Commit
•
0d29304
1
Parent(s):
ebb3615
First commit
Browse files- all_results.json +14 -0
- config.json +28 -0
- eval_results.json +9 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +1 -0
- tokenizer.json +0 -0
- tokenizer_config.json +1 -0
- train_results.json +8 -0
- trainer_state.json +1305 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.71,
|
| 3 |
+
"eval_loss": 1.3815752267837524,
|
| 4 |
+
"eval_runtime": 184.9651,
|
| 5 |
+
"eval_samples": 181300,
|
| 6 |
+
"eval_samples_per_second": 980.185,
|
| 7 |
+
"eval_steps_per_second": 3.833,
|
| 8 |
+
"perplexity": 3.9811679331316605,
|
| 9 |
+
"train_loss": 0.00010499673156234791,
|
| 10 |
+
"train_runtime": 1054.6912,
|
| 11 |
+
"train_samples": 4349088,
|
| 12 |
+
"train_samples_per_second": 19418.005,
|
| 13 |
+
"train_steps_per_second": 9.481
|
| 14 |
+
}
|
config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "finiteautomata/robertuito-base-cased",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"RobertaForMaskedLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"bos_token_id": 0,
|
| 8 |
+
"classifier_dropout": null,
|
| 9 |
+
"eos_token_id": 2,
|
| 10 |
+
"gradient_checkpointing": false,
|
| 11 |
+
"hidden_act": "gelu",
|
| 12 |
+
"hidden_dropout_prob": 0.1,
|
| 13 |
+
"hidden_size": 768,
|
| 14 |
+
"initializer_range": 0.02,
|
| 15 |
+
"intermediate_size": 3072,
|
| 16 |
+
"layer_norm_eps": 1e-12,
|
| 17 |
+
"max_position_embeddings": 130,
|
| 18 |
+
"model_type": "roberta",
|
| 19 |
+
"num_attention_heads": 12,
|
| 20 |
+
"num_hidden_layers": 12,
|
| 21 |
+
"pad_token_id": 1,
|
| 22 |
+
"position_embedding_type": "absolute",
|
| 23 |
+
"torch_dtype": "float32",
|
| 24 |
+
"transformers_version": "4.13.0.dev0",
|
| 25 |
+
"type_vocab_size": 1,
|
| 26 |
+
"use_cache": true,
|
| 27 |
+
"vocab_size": 30000
|
| 28 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.71,
|
| 3 |
+
"eval_loss": 1.3815752267837524,
|
| 4 |
+
"eval_runtime": 184.9651,
|
| 5 |
+
"eval_samples": 181300,
|
| 6 |
+
"eval_samples_per_second": 980.185,
|
| 7 |
+
"eval_steps_per_second": 3.833,
|
| 8 |
+
"perplexity": 3.9811679331316605
|
| 9 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:279137baf3425daf369f597b8586c81d34e7216a962f2549bfff9ad2298a49c3
|
| 3 |
+
size 435341035
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
| 1 |
+
{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": "<mask>"}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
| 1 |
+
{"bos_token": "<s>", "eos_token": "</s>", "sep_token": "</s>", "cls_token": "<s>", "unk_token": "<unk>", "pad_token": "<pad>", "mask_token": "<mask>", "special_tokens_map_file": "models/twerto-base-cased/special_tokens_map.json", "name_or_path": "finiteautomata/robertuito-base-cased", "tokenizer_class": "PreTrainedTokenizerFast"}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.71,
|
| 3 |
+
"train_loss": 0.00010499673156234791,
|
| 4 |
+
"train_runtime": 1054.6912,
|
| 5 |
+
"train_samples": 4349088,
|
| 6 |
+
"train_samples_per_second": 19418.005,
|
| 7 |
+
"train_steps_per_second": 9.481
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1305 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 4.7105774324562955,
|
| 5 |
+
"global_step": 10001,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.02,
|
| 12 |
+
"learning_rate": 2e-05,
|
| 13 |
+
"loss": 1.8914,
|
| 14 |
+
"step": 50
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.05,
|
| 18 |
+
"learning_rate": 4e-05,
|
| 19 |
+
"loss": 1.6734,
|
| 20 |
+
"step": 100
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.07,
|
| 24 |
+
"learning_rate": 6e-05,
|
| 25 |
+
"loss": 1.5728,
|
| 26 |
+
"step": 150
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.09,
|
| 30 |
+
"learning_rate": 8e-05,
|
| 31 |
+
"loss": 1.528,
|
| 32 |
+
"step": 200
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.12,
|
| 36 |
+
"learning_rate": 0.0001,
|
| 37 |
+
"loss": 1.4775,
|
| 38 |
+
"step": 250
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.14,
|
| 42 |
+
"learning_rate": 0.00012,
|
| 43 |
+
"loss": 1.4352,
|
| 44 |
+
"step": 300
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.16,
|
| 48 |
+
"learning_rate": 0.00014,
|
| 49 |
+
"loss": 1.4069,
|
| 50 |
+
"step": 350
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.19,
|
| 54 |
+
"learning_rate": 0.00016,
|
| 55 |
+
"loss": 1.375,
|
| 56 |
+
"step": 400
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.21,
|
| 60 |
+
"learning_rate": 0.00018,
|
| 61 |
+
"loss": 1.367,
|
| 62 |
+
"step": 450
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.24,
|
| 66 |
+
"learning_rate": 0.0002,
|
| 67 |
+
"loss": 1.3565,
|
| 68 |
+
"step": 500
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.26,
|
| 72 |
+
"learning_rate": 0.00022000000000000003,
|
| 73 |
+
"loss": 1.3423,
|
| 74 |
+
"step": 550
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.28,
