Commit
•
5cd6bc6
1
Parent(s):
73b57b1
Upload 10 files
Browse files- config.json +30 -0
- generation_config.json +6 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +1 -0
- trainer_state.json +3485 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "data/models/Mixtral-8x7B-v0.3",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"MixtralForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"bos_token_id": 1,
|
| 8 |
+
"eos_token_id": 2,
|
| 9 |
+
"hidden_act": "silu",
|
| 10 |
+
"hidden_size": 768,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"intermediate_size": 768,
|
| 13 |
+
"max_position_embeddings": 512,
|
| 14 |
+
"model_type": "mixtral",
|
| 15 |
+
"num_attention_heads": 8,
|
| 16 |
+
"num_experts_per_tok": 1,
|
| 17 |
+
"num_hidden_layers": 8,
|
| 18 |
+
"num_key_value_heads": 8,
|
| 19 |
+
"num_local_experts": 8,
|
| 20 |
+
"output_router_logits": false,
|
| 21 |
+
"rms_norm_eps": 1e-05,
|
| 22 |
+
"rope_theta": 1000000.0,
|
| 23 |
+
"router_aux_loss_coef": 0.02,
|
| 24 |
+
"sliding_window": null,
|
| 25 |
+
"tie_word_embeddings": false,
|
| 26 |
+
"torch_dtype": "bfloat16",
|
| 27 |
+
"transformers_version": "4.37.1",
|
| 28 |
+
"use_cache": true,
|
| 29 |
+
"vocab_size": 4096
|
| 30 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.37.1"
|
| 6 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb7b9f653eb8927bce21004e8be819fc5f1a89351bc2700cacd34cbd0a802d5d
|
| 3 |
+
size 276979168
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d5b1c4ee7a87fc2a5f668b10a36944c71efdb9b44487fb42790c588fc67ef49
|
| 3 |
+
size 554108549
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:097c6dd086b8c03f1fe6f9d0797dd61e1755d7bccef8fa20a36fe7669a937b89
|
| 3 |
+
size 14639
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad7c03219c270ae06ab2656ef0b489c2c64d2194c4dd1942cea8d5357486194c
|
| 3 |
+
size 627
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"tokenizer_class": "PreTrainedTokenizerFast", "unk_token": "[UNK]", "cls_token": "[CLS]", "sep_token": "[SEP]", "pad_token": "[PAD]", "mask_token": "[MASK]"}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3485 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 6.110002040863037,
|
| 3 |
+
"best_model_checkpoint": "./results/models/checkpoint-242575",
|
| 4 |
+
"epoch": 34.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 266050,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.06,
|
| 13 |
+
"learning_rate": 0.001997444089456869,
|
| 14 |
+
"loss": 6.3873,
|
| 15 |
+
"step": 500
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.13,
|
| 19 |
+
"learning_rate": 0.001994888178913738,
|
| 20 |
+
"loss": 6.2714,
|
| 21 |
+
"step": 1000
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.19,
|
| 25 |
+
"learning_rate": 0.0019923322683706073,
|
| 26 |
+
"loss": 6.3526,
|
| 27 |
+
"step": 1500
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.26,
|
| 31 |
+
"learning_rate": 0.001989776357827476,
|
| 32 |
+
"loss": 6.3696,
|
| 33 |
+
"step": 2000
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.32,
|
| 37 |
+
"learning_rate": 0.001987220447284345,
|
| 38 |
+
"loss": 6.3501,
|
| 39 |
+
"step": 2500
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.38,
|
| 43 |
+
"learning_rate": 0.001984664536741214,
|
| 44 |
+
"loss": 6.3472,
|
| 45 |
+
"step": 3000
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.45,
|
| 49 |
+
"learning_rate": 0.001982108626198083,
|
| 50 |
+
"loss": 6.347,
|
| 51 |
+
"step": 3500
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.51,
|
| 55 |
+
"learning_rate": 0.0019795527156549523,
|
| 56 |
+
"loss": 6.3555,
|
| 57 |
+
"step": 4000
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.58,
|
| 61 |
+
"learning_rate": 0.0019769968051118214,
|
| 62 |
+
"loss": 6.3432,
|
| 63 |
+
"step": 4500
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.64,
|
| 67 |
+
"learning_rate": 0.00197444089456869,
|
| 68 |
+
"loss": 6.3439,
|
| 69 |
+
"step": 5000
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.7,
|
| 73 |
+
"learning_rate": 0.001971884984025559,
|
| 74 |
+
"loss": 6.3312,
|
| 75 |
+
"step": 5500
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.77,
|
| 79 |
+
"learning_rate": 0.001969329073482428,
|
| 80 |
+
"loss": 6.3279,
|
| 81 |
+
"step": 6000
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.83,
|
| 85 |
+
"learning_rate": 0.0019667731629392973,
|
| 86 |
+
"loss": 6.3177,
|
| 87 |
+
"step": 6500
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.89,
|
| 91 |
+
"learning_rate": 0.001964217252396166,
|
| 92 |
+
"loss": 6.3152,
|
| 93 |
+
"step": 7000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.96,
|
| 97 |
+
"learning_rate": 0.001961661341853035,
|
| 98 |
+
"loss": 6.3086,
|
| 99 |
+
"step": 7500
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 1.0,
|
| 103 |
+
"eval_loss": 6.252776145935059,
|
| 104 |
+
"eval_runtime": 6.9192,
|
| 105 |
+
"eval_samples_per_second": 36.276,
|
| 106 |
+
"eval_steps_per_second": 1.156,
|
| 107 |
+
"step": 7825
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.02,
|
| 111 |
+
"learning_rate": 0.001959105431309904,
|
| 112 |
+
"loss": 6.3102,
|
| 113 |
+
"step": 8000
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 1.09,
|
| 117 |
+
"learning_rate": 0.001956549520766773,
|
| 118 |
+
"loss": 6.2978,
|
| 119 |
+
"step": 8500
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.15,
|
| 123 |
+
"learning_rate": 0.0019539936102236422,
|
| 124 |
+
"loss": 6.2904,
|
| 125 |
+
"step": 9000
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 1.21,
|
| 129 |
+
"learning_rate": 0.001951437699680511,
|
| 130 |
+
"loss": 6.2953,
|
| 131 |
+
"step": 9500
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.28,
|
| 135 |
+
"learning_rate": 0.0019488817891373802,
|
| 136 |
+
"loss": 6.2988,
|
| 137 |
+
"step": 10000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.34,
|
| 141 |
+
"learning_rate": 0.0019463258785942493,
|
| 142 |
+
"loss": 6.2859,
|
| 143 |
+
"step": 10500
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 1.41,
|
| 147 |
+
"learning_rate": 0.0019437699680511184,
|
| 148 |
+
"loss": 6.2906,
|
| 149 |
+
"step": 11000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.47,
|
| 153 |
+
"learning_rate": 0.0019412140575079872,
|
| 154 |
+
"loss": 6.2936,
|
| 155 |
+
"step": 11500
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 1.53,
|
| 159 |
+
"learning_rate": 0.0019386581469648563,
|
| 160 |
+
"loss": 6.288,
|
| 161 |
+
"step": 12000
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.6,
|
| 165 |
+
"learning_rate": 0.0019361022364217254,
|
| 166 |
+
"loss": 6.2888,
|
| 167 |
+
"step": 12500
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.66,
|
| 171 |
+
"learning_rate": 0.0019335463258785943,
|
| 172 |
+
"loss": 6.2866,
|
| 173 |
+
"step": 13000
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 1.73,
|
| 177 |
+
"learning_rate": 0.0019309904153354633,
|
| 178 |
+
"loss": 6.2776,
|
| 179 |
+
"step": 13500
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.79,
|
| 183 |
+
"learning_rate": 0.0019284345047923324,
|
| 184 |
+
"loss": 6.2706,
|
| 185 |
+
"step": 14000
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.85,
|
| 189 |
+
"learning_rate": 0.0019258785942492015,
|
| 190 |
+
"loss": 6.2705,
|
| 191 |
+
"step": 14500
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.92,
|
| 195 |
+
"learning_rate": 0.0019233226837060702,
|
| 196 |
+
"loss": 6.2807,
|
| 197 |
+
"step": 15000
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.98,
|
| 201 |
+
"learning_rate": 0.0019207667731629392,
|
| 202 |
+
"loss": 6.278,
|
| 203 |
+
"step": 15500
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 2.0,
|
| 207 |
+
"eval_loss": 6.216161251068115,
|
| 208 |
+
"eval_runtime": 6.9294,
|
| 209 |
+
"eval_samples_per_second": 36.223,
|
| 210 |
+
"eval_steps_per_second": 1.155,
|
| 211 |
+
"step": 15650
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 2.04,
|
| 215 |
+
"learning_rate": 0.0019182108626198083,
|
| 216 |
+
"loss": 6.2688,
|
| 217 |
+
"step": 16000
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 2.11,
|
| 221 |
+
"learning_rate": 0.0019156549520766772,
|
| 222 |
+
"loss": 6.2732,
|
| 223 |
+
"step": 16500
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 2.17,
|
| 227 |
+
"learning_rate": 0.0019130990415335463,
|
| 228 |
+
"loss": 6.2603,
|
| 229 |
+
"step": 17000
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 2.24,
|
| 233 |
+
"learning_rate": 0.0019105431309904154,
|
| 234 |
+
"loss": 6.2656,
|
| 235 |
+
"step": 17500
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 2.3,
|
| 239 |
+
"learning_rate": 0.0019079872204472844,
|
| 240 |
+
"loss": 6.2551,
|
| 241 |
+
"step": 18000
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 2.36,
|
| 245 |
+
"learning_rate": 0.0019054313099041533,
|
| 246 |
+
"loss": 6.259,
|
| 247 |
+
"step": 18500
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 2.43,
|
| 251 |
+
"learning_rate": 0.0019028753993610224,
|
| 252 |
+
"loss": 6.2607,
|
| 253 |
+
"step": 19000
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 2.49,
|
| 257 |
+
"learning_rate": 0.0019003194888178915,
|
| 258 |
+
"loss": 6.2634,
|
| 259 |
+
"step": 19500
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 2.56,
|
| 263 |
+
"learning_rate": 0.0018977635782747603,
|
| 264 |
+
"loss": 6.26,
|
| 265 |
+
"step": 20000
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 2.62,
|
| 269 |
+
"learning_rate": 0.0018952076677316294,
|
| 270 |
+
"loss": 6.2502,
|
| 271 |
+
"step": 20500
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 2.68,
|
| 275 |
+
"learning_rate": 0.0018926517571884985,
|
| 276 |
+
"loss": 6.2453,
|
| 277 |
+
"step": 21000
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 2.75,
|
| 281 |
+
"learning_rate": 0.0018900958466453676,
|
| 282 |
+
"loss": 6.2529,
|
| 283 |
+
"step": 21500
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 2.81,
|
| 287 |
+
"learning_rate": 0.0018875399361022365,
|
| 288 |
+
"loss": 6.2493,
|
| 289 |
+
"step": 22000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 2.88,
|
| 293 |
+
"learning_rate": 0.0018849840255591055,
|
| 294 |
+
"loss": 6.2534,
|
| 295 |
+
"step": 22500
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 2.94,
|
| 299 |
+
"learning_rate": 0.0018824281150159746,
|
| 300 |
+
"loss": 6.2442,
|
| 301 |
+
"step": 23000
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 3.0,
|
| 305 |
+
"eval_loss": 6.19460391998291,
|
| 306 |
+
"eval_runtime": 6.7538,
|
| 307 |
+
"eval_samples_per_second": 37.164,
|
| 308 |
+
"eval_steps_per_second": 1.185,
|
| 309 |
+
"step": 23475
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 3.0,
|
| 313 |
+
"learning_rate": 0.0018798722044728435,
|
| 314 |
+
"loss": 6.2498,
|
| 315 |
+
"step": 23500
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 3.07,
|
| 319 |
+
"learning_rate": 0.0018773162939297126,
|
| 320 |
+
"loss": 6.2428,
|
| 321 |
+
"step": 24000
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 3.13,
|
| 325 |
+
"learning_rate": 0.0018747603833865817,
|
| 326 |
+
"loss": 6.2447,
|
| 327 |
+
"step": 24500
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 3.19,
|
| 331 |
+
"learning_rate": 0.0018722044728434505,
|
| 332 |
+
"loss": 6.24,
|
| 333 |
+
"step": 25000
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 3.26,
|
| 337 |
+
"learning_rate": 0.0018696485623003194,
|
| 338 |
+
"loss": 6.2476,
|
| 339 |
+
"step": 25500
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 3.32,
|
| 343 |
+
"learning_rate": 0.0018670926517571885,
|
| 344 |
+
"loss": 6.2486,
|
| 345 |
+
"step": 26000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 3.39,
|
| 349 |
+
"learning_rate": 0.0018645367412140576,
|
| 350 |
+
"loss": 6.2428,
|
| 351 |
+
"step": 26500
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 3.45,
|
| 355 |
+
"learning_rate": 0.0018619808306709264,
|
| 356 |
+
"loss": 6.2475,
|
| 357 |
+
"step": 27000
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 3.51,
|
| 361 |
+
"learning_rate": 0.0018594249201277955,
|
| 362 |
+
"loss": 6.2481,
|
| 363 |
+
"step": 27500
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 3.58,
|
| 367 |
+
"learning_rate": 0.0018568690095846646,
|
| 368 |
+
"loss": 6.2381,
|
| 369 |
+
"step": 28000
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 3.64,
|
| 373 |
+
"learning_rate": 0.0018543130990415334,
|
| 374 |
+
"loss": 6.2402,
|
| 375 |
+
"step": 28500
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 3.71,
|
| 379 |
+
"learning_rate": 0.0018517571884984025,
|
| 380 |
+
"loss": 6.2389,
|
| 381 |
+
"step": 29000
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 3.77,
|
| 385 |
+
