Upload folder using huggingface_hub
Browse files- config.json +51 -0
- generation_config.json +6 -0
- latest +1 -0
- merges.txt +0 -0
- model.safetensors +3 -0
- rng_state_0.pth +3 -0
- rng_state_1.pth +3 -0
- rng_state_2.pth +3 -0
- rng_state_3.pth +3 -0
- scheduler.pt +3 -0
- special_tokens_map.json +24 -0
- tokenizer.json +0 -0
- tokenizer_config.json +22 -0
- trainer_state.json +2685 -0
- training_args.bin +3 -0
- vocab.json +0 -0
- zero_to_fp32.py +604 -0
config.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_function": "gelu_new",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"GPTNeoForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_dropout": 0.1,
|
| 7 |
+
"attention_layers": [
|
| 8 |
+
"global",
|
| 9 |
+
"global",
|
| 10 |
+
"global",
|
| 11 |
+
"global",
|
| 12 |
+
"global",
|
| 13 |
+
"global",
|
| 14 |
+
"global",
|
| 15 |
+
"global",
|
| 16 |
+
"global",
|
| 17 |
+
"global",
|
| 18 |
+
"global",
|
| 19 |
+
"global",
|
| 20 |
+
"global",
|
| 21 |
+
"global",
|
| 22 |
+
"global",
|
| 23 |
+
"global"
|
| 24 |
+
],
|
| 25 |
+
"attention_types": [
|
| 26 |
+
[
|
| 27 |
+
[
|
| 28 |
+
"global"
|
| 29 |
+
],
|
| 30 |
+
16
|
| 31 |
+
]
|
| 32 |
+
],
|
| 33 |
+
"bos_token_id": 50256,
|
| 34 |
+
"classifier_dropout": 0.1,
|
| 35 |
+
"embed_dropout": 0.1,
|
| 36 |
+
"eos_token_id": 50256,
|
| 37 |
+
"hidden_size": 1024,
|
| 38 |
+
"initializer_range": 0.02,
|
| 39 |
+
"intermediate_size": null,
|
| 40 |
+
"layer_norm_epsilon": 1e-05,
|
| 41 |
+
"max_position_embeddings": 4096,
|
| 42 |
+
"model_type": "gpt_neo",
|
| 43 |
+
"num_heads": 16,
|
| 44 |
+
"num_layers": 16,
|
| 45 |
+
"resid_dropout": 0.1,
|
| 46 |
+
"torch_dtype": "bfloat16",
|
| 47 |
+
"transformers_version": "4.45.1",
|
| 48 |
+
"use_cache": true,
|
| 49 |
+
"vocab_size": 50257,
|
| 50 |
+
"window_size": 4096
|
| 51 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 50256,
|
| 4 |
+
"eos_token_id": 50256,
|
| 5 |
+
"transformers_version": "4.45.1"
|
| 6 |
+
}
|
latest
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
global_step26040
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cbfe78ac81c86bb4077133cd4d368066e6c84dbe21af4f30c9b7b3fa69ada41e
|
| 3 |
+
size 617249488
|
rng_state_0.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1f2338dc132fa9996cc0309f715d9140f9a5c2d292a430782a88e5aad36e1f9f
|
| 3 |
+
size 15024
|
rng_state_1.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:401d5daae2d1f61c0e1a341c34cd3944037998dd967fe2db5ecd8e521dd41353
|
| 3 |
+
size 15024
|
rng_state_2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df0fd694cc901cbbd9ef57e85409afd692fb87bb1ca2a5ae34cfeacaa7be8286
|
| 3 |
+
size 15024
|
rng_state_3.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7f10a26b8fe2c553a7a20d6d27ec3fbd1d984b5c833e0a62d90a8e8658d10c1
|
| 3 |
+
size 15024
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:272cfba81198575e76e6b9965d7f6f8c7480ede87e02d759cade961d7ff227a2
|
| 3 |
+
size 1064
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<|endoftext|>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": true,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "<|endoftext|>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": true,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": "<|endoftext|>",
|
| 17 |
+
"unk_token": {
|
| 18 |
+
"content": "<|endoftext|>",
|
| 19 |
+
"lstrip": false,
|
| 20 |
+
"normalized": true,
|
| 21 |
+
"rstrip": false,
|
| 22 |
+
"single_word": false
|
| 23 |
+
}
|
| 24 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_prefix_space": false,
|
| 4 |
+
"added_tokens_decoder": {
|
| 5 |
+
"50256": {
|
| 6 |
+
"content": "<|endoftext|>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": true,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"special": true
|
| 12 |
+
}
|
| 13 |
+
},
|
| 14 |
+
"bos_token": "<|endoftext|>",
|
| 15 |
+
"clean_up_tokenization_spaces": false,
|
| 16 |
+
"eos_token": "<|endoftext|>",
|
| 17 |
+
"errors": "replace",
|
| 18 |
+
"model_max_length": 4096,
|
| 19 |
+
"pad_token": "<|endoftext|>",
|
| 20 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 21 |
+
"unk_token": "<|endoftext|>"
|
| 22 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2685 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.17754687368869781,
|
| 3 |
+
"best_model_checkpoint": "results/checkpoint-25000",
|
| 4 |
+
"epoch": 9.998720081914758,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 26040,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.038397542557276336,
|
| 13 |
+
"grad_norm": 1.0079567432403564,
|
| 14 |
+
"learning_rate": 9.999643338380885e-06,
|
| 15 |
+
"loss": 5.5723,
|
| 16 |
+
"step": 100
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.07679508511455267,
|
| 20 |
+
"grad_norm": 0.6461474299430847,
|
| 21 |
+
"learning_rate": 9.998558958654982e-06,
|
| 22 |
+
"loss": 2.2782,
|
| 23 |
+
"step": 200
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.115192627671829,
|
| 27 |
+
"grad_norm": 0.4909125566482544,
|
| 28 |
+
"learning_rate": 9.996746982275233e-06,
|
| 29 |
+
"loss": 1.8047,
|
| 30 |
+
"step": 300
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.15359017022910534,
|
| 34 |
+
"grad_norm": 0.47547289729118347,
|
| 35 |
+
"learning_rate": 9.994207672995245e-06,
|
| 36 |
+
"loss": 1.5821,
|
| 37 |
+
"step": 400
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.19198771278638166,
|