|
| 78 |
+
"learning_rate": 0.00024,
|
| 79 |
+
"loss": 1.3311,
|
| 80 |
+
"step": 600
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.31,
|
| 84 |
+
"learning_rate": 0.00026000000000000003,
|
| 85 |
+
"loss": 1.3245,
|
| 86 |
+
"step": 650
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.33,
|
| 90 |
+
"learning_rate": 0.00028,
|
| 91 |
+
"loss": 1.3214,
|
| 92 |
+
"step": 700
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.35,
|
| 96 |
+
"learning_rate": 0.00030000000000000003,
|
| 97 |
+
"loss": 1.3247,
|
| 98 |
+
"step": 750
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.38,
|
| 102 |
+
"learning_rate": 0.00032,
|
| 103 |
+
"loss": 1.3089,
|
| 104 |
+
"step": 800
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.4,
|
| 108 |
+
"learning_rate": 0.00034,
|
| 109 |
+
"loss": 1.3194,
|
| 110 |
+
"step": 850
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.42,
|
| 114 |
+
"learning_rate": 0.00036,
|
| 115 |
+
"loss": 1.3225,
|
| 116 |
+
"step": 900
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.45,
|
| 120 |
+
"learning_rate": 0.00038,
|
| 121 |
+
"loss": 1.3254,
|
| 122 |
+
"step": 950
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.47,
|
| 126 |
+
"learning_rate": 0.0004,
|
| 127 |
+
"loss": 1.3139,
|
| 128 |
+
"step": 1000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.47,
|
| 132 |
+
"eval_loss": 1.5397893190383911,
|
| 133 |
+
"eval_runtime": 119.9433,
|
| 134 |
+
"eval_samples_per_second": 1511.548,
|
| 135 |
+
"eval_steps_per_second": 5.911,
|
| 136 |
+
"step": 1000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.49,
|
| 140 |
+
"learning_rate": 0.0003977777777777778,
|
| 141 |
+
"loss": 1.3313,
|
| 142 |
+
"step": 1050
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.52,
|
| 146 |
+
"learning_rate": 0.00039555555555555557,
|
| 147 |
+
"loss": 1.318,
|
| 148 |
+
"step": 1100
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 0.54,
|
| 152 |
+
"learning_rate": 0.0003933333333333333,
|
| 153 |
+
"loss": 1.3092,
|
| 154 |
+
"step": 1150
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.57,
|
| 158 |
+
"learning_rate": 0.0003911111111111111,
|
| 159 |
+
"loss": 1.3036,
|
| 160 |
+
"step": 1200
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.59,
|
| 164 |
+
"learning_rate": 0.0003888888888888889,
|
| 165 |
+
"loss": 1.2931,
|
| 166 |
+
"step": 1250
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.61,
|
| 170 |
+
"learning_rate": 0.00038666666666666667,
|
| 171 |
+
"loss": 1.2986,
|
| 172 |
+
"step": 1300
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.64,
|
| 176 |
+
"learning_rate": 0.00038444444444444447,
|
| 177 |
+
"loss": 1.4597,
|
| 178 |
+
"step": 1350
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.66,
|
| 182 |
+
"learning_rate": 0.0003822222222222223,
|
| 183 |
+
"loss": 1.8909,
|
| 184 |
+
"step": 1400
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.68,
|
| 188 |
+
"learning_rate": 0.00038,
|
| 189 |
+
"loss": 1.347,
|
| 190 |
+
"step": 1450
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.71,
|
| 194 |
+
"learning_rate": 0.00037777777777777777,
|
| 195 |
+
"loss": 1.3039,
|
| 196 |
+
"step": 1500
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 0.73,
|
| 200 |
+
"learning_rate": 0.00037555555555555557,
|
| 201 |
+
"loss": 1.2929,
|
| 202 |
+
"step": 1550
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.75,
|
| 206 |
+
"learning_rate": 0.0003733333333333334,
|
| 207 |
+
"loss": 1.2783,
|
| 208 |
+
"step": 1600
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.78,
|
| 212 |
+
"learning_rate": 0.0003711111111111111,
|
| 213 |
+
"loss": 1.2644,
|
| 214 |
+
"step": 1650
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.8,
|
| 218 |
+
"learning_rate": 0.0003688888888888889,
|
| 219 |
+
"loss": 1.255,
|
| 220 |
+
"step": 1700
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.82,
|
| 224 |
+
"learning_rate": 0.00036666666666666667,
|
| 225 |
+
"loss": 1.2509,
|
| 226 |
+
"step": 1750
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.85,
|
| 230 |
+
"learning_rate": 0.00036444444444444447,
|
| 231 |
+
"loss": 1.254,
|
| 232 |
+
"step": 1800
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 0.87,
|
| 236 |
+
"learning_rate": 0.0003622222222222222,
|
| 237 |
+
"loss": 1.2436,
|
| 238 |
+
"step": 1850
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 0.89,
|
| 242 |
+
"learning_rate": 0.00036,
|
| 243 |
+
"loss": 1.2499,
|
| 244 |
+
"step": 1900
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.92,
|
| 248 |
+
"learning_rate": 0.0003577777777777778,
|
| 249 |
+
"loss": 1.5927,
|
| 250 |
+
"step": 1950
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.94,
|
| 254 |
+
"learning_rate": 0.00035555555555555557,
|
| 255 |
+
"loss": 1.4569,
|
| 256 |
+
"step": 2000
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 0.94,
