+
"epoch": 32.684630738522955,
|
| 2096 |
+
"grad_norm": 0.2123280167579651,
|
| 2097 |
+
"learning_rate": 3.657684630738523e-07,
|
| 2098 |
+
"loss": 0.0341,
|
| 2099 |
+
"step": 131000
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 32.80938123752495,
|
| 2103 |
+
"grad_norm": 27.63080596923828,
|
| 2104 |
+
"learning_rate": 3.5953093812375247e-07,
|
| 2105 |
+
"loss": 0.0368,
|
| 2106 |
+
"step": 131500
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 32.93413173652694,
|
| 2110 |
+
"grad_norm": 0.034935545176267624,
|
| 2111 |
+
"learning_rate": 3.5329341317365266e-07,
|
| 2112 |
+
"loss": 0.0394,
|
| 2113 |
+
"step": 132000
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 33.0,
|
| 2117 |
+
"eval_loss": 0.4937605559825897,
|
| 2118 |
+
"eval_runtime": 39.9355,
|
| 2119 |
+
"eval_samples_per_second": 80.279,
|
| 2120 |
+
"eval_steps_per_second": 20.082,
|
| 2121 |
+
"step": 132264
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 33.05888223552894,
|
| 2125 |
+
"grad_norm": 0.003380158683285117,
|
| 2126 |
+
"learning_rate": 3.4705588822355285e-07,
|
| 2127 |
+
"loss": 0.0394,
|
| 2128 |
+
"step": 132500
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"epoch": 33.18363273453094,
|
| 2132 |
+
"grad_norm": 2.721451997756958,
|
| 2133 |
+
"learning_rate": 3.408183632734531e-07,
|
| 2134 |
+
"loss": 0.0365,
|
| 2135 |
+
"step": 133000
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 33.308383233532936,
|
| 2139 |
+
"grad_norm": 0.4309988021850586,
|
| 2140 |
+
"learning_rate": 3.345808383233533e-07,
|
| 2141 |
+
"loss": 0.0302,
|
| 2142 |
+
"step": 133500
|
| 2143 |
+
},
|
| 2144 |
+
{
|
| 2145 |
+
"epoch": 33.43313373253493,
|
| 2146 |
+
"grad_norm": 0.24694228172302246,
|
| 2147 |
+
"learning_rate": 3.283433133732535e-07,
|
| 2148 |
+
"loss": 0.037,
|
| 2149 |
+
"step": 134000
|
| 2150 |
+
},
|
| 2151 |
+
{
|
| 2152 |
+
"epoch": 33.557884231536924,
|
| 2153 |
+
"grad_norm": 0.34988418221473694,
|
| 2154 |
+
"learning_rate": 3.2210578842315366e-07,
|
| 2155 |
+
"loss": 0.0258,
|
| 2156 |
+
"step": 134500
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 33.68263473053892,
|
| 2160 |
+
"grad_norm": 0.19452495872974396,
|
| 2161 |
+
"learning_rate": 3.158682634730539e-07,
|
| 2162 |
+
"loss": 0.035,
|
| 2163 |
+
"step": 135000
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 33.80738522954092,
|
| 2167 |
+
"grad_norm": 0.006651519797742367,
|
| 2168 |
+
"learning_rate": 3.096307385229541e-07,
|
| 2169 |
+
"loss": 0.0368,
|
| 2170 |
+
"step": 135500
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 33.93213572854292,
|
| 2174 |
+
"grad_norm": 0.04128989204764366,
|
| 2175 |
+
"learning_rate": 3.033932135728543e-07,
|
| 2176 |
+
"loss": 0.0301,
|
| 2177 |
+
"step": 136000
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 34.0,
|
| 2181 |
+
"eval_loss": 0.4872666597366333,
|
| 2182 |
+
"eval_runtime": 37.6035,
|
| 2183 |
+
"eval_samples_per_second": 85.258,
|
| 2184 |
+
"eval_steps_per_second": 21.328,
|
| 2185 |
+
"step": 136272
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 34.05688622754491,
|
| 2189 |
+
"grad_norm": 0.05333876982331276,
|
| 2190 |
+
"learning_rate": 2.971556886227545e-07,
|
| 2191 |
+
"loss": 0.0349,
|
| 2192 |
+
"step": 136500