| 41 |
+
"grad_norm": 0.41358837485313416,
|
| 42 |
+
"learning_rate": 9.99094140044013e-06,
|
| 43 |
+
"loss": 1.4754,
|
| 44 |
+
"step": 500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.19198771278638166,
|
| 48 |
+
"eval_valid_loss": 1.4288749694824219,
|
| 49 |
+
"eval_valid_runtime": 4.7117,
|
| 50 |
+
"eval_valid_samples_per_second": 212.238,
|
| 51 |
+
"eval_valid_steps_per_second": 6.792,
|
| 52 |
+
"step": 500
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.19198771278638166,
|
| 56 |
+
"eval_valid_target_loss": 1.4590624570846558,
|
| 57 |
+
"eval_valid_target_runtime": 4.684,
|
| 58 |
+
"eval_valid_target_samples_per_second": 213.493,
|
| 59 |
+
"eval_valid_target_steps_per_second": 6.832,
|
| 60 |
+
"step": 500
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.230385255343658,
|
| 64 |
+
"grad_norm": 0.43229448795318604,
|
| 65 |
+
"learning_rate": 9.986948640052719e-06,
|
| 66 |
+
"loss": 1.4087,
|
| 67 |
+
"step": 600
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.26878279790093434,
|
| 71 |
+
"grad_norm": 0.528977632522583,
|
| 72 |
+
"learning_rate": 9.982229973024328e-06,
|
| 73 |
+
"loss": 1.3245,
|
| 74 |
+
"step": 700
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.3071803404582107,
|
| 78 |
+
"grad_norm": 0.5489594340324402,
|
| 79 |
+
"learning_rate": 9.976786086210186e-06,
|
| 80 |
+
"loss": 1.0455,
|
| 81 |
+
"step": 800
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.34557788301548703,
|
| 85 |
+
"grad_norm": 0.5119125843048096,
|
| 86 |
+
"learning_rate": 9.970617772029439e-06,
|
| 87 |
+
"loss": 0.7605,
|
| 88 |
+
"step": 900
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.3839754255727633,
|
| 92 |
+
"grad_norm": 0.5092576146125793,
|
| 93 |
+
"learning_rate": 9.963725928349814e-06,
|
| 94 |
+
"loss": 0.6005,
|
| 95 |
+
"step": 1000
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.3839754255727633,
|
| 99 |
+
"eval_valid_loss": 0.49165624380111694,
|
| 100 |
+
"eval_valid_runtime": 4.674,
|
| 101 |
+
"eval_valid_samples_per_second": 213.951,
|
| 102 |
+
"eval_valid_steps_per_second": 6.846,
|
| 103 |
+
"step": 1000
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.3839754255727633,
|
| 107 |
+
"eval_valid_target_loss": 0.5142187476158142,
|
| 108 |
+
"eval_valid_target_runtime": 4.6758,
|
| 109 |
+
"eval_valid_target_samples_per_second": 213.869,
|
| 110 |
+
"eval_valid_target_steps_per_second": 6.844,
|
| 111 |
+
"step": 1000
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.42237296813003966,
|
| 115 |
+
"grad_norm": 0.43798330426216125,
|
| 116 |
+
"learning_rate": 9.956111558356915e-06,
|
| 117 |
+
"loss": 0.4887,
|
| 118 |
+
"step": 1100
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.460770510687316,
|
| 122 |
+
"grad_norm": 0.3737218379974365,
|
| 123 |
+
"learning_rate": 9.947775770408207e-06,
|
| 124 |
+
"loss": 0.4307,
|
| 125 |
+
"step": 1200
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.49916805324459235,
|
| 129 |
+
"grad_norm": 0.4303857386112213,
|
| 130 |
+
"learning_rate": 9.938719777871674e-06,
|
| 131 |
+
"loss": 0.4027,
|
| 132 |
+
"step": 1300
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.5375655958018687,
|
| 136 |
+
"grad_norm": 0.3709202706813812,
|
| 137 |
+
"learning_rate": 9.92894489894921e-06,
|
| 138 |
+
"loss": 0.3799,
|
| 139 |
+
"step": 1400
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.575963138359145,
|
| 143 |
+
"grad_norm": 0.3918135464191437,
|
| 144 |
+
"learning_rate": 9.918452556484728e-06,
|
| 145 |
+
"loss": 0.3633,
|
| 146 |
+
"step": 1500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.575963138359145,
|
| 150 |
+
"eval_valid_loss": 0.33228906989097595,
|
| 151 |
+
"eval_valid_runtime": 4.7244,
|
| 152 |
+
"eval_valid_samples_per_second": 211.669,
|
| 153 |
+
"eval_valid_steps_per_second": 6.773,
|
| 154 |
+
"step": 1500
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.575963138359145,
|
| 158 |
+
"eval_valid_target_loss": 0.3428671956062317,
|
| 159 |
+
"eval_valid_target_runtime": 4.6595,
|
| 160 |
+
"eval_valid_target_samples_per_second": 214.617,
|
| 161 |
+
"eval_valid_target_steps_per_second": 6.868,
|
| 162 |
+
"step": 1500
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 0.6143606809164214,
|
| 166 |
+
"grad_norm": 0.3580816686153412,
|
| 167 |
+
"learning_rate": 9.907244277757053e-06,
|
| 168 |
+
"loss": 0.3565,
|
| 169 |
+
"step": 1600
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 0.6527582234736977,
|
| 173 |
+
"grad_norm": 0.34893009066581726,
|
| 174 |
+
"learning_rate": 9.895321694257617e-06,
|
| 175 |
+
"loss": 0.3443,
|
| 176 |
+
"step": 1700
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.6911557660309741,
|
| 180 |
+
"grad_norm": 0.3050221800804138,
|
| 181 |
+
"learning_rate": 9.882686541452967e-06,
|
| 182 |
+
"loss": 0.3339,
|
| 183 |
+
"step": 1800
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.7295533085882504,
|
| 187 |
+
"grad_norm": 0.3123306632041931,
|
| 188 |
+
"learning_rate": 9.869340658532151e-06,
|
| 189 |
+
"loss": 0.3278,
|
| 190 |
+
"step": 1900
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.7679508511455266,
|
| 194 |
+
"grad_norm": 0.31590646505355835,
|
| 195 |
+
"learning_rate": 9.85528598813901e-06,
|
| 196 |
+
"loss": 0.32,
|
| 197 |
+
"step": 2000
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.7679508511455266,
|
| 201 |
+
"eval_valid_loss": 0.29783594608306885,