|
| 260 |
+
"eval_loss": 1.661188006401062,
|
| 261 |
+
"eval_runtime": 106.3952,
|
| 262 |
+
"eval_samples_per_second": 1704.024,
|
| 263 |
+
"eval_steps_per_second": 6.664,
|
| 264 |
+
"step": 2000
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.97,
|
| 268 |
+
"learning_rate": 0.0003533333333333333,
|
| 269 |
+
"loss": 1.333,
|
| 270 |
+
"step": 2050
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 0.99,
|
| 274 |
+
"learning_rate": 0.0003511111111111111,
|
| 275 |
+
"loss": 1.4076,
|
| 276 |
+
"step": 2100
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 1.01,
|
| 280 |
+
"learning_rate": 0.0003488888888888889,
|
| 281 |
+
"loss": 1.4328,
|
| 282 |
+
"step": 2150
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.04,
|
| 286 |
+
"learning_rate": 0.00034666666666666667,
|
| 287 |
+
"loss": 1.3461,
|
| 288 |
+
"step": 2200
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 1.06,
|
| 292 |
+
"learning_rate": 0.0003444444444444445,
|
| 293 |
+
"loss": 1.4831,
|
| 294 |
+
"step": 2250
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 1.08,
|
| 298 |
+
"learning_rate": 0.0003422222222222222,
|
| 299 |
+
"loss": 1.4935,
|
| 300 |
+
"step": 2300
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 1.11,
|
| 304 |
+
"learning_rate": 0.00034,
|
| 305 |
+
"loss": 1.4034,
|
| 306 |
+
"step": 2350
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 1.13,
|
| 310 |
+
"learning_rate": 0.00033777777777777777,
|
| 311 |
+
"loss": 1.3652,
|
| 312 |
+
"step": 2400
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 1.15,
|
| 316 |
+
"learning_rate": 0.0003355555555555556,
|
| 317 |
+
"loss": 1.2921,
|
| 318 |
+
"step": 2450
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.18,
|
| 322 |
+
"learning_rate": 0.0003333333333333334,
|
| 323 |
+
"loss": 1.2564,
|
| 324 |
+
"step": 2500
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.2,
|
| 328 |
+
"learning_rate": 0.0003311111111111111,
|
| 329 |
+
"loss": 1.2344,
|
| 330 |
+
"step": 2550
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 1.22,
|
| 334 |
+
"learning_rate": 0.00032888888888888887,
|
| 335 |
+
"loss": 1.2705,
|
| 336 |
+
"step": 2600
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 1.25,
|
| 340 |
+
"learning_rate": 0.0003266666666666667,
|
| 341 |
+
"loss": 1.3214,
|
| 342 |
+
"step": 2650
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 1.27,
|
| 346 |
+
"learning_rate": 0.0003244444444444445,
|
| 347 |
+
"loss": 1.2318,
|
| 348 |
+
"step": 2700
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 1.3,
|
| 352 |
+
"learning_rate": 0.0003222222222222222,
|
| 353 |
+
"loss": 1.2029,
|
| 354 |
+
"step": 2750
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 1.32,
|
| 358 |
+
"learning_rate": 0.00032,
|
| 359 |
+
"loss": 1.2056,
|
| 360 |
+
"step": 2800
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.34,
|
| 364 |
+
"learning_rate": 0.0003177777777777778,
|
| 365 |
+
"loss": 1.1931,
|
| 366 |
+
"step": 2850
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 1.37,
|
| 370 |
+
"learning_rate": 0.0003155555555555556,
|
| 371 |
+
"loss": 1.1812,
|
| 372 |
+
"step": 2900
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 1.39,
|
| 376 |
+
"learning_rate": 0.0003133333333333333,
|
| 377 |
+
"loss": 1.1863,
|
| 378 |
+
"step": 2950
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 1.41,
|
| 382 |
+
"learning_rate": 0.0003111111111111111,
|
| 383 |
+
"loss": 1.1795,
|
| 384 |
+
"step": 3000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 1.41,
|
| 388 |
+
"eval_loss": 1.5454002618789673,
|
| 389 |
+
"eval_runtime": 104.0653,
|
| 390 |
+
"eval_samples_per_second": 1742.176,
|
| 391 |
+
"eval_steps_per_second": 6.813,
|
| 392 |
+
"step": 3000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 1.44,
|
| 396 |
+
"learning_rate": 0.00030888888888888893,
|
| 397 |
+
"loss": 1.1856,
|
| 398 |
+
"step": 3050
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 1.46,
|
| 402 |
+
"learning_rate": 0.0003066666666666667,
|
| 403 |
+
"loss": 1.1765,
|
| 404 |
+
"step": 3100
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 1.48,
|
| 408 |
+
"learning_rate": 0.0003044444444444444,
|
| 409 |
+
"loss": 1.3244,
|
| 410 |
+
"step": 3150
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 1.51,
|
| 414 |
+
"learning_rate": 0.0003022222222222222,
|
| 415 |
+
"loss": 1.2179,
|
| 416 |
+
"step": 3200
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 1.53,
|
| 420 |
+
"learning_rate": 0.00030000000000000003,
|
| 421 |
+
"loss": 1.2114,
|
| 422 |
+
"step": 3250
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.55,
|
| 426 |
+
"learning_rate": 0.0002977777777777778,
|
| 427 |
+
"loss": 1.1968,
|
| 428 |
+
"step": 3300
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 1.58,
|
| 432 |
+
"learning_rate": 0.0002955555555555556,
|
| 433 |
+
"loss": 1.1798,
|
| 434 |
+