"learning_rate": 0.0018492012779552716,
|
| 386 |
+
"loss": 6.2414,
|
| 387 |
+
"step": 29500
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 3.83,
|
| 391 |
+
"learning_rate": 0.0018466453674121407,
|
| 392 |
+
"loss": 6.2373,
|
| 393 |
+
"step": 30000
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 3.9,
|
| 397 |
+
"learning_rate": 0.0018440894568690096,
|
| 398 |
+
"loss": 6.2408,
|
| 399 |
+
"step": 30500
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 3.96,
|
| 403 |
+
"learning_rate": 0.0018415335463258786,
|
| 404 |
+
"loss": 6.2299,
|
| 405 |
+
"step": 31000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 4.0,
|
| 409 |
+
"eval_loss": 6.177552700042725,
|
| 410 |
+
"eval_runtime": 6.7861,
|
| 411 |
+
"eval_samples_per_second": 36.987,
|
| 412 |
+
"eval_steps_per_second": 1.179,
|
| 413 |
+
"step": 31300
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 4.03,
|
| 417 |
+
"learning_rate": 0.0018389776357827477,
|
| 418 |
+
"loss": 6.2279,
|
| 419 |
+
"step": 31500
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 4.09,
|
| 423 |
+
"learning_rate": 0.0018364217252396166,
|
| 424 |
+
"loss": 6.2341,
|
| 425 |
+
"step": 32000
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 4.15,
|
| 429 |
+
"learning_rate": 0.0018338658146964857,
|
| 430 |
+
"loss": 6.2291,
|
| 431 |
+
"step": 32500
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 4.22,
|
| 435 |
+
"learning_rate": 0.0018313099041533548,
|
| 436 |
+
"loss": 6.2349,
|
| 437 |
+
"step": 33000
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 4.28,
|
| 441 |
+
"learning_rate": 0.0018287539936102238,
|
| 442 |
+
"loss": 6.231,
|
| 443 |
+
"step": 33500
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 4.35,
|
| 447 |
+
"learning_rate": 0.0018261980830670927,
|
| 448 |
+
"loss": 6.2295,
|
| 449 |
+
"step": 34000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 4.41,
|
| 453 |
+
"learning_rate": 0.0018236421725239618,
|
| 454 |
+
"loss": 6.2344,
|
| 455 |
+
"step": 34500
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 4.47,
|
| 459 |
+
"learning_rate": 0.0018210862619808307,
|
| 460 |
+
"loss": 6.2289,
|
| 461 |
+
"step": 35000
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 4.54,
|
| 465 |
+
"learning_rate": 0.0018185303514376995,
|
| 466 |
+
"loss": 6.2309,
|
| 467 |
+
"step": 35500
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 4.6,
|
| 471 |
+
"learning_rate": 0.0018159744408945686,
|
| 472 |
+
"loss": 6.2255,
|
| 473 |
+
"step": 36000
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 4.66,
|
| 477 |
+
"learning_rate": 0.0018134185303514377,
|
| 478 |
+
"loss": 6.2281,
|
| 479 |
+
"step": 36500
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 4.73,
|
| 483 |
+
"learning_rate": 0.0018108626198083068,
|
| 484 |
+
"loss": 6.2269,
|
| 485 |
+
"step": 37000
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 4.79,
|
| 489 |
+
"learning_rate": 0.0018083067092651756,
|
| 490 |
+
"loss": 6.2235,
|
| 491 |
+
"step": 37500
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 4.86,
|
| 495 |
+
"learning_rate": 0.0018057507987220447,
|
| 496 |
+
"loss": 6.2252,
|
| 497 |
+
"step": 38000
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 4.92,
|
| 501 |
+
"learning_rate": 0.0018031948881789138,
|
| 502 |
+
"loss": 6.2243,
|
| 503 |
+
"step": 38500
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 4.98,
|
| 507 |
+
"learning_rate": 0.0018006389776357827,
|
| 508 |
+
"loss": 6.2243,
|
| 509 |
+
"step": 39000
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 5.0,
|
| 513 |
+
"eval_loss": 6.16984748840332,
|
| 514 |
+
"eval_runtime": 6.9467,
|
| 515 |
+
"eval_samples_per_second": 36.132,
|
| 516 |
+
"eval_steps_per_second": 1.152,
|
| 517 |
+
"step": 39125
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 5.05,
|
| 521 |
+
"learning_rate": 0.0017980830670926518,
|
| 522 |
+
"loss": 6.2197,
|
| 523 |
+
"step": 39500
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 5.11,
|
| 527 |
+
"learning_rate": 0.0017955271565495208,
|
| 528 |
+
"loss": 6.2246,
|
| 529 |
+
"step": 40000
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 5.18,
|
| 533 |
+
"learning_rate": 0.00179297124600639,
|
| 534 |
+
"loss": 6.2219,
|
| 535 |
+
"step": 40500
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 5.24,
|
| 539 |
+
"learning_rate": 0.0017904153354632588,
|
| 540 |
+
"loss": 6.2246,
|
| 541 |
+
"step": 41000
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 5.3,
|
| 545 |
+
"learning_rate": 0.0017878594249201279,
|
| 546 |
+
"loss": 6.2203,
|
| 547 |
+
"step": 41500
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 5.37,
|
| 551 |
+
"learning_rate": 0.001785303514376997,
|
| 552 |
+
"loss": 6.2305,
|
| 553 |
+
"step": 42000
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 5.43,
|
| 557 |
+
"learning_rate": 0.0017827476038338658,
|
| 558 |
+
"loss": 6.225,
|
| 559 |
+
"step": 42500
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 5.5,
|
| 563 |
+
"learning_rate": 0.001780191693290735,
|
| 564 |
+
"loss": 6.2216,
|
| 565 |
+
"step": 43000
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 5.56,
|
| 569 |
+
"learning_rate": 0.001777635782747604,
|
| 570 |
+
"loss": 6.2242,
|
| 571 |
+
"step": 43500
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 5.62,
|
| 575 |
+
"learning_rate": 0.001775079872204473,
|
| 576 |
+
"loss": 6.224,
|
| 577 |
+
"step": 44000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 5.69,
|
| 581 |
+
"learning_rate": 0.001772523961661342,
|
| 582 |
+
"loss": 6.2168,
|
| 583 |
+
"step": 44500
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 5.75,
|
| 587 |
+
"learning_rate": 0.0017699680511182108,
|
| 588 |
+
"loss": 6.2249,
|
| 589 |
+
"step": 45000
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 5.81,
|
| 593 |
+
"learning_rate": 0.00176741214057508,
|
| 594 |
+
"loss": 6.2202,
|
| 595 |
+
"step": 45500
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 5.88,
|
| 599 |
+
"learning_rate": 0.0017648562300319488,
|
| 600 |
+
"loss": 6.2227,
|
| 601 |
+
"step": 46000
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 5.94,
|
| 605 |
+
"learning_rate": 0.0017623003194888178,
|
| 606 |
+
"loss": 6.2162,
|
| 607 |
+
"step": 46500
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 6.0,
|
| 611 |
+
"eval_loss": 6.162996768951416,
|
| 612 |
+
"eval_runtime": 6.7939,
|
| 613 |
+
"eval_samples_per_second": 36.945,
|
| 614 |
+
"eval_steps_per_second": 1.178,
|
| 615 |
+
"step": 46950
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 6.01,
|
| 619 |
+
"learning_rate": 0.001759744408945687,
|
| 620 |
+
"loss": 6.2237,
|
| 621 |
+
"step": 47000
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 6.07,
|
| 625 |
+
"learning_rate": 0.0017571884984025558,
|
| 626 |
+
"loss": 6.2159,
|
| 627 |
+
"step": 47500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 6.13,
|
| 631 |
+
"learning_rate": 0.0017546325878594249,
|
| 632 |
+
"loss": 6.2176,
|
| 633 |
+
"step": 48000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 6.2,
|
| 637 |
+
"learning_rate": 0.001752076677316294,
|
| 638 |
+
"loss": 6.2175,
|
| 639 |
+
"step": 48500
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 6.26,
|
| 643 |
+
"learning_rate": 0.001749520766773163,
|
| 644 |
+
"loss": 6.2171,
|
| 645 |
+
"step": 49000
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 6.33,
|
| 649 |
+
"learning_rate": 0.001746964856230032,
|
| 650 |
+
"loss": 6.2105,
|
| 651 |
+
"step": 49500
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 6.39,
|
| 655 |
+
"learning_rate": 0.001744408945686901,
|
| 656 |
+
"loss": 6.221,
|
| 657 |
+
"step": 50000
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 6.45,
|
| 661 |
+
"learning_rate": 0.00174185303514377,
|
| 662 |
+
"loss": 6.2141,
|
| 663 |
+
"step": 50500
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 6.52,
|
| 667 |
+
"learning_rate": 0.001739297124600639,
|
| 668 |
+
"loss": 6.213,
|
| 669 |
+
"step": 51000
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 6.58,
|
| 673 |
+
"learning_rate": 0.001736741214057508,
|
| 674 |
+
"loss": 6.2182,
|
| 675 |
+
"step": 51500
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 6.65,
|
| 679 |
+
"learning_rate": 0.0017341853035143771,
|
| 680 |
+
"loss": 6.2085,
|
| 681 |
+
"step": 52000
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 6.71,
|
| 685 |
+
"learning_rate": 0.0017316293929712462,
|
| 686 |
+
"loss": 6.2142,
|
| 687 |
+
"step": 52500
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 6.77,
|
| 691 |
+
"learning_rate": 0.001729073482428115,
|
| 692 |
+
"loss": 6.2274,
|
| 693 |
+
"step": 53000
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 6.84,
|
| 697 |
+
"learning_rate": 0.0017265175718849841,
|
| 698 |
+
"loss": 6.2124,
|
| 699 |
+
"step": 53500
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 6.9,
|
| 703 |
+
"learning_rate": 0.0017239616613418532,
|
| 704 |
+
"loss": 6.2105,
|
| 705 |
+
"step": 54000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 6.96,
|
| 709 |
+
"learning_rate": 0.0017214057507987219,
|
| 710 |
+
"loss": 6.2079,
|
| 711 |
+
"step": 54500
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 7.0,
|
| 715 |
+
"eval_loss": 6.15675163269043,
|
| 716 |
+
"eval_runtime": 6.7761,
|
| 717 |
+
"eval_samples_per_second": 37.042,
|
| 718 |
+
"eval_steps_per_second": 1.181,
|
| 719 |
+
"step": 54775
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 7.03,
|
| 723 |
+
"learning_rate": 0.001718849840255591,
|
| 724 |
+
"loss": 6.2153,
|
| 725 |
+
"step": 55000
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 7.09,
|
| 729 |
+
"learning_rate": 0.00171629392971246,
|
| 730 |
+
"loss": 6.2105,
|
| 731 |
+
"step": 55500
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 7.16,
|
| 735 |
+
"learning_rate": 0.0017137380191693291,
|
| 736 |
+
"loss": 6.2111,
|
| 737 |
+
"step": 56000
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 7.22,
|
| 741 |
+
"learning_rate": 0.001711182108626198,
|
| 742 |
+
"loss": 6.2167,
|
| 743 |
+
"step": 56500
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 7.28,
|
| 747 |
+
"learning_rate": 0.001708626198083067,
|
| 748 |
+
"loss": 6.216,
|
| 749 |
+
"step": 57000
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 7.35,
|
| 753 |
+
"learning_rate": 0.0017060702875399362,
|
| 754 |
+
"loss": 6.2144,
|
| 755 |
+
"step": 57500
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 7.41,
|
| 759 |
+
"learning_rate": 0.001703514376996805,
|
| 760 |
+
"loss": 6.2141,
|
| 761 |
+
"step": 58000
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 7.48,
|
| 765 |
+
"learning_rate": 0.001700958466453674,
|
| 766 |
+
"loss": 6.2139,
|
| 767 |
+
"step": 58500
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 7.54,
|
| 771 |
+
"learning_rate": 0.0016984025559105432,
|
| 772 |
+
"loss": 6.2117,
|
| 773 |
+
"step": 59000
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 7.6,
|
| 777 |
+
"learning_rate": 0.0016958466453674123,
|
| 778 |
+
"loss": 6.2122,
|
| 779 |
+
"step": 59500
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 7.67,
|
| 783 |
+
"learning_rate": 0.0016932907348242811,
|
| 784 |
+
"loss": 6.2096,
|
| 785 |
+
"step": 60000
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 7.73,
|
| 789 |
+
"learning_rate": 0.0016907348242811502,
|
| 790 |
+
"loss": 6.2055,
|
| 791 |
+
"step": 60500
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 7.8,
|
| 795 |
+
"learning_rate": 0.0016881789137380193,
|
| 796 |
+
"loss": 6.2088,
|
| 797 |
+
"step": 61000
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 7.86,
|
| 801 |
+
"learning_rate": 0.0016856230031948882,
|
| 802 |
+
"loss": 6.2105,
|
| 803 |
+
"step": 61500
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 7.92,
|
| 807 |
+
"learning_rate": 0.0016830670926517573,
|
| 808 |
+
"loss": 6.2161,
|
| 809 |
+
"step": 62000
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 7.99,
|
| 813 |
+
"learning_rate": 0.0016805111821086263,
|
| 814 |
+
"loss": 6.2105,
|
| 815 |
+
"step": 62500
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 8.0,
|
| 819 |
+
"eval_loss": 6.1555399894714355,
|
| 820 |
+
"eval_runtime": 6.9246,
|
| 821 |
+
"eval_samples_per_second": 36.248,
|
| 822 |
+
"eval_steps_per_second": 1.155,
|
| 823 |
+
"step": 62600
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 8.05,
|
| 827 |
+
"learning_rate": 0.0016779552715654954,
|
| 828 |
+
"loss": 6.2111,
|
| 829 |
+
"step": 63000
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 8.12,
|
| 833 |
+
"learning_rate": 0.0016753993610223643,
|
| 834 |
+
"loss": 6.2095,
|
| 835 |
+
"step": 63500
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 8.18,
|
| 839 |