|
| 202 |
+
"eval_valid_runtime": 4.6776,
|
| 203 |
+
"eval_valid_samples_per_second": 213.785,
|
| 204 |
+
"eval_valid_steps_per_second": 6.841,
|
| 205 |
+
"step": 2000
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.7679508511455266,
|
| 209 |
+
"eval_valid_target_loss": 0.3129218816757202,
|
| 210 |
+
"eval_valid_target_runtime": 4.6598,
|
| 211 |
+
"eval_valid_target_samples_per_second": 214.601,
|
| 212 |
+
"eval_valid_target_steps_per_second": 6.867,
|
| 213 |
+
"step": 2000
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.806348393702803,
|
| 217 |
+
"grad_norm": 0.2684693932533264,
|
| 218 |
+
"learning_rate": 9.840524576089392e-06,
|
| 219 |
+
"loss": 0.3194,
|
| 220 |
+
"step": 2100
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.8447459362600793,
|
| 224 |
+
"grad_norm": 0.30888476967811584,
|
| 225 |
+
"learning_rate": 9.82505857107337e-06,
|
| 226 |
+
"loss": 0.3108,
|
| 227 |
+
"step": 2200
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.8831434788173557,
|
| 231 |
+
"grad_norm": 0.2777215242385864,
|
| 232 |
+
"learning_rate": 9.808890224342476e-06,
|
| 233 |
+
"loss": 0.3105,
|
| 234 |
+
"step": 2300
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.921541021374632,
|
| 238 |
+
"grad_norm": 0.30601397156715393,
|
| 239 |
+
"learning_rate": 9.792021889381995e-06,
|
| 240 |
+
"loss": 0.3055,
|
| 241 |
+
"step": 2400
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.9599385639319084,
|
| 245 |
+
"grad_norm": 0.2972748875617981,
|
| 246 |
+
"learning_rate": 9.774456021568404e-06,
|
| 247 |
+
"loss": 0.3008,
|
| 248 |
+
"step": 2500
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.9599385639319084,
|
| 252 |
+
"eval_valid_loss": 0.27842968702316284,
|
| 253 |
+
"eval_valid_runtime": 4.68,
|
| 254 |
+
"eval_valid_samples_per_second": 213.675,
|
| 255 |
+
"eval_valid_steps_per_second": 6.838,
|
| 256 |
+
"step": 2500
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 0.9599385639319084,
|
| 260 |
+
"eval_valid_target_loss": 0.2943359315395355,
|
| 261 |
+
"eval_valid_target_runtime": 4.6764,
|
| 262 |
+
"eval_valid_target_samples_per_second": 213.838,
|
| 263 |
+
"eval_valid_target_steps_per_second": 6.843,
|
| 264 |
+
"step": 2500
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.9983361064891847,
|
| 268 |
+
"grad_norm": 0.2891447842121124,
|
| 269 |
+
"learning_rate": 9.756195177811953e-06,
|
| 270 |
+
"loss": 0.2969,
|
| 271 |
+
"step": 2600
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 1.036733649046461,
|
| 275 |
+
"grad_norm": 0.30049994587898254,
|
| 276 |
+
"learning_rate": 9.737242016184486e-06,
|
| 277 |
+
"loss": 0.2913,
|
| 278 |
+
"step": 2700
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.0751311916037374,
|
| 282 |
+
"grad_norm": 0.25728341937065125,
|
| 283 |
+
"learning_rate": 9.717599295532518e-06,
|
| 284 |
+
"loss": 0.2911,
|
| 285 |
+
"step": 2800
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 1.1135287341610136,
|
| 289 |
+
"grad_norm": 0.31619054079055786,
|
| 290 |
+
"learning_rate": 9.697269875075667e-06,
|
| 291 |
+
"loss": 0.2879,
|
| 292 |
+
"step": 2900
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 1.15192627671829,
|
| 296 |
+
"grad_norm": 0.3005208671092987,
|
| 297 |
+
"learning_rate": 9.676256713990448e-06,
|
| 298 |
+
"loss": 0.2839,
|
| 299 |
+
"step": 3000
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.15192627671829,
|
| 303 |
+
"eval_valid_loss": 0.2648593783378601,
|
| 304 |
+
"eval_valid_runtime": 4.6888,
|
| 305 |
+
"eval_valid_samples_per_second": 213.274,
|
| 306 |
+
"eval_valid_steps_per_second": 6.825,
|
| 307 |
+
"step": 3000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 1.15192627671829,
|
| 311 |
+
"eval_valid_target_loss": 0.2817968726158142,
|
| 312 |
+
"eval_valid_target_runtime": 4.6695,
|
| 313 |
+
"eval_valid_target_samples_per_second": 214.155,
|
| 314 |
+
"eval_valid_target_steps_per_second": 6.853,
|
| 315 |
+
"step": 3000
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 1.1903238192755663,
|
| 319 |
+
"grad_norm": 0.24846772849559784,
|
| 320 |
+
"learning_rate": 9.654562870979545e-06,
|
| 321 |
+
"loss": 0.2803,
|
| 322 |
+
"step": 3100
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 1.2287213618328428,
|
| 326 |
+
"grad_norm": 0.23501113057136536,
|
| 327 |
+
"learning_rate": 9.632191503826574e-06,
|
| 328 |
+
"loss": 0.278,
|
| 329 |
+
"step": 3200
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.267118904390119,
|
| 333 |
+
"grad_norm": 0.27793240547180176,
|
| 334 |
+
"learning_rate": 9.609145868936434e-06,
|
| 335 |
+
"loss": 0.2776,
|
| 336 |
+
"step": 3300
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 1.3055164469473954,
|
| 340 |
+
"grad_norm": 0.2599338889122009,
|
| 341 |
+
"learning_rate": 9.5854293208613e-06,
|
| 342 |
+
"loss": 0.275,
|
| 343 |
+
"step": 3400
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 1.3439139895046717,
|
| 347 |
+
"grad_norm": 0.2431340515613556,
|
| 348 |
+
"learning_rate": 9.561045311812335e-06,
|
| 349 |
+
"loss": 0.2722,
|
| 350 |
+
"step": 3500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.3439139895046717,
|
| 354 |
+
"eval_valid_loss": 0.2545468807220459,
|
| 355 |
+
"eval_valid_runtime": 4.7131,
|
| 356 |
+
"eval_valid_samples_per_second": 212.177,
|
| 357 |
+
"eval_valid_steps_per_second": 6.79,
|
| 358 |
+
"step": 3500