"step": 3350
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.6,
|
| 438 |
+
"learning_rate": 0.0002933333333333333,
|
| 439 |
+
"loss": 1.1851,
|
| 440 |
+
"step": 3400
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 1.62,
|
| 444 |
+
"learning_rate": 0.00029111111111111113,
|
| 445 |
+
"loss": 1.2182,
|
| 446 |
+
"step": 3450
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 1.65,
|
| 450 |
+
"learning_rate": 0.0002888888888888889,
|
| 451 |
+
"loss": 1.1791,
|
| 452 |
+
"step": 3500
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.67,
|
| 456 |
+
"learning_rate": 0.0002866666666666667,
|
| 457 |
+
"loss": 1.1842,
|
| 458 |
+
"step": 3550
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.7,
|
| 462 |
+
"learning_rate": 0.0002844444444444445,
|
| 463 |
+
"loss": 1.2017,
|
| 464 |
+
"step": 3600
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 1.72,
|
| 468 |
+
"learning_rate": 0.00028222222222222223,
|
| 469 |
+
"loss": 1.2376,
|
| 470 |
+
"step": 3650
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.74,
|
| 474 |
+
"learning_rate": 0.00028,
|
| 475 |
+
"loss": 1.3329,
|
| 476 |
+
"step": 3700
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 1.77,
|
| 480 |
+
"learning_rate": 0.0002777777777777778,
|
| 481 |
+
"loss": 1.2929,
|
| 482 |
+
"step": 3750
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.79,
|
| 486 |
+
"learning_rate": 0.0002755555555555556,
|
| 487 |
+
"loss": 1.2387,
|
| 488 |
+
"step": 3800
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 1.81,
|
| 492 |
+
"learning_rate": 0.00027333333333333333,
|
| 493 |
+
"loss": 1.1944,
|
| 494 |
+
"step": 3850
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.84,
|
| 498 |
+
"learning_rate": 0.00027111111111111113,
|
| 499 |
+
"loss": 1.196,
|
| 500 |
+
"step": 3900
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.86,
|
| 504 |
+
"learning_rate": 0.00026888888888888893,
|
| 505 |
+
"loss": 1.1817,
|
| 506 |
+
"step": 3950
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.88,
|
| 510 |
+
"learning_rate": 0.0002666666666666667,
|
| 511 |
+
"loss": 1.1769,
|
| 512 |
+
"step": 4000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 1.88,
|
| 516 |
+
"eval_loss": 1.526174545288086,
|
| 517 |
+
"eval_runtime": 104.8904,
|
| 518 |
+
"eval_samples_per_second": 1728.471,
|
| 519 |
+
"eval_steps_per_second": 6.759,
|
| 520 |
+
"step": 4000
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 1.91,
|
| 524 |
+
"learning_rate": 0.00026444444444444443,
|
| 525 |
+
"loss": 1.1606,
|
| 526 |
+
"step": 4050
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 1.93,
|
| 530 |
+
"learning_rate": 0.00026222222222222223,
|
| 531 |
+
"loss": 1.2703,
|
| 532 |
+
"step": 4100
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 1.95,
|
| 536 |
+
"learning_rate": 0.00026000000000000003,
|
| 537 |
+
"loss": 1.2824,
|
| 538 |
+
"step": 4150
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.98,
|
| 542 |
+
"learning_rate": 0.00025777777777777783,
|
| 543 |
+
"loss": 1.2228,
|
| 544 |
+
"step": 4200
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 2.0,
|
| 548 |
+
"learning_rate": 0.00025555555555555553,
|
| 549 |
+
"loss": 1.2338,
|
| 550 |
+
"step": 4250
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 2.03,
|
| 554 |
+
"learning_rate": 0.00025333333333333333,
|
| 555 |
+
"loss": 1.163,
|
| 556 |
+
"step": 4300
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 2.05,
|
| 560 |
+
"learning_rate": 0.00025111111111111113,
|
| 561 |
+
"loss": 1.1332,
|
| 562 |
+
"step": 4350
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 2.07,
|
| 566 |
+
"learning_rate": 0.0002488888888888889,
|
| 567 |
+
"loss": 1.1206,
|
| 568 |
+
"step": 4400
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 2.1,
|
| 572 |
+
"learning_rate": 0.0002466666666666667,
|
| 573 |
+
"loss": 1.1597,
|
| 574 |
+
"step": 4450
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 2.12,
|
| 578 |
+
"learning_rate": 0.0002444444444444445,
|
| 579 |
+
"loss": 1.1259,
|
| 580 |
+
"step": 4500
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 2.14,
|
| 584 |
+
"learning_rate": 0.0002422222222222222,
|
| 585 |
+
"loss": 1.1178,
|
| 586 |
+
"step": 4550
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 2.17,
|
| 590 |
+
"learning_rate": 0.00024,
|
| 591 |
+
"loss": 1.1119,
|
| 592 |
+
"step": 4600
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 2.19,
|
| 596 |
+
"learning_rate": 0.00023777777777777778,
|
| 597 |
+
"loss": 1.1104,
|
| 598 |
+
"step": 4650
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 2.21,
|
| 602 |
+
"learning_rate": 0.00023555555555555558,
|
| 603 |
+
"loss": 1.1064,
|
| 604 |
+
"step": 4700
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 2.24,
|
| 608 |
+
"learning_rate": 0.00023333333333333336,
|
| 609 |
+
"loss": 1.1111,
|
| 610 |
+
"step": 4750