+
"learning_rate": 0.0016728434504792334,
|
| 840 |
+
"loss": 6.2064,
|
| 841 |
+
"step": 64000
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 8.24,
|
| 845 |
+
"learning_rate": 0.0016702875399361022,
|
| 846 |
+
"loss": 6.2147,
|
| 847 |
+
"step": 64500
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 8.31,
|
| 851 |
+
"learning_rate": 0.001667731629392971,
|
| 852 |
+
"loss": 6.1994,
|
| 853 |
+
"step": 65000
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 8.37,
|
| 857 |
+
"learning_rate": 0.0016651757188498402,
|
| 858 |
+
"loss": 6.2094,
|
| 859 |
+
"step": 65500
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 8.43,
|
| 863 |
+
"learning_rate": 0.0016626198083067093,
|
| 864 |
+
"loss": 6.2075,
|
| 865 |
+
"step": 66000
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 8.5,
|
| 869 |
+
"learning_rate": 0.0016600638977635781,
|
| 870 |
+
"loss": 6.2085,
|
| 871 |
+
"step": 66500
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 8.56,
|
| 875 |
+
"learning_rate": 0.0016575079872204472,
|
| 876 |
+
"loss": 6.2004,
|
| 877 |
+
"step": 67000
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 8.63,
|
| 881 |
+
"learning_rate": 0.0016549520766773163,
|
| 882 |
+
"loss": 6.2112,
|
| 883 |
+
"step": 67500
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 8.69,
|
| 887 |
+
"learning_rate": 0.0016523961661341854,
|
| 888 |
+
"loss": 6.2051,
|
| 889 |
+
"step": 68000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 8.75,
|
| 893 |
+
"learning_rate": 0.0016498402555910543,
|
| 894 |
+
"loss": 6.2024,
|
| 895 |
+
"step": 68500
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 8.82,
|
| 899 |
+
"learning_rate": 0.0016472843450479233,
|
| 900 |
+
"loss": 6.1996,
|
| 901 |
+
"step": 69000
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 8.88,
|
| 905 |
+
"learning_rate": 0.0016447284345047924,
|
| 906 |
+
"loss": 6.2066,
|
| 907 |
+
"step": 69500
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 8.95,
|
| 911 |
+
"learning_rate": 0.0016421725239616613,
|
| 912 |
+
"loss": 6.2013,
|
| 913 |
+
"step": 70000
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 9.0,
|
| 917 |
+
"eval_loss": 6.147340297698975,
|
| 918 |
+
"eval_runtime": 6.7973,
|
| 919 |
+
"eval_samples_per_second": 36.927,
|
| 920 |
+
"eval_steps_per_second": 1.177,
|
| 921 |
+
"step": 70425
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 9.01,
|
| 925 |
+
"learning_rate": 0.0016396166134185304,
|
| 926 |
+
"loss": 6.2085,
|
| 927 |
+
"step": 70500
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 9.07,
|
| 931 |
+
"learning_rate": 0.0016370607028753995,
|
| 932 |
+
"loss": 6.1978,
|
| 933 |
+
"step": 71000
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 9.14,
|
| 937 |
+
"learning_rate": 0.0016345047923322685,
|
| 938 |
+
"loss": 6.2017,
|
| 939 |
+
"step": 71500
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 9.2,
|
| 943 |
+
"learning_rate": 0.0016319488817891374,
|
| 944 |
+
"loss": 6.2047,
|
| 945 |
+
"step": 72000
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 9.27,
|
| 949 |
+
"learning_rate": 0.0016293929712460065,
|
| 950 |
+
"loss": 6.1998,
|
| 951 |
+
"step": 72500
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 9.33,
|
| 955 |
+
"learning_rate": 0.0016268370607028756,
|
| 956 |
+
"loss": 6.2027,
|
| 957 |
+
"step": 73000
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 9.39,
|
| 961 |
+
"learning_rate": 0.0016242811501597444,
|
| 962 |
+
"loss": 6.2024,
|
| 963 |
+
"step": 73500
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 9.46,
|
| 967 |
+
"learning_rate": 0.0016217252396166135,
|
| 968 |
+
"loss": 6.1936,
|
| 969 |
+
"step": 74000
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 9.52,
|
| 973 |
+
"learning_rate": 0.0016191693290734824,
|
| 974 |
+
"loss": 6.2042,
|
| 975 |
+
"step": 74500
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 9.58,
|
| 979 |
+
"learning_rate": 0.0016166134185303515,
|
| 980 |
+
"loss": 6.2021,
|
| 981 |
+
"step": 75000
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 9.65,
|
| 985 |
+
"learning_rate": 0.0016140575079872203,
|
| 986 |
+
"loss": 6.2054,
|
| 987 |
+
"step": 75500
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 9.71,
|
| 991 |
+
"learning_rate": 0.0016115015974440894,
|
| 992 |
+
"loss": 6.2051,
|
| 993 |
+
"step": 76000
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 9.78,
|
| 997 |
+
"learning_rate": 0.0016089456869009585,
|
| 998 |
+
"loss": 6.1963,
|
| 999 |
+
"step": 76500
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 9.84,
|
| 1003 |
+
"learning_rate": 0.0016063897763578274,
|
| 1004 |
+
"loss": 6.2039,
|
| 1005 |
+
"step": 77000
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 9.9,
|
| 1009 |
+
"learning_rate": 0.0016038338658146965,
|
| 1010 |
+
"loss": 6.1993,
|
| 1011 |
+
"step": 77500
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 9.97,
|
| 1015 |
+
"learning_rate": 0.0016012779552715655,
|
| 1016 |
+
"loss": 6.199,
|
| 1017 |
+
"step": 78000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 10.0,
|
| 1021 |
+
"eval_loss": 6.143795967102051,
|
| 1022 |
+
"eval_runtime": 6.7856,
|
| 1023 |
+
"eval_samples_per_second": 36.99,
|
| 1024 |
+
"eval_steps_per_second": 1.179,
|
| 1025 |
+
"step": 78250
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 10.03,
|
| 1029 |
+
"learning_rate": 0.0015987220447284346,
|
| 1030 |
+
"loss": 6.1969,
|
| 1031 |
+
"step": 78500
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 10.1,
|
| 1035 |
+
"learning_rate": 0.0015961661341853035,
|
| 1036 |
+
"loss": 6.2008,
|
| 1037 |
+
"step": 79000
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 10.16,
|
| 1041 |
+
"learning_rate": 0.0015936102236421726,
|
| 1042 |
+
"loss": 6.1974,
|
| 1043 |
+
"step": 79500
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 10.22,
|
| 1047 |
+
"learning_rate": 0.0015910543130990417,
|
| 1048 |
+
"loss": 6.1966,
|
| 1049 |
+
"step": 80000
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 10.29,
|
| 1053 |
+
"learning_rate": 0.0015884984025559105,
|
| 1054 |
+
"loss": 6.199,
|
| 1055 |
+
"step": 80500
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 10.35,
|
| 1059 |
+
"learning_rate": 0.0015859424920127796,
|
| 1060 |
+
"loss": 6.2016,
|
| 1061 |
+
"step": 81000
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 10.42,
|
| 1065 |
+
"learning_rate": 0.0015833865814696487,
|
| 1066 |
+
"loss": 6.1986,
|
| 1067 |
+
"step": 81500
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 10.48,
|
| 1071 |
+
"learning_rate": 0.0015808306709265178,
|
| 1072 |
+
"loss": 6.2013,
|
| 1073 |
+
"step": 82000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 10.54,
|
| 1077 |
+
"learning_rate": 0.0015782747603833866,
|
| 1078 |
+
"loss": 6.1922,
|
| 1079 |
+
"step": 82500
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 10.61,
|
| 1083 |
+
"learning_rate": 0.0015757188498402557,
|
| 1084 |
+
"loss": 6.1999,
|
| 1085 |
+
"step": 83000
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 10.67,
|
| 1089 |
+
"learning_rate": 0.0015731629392971248,
|
| 1090 |
+
"loss": 6.1989,
|
| 1091 |
+
"step": 83500
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 10.73,
|
| 1095 |
+
"learning_rate": 0.0015706070287539937,
|
| 1096 |
+
"loss": 6.1963,
|
| 1097 |
+
"step": 84000
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 10.8,
|
| 1101 |
+
"learning_rate": 0.0015680511182108625,
|
| 1102 |
+
"loss": 6.1966,
|
| 1103 |
+
"step": 84500
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 10.86,
|
| 1107 |
+
"learning_rate": 0.0015654952076677316,
|
| 1108 |
+
"loss": 6.1955,
|
| 1109 |
+
"step": 85000
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 10.93,
|
| 1113 |
+
"learning_rate": 0.0015629392971246005,
|
| 1114 |
+
"loss": 6.1943,
|
| 1115 |
+
"step": 85500
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 10.99,
|
| 1119 |
+
"learning_rate": 0.0015603833865814696,
|
| 1120 |
+
"loss": 6.1953,
|
| 1121 |
+
"step": 86000
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 11.0,
|
| 1125 |
+
"eval_loss": 6.140895366668701,
|
| 1126 |
+
"eval_runtime": 6.7824,
|
| 1127 |
+
"eval_samples_per_second": 37.007,
|
| 1128 |
+
"eval_steps_per_second": 1.18,
|
| 1129 |
+
"step": 86075
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 11.05,
|
| 1133 |
+
"learning_rate": 0.0015578274760383386,
|
| 1134 |
+
"loss": 6.1972,
|
| 1135 |
+
"step": 86500
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 11.12,
|
| 1139 |
+
"learning_rate": 0.0015552715654952077,
|
| 1140 |
+
"loss": 6.1896,
|
| 1141 |
+
"step": 87000
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 11.18,
|
| 1145 |
+
"learning_rate": 0.0015527156549520766,
|
| 1146 |
+
"loss": 6.1901,
|
| 1147 |
+
"step": 87500
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 11.25,
|
| 1151 |
+
"learning_rate": 0.0015501597444089457,
|
| 1152 |
+
"loss": 6.1917,
|
| 1153 |
+
"step": 88000
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 11.31,
|
| 1157 |
+
"learning_rate": 0.0015476038338658148,
|
| 1158 |
+
"loss": 6.194,
|
| 1159 |
+
"step": 88500
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 11.37,
|
| 1163 |
+
"learning_rate": 0.0015450479233226836,
|
| 1164 |
+
"loss": 6.1934,
|
| 1165 |
+
"step": 89000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 11.44,
|
| 1169 |
+
"learning_rate": 0.0015424920127795527,
|
| 1170 |
+
"loss": 6.1912,
|
| 1171 |
+
"step": 89500
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 11.5,
|
| 1175 |
+
"learning_rate": 0.0015399361022364218,
|
| 1176 |
+
"loss": 6.1981,
|
| 1177 |
+
"step": 90000
|
| 1178 |
+
},
|
| 1179 |
+
{
|
| 1180 |
+
"epoch": 11.57,
|
| 1181 |
+
"learning_rate": 0.0015373801916932909,
|
| 1182 |
+
"loss": 6.1942,
|
| 1183 |
+
"step": 90500
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 11.63,
|
| 1187 |
+
"learning_rate": 0.0015348242811501597,
|
| 1188 |
+
"loss": 6.1965,
|
| 1189 |
+
"step": 91000
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 11.69,
|
| 1193 |
+
"learning_rate": 0.0015322683706070288,
|
| 1194 |
+
"loss": 6.1958,
|
| 1195 |
+
"step": 91500
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 11.76,
|
| 1199 |
+
"learning_rate": 0.001529712460063898,
|
| 1200 |
+
"loss": 6.1981,
|
| 1201 |
+
"step": 92000
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 11.82,
|
| 1205 |
+
"learning_rate": 0.0015271565495207668,
|
| 1206 |
+
"loss": 6.1972,
|
| 1207 |
+
"step": 92500
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 11.88,
|
| 1211 |
+
"learning_rate": 0.0015246006389776359,
|
| 1212 |
+
"loss": 6.1959,
|
| 1213 |
+
"step": 93000
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 11.95,
|
| 1217 |
+
"learning_rate": 0.001522044728434505,
|
| 1218 |
+
"loss": 6.1958,
|
| 1219 |
+
"step": 93500
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 12.0,
|
| 1223 |
+
"eval_loss": 6.139165878295898,
|
| 1224 |
+
"eval_runtime": 6.8854,
|
| 1225 |
+
"eval_samples_per_second": 36.454,
|
| 1226 |
+
"eval_steps_per_second": 1.162,
|
| 1227 |
+
"step": 93900
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 12.01,
|
| 1231 |
+
"learning_rate": 0.001519488817891374,
|
| 1232 |
+
"loss": 6.1901,
|
| 1233 |
+
"step": 94000
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 12.08,
|
| 1237 |
+
"learning_rate": 0.0015169329073482427,
|
| 1238 |
+
"loss": 6.1945,
|
| 1239 |
+
"step": 94500
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 12.14,
|
| 1243 |
+
"learning_rate": 0.0015143769968051118,
|
| 1244 |
+
"loss": 6.1949,
|
| 1245 |
+
"step": 95000
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 12.2,
|
| 1249 |
+
"learning_rate": 0.0015118210862619808,
|
| 1250 |
+
"loss": 6.1881,
|
| 1251 |
+
"step": 95500
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 12.27,
|
| 1255 |
+
"learning_rate": 0.0015092651757188497,
|
| 1256 |
+
"loss": 6.189,
|
| 1257 |
+
"step": 96000
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 12.33,
|
| 1261 |
+
"learning_rate": 0.0015067092651757188,
|
| 1262 |
+
"loss": 6.1923,
|
| 1263 |
+
"step": 96500
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 12.4,
|
| 1267 |
+
"learning_rate": 0.0015041533546325879,
|
| 1268 |
+
"loss": 6.1962,
|
| 1269 |
+
"step": 97000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 12.46,
|
| 1273 |
+
"learning_rate": 0.001501597444089457,
|
| 1274 |
+
"loss": 6.1911,
|
| 1275 |
+
"step": 97500
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 12.52,
|
| 1279 |
+
"learning_rate": 0.0014990415335463258,
|
| 1280 |
+
"loss": 6.1948,
|
| 1281 |
+
"step": 98000
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 12.59,
|
| 1285 |
+
"learning_rate": 0.001496485623003195,