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 1.3439139895046717,
|
| 362 |
+
"eval_valid_target_loss": 0.27326563000679016,
|
| 363 |
+
"eval_valid_target_runtime": 4.6635,
|
| 364 |
+
"eval_valid_target_samples_per_second": 214.433,
|
| 365 |
+
"eval_valid_target_steps_per_second": 6.862,
|
| 366 |
+
"step": 3500
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 1.382311532061948,
|
| 370 |
+
"grad_norm": 0.28658226132392883,
|
| 371 |
+
"learning_rate": 9.535997391157174e-06,
|
| 372 |
+
"loss": 0.2693,
|
| 373 |
+
"step": 3600
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 1.4207090746192244,
|
| 377 |
+
"grad_norm": 0.26118528842926025,
|
| 378 |
+
"learning_rate": 9.510289204903273e-06,
|
| 379 |
+
"loss": 0.2667,
|
| 380 |
+
"step": 3700
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 1.4591066171765008,
|
| 384 |
+
"grad_norm": 0.2761940062046051,
|
| 385 |
+
"learning_rate": 9.483924495167204e-06,
|
| 386 |
+
"loss": 0.2654,
|
| 387 |
+
"step": 3800
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.497504159733777,
|
| 391 |
+
"grad_norm": 0.2712952792644501,
|
| 392 |
+
"learning_rate": 9.456907099629933e-06,
|
| 393 |
+
"loss": 0.2642,
|
| 394 |
+
"step": 3900
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 1.5359017022910533,
|
| 398 |
+
"grad_norm": 0.23100949823856354,
|
| 399 |
+
"learning_rate": 9.429240950978212e-06,
|
| 400 |
+
"loss": 0.2622,
|
| 401 |
+
"step": 4000
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.5359017022910533,
|
| 405 |
+
"eval_valid_loss": 0.24485936760902405,
|
| 406 |
+
"eval_valid_runtime": 4.6751,
|
| 407 |
+
"eval_valid_samples_per_second": 213.9,
|
| 408 |
+
"eval_valid_steps_per_second": 6.845,
|
| 409 |
+
"step": 4000
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 1.5359017022910533,
|
| 413 |
+
"eval_valid_target_loss": 0.2641640603542328,
|
| 414 |
+
"eval_valid_target_runtime": 4.6656,
|
| 415 |
+
"eval_valid_target_samples_per_second": 214.335,
|
| 416 |
+
"eval_valid_target_steps_per_second": 6.859,
|
| 417 |
+
"step": 4000
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 1.5742992448483297,
|
| 421 |
+
"grad_norm": 0.2676081359386444,
|
| 422 |
+
"learning_rate": 9.400930076332126e-06,
|
| 423 |
+
"loss": 0.2602,
|
| 424 |
+
"step": 4100
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 1.6126967874056062,
|
| 428 |
+
"grad_norm": 0.24242335557937622,
|
| 429 |
+
"learning_rate": 9.371978596658904e-06,
|
| 430 |
+
"loss": 0.2573,
|
| 431 |
+
"step": 4200
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.6510943299628824,
|
| 435 |
+
"grad_norm": 0.27868130803108215,
|
| 436 |
+
"learning_rate": 9.342390726173065e-06,
|
| 437 |
+
"loss": 0.2574,
|
| 438 |
+
"step": 4300
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 1.6894918725201586,
|
| 442 |
+
"grad_norm": 0.2644180655479431,
|
| 443 |
+
"learning_rate": 9.31217077172299e-06,
|
| 444 |
+
"loss": 0.255,
|
| 445 |
+
"step": 4400
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 1.727889415077435,
|
| 449 |
+
"grad_norm": 0.2352069914340973,
|
| 450 |
+
"learning_rate": 9.281323132164013e-06,
|
| 451 |
+
"loss": 0.2538,
|
| 452 |
+
"step": 4500
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.727889415077435,
|
| 456 |
+
"eval_valid_loss": 0.2354765683412552,
|
| 457 |
+
"eval_valid_runtime": 4.7068,
|
| 458 |
+
"eval_valid_samples_per_second": 212.461,
|
| 459 |
+
"eval_valid_steps_per_second": 6.799,
|
| 460 |
+
"step": 4500
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 1.727889415077435,
|
| 464 |
+
"eval_valid_target_loss": 0.25502344965934753,
|
| 465 |
+
"eval_valid_target_runtime": 4.6612,
|
| 466 |
+
"eval_valid_target_samples_per_second": 214.536,
|
| 467 |
+
"eval_valid_target_steps_per_second": 6.865,
|
| 468 |
+
"step": 4500
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 1.7662869576347113,
|
| 472 |
+
"grad_norm": 0.28041261434555054,
|
| 473 |
+
"learning_rate": 9.249852297718116e-06,
|
| 474 |
+
"loss": 0.2507,
|
| 475 |
+
"step": 4600
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 1.8046845001919878,
|
| 479 |
+
"grad_norm": 0.2735440135002136,
|
| 480 |
+
"learning_rate": 9.217762849320334e-06,
|
| 481 |
+
"loss": 0.2496,
|
| 482 |
+
"step": 4700
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.843082042749264,
|
| 486 |
+
"grad_norm": 0.26316097378730774,
|
| 487 |
+
"learning_rate": 9.185059457951933e-06,
|
| 488 |
+
"loss": 0.2479,
|
| 489 |
+
"step": 4800
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.8814795853065402,
|
| 493 |
+
"grad_norm": 0.23891638219356537,
|
| 494 |
+
"learning_rate": 9.151746883960512e-06,
|
| 495 |
+
"loss": 0.2457,
|
| 496 |
+
"step": 4900
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 1.9198771278638167,
|
| 500 |
+
"grad_norm": 0.22432874143123627,
|
| 501 |
+
"learning_rate": 9.117829976367072e-06,
|
| 502 |
+
"loss": 0.2446,
|
| 503 |
+
"step": 5000
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 1.9198771278638167,
|
| 507 |
+
"eval_valid_loss": 0.2283046841621399,
|
| 508 |
+
"eval_valid_runtime": 4.6829,
|
| 509 |
+
"eval_valid_samples_per_second": 213.544,
|
| 510 |
+
"eval_valid_steps_per_second": 6.833,
|
| 511 |
+
"step": 5000
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 1.9198771278638167,
|
| 515 |
+
"eval_valid_target_loss": 0.24809375405311584,
|
| 516 |
+
"eval_valid_target_runtime": 4.6694,
|
| 517 |
+