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 34.181636726546905,
|
| 2196 |
+
"grad_norm": 1.3579726219177246,
|
| 2197 |
+
"learning_rate": 2.909181636726547e-07,
|
| 2198 |
+
"loss": 0.0285,
|
| 2199 |
+
"step": 137000
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 34.3063872255489,
|
| 2203 |
+
"grad_norm": 0.6725994348526001,
|
| 2204 |
+
"learning_rate": 2.8468063872255486e-07,
|
| 2205 |
+
"loss": 0.0361,
|
| 2206 |
+
"step": 137500
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 34.4311377245509,
|
| 2210 |
+
"grad_norm": 0.03919246420264244,
|
| 2211 |
+
"learning_rate": 2.7844311377245504e-07,
|
| 2212 |
+
"loss": 0.0274,
|
| 2213 |
+
"step": 138000
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 34.55588822355289,
|
| 2217 |
+
"grad_norm": 35.5837287902832,
|
| 2218 |
+
"learning_rate": 2.722055888223553e-07,
|
| 2219 |
+
"loss": 0.0363,
|
| 2220 |
+
"step": 138500
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 34.68063872255489,
|
| 2224 |
+
"grad_norm": 0.007728968746960163,
|
| 2225 |
+
"learning_rate": 2.659680638722555e-07,
|
| 2226 |
+
"loss": 0.0391,
|
| 2227 |
+
"step": 139000
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 34.80538922155689,
|
| 2231 |
+
"grad_norm": 0.07272203266620636,
|
| 2232 |
+
"learning_rate": 2.5973053892215567e-07,
|
| 2233 |
+
"loss": 0.0268,
|
| 2234 |
+
"step": 139500
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 34.930139720558884,
|
| 2238 |
+
"grad_norm": 0.33094656467437744,
|
| 2239 |
+
"learning_rate": 2.5349301397205586e-07,
|
| 2240 |
+
"loss": 0.0365,
|
| 2241 |
+
"step": 140000
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 35.0,
|
| 2245 |
+
"eval_loss": 0.4920032024383545,
|
| 2246 |
+
"eval_runtime": 40.4781,
|
| 2247 |
+
"eval_samples_per_second": 79.203,
|
| 2248 |
+
"eval_steps_per_second": 19.813,
|
| 2249 |
+
"step": 140280
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 35.054890219560875,
|
| 2253 |
+
"grad_norm": 191.99266052246094,
|
| 2254 |
+
"learning_rate": 2.472554890219561e-07,
|
| 2255 |
+
"loss": 0.0333,
|
| 2256 |
+
"step": 140500
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 35.17964071856287,
|
| 2260 |
+
"grad_norm": 0.002573936013504863,
|
| 2261 |
+
"learning_rate": 2.410179640718563e-07,
|
| 2262 |
+
"loss": 0.0327,
|
| 2263 |
+
"step": 141000
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 35.30439121756487,
|
| 2267 |
+
"grad_norm": 0.04750495404005051,
|
| 2268 |
+
"learning_rate": 2.3478043912175645e-07,
|
| 2269 |
+
"loss": 0.0345,
|
| 2270 |
+
"step": 141500
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 35.42914171656687,
|
| 2274 |
+
"grad_norm": 193.8626251220703,
|
| 2275 |
+
"learning_rate": 2.2854291417165667e-07,
|
| 2276 |
+
"loss": 0.0321,
|
| 2277 |
+
"step": 142000
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 35.553892215568865,
|
| 2281 |
+
"grad_norm": 0.0009173236903734505,
|
| 2282 |
+
"learning_rate": 2.2230538922155686e-07,
|
| 2283 |
+
"loss": 0.0359,
|
| 2284 |
+
"step": 142500
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 35.678642714570856,
|
| 2288 |
+
"grad_norm": 0.12355954945087433,
|
| 2289 |
+
"learning_rate": 2.1606786427145708e-07,
|
| 2290 |
+
"loss": 0.0347,
|
| 2291 |
+