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 2.26,
|
| 614 |
+
"learning_rate": 0.0002311111111111111,
|
| 615 |
+
"loss": 1.0994,
|
| 616 |
+
"step": 4800
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 2.28,
|
| 620 |
+
"learning_rate": 0.00022888888888888888,
|
| 621 |
+
"loss": 1.1002,
|
| 622 |
+
"step": 4850
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 2.31,
|
| 626 |
+
"learning_rate": 0.00022666666666666668,
|
| 627 |
+
"loss": 1.1486,
|
| 628 |
+
"step": 4900
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 2.33,
|
| 632 |
+
"learning_rate": 0.00022444444444444446,
|
| 633 |
+
"loss": 1.1115,
|
| 634 |
+
"step": 4950
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 2.36,
|
| 638 |
+
"learning_rate": 0.00022222222222222223,
|
| 639 |
+
"loss": 1.1026,
|
| 640 |
+
"step": 5000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 2.36,
|
| 644 |
+
"eval_loss": 1.4830693006515503,
|
| 645 |
+
"eval_runtime": 104.1599,
|
| 646 |
+
"eval_samples_per_second": 1740.593,
|
| 647 |
+
"eval_steps_per_second": 6.807,
|
| 648 |
+
"step": 5000
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 2.38,
|
| 652 |
+
"learning_rate": 0.00022000000000000003,
|
| 653 |
+
"loss": 1.1098,
|
| 654 |
+
"step": 5050
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 2.4,
|
| 658 |
+
"learning_rate": 0.00021777777777777778,
|
| 659 |
+
"loss": 1.0984,
|
| 660 |
+
"step": 5100
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 2.43,
|
| 664 |
+
"learning_rate": 0.00021555555555555556,
|
| 665 |
+
"loss": 1.0858,
|
| 666 |
+
"step": 5150
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 2.45,
|
| 670 |
+
"learning_rate": 0.00021333333333333333,
|
| 671 |
+
"loss": 1.1243,
|
| 672 |
+
"step": 5200
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 2.47,
|
| 676 |
+
"learning_rate": 0.00021111111111111113,
|
| 677 |
+
"loss": 1.125,
|
| 678 |
+
"step": 5250
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 2.5,
|
| 682 |
+
"learning_rate": 0.0002088888888888889,
|
| 683 |
+
"loss": 1.108,
|
| 684 |
+
"step": 5300
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 2.52,
|
| 688 |
+
"learning_rate": 0.00020666666666666668,
|
| 689 |
+
"loss": 1.0821,
|
| 690 |
+
"step": 5350
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 2.54,
|
| 694 |
+
"learning_rate": 0.00020444444444444443,
|
| 695 |
+
"loss": 1.087,
|
| 696 |
+
"step": 5400
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 2.57,
|
| 700 |
+
"learning_rate": 0.00020222222222222223,
|
| 701 |
+
"loss": 1.0814,
|
| 702 |
+
"step": 5450
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 2.59,
|
| 706 |
+
"learning_rate": 0.0002,
|
| 707 |
+
"loss": 1.0799,
|
| 708 |
+
"step": 5500
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 2.61,
|
| 712 |
+
"learning_rate": 0.00019777777777777778,
|
| 713 |
+
"loss": 1.0771,
|
| 714 |
+
"step": 5550
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 2.64,
|
| 718 |
+
"learning_rate": 0.00019555555555555556,
|
| 719 |
+
"loss": 1.0777,
|
| 720 |
+
"step": 5600
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 2.66,
|
| 724 |
+
"learning_rate": 0.00019333333333333333,
|
| 725 |
+
"loss": 1.0765,
|
| 726 |
+
"step": 5650
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 2.68,
|
| 730 |
+
"learning_rate": 0.00019111111111111114,
|
| 731 |
+
"loss": 1.0703,
|
| 732 |
+
"step": 5700
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 2.71,
|
| 736 |
+
"learning_rate": 0.00018888888888888888,
|
| 737 |
+
"loss": 1.0722,
|
| 738 |
+
"step": 5750
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 2.73,
|
| 742 |
+
"learning_rate": 0.0001866666666666667,
|
| 743 |
+
"loss": 1.0917,
|
| 744 |
+
"step": 5800
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 2.76,
|
| 748 |
+
"learning_rate": 0.00018444444444444446,
|
| 749 |
+
"loss": 1.0741,
|
| 750 |
+
"step": 5850
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 2.78,
|
| 754 |
+
"learning_rate": 0.00018222222222222224,
|
| 755 |
+
"loss": 1.077,
|
| 756 |
+
"step": 5900
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 2.8,
|
| 760 |
+
"learning_rate": 0.00018,
|
| 761 |
+
"loss": 1.0683,
|
| 762 |
+
"step": 5950
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 2.83,
|
| 766 |
+
"learning_rate": 0.00017777777777777779,
|
| 767 |
+
"loss": 1.0711,
|
| 768 |
+
"step": 6000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 2.83,
|
| 772 |
+
"eval_loss": 1.4865516424179077,
|
| 773 |
+
"eval_runtime": 104.1916,
|
| 774 |
+
"eval_samples_per_second": 1740.063,
|
| 775 |
+
"eval_steps_per_second": 6.805,
|
| 776 |
+
"step": 6000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 2.85,
|
| 780 |
+
"learning_rate": 0.00017555555555555556,
|
| 781 |
+
"loss": 1.0955,
|
| 782 |
+
"step": 6050
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 2.87,
|
| 786 |
+