|
| 1286 |
+
"loss": 6.197,
|
| 1287 |
+
"step": 98500
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 12.65,
|
| 1291 |
+
"learning_rate": 0.001493929712460064,
|
| 1292 |
+
"loss": 6.1898,
|
| 1293 |
+
"step": 99000
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 12.72,
|
| 1297 |
+
"learning_rate": 0.0014913738019169329,
|
| 1298 |
+
"loss": 6.1924,
|
| 1299 |
+
"step": 99500
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 12.78,
|
| 1303 |
+
"learning_rate": 0.001488817891373802,
|
| 1304 |
+
"loss": 6.1898,
|
| 1305 |
+
"step": 100000
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 12.84,
|
| 1309 |
+
"learning_rate": 0.001486261980830671,
|
| 1310 |
+
"loss": 6.1934,
|
| 1311 |
+
"step": 100500
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 12.91,
|
| 1315 |
+
"learning_rate": 0.0014837060702875401,
|
| 1316 |
+
"loss": 6.1944,
|
| 1317 |
+
"step": 101000
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 12.97,
|
| 1321 |
+
"learning_rate": 0.001481150159744409,
|
| 1322 |
+
"loss": 6.1935,
|
| 1323 |
+
"step": 101500
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 13.0,
|
| 1327 |
+
"eval_loss": 6.138128280639648,
|
| 1328 |
+
"eval_runtime": 6.8154,
|
| 1329 |
+
"eval_samples_per_second": 36.829,
|
| 1330 |
+
"eval_steps_per_second": 1.174,
|
| 1331 |
+
"step": 101725
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 13.04,
|
| 1335 |
+
"learning_rate": 0.001478594249201278,
|
| 1336 |
+
"loss": 6.1931,
|
| 1337 |
+
"step": 102000
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 13.1,
|
| 1341 |
+
"learning_rate": 0.0014760383386581471,
|
| 1342 |
+
"loss": 6.1947,
|
| 1343 |
+
"step": 102500
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 13.16,
|
| 1347 |
+
"learning_rate": 0.001473482428115016,
|
| 1348 |
+
"loss": 6.1916,
|
| 1349 |
+
"step": 103000
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 13.23,
|
| 1353 |
+
"learning_rate": 0.001470926517571885,
|
| 1354 |
+
"loss": 6.1923,
|
| 1355 |
+
"step": 103500
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 13.29,
|
| 1359 |
+
"learning_rate": 0.0014683706070287542,
|
| 1360 |
+
"loss": 6.189,
|
| 1361 |
+
"step": 104000
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 13.35,
|
| 1365 |
+
"learning_rate": 0.0014658146964856228,
|
| 1366 |
+
"loss": 6.185,
|
| 1367 |
+
"step": 104500
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 13.42,
|
| 1371 |
+
"learning_rate": 0.001463258785942492,
|
| 1372 |
+
"loss": 6.1979,
|
| 1373 |
+
"step": 105000
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 13.48,
|
| 1377 |
+
"learning_rate": 0.001460702875399361,
|
| 1378 |
+
"loss": 6.1919,
|
| 1379 |
+
"step": 105500
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 13.55,
|
| 1383 |
+
"learning_rate": 0.00145814696485623,
|
| 1384 |
+
"loss": 6.1907,
|
| 1385 |
+
"step": 106000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 13.61,
|
| 1389 |
+
"learning_rate": 0.001455591054313099,
|
| 1390 |
+
"loss": 6.1877,
|
| 1391 |
+
"step": 106500
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 13.67,
|
| 1395 |
+
"learning_rate": 0.001453035143769968,
|
| 1396 |
+
"loss": 6.1884,
|
| 1397 |
+
"step": 107000
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 13.74,
|
| 1401 |
+
"learning_rate": 0.0014504792332268371,
|
| 1402 |
+
"loss": 6.1904,
|
| 1403 |
+
"step": 107500
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 13.8,
|
| 1407 |
+
"learning_rate": 0.001447923322683706,
|
| 1408 |
+
"loss": 6.1909,
|
| 1409 |
+
"step": 108000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 13.87,
|
| 1413 |
+
"learning_rate": 0.001445367412140575,
|
| 1414 |
+
"loss": 6.1957,
|
| 1415 |
+
"step": 108500
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 13.93,
|
| 1419 |
+
"learning_rate": 0.0014428115015974441,
|
| 1420 |
+
"loss": 6.1914,
|
| 1421 |
+
"step": 109000
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 13.99,
|
| 1425 |
+
"learning_rate": 0.0014402555910543132,
|
| 1426 |
+
"loss": 6.1865,
|
| 1427 |
+
"step": 109500
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 14.0,
|
| 1431 |
+
"eval_loss": 6.134657859802246,
|
| 1432 |
+
"eval_runtime": 7.8402,
|
| 1433 |
+
"eval_samples_per_second": 32.015,
|
| 1434 |
+
"eval_steps_per_second": 1.02,
|
| 1435 |
+
"step": 109550
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 14.06,
|
| 1439 |
+
"learning_rate": 0.001437699680511182,
|
| 1440 |
+
"loss": 6.189,
|
| 1441 |
+
"step": 110000
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 14.12,
|
| 1445 |
+
"learning_rate": 0.0014351437699680512,
|
| 1446 |
+
"loss": 6.1883,
|
| 1447 |
+
"step": 110500
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 14.19,
|
| 1451 |
+
"learning_rate": 0.0014325878594249203,
|
| 1452 |
+
"loss": 6.1902,
|
| 1453 |
+
"step": 111000
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 14.25,
|
| 1457 |
+
"learning_rate": 0.0014300319488817891,
|
| 1458 |
+
"loss": 6.191,
|
| 1459 |
+
"step": 111500
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 14.31,
|
| 1463 |
+
"learning_rate": 0.0014274760383386582,
|
| 1464 |
+
"loss": 6.1934,
|
| 1465 |
+
"step": 112000
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 14.38,
|
| 1469 |
+
"learning_rate": 0.0014249201277955273,
|
| 1470 |
+
"loss": 6.1841,
|
| 1471 |
+
"step": 112500
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 14.44,
|
| 1475 |
+
"learning_rate": 0.0014223642172523964,
|
| 1476 |
+
"loss": 6.1861,
|
| 1477 |
+
"step": 113000
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 14.5,
|
| 1481 |
+
"learning_rate": 0.0014198083067092652,
|
| 1482 |
+
"loss": 6.1883,
|
| 1483 |
+
"step": 113500
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 14.57,
|
| 1487 |
+
"learning_rate": 0.001417252396166134,
|
| 1488 |
+
"loss": 6.1884,
|
| 1489 |
+
"step": 114000
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 14.63,
|
| 1493 |
+
"learning_rate": 0.0014146964856230032,
|
| 1494 |
+
"loss": 6.1797,
|
| 1495 |
+
"step": 114500
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 14.7,
|
| 1499 |
+
"learning_rate": 0.001412140575079872,
|
| 1500 |
+
"loss": 6.1894,
|
| 1501 |
+
"step": 115000
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 14.76,
|
| 1505 |
+
"learning_rate": 0.0014095846645367411,
|
| 1506 |
+
"loss": 6.187,
|
| 1507 |
+
"step": 115500
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 14.82,
|
| 1511 |
+
"learning_rate": 0.0014070287539936102,
|
| 1512 |
+
"loss": 6.1863,
|
| 1513 |
+
"step": 116000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 14.89,
|
| 1517 |
+
"learning_rate": 0.0014044728434504793,
|
| 1518 |
+
"loss": 6.1849,
|
| 1519 |
+
"step": 116500
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 14.95,
|
| 1523 |
+
"learning_rate": 0.0014019169329073482,
|
| 1524 |
+
"loss": 6.1896,
|
| 1525 |
+
"step": 117000
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 15.0,
|
| 1529 |
+
"eval_loss": 6.130258083343506,
|
| 1530 |
+
"eval_runtime": 6.8222,
|
| 1531 |
+
"eval_samples_per_second": 36.792,
|
| 1532 |
+
"eval_steps_per_second": 1.173,
|
| 1533 |
+
"step": 117375
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 15.02,
|
| 1537 |
+
"learning_rate": 0.0013993610223642173,
|
| 1538 |
+
"loss": 6.1876,
|
| 1539 |
+
"step": 117500
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 15.08,
|
| 1543 |
+
"learning_rate": 0.0013968051118210863,
|
| 1544 |
+
"loss": 6.1846,
|
| 1545 |
+
"step": 118000
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 15.14,
|
| 1549 |
+
"learning_rate": 0.0013942492012779552,
|
| 1550 |
+
"loss": 6.1883,
|
| 1551 |
+
"step": 118500
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 15.21,
|
| 1555 |
+
"learning_rate": 0.0013916932907348243,
|
| 1556 |
+
"loss": 6.1785,
|
| 1557 |
+
"step": 119000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 15.27,
|
| 1561 |
+
"learning_rate": 0.0013891373801916934,
|
| 1562 |
+
"loss": 6.1827,
|
| 1563 |
+
"step": 119500
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 15.34,
|
| 1567 |
+
"learning_rate": 0.0013865814696485625,
|
| 1568 |
+
"loss": 6.1879,
|
| 1569 |
+
"step": 120000
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 15.4,
|
| 1573 |
+
"learning_rate": 0.0013840255591054313,
|
| 1574 |
+
"loss": 6.1734,
|
| 1575 |
+
"step": 120500
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 15.46,
|
| 1579 |
+
"learning_rate": 0.0013814696485623004,
|
| 1580 |
+
"loss": 6.1852,
|
| 1581 |
+
"step": 121000
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 15.53,
|
| 1585 |
+
"learning_rate": 0.0013789137380191695,
|
| 1586 |
+
"loss": 6.1903,
|
| 1587 |
+
"step": 121500
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 15.59,
|
| 1591 |
+
"learning_rate": 0.0013763578274760384,
|
| 1592 |
+
"loss": 6.1877,
|
| 1593 |
+
"step": 122000
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 15.65,
|
| 1597 |
+
"learning_rate": 0.0013738019169329074,
|
| 1598 |
+
"loss": 6.1779,
|
| 1599 |
+
"step": 122500
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 15.72,
|
| 1603 |
+
"learning_rate": 0.0013712460063897765,
|
| 1604 |
+
"loss": 6.185,
|
| 1605 |
+
"step": 123000
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 15.78,
|
| 1609 |
+
"learning_rate": 0.0013686900958466456,
|
| 1610 |
+
"loss": 6.1835,
|
| 1611 |
+
"step": 123500
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 15.85,
|
| 1615 |
+
"learning_rate": 0.0013661341853035143,
|
| 1616 |
+
"loss": 6.1792,
|
| 1617 |
+
"step": 124000
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 15.91,
|
| 1621 |
+
"learning_rate": 0.0013635782747603833,
|
| 1622 |
+
"loss": 6.182,
|
| 1623 |
+
"step": 124500
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 15.97,
|
| 1627 |
+
"learning_rate": 0.0013610223642172524,
|
| 1628 |
+
"loss": 6.1884,
|
| 1629 |
+
"step": 125000
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 16.0,
|
| 1633 |
+
"eval_loss": 6.127689361572266,
|
| 1634 |
+
"eval_runtime": 6.7878,
|
| 1635 |
+
"eval_samples_per_second": 36.978,
|
| 1636 |
+
"eval_steps_per_second": 1.179,
|
| 1637 |
+
"step": 125200
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 16.04,
|
| 1641 |
+
"learning_rate": 0.0013584664536741213,
|
| 1642 |
+
"loss": 6.1881,
|
| 1643 |
+
"step": 125500
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 16.1,
|
| 1647 |
+
"learning_rate": 0.0013559105431309904,
|
| 1648 |
+
"loss": 6.1847,
|
| 1649 |
+
"step": 126000
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 16.17,
|
| 1653 |
+
"learning_rate": 0.0013533546325878595,
|
| 1654 |
+
"loss": 6.1822,
|
| 1655 |
+
"step": 126500
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 16.23,
|
| 1659 |
+
"learning_rate": 0.0013507987220447283,
|
| 1660 |
+
"loss": 6.1777,
|
| 1661 |
+
"step": 127000
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 16.29,
|
| 1665 |
+
"learning_rate": 0.0013482428115015974,
|
| 1666 |
+
"loss": 6.1886,
|
| 1667 |
+
"step": 127500
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 16.36,
|
| 1671 |
+
"learning_rate": 0.0013456869009584665,
|
| 1672 |
+
"loss": 6.1844,
|
| 1673 |
+
"step": 128000
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 16.42,
|
| 1677 |
+
"learning_rate": 0.0013431309904153356,
|
| 1678 |
+
"loss": 6.1775,
|
| 1679 |
+
"step": 128500
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 16.49,
|
| 1683 |
+
"learning_rate": 0.0013405750798722044,
|
| 1684 |
+
"loss": 6.1829,
|
| 1685 |
+
"step": 129000
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 16.55,
|
| 1689 |
+
"learning_rate": 0.0013380191693290735,
|
| 1690 |
+
"loss": 6.1766,
|
| 1691 |
+
"step": 129500
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 16.61,
|
| 1695 |
+
"learning_rate": 0.0013354632587859426,
|
| 1696 |
+
"loss": 6.1847,
|
| 1697 |
+
"step": 130000
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 16.68,
|
| 1701 |
+
"learning_rate": 0.0013329073482428115,
|
| 1702 |
+
"loss": 6.1804,
|
| 1703 |
+
"step": 130500
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 16.74,
|
| 1707 |
+
"learning_rate": 0.0013303514376996806,
|
| 1708 |
+
"loss": 6.1774,
|
| 1709 |
+
"step": 131000
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 16.81,
|
| 1713 |
+
"learning_rate": 0.0013277955271565496,
|
| 1714 |
+
"loss": 6.1767,
|
| 1715 |
+
"step": 131500
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 16.87,
|
| 1719 |
+
"learning_rate": 0.0013252396166134187,
|
| 1720 |
+
"loss": 6.178,
|
| 1721 |
+
"step": 132000
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 16.93,
|
| 1725 |
+
"learning_rate": 0.0013226837060702876,
|
| 1726 |
+
"loss": 6.1806,