"step": 143000
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 35.80339321357285,
|
| 2295 |
+
"grad_norm": 0.24140344560146332,
|
| 2296 |
+
"learning_rate": 2.0983033932135726e-07,
|
| 2297 |
+
"loss": 0.031,
|
| 2298 |
+
"step": 143500
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 35.92814371257485,
|
| 2302 |
+
"grad_norm": 0.007129414472728968,
|
| 2303 |
+
"learning_rate": 2.0359281437125748e-07,
|
| 2304 |
+
"loss": 0.0214,
|
| 2305 |
+
"step": 144000
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 36.0,
|
| 2309 |
+
"eval_loss": 0.4941750466823578,
|
| 2310 |
+
"eval_runtime": 38.7085,
|
| 2311 |
+
"eval_samples_per_second": 82.824,
|
| 2312 |
+
"eval_steps_per_second": 20.719,
|
| 2313 |
+
"step": 144288
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 36.05289421157685,
|
| 2317 |
+
"grad_norm": 0.27973344922065735,
|
| 2318 |
+
"learning_rate": 1.9735528942115767e-07,
|
| 2319 |
+
"loss": 0.0331,
|
| 2320 |
+
"step": 144500
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 36.17764471057884,
|
| 2324 |
+
"grad_norm": 0.05331612005829811,
|
| 2325 |
+
"learning_rate": 1.911177644710579e-07,
|
| 2326 |
+
"loss": 0.0303,
|
| 2327 |
+
"step": 145000
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 36.30239520958084,
|
| 2331 |
+
"grad_norm": 1.8135106563568115,
|
| 2332 |
+
"learning_rate": 1.8488023952095808e-07,
|
| 2333 |
+
"loss": 0.0349,
|
| 2334 |
+
"step": 145500
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 36.427145708582835,
|
| 2338 |
+
"grad_norm": 0.13009090721607208,
|
| 2339 |
+
"learning_rate": 1.7864271457085827e-07,
|
| 2340 |
+
"loss": 0.0405,
|
| 2341 |
+
"step": 146000
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 36.55189620758483,
|
| 2345 |
+
"grad_norm": 0.07144490629434586,
|
| 2346 |
+
"learning_rate": 1.7240518962075848e-07,
|
| 2347 |
+
"loss": 0.0377,
|
| 2348 |
+
"step": 146500
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 36.67664670658683,
|
| 2352 |
+
"grad_norm": 74.39689636230469,
|
| 2353 |
+
"learning_rate": 1.6616766467065867e-07,
|
| 2354 |
+
"loss": 0.0278,
|
| 2355 |
+
"step": 147000
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 36.80139720558882,
|
| 2359 |
+
"grad_norm": 0.08526595681905746,
|
| 2360 |
+
"learning_rate": 1.599301397205589e-07,
|
| 2361 |
+
"loss": 0.0306,
|
| 2362 |
+
"step": 147500
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 36.92614770459082,
|
| 2366 |
+
"grad_norm": 12.262850761413574,
|
| 2367 |
+
"learning_rate": 1.5369261477045908e-07,
|
| 2368 |
+
"loss": 0.0314,
|
| 2369 |
+
"step": 148000
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 37.0,
|
| 2373 |
+
"eval_loss": 0.49442577362060547,
|
| 2374 |
+
"eval_runtime": 42.7404,
|
| 2375 |
+
"eval_samples_per_second": 75.011,
|
| 2376 |
+
"eval_steps_per_second": 18.764,
|
| 2377 |
+
"step": 148296
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 37.050898203592816,
|
| 2381 |
+
"grad_norm": 0.02493446320295334,
|
| 2382 |
+
"learning_rate": 1.474550898203593e-07,
|
| 2383 |
+
"loss": 0.0262,
|
| 2384 |
+
"step": 148500
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 37.17564870259481,
|
| 2388 |
+
"grad_norm": 0.14130648970603943,
|
| 2389 |
+
"learning_rate": 1.4121756487025949e-07,
|