"eval_valid_target_samples_per_second": 214.162,
|
| 518 |
+
"eval_valid_target_steps_per_second": 6.853,
|
| 519 |
+
"step": 5000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 1.9582746704210932,
|
| 523 |
+
"grad_norm": 0.27488961815834045,
|
| 524 |
+
"learning_rate": 9.08331367216019e-06,
|
| 525 |
+
"loss": 0.2434,
|
| 526 |
+
"step": 5100
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 1.9966722129783694,
|
| 530 |
+
"grad_norm": 0.2284267097711563,
|
| 531 |
+
"learning_rate": 9.048202995577383e-06,
|
| 532 |
+
"loss": 0.24,
|
| 533 |
+
"step": 5200
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 2.0350697555356456,
|
| 537 |
+
"grad_norm": 0.2710357904434204,
|
| 538 |
+
"learning_rate": 9.012503057373769e-06,
|
| 539 |
+
"loss": 0.2399,
|
| 540 |
+
"step": 5300
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 2.073467298092922,
|
| 544 |
+
"grad_norm": 0.24398750066757202,
|
| 545 |
+
"learning_rate": 8.976219054078147e-06,
|
| 546 |
+
"loss": 0.2391,
|
| 547 |
+
"step": 5400
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 2.1118648406501985,
|
| 551 |
+
"grad_norm": 0.24732039868831635,
|
| 552 |
+
"learning_rate": 8.939356267236582e-06,
|
| 553 |
+
"loss": 0.2374,
|
| 554 |
+
"step": 5500
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 2.1118648406501985,
|
| 558 |
+
"eval_valid_loss": 0.22253906726837158,
|
| 559 |
+
"eval_valid_runtime": 4.6969,
|
| 560 |
+
"eval_valid_samples_per_second": 212.904,
|
| 561 |
+
"eval_valid_steps_per_second": 6.813,
|
| 562 |
+
"step": 5500
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 2.1118648406501985,
|
| 566 |
+
"eval_valid_target_loss": 0.24240624904632568,
|
| 567 |
+
"eval_valid_target_runtime": 4.6761,
|
| 568 |
+
"eval_valid_target_samples_per_second": 213.853,
|
| 569 |
+
"eval_valid_target_steps_per_second": 6.843,
|
| 570 |
+
"step": 5500
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 2.1502623832074748,
|
| 574 |
+
"grad_norm": 0.23949123919010162,
|
| 575 |
+
"learning_rate": 8.901920062643607e-06,
|
| 576 |
+
"loss": 0.2368,
|
| 577 |
+
"step": 5600
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 2.188659925764751,
|
| 581 |
+
"grad_norm": 0.26010605692863464,
|
| 582 |
+
"learning_rate": 8.863915889561188e-06,
|
| 583 |
+
"loss": 0.2351,
|
| 584 |
+
"step": 5700
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.2270574683220272,
|
| 588 |
+
"grad_norm": 0.2524034380912781,
|
| 589 |
+
"learning_rate": 8.825349279925506e-06,
|
| 590 |
+
"loss": 0.2333,
|
| 591 |
+
"step": 5800
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 2.265455010879304,
|
| 595 |
+
"grad_norm": 0.24745632708072662,
|
| 596 |
+
"learning_rate": 8.78622584754173e-06,
|
| 597 |
+
"loss": 0.2323,
|
| 598 |
+
"step": 5900
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 2.30385255343658,
|
| 602 |
+
"grad_norm": 0.2586907148361206,
|
| 603 |
+
"learning_rate": 8.746551287266863e-06,
|
| 604 |
+
"loss": 0.2312,
|
| 605 |
+
"step": 6000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 2.30385255343658,
|
| 609 |
+
"eval_valid_loss": 0.216859370470047,
|
| 610 |
+
"eval_valid_runtime": 4.6709,
|
| 611 |
+
"eval_valid_samples_per_second": 214.092,
|
| 612 |
+
"eval_valid_steps_per_second": 6.851,
|
| 613 |
+
"step": 6000
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 2.30385255343658,
|
| 617 |
+
"eval_valid_target_loss": 0.23771093785762787,
|
| 618 |
+
"eval_valid_target_runtime": 4.6848,
|
| 619 |
+
"eval_valid_target_samples_per_second": 213.455,
|
| 620 |
+
"eval_valid_target_steps_per_second": 6.831,
|
| 621 |
+
"step": 6000
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 2.3422500959938564,
|
| 625 |
+
"grad_norm": 0.24499697983264923,
|
| 626 |
+
"learning_rate": 8.706331374180792e-06,
|
| 627 |
+
"loss": 0.2301,
|
| 628 |
+
"step": 6100
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 2.3806476385511326,
|
| 632 |
+
"grad_norm": 0.24237163364887238,
|
| 633 |
+
"learning_rate": 8.665571962745655e-06,
|
| 634 |
+
"loss": 0.2304,
|
| 635 |
+
"step": 6200
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 2.419045181108409,
|
| 639 |
+
"grad_norm": 0.27395910024642944,
|
| 640 |
+
"learning_rate": 8.624278985953665e-06,
|
| 641 |
+
"loss": 0.2287,
|
| 642 |
+
"step": 6300
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 2.4574427236656855,
|
| 646 |
+
"grad_norm": 0.2500033378601074,
|
| 647 |
+
"learning_rate": 8.582458454463493e-06,
|
| 648 |
+
"loss": 0.2279,
|
| 649 |
+
"step": 6400
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 2.4958402662229617,
|
| 653 |
+
"grad_norm": 0.2605977952480316,
|
| 654 |
+
"learning_rate": 8.540116455725346e-06,
|
| 655 |
+
"loss": 0.2277,
|
| 656 |
+
"step": 6500
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 2.4958402662229617,
|
| 660 |
+
"eval_valid_loss": 0.21196874976158142,
|
| 661 |
+
"eval_valid_runtime": 4.6941,
|
| 662 |
+
"eval_valid_samples_per_second": 213.035,
|
| 663 |
+
"eval_valid_steps_per_second": 6.817,
|
| 664 |
+
"step": 6500
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 2.4958402662229617,
|
| 668 |
+
"eval_valid_target_loss": 0.23328906297683716,
|
| 669 |
+
"eval_valid_target_runtime": 4.6792,
|
| 670 |
+
"eval_valid_target_samples_per_second": 213.712,
|
| 671 |
+
"eval_valid_target_steps_per_second": 6.839,
|
| 672 |
+
"step": 6500
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 2.534237808780238,