|
| 1727 |
+
"step": 132500
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 17.0,
|
| 1731 |
+
"learning_rate": 0.0013201277955271567,
|
| 1732 |
+
"loss": 6.1809,
|
| 1733 |
+
"step": 133000
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 17.0,
|
| 1737 |
+
"eval_loss": 6.124339580535889,
|
| 1738 |
+
"eval_runtime": 6.8363,
|
| 1739 |
+
"eval_samples_per_second": 36.716,
|
| 1740 |
+
"eval_steps_per_second": 1.17,
|
| 1741 |
+
"step": 133025
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 17.06,
|
| 1745 |
+
"learning_rate": 0.0013175718849840258,
|
| 1746 |
+
"loss": 6.1733,
|
| 1747 |
+
"step": 133500
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 17.12,
|
| 1751 |
+
"learning_rate": 0.0013150159744408944,
|
| 1752 |
+
"loss": 6.1825,
|
| 1753 |
+
"step": 134000
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 17.19,
|
| 1757 |
+
"learning_rate": 0.0013124600638977635,
|
| 1758 |
+
"loss": 6.1752,
|
| 1759 |
+
"step": 134500
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 17.25,
|
| 1763 |
+
"learning_rate": 0.0013099041533546326,
|
| 1764 |
+
"loss": 6.1785,
|
| 1765 |
+
"step": 135000
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 17.32,
|
| 1769 |
+
"learning_rate": 0.0013073482428115017,
|
| 1770 |
+
"loss": 6.1761,
|
| 1771 |
+
"step": 135500
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 17.38,
|
| 1775 |
+
"learning_rate": 0.0013047923322683705,
|
| 1776 |
+
"loss": 6.1833,
|
| 1777 |
+
"step": 136000
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 17.44,
|
| 1781 |
+
"learning_rate": 0.0013022364217252396,
|
| 1782 |
+
"loss": 6.1727,
|
| 1783 |
+
"step": 136500
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 17.51,
|
| 1787 |
+
"learning_rate": 0.0012996805111821087,
|
| 1788 |
+
"loss": 6.1744,
|
| 1789 |
+
"step": 137000
|
| 1790 |
+
},
|
| 1791 |
+
{
|
| 1792 |
+
"epoch": 17.57,
|
| 1793 |
+
"learning_rate": 0.0012971246006389776,
|
| 1794 |
+
"loss": 6.1766,
|
| 1795 |
+
"step": 137500
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 17.64,
|
| 1799 |
+
"learning_rate": 0.0012945686900958466,
|
| 1800 |
+
"loss": 6.1754,
|
| 1801 |
+
"step": 138000
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 17.7,
|
| 1805 |
+
"learning_rate": 0.0012920127795527157,
|
| 1806 |
+
"loss": 6.1822,
|
| 1807 |
+
"step": 138500
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 17.76,
|
| 1811 |
+
"learning_rate": 0.0012894568690095848,
|
| 1812 |
+
"loss": 6.1855,
|
| 1813 |
+
"step": 139000
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 17.83,
|
| 1817 |
+
"learning_rate": 0.0012869009584664537,
|
| 1818 |
+
"loss": 6.1797,
|
| 1819 |
+
"step": 139500
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 17.89,
|
| 1823 |
+
"learning_rate": 0.0012843450479233227,
|
| 1824 |
+
"loss": 6.1796,
|
| 1825 |
+
"step": 140000
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 17.96,
|
| 1829 |
+
"learning_rate": 0.0012817891373801918,
|
| 1830 |
+
"loss": 6.1814,
|
| 1831 |
+
"step": 140500
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 18.0,
|
| 1835 |
+
"eval_loss": 6.123664379119873,
|
| 1836 |
+
"eval_runtime": 6.845,
|
| 1837 |
+
"eval_samples_per_second": 36.669,
|
| 1838 |
+
"eval_steps_per_second": 1.169,
|
| 1839 |
+
"step": 140850
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 18.02,
|
| 1843 |
+
"learning_rate": 0.0012792332268370607,
|
| 1844 |
+
"loss": 6.1872,
|
| 1845 |
+
"step": 141000
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 18.08,
|
| 1849 |
+
"learning_rate": 0.0012766773162939298,
|
| 1850 |
+
"loss": 6.1834,
|
| 1851 |
+
"step": 141500
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 18.15,
|
| 1855 |
+
"learning_rate": 0.0012741214057507989,
|
| 1856 |
+
"loss": 6.1793,
|
| 1857 |
+
"step": 142000
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 18.21,
|
| 1861 |
+
"learning_rate": 0.001271565495207668,
|
| 1862 |
+
"loss": 6.1731,
|
| 1863 |
+
"step": 142500
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 18.27,
|
| 1867 |
+
"learning_rate": 0.0012690095846645368,
|
| 1868 |
+
"loss": 6.183,
|
| 1869 |
+
"step": 143000
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 18.34,
|
| 1873 |
+
"learning_rate": 0.001266453674121406,
|
| 1874 |
+
"loss": 6.1745,
|
| 1875 |
+
"step": 143500
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 18.4,
|
| 1879 |
+
"learning_rate": 0.0012638977635782748,
|
| 1880 |
+
"loss": 6.1846,
|
| 1881 |
+
"step": 144000
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 18.47,
|
| 1885 |
+
"learning_rate": 0.0012613418530351436,
|
| 1886 |
+
"loss": 6.1767,
|
| 1887 |
+
"step": 144500
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 18.53,
|
| 1891 |
+
"learning_rate": 0.0012587859424920127,
|
| 1892 |
+
"loss": 6.1747,
|
| 1893 |
+
"step": 145000
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 18.59,
|
| 1897 |
+
"learning_rate": 0.0012562300319488818,
|
| 1898 |
+
"loss": 6.1804,
|
| 1899 |
+
"step": 145500
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 18.66,
|
| 1903 |
+
"learning_rate": 0.0012536741214057507,
|
| 1904 |
+
"loss": 6.1729,
|
| 1905 |
+
"step": 146000
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 18.72,
|
| 1909 |
+
"learning_rate": 0.0012511182108626197,
|
| 1910 |
+
"loss": 6.1791,
|
| 1911 |
+
"step": 146500
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 18.79,
|
| 1915 |
+
"learning_rate": 0.0012485623003194888,
|
| 1916 |
+
"loss": 6.173,
|
| 1917 |
+
"step": 147000
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 18.85,
|
| 1921 |
+
"learning_rate": 0.001246006389776358,
|
| 1922 |
+
"loss": 6.1732,
|
| 1923 |
+
"step": 147500
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 18.91,
|
| 1927 |
+
"learning_rate": 0.0012434504792332268,
|
| 1928 |
+
"loss": 6.1771,
|
| 1929 |
+
"step": 148000
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 18.98,
|
| 1933 |
+
"learning_rate": 0.0012408945686900959,
|
| 1934 |
+
"loss": 6.1754,
|
| 1935 |
+
"step": 148500
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 19.0,
|
| 1939 |
+
"eval_loss": 6.119473934173584,
|
| 1940 |
+
"eval_runtime": 6.9262,
|
| 1941 |
+
"eval_samples_per_second": 36.239,
|
| 1942 |
+
"eval_steps_per_second": 1.155,
|
| 1943 |
+
"step": 148675
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 19.04,
|
| 1947 |
+
"learning_rate": 0.001238338658146965,
|
| 1948 |
+
"loss": 6.1773,
|
| 1949 |
+
"step": 149000
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 19.11,
|
| 1953 |
+
"learning_rate": 0.0012357827476038338,
|
| 1954 |
+
"loss": 6.176,
|
| 1955 |
+
"step": 149500
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 19.17,
|
| 1959 |
+
"learning_rate": 0.001233226837060703,
|
| 1960 |
+
"loss": 6.1746,
|
| 1961 |
+
"step": 150000
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 19.23,
|
| 1965 |
+
"learning_rate": 0.001230670926517572,
|
| 1966 |
+
"loss": 6.1773,
|
| 1967 |
+
"step": 150500
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 19.3,
|
| 1971 |
+
"learning_rate": 0.001228115015974441,
|
| 1972 |
+
"loss": 6.1771,
|
| 1973 |
+
"step": 151000
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 19.36,
|
| 1977 |
+
"learning_rate": 0.00122555910543131,
|
| 1978 |
+
"loss": 6.1736,
|
| 1979 |
+
"step": 151500
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 19.42,
|
| 1983 |
+
"learning_rate": 0.001223003194888179,
|
| 1984 |
+
"loss": 6.1724,
|
| 1985 |
+
"step": 152000
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 19.49,
|
| 1989 |
+
"learning_rate": 0.001220447284345048,
|
| 1990 |
+
"loss": 6.1809,
|
| 1991 |
+
"step": 152500
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 19.55,
|
| 1995 |
+
"learning_rate": 0.001217891373801917,
|
| 1996 |
+
"loss": 6.1769,
|
| 1997 |
+
"step": 153000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 19.62,
|
| 2001 |
+
"learning_rate": 0.001215335463258786,
|
| 2002 |
+
"loss": 6.1712,
|
| 2003 |
+
"step": 153500
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 19.68,
|
| 2007 |
+
"learning_rate": 0.001212779552715655,
|
| 2008 |
+
"loss": 6.1779,
|
| 2009 |
+
"step": 154000
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 19.74,
|
| 2013 |
+
"learning_rate": 0.001210223642172524,
|
| 2014 |
+
"loss": 6.1758,
|
| 2015 |
+
"step": 154500
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 19.81,
|
| 2019 |
+
"learning_rate": 0.0012076677316293929,
|
| 2020 |
+
"loss": 6.1681,
|
| 2021 |
+
"step": 155000
|
| 2022 |
+
},
|
| 2023 |
+
{
|
| 2024 |
+
"epoch": 19.87,
|
| 2025 |
+
"learning_rate": 0.001205111821086262,
|
| 2026 |
+
"loss": 6.1728,
|
| 2027 |
+
"step": 155500
|
| 2028 |
+
},
|
| 2029 |
+
{
|
| 2030 |
+
"epoch": 19.94,
|
| 2031 |
+
"learning_rate": 0.001202555910543131,
|
| 2032 |
+
"loss": 6.1737,
|
| 2033 |
+
"step": 156000
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 20.0,
|
| 2037 |
+
"learning_rate": 0.0012,
|
| 2038 |
+
"loss": 6.1769,
|
| 2039 |
+
"step": 156500
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 20.0,
|
| 2043 |
+
"eval_loss": 6.11764669418335,
|
| 2044 |
+
"eval_runtime": 6.8248,
|
| 2045 |
+
"eval_samples_per_second": 36.778,
|
| 2046 |
+
"eval_steps_per_second": 1.172,
|
| 2047 |
+
"step": 156500
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 20.06,
|
| 2051 |
+
"learning_rate": 0.001197444089456869,
|
| 2052 |
+
"loss": 6.1785,
|
| 2053 |
+
"step": 157000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 20.13,
|
| 2057 |
+
"learning_rate": 0.001194888178913738,
|
| 2058 |
+
"loss": 6.1772,
|
| 2059 |
+
"step": 157500
|
| 2060 |
+
},
|
| 2061 |
+
{
|
| 2062 |
+
"epoch": 20.19,
|
| 2063 |
+
"learning_rate": 0.0011923322683706071,
|
| 2064 |
+
"loss": 6.1648,
|
| 2065 |
+
"step": 158000
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 20.26,
|
| 2069 |
+
"learning_rate": 0.001189776357827476,
|
| 2070 |
+
"loss": 6.1715,
|
| 2071 |
+
"step": 158500
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 20.32,
|
| 2075 |
+
"learning_rate": 0.001187220447284345,
|
| 2076 |
+
"loss": 6.1745,
|
| 2077 |
+
"step": 159000
|
| 2078 |
+
},
|
| 2079 |
+
{
|
| 2080 |
+
"epoch": 20.38,
|
| 2081 |
+
"learning_rate": 0.0011846645367412142,
|
| 2082 |
+
"loss": 6.1769,
|
| 2083 |
+
"step": 159500
|
| 2084 |
+
},
|
| 2085 |
+
{
|
| 2086 |
+
"epoch": 20.45,
|
| 2087 |
+
"learning_rate": 0.001182108626198083,
|
| 2088 |
+
"loss": 6.1783,
|
| 2089 |
+
"step": 160000
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 20.51,
|
| 2093 |
+
"learning_rate": 0.0011795527156549521,
|
| 2094 |
+
"loss": 6.1693,
|
| 2095 |
+
"step": 160500
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 20.58,
|
| 2099 |
+
"learning_rate": 0.0011769968051118212,
|
| 2100 |
+
"loss": 6.1726,
|
| 2101 |
+
"step": 161000
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 20.64,
|
| 2105 |
+
"learning_rate": 0.0011744408945686903,
|
| 2106 |
+
"loss": 6.1676,
|
| 2107 |
+
"step": 161500
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 20.7,
|
| 2111 |
+
"learning_rate": 0.0011718849840255592,
|
| 2112 |
+
"loss": 6.1669,
|
| 2113 |
+
"step": 162000
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 20.77,
|
| 2117 |
+
"learning_rate": 0.0011693290734824282,
|
| 2118 |
+
"loss": 6.1636,
|
| 2119 |
+
"step": 162500
|
| 2120 |
+
},
|
| 2121 |
+
{
|
| 2122 |
+
"epoch": 20.83,
|
| 2123 |
+
"learning_rate": 0.0011667731629392973,
|
| 2124 |
+
"loss": 6.1791,
|
| 2125 |
+
"step": 163000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 20.89,
|
| 2129 |
+
"learning_rate": 0.001164217252396166,
|
| 2130 |
+
"loss": 6.1757,
|
| 2131 |
+
"step": 163500
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 20.96,
|
| 2135 |
+
"learning_rate": 0.001161661341853035,
|
| 2136 |
+
"loss": 6.1716,
|
| 2137 |
+
"step": 164000
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 21.0,
|
| 2141 |
+
"eval_loss": 6.116322994232178,
|
| 2142 |
+
"eval_runtime": 6.8185,
|
| 2143 |
+
"eval_samples_per_second": 36.812,
|
| 2144 |
+
"eval_steps_per_second": 1.173,
|
| 2145 |
+
"step": 164325
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 21.02,
|
| 2149 |
+
"learning_rate": 0.0011591054313099041,
|
| 2150 |
+
"loss": 6.1742,
|
| 2151 |
+
"step": 164500
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 21.09,
|
| 2155 |
+
"learning_rate": 0.001156549520766773,
|
| 2156 |
+
"loss": 6.1689,
|
| 2157 |
+
"step": 165000
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 21.15,
|
| 2161 |
+
"learning_rate": 0.001153993610223642,
|
| 2162 |
+
"loss": 6.162,
|
| 2163 |
+
"step": 165500
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 21.21,
|
| 2167 |
+
"learning_rate": 0.0011514376996805112,