| 2390 |
+
"loss": 0.0281,
|
| 2391 |
+
"step": 149000
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 37.300399201596804,
|
| 2395 |
+
"grad_norm": 0.035768117755651474,
|
| 2396 |
+
"learning_rate": 1.3498003992015965e-07,
|
| 2397 |
+
"loss": 0.0255,
|
| 2398 |
+
"step": 149500
|
| 2399 |
+
},
|
| 2400 |
+
{
|
| 2401 |
+
"epoch": 37.4251497005988,
|
| 2402 |
+
"grad_norm": 0.18820720911026,
|
| 2403 |
+
"learning_rate": 1.2874251497005986e-07,
|
| 2404 |
+
"loss": 0.032,
|
| 2405 |
+
"step": 150000
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 37.5499001996008,
|
| 2409 |
+
"grad_norm": 0.37001463770866394,
|
| 2410 |
+
"learning_rate": 1.2250499001996008e-07,
|
| 2411 |
+
"loss": 0.0301,
|
| 2412 |
+
"step": 150500
|
| 2413 |
+
},
|
| 2414 |
+
{
|
| 2415 |
+
"epoch": 37.6746506986028,
|
| 2416 |
+
"grad_norm": 0.06626907736063004,
|
| 2417 |
+
"learning_rate": 1.1626746506986028e-07,
|
| 2418 |
+
"loss": 0.0238,
|
| 2419 |
+
"step": 151000
|
| 2420 |
+
},
|
| 2421 |
+
{
|
| 2422 |
+
"epoch": 37.79940119760479,
|
| 2423 |
+
"grad_norm": 19.17169189453125,
|
| 2424 |
+
"learning_rate": 1.1002994011976049e-07,
|
| 2425 |
+
"loss": 0.0385,
|
| 2426 |
+
"step": 151500
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 37.924151696606785,
|
| 2430 |
+
"grad_norm": 4.972864627838135,
|
| 2431 |
+
"learning_rate": 1.0379241516966066e-07,
|
| 2432 |
+
"loss": 0.0337,
|
| 2433 |
+
"step": 152000
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 38.0,
|
| 2437 |
+
"eval_loss": 0.48605817556381226,
|
| 2438 |
+
"eval_runtime": 40.1954,
|
| 2439 |
+
"eval_samples_per_second": 79.76,
|
| 2440 |
+
"eval_steps_per_second": 19.953,
|
| 2441 |
+
"step": 152304
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 38.04890219560878,
|
| 2445 |
+
"grad_norm": 0.002587054157629609,
|
| 2446 |
+
"learning_rate": 9.755489021956087e-08,
|
| 2447 |
+
"loss": 0.0334,
|
| 2448 |
+
"step": 152500
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 38.17365269461078,
|
| 2452 |
+
"grad_norm": 70.7108383178711,
|
| 2453 |
+
"learning_rate": 9.131736526946107e-08,
|
| 2454 |
+
"loss": 0.0319,
|
| 2455 |
+
"step": 153000
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 38.29840319361278,
|
| 2459 |
+
"grad_norm": 0.5694107413291931,
|
| 2460 |
+
"learning_rate": 8.507984031936127e-08,
|
| 2461 |
+
"loss": 0.0313,
|
| 2462 |
+
"step": 153500
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 38.42315369261477,
|
| 2466 |
+
"grad_norm": 0.003176228841766715,
|
| 2467 |
+
"learning_rate": 7.884231536926148e-08,
|
| 2468 |
+
"loss": 0.0298,
|
| 2469 |
+
"step": 154000
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 38.547904191616766,
|
| 2473 |
+
"grad_norm": 0.004230498801916838,
|
| 2474 |
+
"learning_rate": 7.260479041916168e-08,
|
| 2475 |
+
"loss": 0.0284,
|
| 2476 |
+
"step": 154500
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 38.672654690618764,
|
| 2480 |
+
"grad_norm": 0.13844607770442963,
|
| 2481 |
+
"learning_rate": 6.636726546906188e-08,
|
| 2482 |
+
"loss": 0.0305,
|
| 2483 |
+
"step": 155000
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 38.79740518962076,
|
| 2487 |
+
"grad_norm": 0.05394995957612991,
|
| 2488 |
+
"learning_rate": 6.012974051896207e-08,
|
| 2489 |