"learning_rate": 0.00017333333333333334,
|
| 787 |
+
"loss": 1.1421,
|
| 788 |
+
"step": 6100
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 2.9,
|
| 792 |
+
"learning_rate": 0.0001711111111111111,
|
| 793 |
+
"loss": 1.1308,
|
| 794 |
+
"step": 6150
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 2.92,
|
| 798 |
+
"learning_rate": 0.00016888888888888889,
|
| 799 |
+
"loss": 1.0915,
|
| 800 |
+
"step": 6200
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 2.94,
|
| 804 |
+
"learning_rate": 0.0001666666666666667,
|
| 805 |
+
"loss": 1.0693,
|
| 806 |
+
"step": 6250
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 2.97,
|
| 810 |
+
"learning_rate": 0.00016444444444444444,
|
| 811 |
+
"loss": 1.0804,
|
| 812 |
+
"step": 6300
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 2.99,
|
| 816 |
+
"learning_rate": 0.00016222222222222224,
|
| 817 |
+
"loss": 1.0684,
|
| 818 |
+
"step": 6350
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 3.01,
|
| 822 |
+
"learning_rate": 0.00016,
|
| 823 |
+
"loss": 1.0886,
|
| 824 |
+
"step": 6400
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 3.04,
|
| 828 |
+
"learning_rate": 0.0001577777777777778,
|
| 829 |
+
"loss": 1.0576,
|
| 830 |
+
"step": 6450
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 3.06,
|
| 834 |
+
"learning_rate": 0.00015555555555555556,
|
| 835 |
+
"loss": 1.0446,
|
| 836 |
+
"step": 6500
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 3.09,
|
| 840 |
+
"learning_rate": 0.00015333333333333334,
|
| 841 |
+
"loss": 1.0426,
|
| 842 |
+
"step": 6550
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 3.11,
|
| 846 |
+
"learning_rate": 0.0001511111111111111,
|
| 847 |
+
"loss": 1.0404,
|
| 848 |
+
"step": 6600
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 3.13,
|
| 852 |
+
"learning_rate": 0.0001488888888888889,
|
| 853 |
+
"loss": 1.0446,
|
| 854 |
+
"step": 6650
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 3.16,
|
| 858 |
+
"learning_rate": 0.00014666666666666666,
|
| 859 |
+
"loss": 1.0564,
|
| 860 |
+
"step": 6700
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 3.18,
|
| 864 |
+
"learning_rate": 0.00014444444444444444,
|
| 865 |
+
"loss": 1.0452,
|
| 866 |
+
"step": 6750
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 3.2,
|
| 870 |
+
"learning_rate": 0.00014222222222222224,
|
| 871 |
+
"loss": 1.0327,
|
| 872 |
+
"step": 6800
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 3.23,
|
| 876 |
+
"learning_rate": 0.00014,
|
| 877 |
+
"loss": 1.0353,
|
| 878 |
+
"step": 6850
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 3.25,
|
| 882 |
+
"learning_rate": 0.0001377777777777778,
|
| 883 |
+
"loss": 1.0259,
|
| 884 |
+
"step": 6900
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 3.27,
|
| 888 |
+
"learning_rate": 0.00013555555555555556,
|
| 889 |
+
"loss": 1.0328,
|
| 890 |
+
"step": 6950
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 3.3,
|
| 894 |
+
"learning_rate": 0.00013333333333333334,
|
| 895 |
+
"loss": 1.0321,
|
| 896 |
+
"step": 7000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 3.3,
|
| 900 |
+
"eval_loss": 1.4409459829330444,
|
| 901 |
+
"eval_runtime": 104.0221,
|
| 902 |
+
"eval_samples_per_second": 1742.9,
|
| 903 |
+
"eval_steps_per_second": 6.816,
|
| 904 |
+
"step": 7000
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 3.32,
|
| 908 |
+
"learning_rate": 0.00013111111111111111,
|
| 909 |
+
"loss": 1.0295,
|
| 910 |
+
"step": 7050
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 3.34,
|
| 914 |
+
"learning_rate": 0.00012888888888888892,
|
| 915 |
+
"loss": 1.0262,
|
| 916 |
+
"step": 7100
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 3.37,
|
| 920 |
+
"learning_rate": 0.00012666666666666666,
|
| 921 |
+
"loss": 1.0278,
|
| 922 |
+
"step": 7150
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 3.39,
|
| 926 |
+
"learning_rate": 0.00012444444444444444,
|
| 927 |
+
"loss": 1.0227,
|
| 928 |
+
"step": 7200
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 3.41,
|
| 932 |
+
"learning_rate": 0.00012222222222222224,
|
| 933 |
+
"loss": 1.0262,
|
| 934 |
+
"step": 7250
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 3.44,
|
| 938 |
+
"learning_rate": 0.00012,
|
| 939 |
+
"loss": 1.0205,
|
| 940 |
+
"step": 7300
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 3.46,
|
| 944 |
+
"learning_rate": 0.00011777777777777779,
|
| 945 |
+
"loss": 1.0107,
|
| 946 |
+
"step": 7350
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 3.49,
|
| 950 |
+
"learning_rate": 0.00011555555555555555,
|
| 951 |
+
"loss": 1.0218,
|
| 952 |
+
"step": 7400
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 3.51,
|
| 956 |
+
"learning_rate": 0.00011333333333333334,
|
| 957 |
+
"loss": 1.0205,
|
| 958 |
+
"step": 7450
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 3.53,
|
| 962 |
+
"learning_rate": 0.00011111111111111112,