|
| 2168 |
+
"loss": 6.1729,
|
| 2169 |
+
"step": 166000
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 21.28,
|
| 2173 |
+
"learning_rate": 0.0011488817891373803,
|
| 2174 |
+
"loss": 6.1756,
|
| 2175 |
+
"step": 166500
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 21.34,
|
| 2179 |
+
"learning_rate": 0.0011463258785942491,
|
| 2180 |
+
"loss": 6.1694,
|
| 2181 |
+
"step": 167000
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 21.41,
|
| 2185 |
+
"learning_rate": 0.0011437699680511182,
|
| 2186 |
+
"loss": 6.1755,
|
| 2187 |
+
"step": 167500
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 21.47,
|
| 2191 |
+
"learning_rate": 0.0011412140575079873,
|
| 2192 |
+
"loss": 6.1735,
|
| 2193 |
+
"step": 168000
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 21.53,
|
| 2197 |
+
"learning_rate": 0.0011386581469648562,
|
| 2198 |
+
"loss": 6.1676,
|
| 2199 |
+
"step": 168500
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 21.6,
|
| 2203 |
+
"learning_rate": 0.0011361022364217252,
|
| 2204 |
+
"loss": 6.1716,
|
| 2205 |
+
"step": 169000
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 21.66,
|
| 2209 |
+
"learning_rate": 0.0011335463258785943,
|
| 2210 |
+
"loss": 6.1737,
|
| 2211 |
+
"step": 169500
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 21.73,
|
| 2215 |
+
"learning_rate": 0.0011309904153354634,
|
| 2216 |
+
"loss": 6.1743,
|
| 2217 |
+
"step": 170000
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 21.79,
|
| 2221 |
+
"learning_rate": 0.0011284345047923323,
|
| 2222 |
+
"loss": 6.1749,
|
| 2223 |
+
"step": 170500
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 21.85,
|
| 2227 |
+
"learning_rate": 0.0011258785942492014,
|
| 2228 |
+
"loss": 6.1724,
|
| 2229 |
+
"step": 171000
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 21.92,
|
| 2233 |
+
"learning_rate": 0.0011233226837060704,
|
| 2234 |
+
"loss": 6.1676,
|
| 2235 |
+
"step": 171500
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 21.98,
|
| 2239 |
+
"learning_rate": 0.0011207667731629393,
|
| 2240 |
+
"loss": 6.1823,
|
| 2241 |
+
"step": 172000
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 22.0,
|
| 2245 |
+
"eval_loss": 6.115005016326904,
|
| 2246 |
+
"eval_runtime": 57.523,
|
| 2247 |
+
"eval_samples_per_second": 4.363,
|
| 2248 |
+
"eval_steps_per_second": 0.139,
|
| 2249 |
+
"step": 172150
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 22.04,
|
| 2253 |
+
"learning_rate": 0.0011182108626198084,
|
| 2254 |
+
"loss": 6.1729,
|
| 2255 |
+
"step": 172500
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 22.11,
|
| 2259 |
+
"learning_rate": 0.0011156549520766775,
|
| 2260 |
+
"loss": 6.1746,
|
| 2261 |
+
"step": 173000
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 22.17,
|
| 2265 |
+
"learning_rate": 0.0011130990415335463,
|
| 2266 |
+
"loss": 6.1732,
|
| 2267 |
+
"step": 173500
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 22.24,
|
| 2271 |
+
"learning_rate": 0.0011105431309904152,
|
| 2272 |
+
"loss": 6.1712,
|
| 2273 |
+
"step": 174000
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 22.3,
|
| 2277 |
+
"learning_rate": 0.0011079872204472843,
|
| 2278 |
+
"loss": 6.1677,
|
| 2279 |
+
"step": 174500
|
| 2280 |
+
},
|
| 2281 |
+
{
|
| 2282 |
+
"epoch": 22.36,
|
| 2283 |
+
"learning_rate": 0.0011054313099041534,
|
| 2284 |
+
"loss": 6.1683,
|
| 2285 |
+
"step": 175000
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 22.43,
|
| 2289 |
+
"learning_rate": 0.0011028753993610222,
|
| 2290 |
+
"loss": 6.1741,
|
| 2291 |
+
"step": 175500
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 22.49,
|
| 2295 |
+
"learning_rate": 0.0011003194888178913,
|
| 2296 |
+
"loss": 6.1663,
|
| 2297 |
+
"step": 176000
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 22.56,
|
| 2301 |
+
"learning_rate": 0.0010977635782747604,
|
| 2302 |
+
"loss": 6.1681,
|
| 2303 |
+
"step": 176500
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 22.62,
|
| 2307 |
+
"learning_rate": 0.0010952076677316295,
|
| 2308 |
+
"loss": 6.174,
|
| 2309 |
+
"step": 177000
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 22.68,
|
| 2313 |
+
"learning_rate": 0.0010926517571884984,
|
| 2314 |
+
"loss": 6.1674,
|
| 2315 |
+
"step": 177500
|
| 2316 |
+
},
|
| 2317 |
+
{
|
| 2318 |
+
"epoch": 22.75,
|
| 2319 |
+
"learning_rate": 0.0010900958466453674,
|
| 2320 |
+
"loss": 6.1701,
|
| 2321 |
+
"step": 178000
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 22.81,
|
| 2325 |
+
"learning_rate": 0.0010875399361022365,
|
| 2326 |
+
"loss": 6.1724,
|
| 2327 |
+
"step": 178500
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 22.88,
|
| 2331 |
+
"learning_rate": 0.0010849840255591054,
|
| 2332 |
+
"loss": 6.1712,
|
| 2333 |
+
"step": 179000
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 22.94,
|
| 2337 |
+
"learning_rate": 0.0010824281150159745,
|
| 2338 |
+
"loss": 6.1679,
|
| 2339 |
+
"step": 179500
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 23.0,
|
| 2343 |
+
"eval_loss": 6.114222049713135,
|
| 2344 |
+
"eval_runtime": 6.9135,
|
| 2345 |
+
"eval_samples_per_second": 36.306,
|
| 2346 |
+
"eval_steps_per_second": 1.157,
|
| 2347 |
+
"step": 179975
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 23.0,
|
| 2351 |
+
"learning_rate": 0.0010798722044728436,
|
| 2352 |
+
"loss": 6.1666,
|
| 2353 |
+
"step": 180000
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 23.07,
|
| 2357 |
+
"learning_rate": 0.0010773162939297126,
|
| 2358 |
+
"loss": 6.1702,
|
| 2359 |
+
"step": 180500
|
| 2360 |
+
},
|
| 2361 |
+
{
|
| 2362 |
+
"epoch": 23.13,
|
| 2363 |
+
"learning_rate": 0.0010747603833865815,
|
| 2364 |
+
"loss": 6.166,
|
| 2365 |
+
"step": 181000
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 23.19,
|
| 2369 |
+
"learning_rate": 0.0010722044728434506,
|
| 2370 |
+
"loss": 6.1651,
|
| 2371 |
+
"step": 181500
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 23.26,
|
| 2375 |
+
"learning_rate": 0.0010696485623003197,
|
| 2376 |
+
"loss": 6.1721,
|
| 2377 |
+
"step": 182000
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 23.32,
|
| 2381 |
+
"learning_rate": 0.0010670926517571885,
|
| 2382 |
+
"loss": 6.1721,
|
| 2383 |
+
"step": 182500
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 23.39,
|
| 2387 |
+
"learning_rate": 0.0010645367412140576,
|
| 2388 |
+
"loss": 6.168,
|
| 2389 |
+
"step": 183000
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 23.45,
|
| 2393 |
+
"learning_rate": 0.0010619808306709265,
|
| 2394 |
+
"loss": 6.1732,
|
| 2395 |
+
"step": 183500
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 23.51,
|
| 2399 |
+
"learning_rate": 0.0010594249201277954,
|
| 2400 |
+
"loss": 6.171,
|
| 2401 |
+
"step": 184000
|
| 2402 |
+
},
|
| 2403 |
+
{
|
| 2404 |
+
"epoch": 23.58,
|
| 2405 |
+
"learning_rate": 0.0010568690095846644,
|
| 2406 |
+
"loss": 6.1697,
|
| 2407 |
+
"step": 184500
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 23.64,
|
| 2411 |
+
"learning_rate": 0.0010543130990415335,
|
| 2412 |
+
"loss": 6.1667,
|
| 2413 |
+
"step": 185000
|
| 2414 |
+
},
|
| 2415 |
+
{
|
| 2416 |
+
"epoch": 23.71,
|
| 2417 |
+
"learning_rate": 0.0010517571884984026,
|
| 2418 |
+
"loss": 6.1762,
|
| 2419 |
+
"step": 185500
|
| 2420 |
+
},
|
| 2421 |
+
{
|
| 2422 |
+
"epoch": 23.77,
|
| 2423 |
+
"learning_rate": 0.0010492012779552715,
|
| 2424 |
+
"loss": 6.1661,
|
| 2425 |
+
"step": 186000
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 23.83,
|
| 2429 |
+
"learning_rate": 0.0010466453674121406,
|
| 2430 |
+
"loss": 6.1741,
|
| 2431 |
+
"step": 186500
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 23.9,
|
| 2435 |
+
"learning_rate": 0.0010440894568690096,
|
| 2436 |
+
"loss": 6.166,
|
| 2437 |
+
"step": 187000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 23.96,
|
| 2441 |
+
"learning_rate": 0.0010415335463258785,
|
| 2442 |
+
"loss": 6.1697,
|
| 2443 |
+
"step": 187500
|
| 2444 |
+
},
|
| 2445 |
+
{
|
| 2446 |
+
"epoch": 24.0,
|
| 2447 |
+
"eval_loss": 6.112667083740234,
|
| 2448 |
+
"eval_runtime": 6.8444,
|
| 2449 |
+
"eval_samples_per_second": 36.672,
|
| 2450 |
+
"eval_steps_per_second": 1.169,
|
| 2451 |
+
"step": 187800
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 24.03,
|
| 2455 |
+
"learning_rate": 0.0010389776357827476,
|
| 2456 |
+
"loss": 6.1674,
|
| 2457 |
+
"step": 188000
|
| 2458 |
+
},
|
| 2459 |
+
{
|
| 2460 |
+
"epoch": 24.09,
|
| 2461 |
+
"learning_rate": 0.0010364217252396167,
|
| 2462 |
+
"loss": 6.169,
|
| 2463 |
+
"step": 188500
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 24.15,
|
| 2467 |
+
"learning_rate": 0.0010338658146964858,
|
| 2468 |
+
"loss": 6.1648,
|
| 2469 |
+
"step": 189000
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 24.22,
|
| 2473 |
+
"learning_rate": 0.0010313099041533546,
|
| 2474 |
+
"loss": 6.1711,
|
| 2475 |
+
"step": 189500
|
| 2476 |
+
},
|
| 2477 |
+
{
|
| 2478 |
+
"epoch": 24.28,
|
| 2479 |
+
"learning_rate": 0.0010287539936102237,
|
| 2480 |
+
"loss": 6.1621,
|
| 2481 |
+
"step": 190000
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 24.35,
|
| 2485 |
+
"learning_rate": 0.0010261980830670928,
|
| 2486 |
+
"loss": 6.1666,
|
| 2487 |
+
"step": 190500
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 24.41,
|
| 2491 |
+
"learning_rate": 0.0010236421725239617,
|
| 2492 |
+
"loss": 6.1685,
|
| 2493 |
+
"step": 191000
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 24.47,
|
| 2497 |
+
"learning_rate": 0.0010210862619808307,
|
| 2498 |
+
"loss": 6.171,
|
| 2499 |
+
"step": 191500
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 24.54,
|
| 2503 |
+
"learning_rate": 0.0010185303514376998,
|
| 2504 |
+
"loss": 6.1777,
|
| 2505 |
+
"step": 192000
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 24.6,
|
| 2509 |
+
"learning_rate": 0.001015974440894569,
|
| 2510 |
+
"loss": 6.1717,
|
| 2511 |
+
"step": 192500
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 24.66,
|
| 2515 |
+
"learning_rate": 0.0010134185303514378,
|
| 2516 |
+
"loss": 6.1702,
|
| 2517 |
+
"step": 193000
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 24.73,
|
| 2521 |
+
"learning_rate": 0.0010108626198083066,
|
| 2522 |
+
"loss": 6.1728,
|
| 2523 |
+
"step": 193500
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 24.79,
|
| 2527 |
+
"learning_rate": 0.0010083067092651757,
|
| 2528 |
+
"loss": 6.1671,
|
| 2529 |
+
"step": 194000
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 24.86,
|
| 2533 |
+
"learning_rate": 0.0010057507987220446,
|
| 2534 |
+
"loss": 6.16,
|
| 2535 |
+
"step": 194500
|
| 2536 |
+
},
|
| 2537 |
+
{
|
| 2538 |
+
"epoch": 24.92,
|
| 2539 |
+
"learning_rate": 0.0010031948881789137,
|
| 2540 |
+
"loss": 6.1631,
|
| 2541 |
+
"step": 195000
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 24.98,
|
| 2545 |
+
"learning_rate": 0.0010006389776357828,
|
| 2546 |
+
"loss": 6.1719,
|
| 2547 |
+
"step": 195500
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 25.0,
|
| 2551 |
+
"eval_loss": 6.112457275390625,
|
| 2552 |
+
"eval_runtime": 6.8286,
|
| 2553 |
+
"eval_samples_per_second": 36.757,
|
| 2554 |
+
"eval_steps_per_second": 1.172,
|
| 2555 |
+
"step": 195625
|
| 2556 |
+
},
|
| 2557 |
+
{
|
| 2558 |
+
"epoch": 25.05,
|
| 2559 |
+
"learning_rate": 0.0009980830670926518,
|
| 2560 |
+
"loss": 6.1638,
|
| 2561 |
+
"step": 196000
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 25.11,
|
| 2565 |
+
"learning_rate": 0.0009955271565495207,
|
| 2566 |
+
"loss": 6.1723,
|
| 2567 |
+
"step": 196500
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 25.18,
|
| 2571 |
+
"learning_rate": 0.0009929712460063898,
|
| 2572 |
+
"loss": 6.1753,
|
| 2573 |
+
"step": 197000
|
| 2574 |
+
},
|
| 2575 |
+
{
|
| 2576 |
+
"epoch": 25.24,
|
| 2577 |
+
"learning_rate": 0.0009904153354632589,
|
| 2578 |
+
"loss": 6.1636,
|
| 2579 |
+
"step": 197500
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 25.3,
|
| 2583 |
+
"learning_rate": 0.0009878594249201277,
|
| 2584 |
+
"loss": 6.1734,
|
| 2585 |
+
"step": 198000
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 25.37,
|
| 2589 |
+
"learning_rate": 0.0009853035143769968,
|
| 2590 |
+
"loss": 6.1705,
|
| 2591 |
+
"step": 198500
|
| 2592 |
+
},
|
| 2593 |
+
{
|
| 2594 |
+
"epoch": 25.43,
|
| 2595 |
+
"learning_rate": 0.000982747603833866,
|
| 2596 |
+
"loss": 6.1615,
|
| 2597 |
+
"step": 199000
|
| 2598 |
+
},
|
| 2599 |
+
{
|
| 2600 |
+
"epoch": 25.5,
|
| 2601 |
+
"learning_rate": 0.000980191693290735,
|
| 2602 |
+
"loss": 6.1636,
|
| 2603 |
+
"step": 199500
|
| 2604 |
+
},
|
| 2605 |
+
{
|
| 2606 |
+
"epoch": 25.56,
|
| 2607 |
+
"learning_rate": 0.0009776357827476038,
|
| 2608 |
+
"loss": 6.1756,
|