|
| 676 |
+
"grad_norm": 0.2220095992088318,
|
| 677 |
+
"learning_rate": 8.497259153094875e-06,
|
| 678 |
+
"loss": 0.2254,
|
| 679 |
+
"step": 6600
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 2.5726353513375146,
|
| 683 |
+
"grad_norm": 0.24707047641277313,
|
| 684 |
+
"learning_rate": 8.453892784936022e-06,
|
| 685 |
+
"loss": 0.2239,
|
| 686 |
+
"step": 6700
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 2.611032893894791,
|
| 690 |
+
"grad_norm": 0.23103290796279907,
|
| 691 |
+
"learning_rate": 8.41002366371297e-06,
|
| 692 |
+
"loss": 0.224,
|
| 693 |
+
"step": 6800
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 2.649430436452067,
|
| 697 |
+
"grad_norm": 0.2249547839164734,
|
| 698 |
+
"learning_rate": 8.36565817507127e-06,
|
| 699 |
+
"loss": 0.2227,
|
| 700 |
+
"step": 6900
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 2.6878279790093433,
|
| 704 |
+
"grad_norm": 0.24457262456417084,
|
| 705 |
+
"learning_rate": 8.32080277690836e-06,
|
| 706 |
+
"loss": 0.2209,
|
| 707 |
+
"step": 7000
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 2.6878279790093433,
|
| 711 |
+
"eval_valid_loss": 0.20793749392032623,
|
| 712 |
+
"eval_valid_runtime": 4.6727,
|
| 713 |
+
"eval_valid_samples_per_second": 214.01,
|
| 714 |
+
"eval_valid_steps_per_second": 6.848,
|
| 715 |
+
"step": 7000
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 2.6878279790093433,
|
| 719 |
+
"eval_valid_target_loss": 0.22950781881809235,
|
| 720 |
+
"eval_valid_target_runtime": 4.6848,
|
| 721 |
+
"eval_valid_target_samples_per_second": 213.456,
|
| 722 |
+
"eval_valid_target_steps_per_second": 6.831,
|
| 723 |
+
"step": 7000
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 2.7262255215666196,
|
| 727 |
+
"grad_norm": 0.23176012933254242,
|
| 728 |
+
"learning_rate": 8.275463998433537e-06,
|
| 729 |
+
"loss": 0.2206,
|
| 730 |
+
"step": 7100
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 2.764623064123896,
|
| 734 |
+
"grad_norm": 0.21723733842372894,
|
| 735 |
+
"learning_rate": 8.229648439217552e-06,
|
| 736 |
+
"loss": 0.2203,
|
| 737 |
+
"step": 7200
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 2.8030206066811725,
|
| 741 |
+
"grad_norm": 0.2428179383277893,
|
| 742 |
+
"learning_rate": 8.183362768231971e-06,
|
| 743 |
+
"loss": 0.2192,
|
| 744 |
+
"step": 7300
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 2.8414181492384487,
|
| 748 |
+
"grad_norm": 0.2162482738494873,
|
| 749 |
+
"learning_rate": 8.136613722878437e-06,
|
| 750 |
+
"loss": 0.2183,
|
| 751 |
+
"step": 7400
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 2.879815691795725,
|
| 755 |
+
"grad_norm": 0.22231200337409973,
|
| 756 |
+
"learning_rate": 8.08940810800796e-06,
|
| 757 |
+
"loss": 0.2177,
|
| 758 |
+
"step": 7500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 2.879815691795725,
|
| 762 |
+
"eval_valid_loss": 0.20469531416893005,
|
| 763 |
+
"eval_valid_runtime": 4.6819,
|
| 764 |
+
"eval_valid_samples_per_second": 213.587,
|
| 765 |
+
"eval_valid_steps_per_second": 6.835,
|
| 766 |
+
"step": 7500
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 2.879815691795725,
|
| 770 |
+
"eval_valid_target_loss": 0.2264062464237213,
|
| 771 |
+
"eval_valid_target_runtime": 4.6647,
|
| 772 |
+
"eval_valid_target_samples_per_second": 214.376,
|
| 773 |
+
"eval_valid_target_steps_per_second": 6.86,
|
| 774 |
+
"step": 7500
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 2.9182132343530016,
|
| 778 |
+
"grad_norm": 0.2663327157497406,
|
| 779 |
+
"learning_rate": 8.041752794930389e-06,
|
| 780 |
+
"loss": 0.2172,
|
| 781 |
+
"step": 7600
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 2.956610776910278,
|
| 785 |
+
"grad_norm": 0.2545444369316101,
|
| 786 |
+
"learning_rate": 7.993654720414227e-06,
|
| 787 |
+
"loss": 0.216,
|
| 788 |
+
"step": 7700
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 2.995008319467554,
|
| 792 |
+
"grad_norm": 0.2252371460199356,
|
| 793 |
+
"learning_rate": 7.9451208856769e-06,
|
| 794 |
+
"loss": 0.2154,
|
| 795 |
+
"step": 7800
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 3.0334058620248303,
|
| 799 |
+
"grad_norm": 0.2507840394973755,
|
| 800 |
+
"learning_rate": 7.896158355365643e-06,
|
| 801 |
+
"loss": 0.2151,
|
| 802 |
+
"step": 7900
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 3.0718034045821065,
|
| 806 |
+
"grad_norm": 0.22570189833641052,
|
| 807 |
+
"learning_rate": 7.846774256529178e-06,
|
| 808 |
+
"loss": 0.2131,
|
| 809 |
+
"step": 8000
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 3.0718034045821065,
|
| 813 |
+
"eval_valid_loss": 0.2014453113079071,
|
| 814 |
+
"eval_valid_runtime": 4.6924,
|
| 815 |
+
"eval_valid_samples_per_second": 213.111,
|
| 816 |
+
"eval_valid_steps_per_second": 6.82,
|
| 817 |
+
"step": 8000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 3.0718034045821065,
|
| 821 |
+
"eval_valid_target_loss": 0.22346094250679016,
|
| 822 |
+
"eval_valid_target_runtime": 4.6655,
|
| 823 |
+
"eval_valid_target_samples_per_second": 214.339,
|
| 824 |
+
"eval_valid_target_steps_per_second": 6.859,
|
| 825 |
+
"step": 8000
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 3.110200947139383,
|
| 829 |
+
"grad_norm": 0.24750301241874695,
|
| 830 |
+
"learning_rate": 7.796975777580276e-06,
|
| 831 |
+
"loss": 0.2133,
|
| 832 |
+
"step": 8100