|
| 963 |
+
"loss": 1.0196,
|
| 964 |
+
"step": 7500
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 3.56,
|
| 968 |
+
"learning_rate": 0.00010888888888888889,
|
| 969 |
+
"loss": 1.0197,
|
| 970 |
+
"step": 7550
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 3.58,
|
| 974 |
+
"learning_rate": 0.00010666666666666667,
|
| 975 |
+
"loss": 1.0139,
|
| 976 |
+
"step": 7600
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 3.6,
|
| 980 |
+
"learning_rate": 0.00010444444444444445,
|
| 981 |
+
"loss": 1.0011,
|
| 982 |
+
"step": 7650
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 3.63,
|
| 986 |
+
"learning_rate": 0.00010222222222222222,
|
| 987 |
+
"loss": 1.0056,
|
| 988 |
+
"step": 7700
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 3.65,
|
| 992 |
+
"learning_rate": 0.0001,
|
| 993 |
+
"loss": 1.0152,
|
| 994 |
+
"step": 7750
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 3.67,
|
| 998 |
+
"learning_rate": 9.777777777777778e-05,
|
| 999 |
+
"loss": 1.0078,
|
| 1000 |
+
"step": 7800
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 3.7,
|
| 1004 |
+
"learning_rate": 9.555555555555557e-05,
|
| 1005 |
+
"loss": 1.0101,
|
| 1006 |
+
"step": 7850
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 3.72,
|
| 1010 |
+
"learning_rate": 9.333333333333334e-05,
|
| 1011 |
+
"loss": 0.9983,
|
| 1012 |
+
"step": 7900
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 3.74,
|
| 1016 |
+
"learning_rate": 9.111111111111112e-05,
|
| 1017 |
+
"loss": 0.9922,
|
| 1018 |
+
"step": 7950
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 3.77,
|
| 1022 |
+
"learning_rate": 8.888888888888889e-05,
|
| 1023 |
+
"loss": 0.9994,
|
| 1024 |
+
"step": 8000
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 3.77,
|
| 1028 |
+
"eval_loss": 1.420628547668457,
|
| 1029 |
+
"eval_runtime": 104.1119,
|
| 1030 |
+
"eval_samples_per_second": 1741.396,
|
| 1031 |
+
"eval_steps_per_second": 6.81,
|
| 1032 |
+
"step": 8000
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 3.79,
|
| 1036 |
+
"learning_rate": 8.666666666666667e-05,
|
| 1037 |
+
"loss": 1.0066,
|
| 1038 |
+
"step": 8050
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 3.82,
|
| 1042 |
+
"learning_rate": 8.444444444444444e-05,
|
| 1043 |
+
"loss": 1.0028,
|
| 1044 |
+
"step": 8100
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 3.84,
|
| 1048 |
+
"learning_rate": 8.222222222222222e-05,
|
| 1049 |
+
"loss": 1.0007,
|
| 1050 |
+
"step": 8150
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 3.86,
|
| 1054 |
+
"learning_rate": 8e-05,
|
| 1055 |
+
"loss": 1.0081,
|
| 1056 |
+
"step": 8200
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 3.89,
|
| 1060 |
+
"learning_rate": 7.777777777777778e-05,
|
| 1061 |
+
"loss": 0.9953,
|
| 1062 |
+
"step": 8250
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 3.91,
|
| 1066 |
+
"learning_rate": 7.555555555555556e-05,
|
| 1067 |
+
"loss": 0.9866,
|
| 1068 |
+
"step": 8300
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 3.93,
|
| 1072 |
+
"learning_rate": 7.333333333333333e-05,
|
| 1073 |
+
"loss": 0.9961,
|
| 1074 |
+
"step": 8350
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 3.96,
|
| 1078 |
+
"learning_rate": 7.111111111111112e-05,
|
| 1079 |
+
"loss": 0.9891,
|
| 1080 |
+
"step": 8400
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 3.98,
|
| 1084 |
+
"learning_rate": 6.88888888888889e-05,
|
| 1085 |
+
"loss": 0.9917,
|
| 1086 |
+
"step": 8450
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 4.0,
|
| 1090 |
+
"learning_rate": 6.666666666666667e-05,
|
| 1091 |
+
"loss": 1.0035,
|
| 1092 |
+
"step": 8500
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 4.03,
|
| 1096 |
+
"learning_rate": 6.444444444444446e-05,
|
| 1097 |
+
"loss": 0.9899,
|
| 1098 |
+
"step": 8550
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 4.05,
|
| 1102 |
+
"learning_rate": 6.222222222222222e-05,
|
| 1103 |
+
"loss": 0.9769,
|
| 1104 |
+
"step": 8600
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 4.07,
|
| 1108 |
+
"learning_rate": 6e-05,
|
| 1109 |
+
"loss": 0.9781,
|
| 1110 |
+
"step": 8650
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 4.1,
|
| 1114 |
+
"learning_rate": 5.7777777777777776e-05,
|
| 1115 |
+
"loss": 0.9802,
|
| 1116 |
+
"step": 8700
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 4.12,
|
| 1120 |
+
"learning_rate": 5.555555555555556e-05,
|
| 1121 |
+
"loss": 0.9727,
|
| 1122 |
+
"step": 8750
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 4.15,
|
| 1126 |
+
"learning_rate": 5.333333333333333e-05,
|
| 1127 |
+
"loss": 0.9742,
|
| 1128 |
+
"step": 8800
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 4.17,
|
| 1132 |
+
"learning_rate": 5.111111111111111e-05,
|
| 1133 |
+
"loss": 0.9723,
|
| 1134 |
+
"step": 8850
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 4.19,
|
| 1138 |