| 2609 |
+
"step": 200000
|
| 2610 |
+
},
|
| 2611 |
+
{
|
| 2612 |
+
"epoch": 25.62,
|
| 2613 |
+
"learning_rate": 0.0009750798722044729,
|
| 2614 |
+
"loss": 6.1726,
|
| 2615 |
+
"step": 200500
|
| 2616 |
+
},
|
| 2617 |
+
{
|
| 2618 |
+
"epoch": 25.69,
|
| 2619 |
+
"learning_rate": 0.0009725239616613418,
|
| 2620 |
+
"loss": 6.1758,
|
| 2621 |
+
"step": 201000
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 25.75,
|
| 2625 |
+
"learning_rate": 0.0009699680511182109,
|
| 2626 |
+
"loss": 6.1662,
|
| 2627 |
+
"step": 201500
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 25.81,
|
| 2631 |
+
"learning_rate": 0.0009674121405750799,
|
| 2632 |
+
"loss": 6.1656,
|
| 2633 |
+
"step": 202000
|
| 2634 |
+
},
|
| 2635 |
+
{
|
| 2636 |
+
"epoch": 25.88,
|
| 2637 |
+
"learning_rate": 0.0009648562300319489,
|
| 2638 |
+
"loss": 6.1568,
|
| 2639 |
+
"step": 202500
|
| 2640 |
+
},
|
| 2641 |
+
{
|
| 2642 |
+
"epoch": 25.94,
|
| 2643 |
+
"learning_rate": 0.0009623003194888179,
|
| 2644 |
+
"loss": 6.1678,
|
| 2645 |
+
"step": 203000
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 26.0,
|
| 2649 |
+
"eval_loss": 6.112171649932861,
|
| 2650 |
+
"eval_runtime": 6.8367,
|
| 2651 |
+
"eval_samples_per_second": 36.714,
|
| 2652 |
+
"eval_steps_per_second": 1.17,
|
| 2653 |
+
"step": 203450
|
| 2654 |
+
},
|
| 2655 |
+
{
|
| 2656 |
+
"epoch": 26.01,
|
| 2657 |
+
"learning_rate": 0.0009597444089456869,
|
| 2658 |
+
"loss": 6.1698,
|
| 2659 |
+
"step": 203500
|
| 2660 |
+
},
|
| 2661 |
+
{
|
| 2662 |
+
"epoch": 26.07,
|
| 2663 |
+
"learning_rate": 0.000957188498402556,
|
| 2664 |
+
"loss": 6.158,
|
| 2665 |
+
"step": 204000
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 26.13,
|
| 2669 |
+
"learning_rate": 0.000954632587859425,
|
| 2670 |
+
"loss": 6.1723,
|
| 2671 |
+
"step": 204500
|
| 2672 |
+
},
|
| 2673 |
+
{
|
| 2674 |
+
"epoch": 26.2,
|
| 2675 |
+
"learning_rate": 0.000952076677316294,
|
| 2676 |
+
"loss": 6.1665,
|
| 2677 |
+
"step": 205000
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 26.26,
|
| 2681 |
+
"learning_rate": 0.000949520766773163,
|
| 2682 |
+
"loss": 6.1657,
|
| 2683 |
+
"step": 205500
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 26.33,
|
| 2687 |
+
"learning_rate": 0.0009469648562300319,
|
| 2688 |
+
"loss": 6.1692,
|
| 2689 |
+
"step": 206000
|
| 2690 |
+
},
|
| 2691 |
+
{
|
| 2692 |
+
"epoch": 26.39,
|
| 2693 |
+
"learning_rate": 0.000944408945686901,
|
| 2694 |
+
"loss": 6.1713,
|
| 2695 |
+
"step": 206500
|
| 2696 |
+
},
|
| 2697 |
+
{
|
| 2698 |
+
"epoch": 26.45,
|
| 2699 |
+
"learning_rate": 0.0009418530351437699,
|
| 2700 |
+
"loss": 6.1642,
|
| 2701 |
+
"step": 207000
|
| 2702 |
+
},
|
| 2703 |
+
{
|
| 2704 |
+
"epoch": 26.52,
|
| 2705 |
+
"learning_rate": 0.000939297124600639,
|
| 2706 |
+
"loss": 6.1679,
|
| 2707 |
+
"step": 207500
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 26.58,
|
| 2711 |
+
"learning_rate": 0.000936741214057508,
|
| 2712 |
+
"loss": 6.1662,
|
| 2713 |
+
"step": 208000
|
| 2714 |
+
},
|
| 2715 |
+
{
|
| 2716 |
+
"epoch": 26.65,
|
| 2717 |
+
"learning_rate": 0.0009341853035143771,
|
| 2718 |
+
"loss": 6.1701,
|
| 2719 |
+
"step": 208500
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 26.71,
|
| 2723 |
+
"learning_rate": 0.000931629392971246,
|
| 2724 |
+
"loss": 6.1702,
|
| 2725 |
+
"step": 209000
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 26.77,
|
| 2729 |
+
"learning_rate": 0.000929073482428115,
|
| 2730 |
+
"loss": 6.1672,
|
| 2731 |
+
"step": 209500
|
| 2732 |
+
},
|
| 2733 |
+
{
|
| 2734 |
+
"epoch": 26.84,
|
| 2735 |
+
"learning_rate": 0.0009265175718849841,
|
| 2736 |
+
"loss": 6.1697,
|
| 2737 |
+
"step": 210000
|
| 2738 |
+
},
|
| 2739 |
+
{
|
| 2740 |
+
"epoch": 26.9,
|
| 2741 |
+
"learning_rate": 0.000923961661341853,
|
| 2742 |
+
"loss": 6.1657,
|
| 2743 |
+
"step": 210500
|
| 2744 |
+
},
|
| 2745 |
+
{
|
| 2746 |
+
"epoch": 26.96,
|
| 2747 |
+
"learning_rate": 0.000921405750798722,
|
| 2748 |
+
"loss": 6.1664,
|
| 2749 |
+
"step": 211000
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 27.0,
|
| 2753 |
+
"eval_loss": 6.112409591674805,
|
| 2754 |
+
"eval_runtime": 6.9228,
|
| 2755 |
+
"eval_samples_per_second": 36.257,
|
| 2756 |
+
"eval_steps_per_second": 1.156,
|
| 2757 |
+
"step": 211275
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 27.03,
|
| 2761 |
+
"learning_rate": 0.000918849840255591,
|
| 2762 |
+
"loss": 6.164,
|
| 2763 |
+
"step": 211500
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 27.09,
|
| 2767 |
+
"learning_rate": 0.0009162939297124601,
|
| 2768 |
+
"loss": 6.1717,
|
| 2769 |
+
"step": 212000
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 27.16,
|
| 2773 |
+
"learning_rate": 0.0009137380191693291,
|
| 2774 |
+
"loss": 6.1646,
|
| 2775 |
+
"step": 212500
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 27.22,
|
| 2779 |
+
"learning_rate": 0.0009111821086261981,
|
| 2780 |
+
"loss": 6.1641,
|
| 2781 |
+
"step": 213000
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 27.28,
|
| 2785 |
+
"learning_rate": 0.0009086261980830671,
|
| 2786 |
+
"loss": 6.1634,
|
| 2787 |
+
"step": 213500
|
| 2788 |
+
},
|
| 2789 |
+
{
|
| 2790 |
+
"epoch": 27.35,
|
| 2791 |
+
"learning_rate": 0.0009060702875399361,
|
| 2792 |
+
"loss": 6.17,
|
| 2793 |
+
"step": 214000
|
| 2794 |
+
},
|
| 2795 |
+
{
|
| 2796 |
+
"epoch": 27.41,
|
| 2797 |
+
"learning_rate": 0.0009035143769968052,
|
| 2798 |
+
"loss": 6.1703,
|
| 2799 |
+
"step": 214500
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 27.48,
|
| 2803 |
+
"learning_rate": 0.0009009584664536742,
|
| 2804 |
+
"loss": 6.1721,
|
| 2805 |
+
"step": 215000
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 27.54,
|
| 2809 |
+
"learning_rate": 0.000898402555910543,
|
| 2810 |
+
"loss": 6.1701,
|
| 2811 |
+
"step": 215500
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 27.6,
|
| 2815 |
+
"learning_rate": 0.0008958466453674121,
|
| 2816 |
+
"loss": 6.1649,
|
| 2817 |
+
"step": 216000
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 27.67,
|
| 2821 |
+
"learning_rate": 0.0008932907348242811,
|
| 2822 |
+
"loss": 6.1727,
|
| 2823 |
+
"step": 216500
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 27.73,
|
| 2827 |
+
"learning_rate": 0.0008907348242811502,
|
| 2828 |
+
"loss": 6.1688,
|
| 2829 |
+
"step": 217000
|
| 2830 |
+
},
|
| 2831 |
+
{
|
| 2832 |
+
"epoch": 27.8,
|
| 2833 |
+
"learning_rate": 0.0008881789137380192,
|
| 2834 |
+
"loss": 6.166,
|
| 2835 |
+
"step": 217500
|
| 2836 |
+
},
|
| 2837 |
+
{
|
| 2838 |
+
"epoch": 27.86,
|
| 2839 |
+
"learning_rate": 0.0008856230031948882,
|
| 2840 |
+
"loss": 6.1683,
|
| 2841 |
+
"step": 218000
|
| 2842 |
+
},
|
| 2843 |
+
{
|
| 2844 |
+
"epoch": 27.92,
|
| 2845 |
+
"learning_rate": 0.0008830670926517572,
|
| 2846 |
+
"loss": 6.1647,
|
| 2847 |
+
"step": 218500
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 27.99,
|
| 2851 |
+
"learning_rate": 0.0008805111821086262,
|
| 2852 |
+
"loss": 6.1643,
|
| 2853 |
+
"step": 219000
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 28.0,
|
| 2857 |
+
"eval_loss": 6.111097812652588,
|
| 2858 |
+
"eval_runtime": 6.8343,
|
| 2859 |
+
"eval_samples_per_second": 36.727,
|
| 2860 |
+
"eval_steps_per_second": 1.171,
|
| 2861 |
+
"step": 219100
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 28.05,
|
| 2865 |
+
"learning_rate": 0.0008779552715654953,
|
| 2866 |
+
"loss": 6.1691,
|
| 2867 |
+
"step": 219500
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 28.12,
|
| 2871 |
+
"learning_rate": 0.0008753993610223643,
|
| 2872 |
+
"loss": 6.1653,
|
| 2873 |
+
"step": 220000
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 28.18,
|
| 2877 |
+
"learning_rate": 0.0008728434504792332,
|
| 2878 |
+
"loss": 6.1633,
|
| 2879 |
+
"step": 220500
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 28.24,
|
| 2883 |
+
"learning_rate": 0.0008702875399361022,
|
| 2884 |
+
"loss": 6.1705,
|
| 2885 |
+
"step": 221000
|
| 2886 |
+
},
|
| 2887 |
+
{
|
| 2888 |
+
"epoch": 28.31,
|
| 2889 |
+
"learning_rate": 0.0008677316293929713,
|
| 2890 |
+
"loss": 6.1695,
|
| 2891 |
+
"step": 221500
|
| 2892 |
+
},
|
| 2893 |
+
{
|
| 2894 |
+
"epoch": 28.37,
|
| 2895 |
+
"learning_rate": 0.0008651757188498403,
|
| 2896 |
+
"loss": 6.168,
|
| 2897 |
+
"step": 222000
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 28.43,
|
| 2901 |
+
"learning_rate": 0.0008626198083067092,
|
| 2902 |
+
"loss": 6.1652,
|
| 2903 |
+
"step": 222500
|
| 2904 |
+
},
|
| 2905 |
+
{
|
| 2906 |
+
"epoch": 28.5,
|
| 2907 |
+
"learning_rate": 0.0008600638977635783,
|
| 2908 |
+
"loss": 6.1695,
|
| 2909 |
+
"step": 223000
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 28.56,
|
| 2913 |
+
"learning_rate": 0.0008575079872204473,
|
| 2914 |
+
"loss": 6.1675,
|
| 2915 |
+
"step": 223500
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 28.63,
|
| 2919 |
+
"learning_rate": 0.0008549520766773164,
|
| 2920 |
+
"loss": 6.1634,
|
| 2921 |
+
"step": 224000
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 28.69,
|
| 2925 |
+
"learning_rate": 0.0008523961661341853,
|
| 2926 |
+
"loss": 6.1728,
|
| 2927 |
+
"step": 224500
|
| 2928 |
+
},
|
| 2929 |
+
{
|
| 2930 |
+
"epoch": 28.75,
|
| 2931 |
+
"learning_rate": 0.0008498402555910543,
|
| 2932 |
+
"loss": 6.1689,
|
| 2933 |
+
"step": 225000
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 28.82,
|
| 2937 |
+
"learning_rate": 0.0008472843450479233,
|
| 2938 |
+
"loss": 6.1664,
|
| 2939 |
+
"step": 225500
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 28.88,
|
| 2943 |
+
"learning_rate": 0.0008447284345047923,
|
| 2944 |
+
"loss": 6.1696,
|
| 2945 |
+
"step": 226000
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 28.95,
|
| 2949 |
+
"learning_rate": 0.0008421725239616614,
|
| 2950 |
+
"loss": 6.1629,
|
| 2951 |
+
"step": 226500
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 29.0,
|
| 2955 |
+
"eval_loss": 6.1112260818481445,
|
| 2956 |
+
"eval_runtime": 6.8361,
|
| 2957 |
+
"eval_samples_per_second": 36.717,
|
| 2958 |
+
"eval_steps_per_second": 1.17,
|
| 2959 |
+
"step": 226925
|
| 2960 |
+
},
|
| 2961 |
+
{
|
| 2962 |
+
"epoch": 29.01,
|
| 2963 |
+
"learning_rate": 0.0008396166134185303,
|
| 2964 |
+
"loss": 6.1625,
|
| 2965 |
+
"step": 227000
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 29.07,
|
| 2969 |
+
"learning_rate": 0.0008370607028753994,
|
| 2970 |
+
"loss": 6.167,
|
| 2971 |
+
"step": 227500
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 29.14,
|
| 2975 |
+
"learning_rate": 0.0008345047923322684,
|
| 2976 |
+
"loss": 6.1627,
|
| 2977 |
+
"step": 228000
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 29.2,
|
| 2981 |
+
"learning_rate": 0.0008319488817891374,
|
| 2982 |
+
"loss": 6.1653,
|
| 2983 |
+
"step": 228500
|
| 2984 |
+
},
|
| 2985 |
+
{
|
| 2986 |
+
"epoch": 29.27,
|
| 2987 |
+
"learning_rate": 0.0008293929712460064,
|
| 2988 |
+
"loss": 6.1711,
|
| 2989 |
+
"step": 229000
|
| 2990 |
+
},
|
| 2991 |
+
{
|
| 2992 |
+
"epoch": 29.33,
|
| 2993 |
+
"learning_rate": 0.0008268370607028754,
|
| 2994 |
+
"loss": 6.1661,
|
| 2995 |
+
"step": 229500
|
| 2996 |
+
},
|
| 2997 |
+
{
|
| 2998 |
+
"epoch": 29.39,
|
| 2999 |
+
"learning_rate": 0.0008242811501597445,
|
| 3000 |
+
"loss": 6.1655,
|
| 3001 |
+
"step": 230000
|
| 3002 |
+
},
|
| 3003 |
+
{
|
| 3004 |
+
"epoch": 29.46,
|
| 3005 |
+
"learning_rate": 0.0008217252396166134,
|
| 3006 |
+
"loss": 6.1738,
|
| 3007 |
+
"step": 230500
|
| 3008 |
+
},
|
| 3009 |
+
{
|
| 3010 |
+
"epoch": 29.52,
|
| 3011 |
+
"learning_rate": 0.0008191693290734825,
|
| 3012 |
+
"loss": 6.1657,
|
| 3013 |
+
"step": 231000
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 29.58,
|
| 3017 |
+
"learning_rate": 0.0008166134185303514,
|
| 3018 |
+
"loss": 6.1686,
|
| 3019 |
+
"step": 231500
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 29.65,
|
| 3023 |
+
"learning_rate": 0.0008140575079872204,
|
| 3024 |
+
"loss": 6.1636,
|
| 3025 |
+
"step": 232000
|
| 3026 |
+
},
|
| 3027 |
+
{
|
| 3028 |
+
"epoch": 29.71,
|
| 3029 |
+
"learning_rate": 0.0008115015974440895,
|
| 3030 |
+
"loss": 6.1645,
|
| 3031 |
+
"step": 232500
|
| 3032 |
+
},
|
| 3033 |
+
{
|
| 3034 |
+
"epoch": 29.78,
|
| 3035 |
+
"learning_rate": 0.0008089456869009585,
|
| 3036 |
+
"loss": 6.1653,
|
| 3037 |
+
"step": 233000
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 29.84,
|
| 3041 |
+
"learning_rate": 0.0008063897763578275,
|
| 3042 |
+
"loss": 6.1602,
|
| 3043 |
+
"step": 233500
|
| 3044 |
+
},
|
| 3045 |
+
{
|
| 3046 |
+
"epoch": 29.9,
|
| 3047 |
+
"learning_rate": 0.0008038338658146965,
|
| 3048 |
+
"loss": 6.1667,
|
| 3049 |
+
"step": 234000
|
| 3050 |
+
},
|
| 3051 |
+
{