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 3.1485984896966595,
|
| 836 |
+
"grad_norm": 0.2118765264749527,
|
| 837 |
+
"learning_rate": 7.746770167249413e-06,
|
| 838 |
+
"loss": 0.2124,
|
| 839 |
+
"step": 8200
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 3.1869960322539357,
|
| 843 |
+
"grad_norm": 0.22295965254306793,
|
| 844 |
+
"learning_rate": 7.696164733529628e-06,
|
| 845 |
+
"loss": 0.2123,
|
| 846 |
+
"step": 8300
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 3.225393574811212,
|
| 850 |
+
"grad_norm": 0.2226712554693222,
|
| 851 |
+
"learning_rate": 7.645166842612766e-06,
|
| 852 |
+
"loss": 0.2115,
|
| 853 |
+
"step": 8400
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 3.2637911173684886,
|
| 857 |
+
"grad_norm": 0.22712872922420502,
|
| 858 |
+
"learning_rate": 7.593783917817248e-06,
|
| 859 |
+
"loss": 0.211,
|
| 860 |
+
"step": 8500
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 3.2637911173684886,
|
| 864 |
+
"eval_valid_loss": 0.19893750548362732,
|
| 865 |
+
"eval_valid_runtime": 4.6876,
|
| 866 |
+
"eval_valid_samples_per_second": 213.327,
|
| 867 |
+
"eval_valid_steps_per_second": 6.826,
|
| 868 |
+
"step": 8500
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 3.2637911173684886,
|
| 872 |
+
"eval_valid_target_loss": 0.22138281166553497,
|
| 873 |
+
"eval_valid_target_runtime": 4.6684,
|
| 874 |
+
"eval_valid_target_samples_per_second": 214.206,
|
| 875 |
+
"eval_valid_target_steps_per_second": 6.855,
|
| 876 |
+
"step": 8500
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 3.302188659925765,
|
| 880 |
+
"grad_norm": 0.20663662254810333,
|
| 881 |
+
"learning_rate": 7.5420234385075155e-06,
|
| 882 |
+
"loss": 0.211,
|
| 883 |
+
"step": 8600
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 3.340586202483041,
|
| 887 |
+
"grad_norm": 0.24639233946800232,
|
| 888 |
+
"learning_rate": 7.489892939005333e-06,
|
| 889 |
+
"loss": 0.2099,
|
| 890 |
+
"step": 8700
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 3.3789837450403173,
|
| 894 |
+
"grad_norm": 0.21435491740703583,
|
| 895 |
+
"learning_rate": 7.437400007493079e-06,
|
| 896 |
+
"loss": 0.209,
|
| 897 |
+
"step": 8800
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 3.4173812875975935,
|
| 901 |
+
"grad_norm": 0.21131959557533264,
|
| 902 |
+
"learning_rate": 7.384552284909195e-06,
|
| 903 |
+
"loss": 0.2081,
|
| 904 |
+
"step": 8900
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 3.45577883015487,
|
| 908 |
+
"grad_norm": 0.2295517921447754,
|
| 909 |
+
"learning_rate": 7.3313574638359734e-06,
|
| 910 |
+
"loss": 0.2084,
|
| 911 |
+
"step": 9000
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 3.45577883015487,
|
| 915 |
+
"eval_valid_loss": 0.19658593833446503,
|
| 916 |
+
"eval_valid_runtime": 4.6935,
|
| 917 |
+
"eval_valid_samples_per_second": 213.059,
|
| 918 |
+
"eval_valid_steps_per_second": 6.818,
|
| 919 |
+
"step": 9000
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 3.45577883015487,
|
| 923 |
+
"eval_valid_target_loss": 0.2188750058412552,
|
| 924 |
+
"eval_valid_target_runtime": 4.6686,
|
| 925 |
+
"eval_valid_target_samples_per_second": 214.199,
|
| 926 |
+
"eval_valid_target_steps_per_second": 6.854,
|
| 927 |
+
"step": 9000
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 3.4941763727121464,
|
| 931 |
+
"grad_norm": 0.2244088351726532,
|
| 932 |
+
"learning_rate": 7.277823287379801e-06,
|
| 933 |
+
"loss": 0.2084,
|
| 934 |
+
"step": 9100
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 3.5325739152694227,
|
| 938 |
+
"grad_norm": 0.2267696112394333,
|
| 939 |
+
"learning_rate": 7.2239575480440774e-06,
|
| 940 |
+
"loss": 0.2085,
|
| 941 |
+
"step": 9200
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 3.5709714578266993,
|
| 945 |
+
"grad_norm": 0.20846766233444214,
|
| 946 |
+
"learning_rate": 7.169768086594913e-06,
|
| 947 |
+
"loss": 0.2063,
|
| 948 |
+
"step": 9300
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 3.6093690003839756,
|
| 952 |
+
"grad_norm": 0.23632733523845673,
|
| 953 |
+
"learning_rate": 7.115262790919827e-06,
|
| 954 |
+
"loss": 0.2068,
|
| 955 |
+
"step": 9400
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 3.647766542941252,
|
| 959 |
+
"grad_norm": 0.20877471566200256,
|
| 960 |
+
"learning_rate": 7.060449594879573e-06,
|
| 961 |
+
"loss": 0.2059,
|
| 962 |
+
"step": 9500
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 3.647766542941252,
|
| 966 |
+
"eval_valid_loss": 0.19441406428813934,
|
| 967 |
+
"eval_valid_runtime": 4.6671,
|
| 968 |
+
"eval_valid_samples_per_second": 214.264,
|
| 969 |
+
"eval_valid_steps_per_second": 6.856,
|
| 970 |
+
"step": 9500
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 3.647766542941252,
|
| 974 |
+
"eval_valid_target_loss": 0.21704687178134918,
|
| 975 |
+
"eval_valid_target_runtime": 4.6648,
|
| 976 |
+
"eval_valid_target_samples_per_second": 214.371,
|
| 977 |
+
"eval_valid_target_steps_per_second": 6.86,
|
| 978 |
+
"step": 9500
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 3.686164085498528,
|
| 982 |
+
"grad_norm": 0.20587915182113647,
|
| 983 |
+
"learning_rate": 7.0053364771532805e-06,
|
| 984 |
+
"loss": 0.2058,
|
| 985 |
+
"step": 9600
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 3.7245616280558043,
|
| 989 |
+
"grad_norm": 0.208708256483078,
|
| 990 |
+
"learning_rate": 6.949931460077058e-06,
|
| 991 |
+
"loss": 0.2052,
|
| 992 |
+