+
"loss": 0.0269,
|
| 2490 |
+
"step": 155500
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 38.92215568862275,
|
| 2494 |
+
"grad_norm": 0.11763022094964981,
|
| 2495 |
+
"learning_rate": 5.3892215568862274e-08,
|
| 2496 |
+
"loss": 0.0279,
|
| 2497 |
+
"step": 156000
|
| 2498 |
+
},
|
| 2499 |
+
{
|
| 2500 |
+
"epoch": 39.0,
|
| 2501 |
+
"eval_loss": 0.4873499870300293,
|
| 2502 |
+
"eval_runtime": 44.0281,
|
| 2503 |
+
"eval_samples_per_second": 72.817,
|
| 2504 |
+
"eval_steps_per_second": 18.216,
|
| 2505 |
+
"step": 156312
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 39.04690618762475,
|
| 2509 |
+
"grad_norm": 0.22139760851860046,
|
| 2510 |
+
"learning_rate": 4.765469061876248e-08,
|
| 2511 |
+
"loss": 0.0255,
|
| 2512 |
+
"step": 156500
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 39.17165668662675,
|
| 2516 |
+
"grad_norm": 0.002428988926112652,
|
| 2517 |
+
"learning_rate": 4.1417165668662674e-08,
|
| 2518 |
+
"loss": 0.0302,
|
| 2519 |
+
"step": 157000
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 39.296407185628745,
|
| 2523 |
+
"grad_norm": 0.07879871129989624,
|
| 2524 |
+
"learning_rate": 3.517964071856287e-08,
|
| 2525 |
+
"loss": 0.027,
|
| 2526 |
+
"step": 157500
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 39.421157684630735,
|
| 2530 |
+
"grad_norm": 0.03594490885734558,
|
| 2531 |
+
"learning_rate": 2.8942115768463073e-08,
|
| 2532 |
+
"loss": 0.033,
|
| 2533 |
+
"step": 158000
|
| 2534 |
+
},
|
| 2535 |
+
{
|
| 2536 |
+
"epoch": 39.54590818363273,
|
| 2537 |
+
"grad_norm": 0.12444847822189331,
|
| 2538 |
+
"learning_rate": 2.2704590818363273e-08,
|
| 2539 |
+
"loss": 0.0271,
|
| 2540 |
+
"step": 158500
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 39.67065868263473,
|
| 2544 |
+
"grad_norm": 47.82669448852539,
|
| 2545 |
+
"learning_rate": 1.6467065868263473e-08,
|
| 2546 |
+
"loss": 0.0276,
|
| 2547 |
+
"step": 159000
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 39.79540918163673,
|
| 2551 |
+
"grad_norm": 0.1385308802127838,
|
| 2552 |
+
"learning_rate": 1.0229540918163672e-08,
|
| 2553 |
+
"loss": 0.03,
|
| 2554 |
+
"step": 159500
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 39.920159680638726,
|
| 2558 |
+
"grad_norm": 0.1429419070482254,
|
| 2559 |
+
"learning_rate": 3.992015968063871e-09,
|
| 2560 |
+
"loss": 0.0303,
|
| 2561 |
+
"step": 160000
|
| 2562 |
+
}
|
| 2563 |
+
],
|
| 2564 |
+
"logging_steps": 500,
|
| 2565 |
+
"max_steps": 160320,
|
| 2566 |
+
"num_input_tokens_seen": 0,
|
| 2567 |
+
"num_train_epochs": 40,
|
| 2568 |
+
"save_steps": 500,
|
| 2569 |
+
"stateful_callbacks": {
|
| 2570 |
+
"TrainerControl": {
|
| 2571 |
+
"args": {
|
| 2572 |
+
"should_epoch_stop": false,
|
| 2573 |
+
"should_evaluate": false,
|
| 2574 |
+
"should_log": false,
|
| 2575 |
+
"should_save": true,
|
| 2576 |
+
"should_training_stop": true
|
| 2577 |
+
},
|
| 2578 |
+
"attributes": {}
|
| 2579 |
+
}
|
| 2580 |
+
},
|
| 2581 |
+
"total_flos": 4.21770798769152e+16,
|
| 2582 |
+
"train_batch_size": 4,
|
| 2583 |
+
"trial_name": null,
|
| 2584 |
+
"trial_params": null
|
| 2585 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:162afc2cd7810ba9eb3e571711d88bc2d858afcfe980344a5610c4caccebeac1
|
| 3 |
+
size 5112
|