|
| 3052 |
+
"epoch": 29.97,
|
| 3053 |
+
"learning_rate": 0.0008012779552715655,
|
| 3054 |
+
"loss": 6.1712,
|
| 3055 |
+
"step": 234500
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 30.0,
|
| 3059 |
+
"eval_loss": 6.1112165451049805,
|
| 3060 |
+
"eval_runtime": 6.8132,
|
| 3061 |
+
"eval_samples_per_second": 36.84,
|
| 3062 |
+
"eval_steps_per_second": 1.174,
|
| 3063 |
+
"step": 234750
|
| 3064 |
+
},
|
| 3065 |
+
{
|
| 3066 |
+
"epoch": 30.03,
|
| 3067 |
+
"learning_rate": 0.0007987220447284346,
|
| 3068 |
+
"loss": 6.1752,
|
| 3069 |
+
"step": 235000
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 30.1,
|
| 3073 |
+
"learning_rate": 0.0007961661341853034,
|
| 3074 |
+
"loss": 6.1726,
|
| 3075 |
+
"step": 235500
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 30.16,
|
| 3079 |
+
"learning_rate": 0.0007936102236421725,
|
| 3080 |
+
"loss": 6.1653,
|
| 3081 |
+
"step": 236000
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 30.22,
|
| 3085 |
+
"learning_rate": 0.0007910543130990415,
|
| 3086 |
+
"loss": 6.1677,
|
| 3087 |
+
"step": 236500
|
| 3088 |
+
},
|
| 3089 |
+
{
|
| 3090 |
+
"epoch": 30.29,
|
| 3091 |
+
"learning_rate": 0.0007884984025559106,
|
| 3092 |
+
"loss": 6.1666,
|
| 3093 |
+
"step": 237000
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 30.35,
|
| 3097 |
+
"learning_rate": 0.0007859424920127796,
|
| 3098 |
+
"loss": 6.1628,
|
| 3099 |
+
"step": 237500
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 30.42,
|
| 3103 |
+
"learning_rate": 0.0007833865814696485,
|
| 3104 |
+
"loss": 6.1717,
|
| 3105 |
+
"step": 238000
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 30.48,
|
| 3109 |
+
"learning_rate": 0.0007808306709265176,
|
| 3110 |
+
"loss": 6.1636,
|
| 3111 |
+
"step": 238500
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 30.54,
|
| 3115 |
+
"learning_rate": 0.0007782747603833866,
|
| 3116 |
+
"loss": 6.1666,
|
| 3117 |
+
"step": 239000
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 30.61,
|
| 3121 |
+
"learning_rate": 0.0007757188498402557,
|
| 3122 |
+
"loss": 6.1612,
|
| 3123 |
+
"step": 239500
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 30.67,
|
| 3127 |
+
"learning_rate": 0.0007731629392971247,
|
| 3128 |
+
"loss": 6.1623,
|
| 3129 |
+
"step": 240000
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 30.73,
|
| 3133 |
+
"learning_rate": 0.0007706070287539936,
|
| 3134 |
+
"loss": 6.1657,
|
| 3135 |
+
"step": 240500
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 30.8,
|
| 3139 |
+
"learning_rate": 0.0007680511182108626,
|
| 3140 |
+
"loss": 6.1634,
|
| 3141 |
+
"step": 241000
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 30.86,
|
| 3145 |
+
"learning_rate": 0.0007654952076677316,
|
| 3146 |
+
"loss": 6.1704,
|
| 3147 |
+
"step": 241500
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 30.93,
|
| 3151 |
+
"learning_rate": 0.0007629392971246007,
|
| 3152 |
+
"loss": 6.1684,
|
| 3153 |
+
"step": 242000
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 30.99,
|
| 3157 |
+
"learning_rate": 0.0007603833865814696,
|
| 3158 |
+
"loss": 6.1588,
|
| 3159 |
+
"step": 242500
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 31.0,
|
| 3163 |
+
"eval_loss": 6.110002040863037,
|
| 3164 |
+
"eval_runtime": 6.9156,
|
| 3165 |
+
"eval_samples_per_second": 36.295,
|
| 3166 |
+
"eval_steps_per_second": 1.157,
|
| 3167 |
+
"step": 242575
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 31.05,
|
| 3171 |
+
"learning_rate": 0.0007578274760383387,
|
| 3172 |
+
"loss": 6.167,
|
| 3173 |
+
"step": 243000
|
| 3174 |
+
},
|
| 3175 |
+
{
|
| 3176 |
+
"epoch": 31.12,
|
| 3177 |
+
"learning_rate": 0.0007552715654952077,
|
| 3178 |
+
"loss": 6.1609,
|
| 3179 |
+
"step": 243500
|
| 3180 |
+
},
|
| 3181 |
+
{
|
| 3182 |
+
"epoch": 31.18,
|
| 3183 |
+
"learning_rate": 0.0007527156549520767,
|
| 3184 |
+
"loss": 6.1638,
|
| 3185 |
+
"step": 244000
|
| 3186 |
+
},
|
| 3187 |
+
{
|
| 3188 |
+
"epoch": 31.25,
|
| 3189 |
+
"learning_rate": 0.0007501597444089458,
|
| 3190 |
+
"loss": 6.1655,
|
| 3191 |
+
"step": 244500
|
| 3192 |
+
},
|
| 3193 |
+
{
|
| 3194 |
+
"epoch": 31.31,
|
| 3195 |
+
"learning_rate": 0.0007476038338658147,
|
| 3196 |
+
"loss": 6.1675,
|
| 3197 |
+
"step": 245000
|
| 3198 |
+
},
|
| 3199 |
+
{
|
| 3200 |
+
"epoch": 31.37,
|
| 3201 |
+
"learning_rate": 0.0007450479233226837,
|
| 3202 |
+
"loss": 6.1714,
|
| 3203 |
+
"step": 245500
|
| 3204 |
+
},
|
| 3205 |
+
{
|
| 3206 |
+
"epoch": 31.44,
|
| 3207 |
+
"learning_rate": 0.0007424920127795527,
|
| 3208 |
+
"loss": 6.1628,
|
| 3209 |
+
"step": 246000
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 31.5,
|
| 3213 |
+
"learning_rate": 0.0007399361022364218,
|
| 3214 |
+
"loss": 6.1674,
|
| 3215 |
+
"step": 246500
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 31.57,
|
| 3219 |
+
"learning_rate": 0.0007373801916932907,
|
| 3220 |
+
"loss": 6.1589,
|
| 3221 |
+
"step": 247000
|
| 3222 |
+
},
|
| 3223 |
+
{
|
| 3224 |
+
"epoch": 31.63,
|
| 3225 |
+
"learning_rate": 0.0007348242811501597,
|
| 3226 |
+
"loss": 6.1727,
|
| 3227 |
+
"step": 247500
|
| 3228 |
+
},
|
| 3229 |
+
{
|
| 3230 |
+
"epoch": 31.69,
|
| 3231 |
+
"learning_rate": 0.0007322683706070288,
|
| 3232 |
+
"loss": 6.1688,
|
| 3233 |
+
"step": 248000
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 31.76,
|
| 3237 |
+
"learning_rate": 0.0007297124600638978,
|
| 3238 |
+
"loss": 6.1678,
|
| 3239 |
+
"step": 248500
|
| 3240 |
+
},
|
| 3241 |
+
{
|
| 3242 |
+
"epoch": 31.82,
|
| 3243 |
+
"learning_rate": 0.0007271565495207669,
|
| 3244 |
+
"loss": 6.162,
|
| 3245 |
+
"step": 249000
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 31.88,
|
| 3249 |
+
"learning_rate": 0.0007246006389776358,
|
| 3250 |
+
"loss": 6.1671,
|
| 3251 |
+
"step": 249500
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 31.95,
|
| 3255 |
+
"learning_rate": 0.0007220447284345049,
|
| 3256 |
+
"loss": 6.1686,
|
| 3257 |
+
"step": 250000
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 32.0,
|
| 3261 |
+
"eval_loss": 6.110647201538086,
|
| 3262 |
+
"eval_runtime": 6.86,
|
| 3263 |
+
"eval_samples_per_second": 36.589,
|
| 3264 |
+
"eval_steps_per_second": 1.166,
|
| 3265 |
+
"step": 250400
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 32.01,
|
| 3269 |
+
"learning_rate": 0.0007194888178913738,
|
| 3270 |
+
"loss": 6.1651,
|
| 3271 |
+
"step": 250500
|
| 3272 |
+
},
|
| 3273 |
+
{
|
| 3274 |
+
"epoch": 32.08,
|
| 3275 |
+
"learning_rate": 0.0007169329073482428,
|
| 3276 |
+
"loss": 6.1736,
|
| 3277 |
+
"step": 251000
|
| 3278 |
+
},
|
| 3279 |
+
{
|
| 3280 |
+
"epoch": 32.14,
|
| 3281 |
+
"learning_rate": 0.0007143769968051118,
|
| 3282 |
+
"loss": 6.1687,
|
| 3283 |
+
"step": 251500
|
| 3284 |
+
},
|
| 3285 |
+
{
|
| 3286 |
+
"epoch": 32.2,
|
| 3287 |
+
"learning_rate": 0.0007118210862619808,
|
| 3288 |
+
"loss": 6.1646,
|
| 3289 |
+
"step": 252000
|
| 3290 |
+
},
|
| 3291 |
+
{
|
| 3292 |
+
"epoch": 32.27,
|
| 3293 |
+
"learning_rate": 0.0007092651757188499,
|
| 3294 |
+
"loss": 6.1676,
|
| 3295 |
+
"step": 252500
|
| 3296 |
+
},
|
| 3297 |
+
{
|
| 3298 |
+
"epoch": 32.33,
|
| 3299 |
+
"learning_rate": 0.0007067092651757189,
|
| 3300 |
+
"loss": 6.1536,
|
| 3301 |
+
"step": 253000
|
| 3302 |
+
},
|
| 3303 |
+
{
|
| 3304 |
+
"epoch": 32.4,
|
| 3305 |
+
"learning_rate": 0.0007041533546325878,
|
| 3306 |
+
"loss": 6.1628,
|
| 3307 |
+
"step": 253500
|
| 3308 |
+
},
|
| 3309 |
+
{
|
| 3310 |
+
"epoch": 32.46,
|
| 3311 |
+
"learning_rate": 0.0007015974440894569,
|
| 3312 |
+
"loss": 6.1662,
|
| 3313 |
+
"step": 254000
|
| 3314 |
+
},
|
| 3315 |
+
{
|
| 3316 |
+
"epoch": 32.52,
|
| 3317 |
+
"learning_rate": 0.0006990415335463259,
|
| 3318 |
+
"loss": 6.171,
|
| 3319 |
+
"step": 254500
|
| 3320 |
+
},
|
| 3321 |
+
{
|
| 3322 |
+
"epoch": 32.59,
|
| 3323 |
+
"learning_rate": 0.000696485623003195,
|
| 3324 |
+
"loss": 6.1691,
|
| 3325 |
+
"step": 255000
|
| 3326 |
+
},
|
| 3327 |
+
{
|
| 3328 |
+
"epoch": 32.65,
|
| 3329 |
+
"learning_rate": 0.0006939297124600638,
|
| 3330 |
+
"loss": 6.1668,
|
| 3331 |
+
"step": 255500
|
| 3332 |
+
},
|
| 3333 |
+
{
|
| 3334 |
+
"epoch": 32.72,
|
| 3335 |
+
"learning_rate": 0.0006913738019169329,
|
| 3336 |
+
"loss": 6.1738,
|
| 3337 |
+
"step": 256000
|
| 3338 |
+
},
|
| 3339 |
+
{
|
| 3340 |
+
"epoch": 32.78,
|
| 3341 |
+
"learning_rate": 0.0006888178913738019,
|
| 3342 |
+
"loss": 6.1652,
|
| 3343 |
+
"step": 256500
|
| 3344 |
+
},
|
| 3345 |
+
{
|
| 3346 |
+
"epoch": 32.84,
|
| 3347 |
+
"learning_rate": 0.0006862619808306709,
|
| 3348 |
+
"loss": 6.164,
|
| 3349 |
+
"step": 257000
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 32.91,
|
| 3353 |
+
"learning_rate": 0.00068370607028754,
|
| 3354 |
+
"loss": 6.1592,
|
| 3355 |
+
"step": 257500
|
| 3356 |
+
},
|
| 3357 |
+
{
|
| 3358 |
+
"epoch": 32.97,
|
| 3359 |
+
"learning_rate": 0.0006811501597444089,
|
| 3360 |
+
"loss": 6.167,
|
| 3361 |
+
"step": 258000
|
| 3362 |
+
},
|
| 3363 |
+
{
|
| 3364 |
+
"epoch": 33.0,
|
| 3365 |
+
"eval_loss": 6.110122203826904,
|
| 3366 |
+
"eval_runtime": 6.8168,
|
| 3367 |
+
"eval_samples_per_second": 36.821,
|
| 3368 |
+
"eval_steps_per_second": 1.174,
|
| 3369 |
+
"step": 258225
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 33.04,
|
| 3373 |
+
"learning_rate": 0.000678594249201278,
|
| 3374 |
+
"loss": 6.1653,
|
| 3375 |
+
"step": 258500
|
| 3376 |
+
},
|
| 3377 |
+
{
|
| 3378 |
+
"epoch": 33.1,
|
| 3379 |
+
"learning_rate": 0.000676038338658147,
|
| 3380 |
+
"loss": 6.1616,
|
| 3381 |
+
"step": 259000
|
| 3382 |
+
},
|
| 3383 |
+
{
|
| 3384 |
+
"epoch": 33.16,
|
| 3385 |
+
"learning_rate": 0.0006734824281150161,
|
| 3386 |
+
"loss": 6.1651,
|
| 3387 |
+
"step": 259500
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 33.23,
|
| 3391 |
+
"learning_rate": 0.0006709265175718851,
|
| 3392 |
+
"loss": 6.1698,
|
| 3393 |
+
"step": 260000
|
| 3394 |
+
},
|
| 3395 |
+
{
|
| 3396 |
+
"epoch": 33.29,
|
| 3397 |
+
"learning_rate": 0.0006683706070287539,
|
| 3398 |
+
"loss": 6.1654,
|
| 3399 |
+
"step": 260500
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 33.35,
|
| 3403 |
+
"learning_rate": 0.000665814696485623,
|
| 3404 |
+
"loss": 6.1634,
|
| 3405 |
+
"step": 261000
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 33.42,
|
| 3409 |
+
"learning_rate": 0.000663258785942492,
|
| 3410 |
+
"loss": 6.1647,
|
| 3411 |
+
"step": 261500
|
| 3412 |
+
},
|
| 3413 |
+
{
|
| 3414 |
+
"epoch": 33.48,
|
| 3415 |
+
"learning_rate": 0.0006607028753993611,
|
| 3416 |
+
"loss": 6.1584,
|
| 3417 |
+
"step": 262000
|
| 3418 |
+
},
|
| 3419 |
+
{
|
| 3420 |
+
"epoch": 33.55,
|
| 3421 |
+
"learning_rate": 0.00065814696485623,
|
| 3422 |
+
"loss": 6.1657,
|
| 3423 |
+
"step": 262500
|
| 3424 |
+
},
|
| 3425 |
+
{
|
| 3426 |
+
"epoch": 33.61,
|
| 3427 |
+
"learning_rate": 0.000655591054313099,
|
| 3428 |
+
"loss": 6.1676,
|
| 3429 |
+
"step": 263000
|
| 3430 |
+
},
|
| 3431 |
+
{
|
| 3432 |
+
"epoch": 33.67,
|
| 3433 |
+
"learning_rate": 0.0006530351437699681,
|
| 3434 |
+
"loss": 6.1595,
|
| 3435 |
+
"step": 263500
|
| 3436 |
+
},
|
| 3437 |
+
{
|
| 3438 |
+
"epoch": 33.74,
|
| 3439 |
+
"learning_rate": 0.0006504792332268371,
|
| 3440 |
+
"loss": 6.1643,
|
| 3441 |
+
"step": 264000
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 33.8,
|
| 3445 |
+
"learning_rate": 0.0006479233226837062,
|
| 3446 |
+
"loss": 6.1737,
|
| 3447 |
+
"step": 264500
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 33.87,
|
| 3451 |
+
"learning_rate": 0.000645367412140575,
|
| 3452 |
+
"loss": 6.1674,
|
| 3453 |
+
"step": 265000
|
| 3454 |
+
},
|
| 3455 |
+
{
|
| 3456 |
+
"epoch": 33.93,
|
| 3457 |
+
"learning_rate": 0.0006428115015974441,
|
| 3458 |
+
"loss": 6.1719,
|
| 3459 |
+
"step": 265500
|
| 3460 |
+
},
|
| 3461 |
+
{
|
| 3462 |
+
"epoch": 33.99,
|
| 3463 |
+
"learning_rate": 0.0006402555910543131,
|
| 3464 |
+
"loss": 6.1695,
|
| 3465 |
+
"step": 266000
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 34.0,
|
| 3469 |
+
"eval_loss": 6.110330104827881,
|
| 3470 |
+
"eval_runtime": 6.7957,
|
| 3471 |
+
"eval_samples_per_second": 36.935,
|
| 3472 |
+
"eval_steps_per_second": 1.177,
|
| 3473 |
+
"step": 266050
|
| 3474 |
+
}
|
| 3475 |
+
],
|
| 3476 |
+
"logging_steps": 500,
|
| 3477 |
+
"max_steps": 391250,
|
| 3478 |
+
"num_input_tokens_seen": 0,
|
| 3479 |
+
"num_train_epochs": 50,
|
| 3480 |
+
"save_steps": 500,
|
| 3481 |
+
"total_flos": 4.575802027758605e+19,
|
| 3482 |
+
"train_batch_size": 32,
|
| 3483 |
+
"trial_name": null,
|
| 3484 |
+
"trial_params": null
|
| 3485 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d51e4081e7b4ed59fc54a424ce2eeae083ef712ec31040c6b906df43761a7f48
|
| 3 |
+
size 4155
|