+
"learning_rate": 4.888888888888889e-05,
|
| 1139 |
+
"loss": 0.9809,
|
| 1140 |
+
"step": 8900
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 4.22,
|
| 1144 |
+
"learning_rate": 4.666666666666667e-05,
|
| 1145 |
+
"loss": 0.9753,
|
| 1146 |
+
"step": 8950
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 4.24,
|
| 1150 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 1151 |
+
"loss": 0.9774,
|
| 1152 |
+
"step": 9000
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 4.24,
|
| 1156 |
+
"eval_loss": 1.3941148519515991,
|
| 1157 |
+
"eval_runtime": 104.2747,
|
| 1158 |
+
"eval_samples_per_second": 1738.677,
|
| 1159 |
+
"eval_steps_per_second": 6.799,
|
| 1160 |
+
"step": 9000
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 4.26,
|
| 1164 |
+
"learning_rate": 4.222222222222222e-05,
|
| 1165 |
+
"loss": 0.9603,
|
| 1166 |
+
"step": 9050
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 4.29,
|
| 1170 |
+
"learning_rate": 4e-05,
|
| 1171 |
+
"loss": 0.9632,
|
| 1172 |
+
"step": 9100
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 4.31,
|
| 1176 |
+
"learning_rate": 3.777777777777778e-05,
|
| 1177 |
+
"loss": 0.9701,
|
| 1178 |
+
"step": 9150
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 4.33,
|
| 1182 |
+
"learning_rate": 3.555555555555556e-05,
|
| 1183 |
+
"loss": 0.966,
|
| 1184 |
+
"step": 9200
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 4.36,
|
| 1188 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 1189 |
+
"loss": 0.9725,
|
| 1190 |
+
"step": 9250
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 4.38,
|
| 1194 |
+
"learning_rate": 3.111111111111111e-05,
|
| 1195 |
+
"loss": 0.9644,
|
| 1196 |
+
"step": 9300
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 4.4,
|
| 1200 |
+
"learning_rate": 2.8888888888888888e-05,
|
| 1201 |
+
"loss": 0.9613,
|
| 1202 |
+
"step": 9350
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 4.43,
|
| 1206 |
+
"learning_rate": 2.6666666666666667e-05,
|
| 1207 |
+
"loss": 0.971,
|
| 1208 |
+
"step": 9400
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 4.45,
|
| 1212 |
+
"learning_rate": 2.4444444444444445e-05,
|
| 1213 |
+
"loss": 0.9592,
|
| 1214 |
+
"step": 9450
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 4.47,
|
| 1218 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 1219 |
+
"loss": 0.9692,
|
| 1220 |
+
"step": 9500
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 4.5,
|
| 1224 |
+
"learning_rate": 2e-05,
|
| 1225 |
+
"loss": 0.9639,
|
| 1226 |
+
"step": 9550
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 4.52,
|
| 1230 |
+
"learning_rate": 1.777777777777778e-05,
|
| 1231 |
+
"loss": 0.9562,
|
| 1232 |
+
"step": 9600
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 4.55,
|
| 1236 |
+
"learning_rate": 1.5555555555555555e-05,
|
| 1237 |
+
"loss": 0.9645,
|
| 1238 |
+
"step": 9650
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 4.57,
|
| 1242 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 1243 |
+
"loss": 0.9638,
|
| 1244 |
+
"step": 9700
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 4.59,
|
| 1248 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 1249 |
+
"loss": 0.9614,
|
| 1250 |
+
"step": 9750
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 4.62,
|
| 1254 |
+
"learning_rate": 8.88888888888889e-06,
|
| 1255 |
+
"loss": 0.9607,
|
| 1256 |
+
"step": 9800
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 4.64,
|
| 1260 |
+
"learning_rate": 6.666666666666667e-06,
|
| 1261 |
+
"loss": 0.9602,
|
| 1262 |
+
"step": 9850
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 4.66,
|
| 1266 |
+
"learning_rate": 4.444444444444445e-06,
|
| 1267 |
+
"loss": 0.9643,
|
| 1268 |
+
"step": 9900
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 4.69,
|
| 1272 |
+
"learning_rate": 2.2222222222222225e-06,
|
| 1273 |
+
"loss": 0.9525,
|
| 1274 |
+
"step": 9950
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 4.71,
|
| 1278 |
+
"learning_rate": 0.0,
|
| 1279 |
+
"loss": 0.9618,
|
| 1280 |
+
"step": 10000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 4.71,
|
| 1284 |
+
"eval_loss": 1.3830920457839966,
|
| 1285 |
+
"eval_runtime": 104.4628,
|
| 1286 |
+
"eval_samples_per_second": 1735.547,
|
| 1287 |
+
"eval_steps_per_second": 6.787,
|
| 1288 |
+
"step": 10000
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 4.71,
|
| 1292 |
+
"step": 10001,
|
| 1293 |
+
"total_flos": 1.685090114171044e+17,
|
| 1294 |
+
"train_loss": 0.00010499673156234791,
|
| 1295 |
+
"train_runtime": 1054.6912,
|
| 1296 |
+
"train_samples_per_second": 19418.005,
|
| 1297 |
+
"train_steps_per_second": 9.481
|
| 1298 |
+
}
|
| 1299 |
+
],
|
| 1300 |
+
"max_steps": 10000,
|
| 1301 |
+
"num_train_epochs": 5,
|
| 1302 |
+
"total_flos": 1.685090114171044e+17,
|
| 1303 |
+
"trial_name": null,
|
| 1304 |
+
"trial_params": null
|
| 1305 |
+
}
|