"step": 9700
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 3.7629591706130805,
|
| 996 |
+
"grad_norm": 0.21517980098724365,
|
| 997 |
+
"learning_rate": 6.894242608476263e-06,
|
| 998 |
+
"loss": 0.2049,
|
| 999 |
+
"step": 9800
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 3.801356713170357,
|
| 1003 |
+
"grad_norm": 0.22570070624351501,
|
| 1004 |
+
"learning_rate": 6.8382780284915685e-06,
|
| 1005 |
+
"loss": 0.2047,
|
| 1006 |
+
"step": 9900
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 3.8397542557276334,
|
| 1010 |
+
"grad_norm": 0.22346258163452148,
|
| 1011 |
+
"learning_rate": 6.782045866399023e-06,
|
| 1012 |
+
"loss": 0.2037,
|
| 1013 |
+
"step": 10000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 3.8397542557276334,
|
| 1017 |
+
"eval_valid_loss": 0.1928359419107437,
|
| 1018 |
+
"eval_valid_runtime": 4.6748,
|
| 1019 |
+
"eval_valid_samples_per_second": 213.912,
|
| 1020 |
+
"eval_valid_steps_per_second": 6.845,
|
| 1021 |
+
"step": 10000
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 3.8397542557276334,
|
| 1025 |
+
"eval_valid_target_loss": 0.21531249582767487,
|
| 1026 |
+
"eval_valid_target_runtime": 4.6773,
|
| 1027 |
+
"eval_valid_target_samples_per_second": 213.8,
|
| 1028 |
+
"eval_valid_target_steps_per_second": 6.842,
|
| 1029 |
+
"step": 10000
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 3.8781517982849096,
|
| 1033 |
+
"grad_norm": 0.2544507086277008,
|
| 1034 |
+
"learning_rate": 6.725554307424274e-06,
|
| 1035 |
+
"loss": 0.2036,
|
| 1036 |
+
"step": 10100
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 3.9165493408421863,
|
| 1040 |
+
"grad_norm": 0.27723318338394165,
|
| 1041 |
+
"learning_rate": 6.668811574551106e-06,
|
| 1042 |
+
"loss": 0.2039,
|
| 1043 |
+
"step": 10200
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 3.9549468833994625,
|
| 1047 |
+
"grad_norm": 0.22496485710144043,
|
| 1048 |
+
"learning_rate": 6.6118259273245065e-06,
|
| 1049 |
+
"loss": 0.2032,
|
| 1050 |
+
"step": 10300
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 3.9933444259567388,
|
| 1054 |
+
"grad_norm": 0.22093619406223297,
|
| 1055 |
+
"learning_rate": 6.55460566064838e-06,
|
| 1056 |
+
"loss": 0.2027,
|
| 1057 |
+
"step": 10400
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 4.031741968514015,
|
| 1061 |
+
"grad_norm": 0.2137976437807083,
|
| 1062 |
+
"learning_rate": 6.497159103578143e-06,
|
| 1063 |
+
"loss": 0.2016,
|
| 1064 |
+
"step": 10500
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 4.031741968514015,
|
| 1068 |
+
"eval_valid_loss": 0.19111718237400055,
|
| 1069 |
+
"eval_valid_runtime": 4.6833,
|
| 1070 |
+
"eval_valid_samples_per_second": 213.523,
|
| 1071 |
+
"eval_valid_steps_per_second": 6.833,
|
| 1072 |
+
"step": 10500
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 4.031741968514015,
|
| 1076 |
+
"eval_valid_target_loss": 0.2142656296491623,
|
| 1077 |
+
"eval_valid_target_runtime": 4.6587,
|
| 1078 |
+
"eval_valid_target_samples_per_second": 214.65,
|
| 1079 |
+
"eval_valid_target_steps_per_second": 6.869,
|
| 1080 |
+
"step": 10500
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 4.070139511071291,
|
| 1084 |
+
"grad_norm": 0.20360158383846283,
|
| 1085 |
+
"learning_rate": 6.439494618108332e-06,
|
| 1086 |
+
"loss": 0.2013,
|
| 1087 |
+
"step": 10600
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 4.1085370536285675,
|
| 1091 |
+
"grad_norm": 0.21878282725811005,
|
| 1092 |
+
"learning_rate": 6.38162059795542e-06,
|
| 1093 |
+
"loss": 0.2006,
|
| 1094 |
+
"step": 10700
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 4.146934596185844,
|
| 1098 |
+
"grad_norm": 0.2319776862859726,
|
| 1099 |
+
"learning_rate": 6.323545467336017e-06,
|
| 1100 |
+
"loss": 0.2012,
|
| 1101 |
+
"step": 10800
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 4.185332138743121,
|
| 1105 |
+
"grad_norm": 0.20898312330245972,
|
| 1106 |
+
"learning_rate": 6.26527767974063e-06,
|
| 1107 |
+
"loss": 0.2005,
|
| 1108 |
+
"step": 10900
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 4.223729681300397,
|
| 1112 |
+
"grad_norm": 0.21366915106773376,
|
| 1113 |
+
"learning_rate": 6.206825716703166e-06,
|
| 1114 |
+
"loss": 0.2,
|
| 1115 |
+
"step": 11000
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 4.223729681300397,
|
| 1119 |
+
"eval_valid_loss": 0.18977344036102295,
|
| 1120 |
+
"eval_valid_runtime": 4.7328,
|
| 1121 |
+
"eval_valid_samples_per_second": 211.293,
|
| 1122 |
+
"eval_valid_steps_per_second": 6.761,
|
| 1123 |
+
"step": 11000
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 4.223729681300397,
|
| 1127 |
+
"eval_valid_target_loss": 0.21274219453334808,
|
| 1128 |
+
"eval_valid_target_runtime": 4.6506,
|
| 1129 |
+
"eval_valid_target_samples_per_second": 215.026,
|
| 1130 |
+
"eval_valid_target_steps_per_second": 6.881,
|
| 1131 |
+
"step": 11000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 4.262127223857673,
|
| 1135 |
+
"grad_norm": 0.20968745648860931,
|
| 1136 |
+
"learning_rate": 6.1481980865663405e-06,
|
| 1137 |
+
"loss": 0.1993,
|
| 1138 |
+
"step": 11100
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 4.3005247664149495,
|
| 1142 |
+
"grad_norm": 0.20683012902736664,
|
| 1143 |
+
"learning_rate": 6.089403323243203e-06,
|
| 1144 |
+
"loss": 0.1992,
|
| 1145 |
+
"step": 11200
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 4.338922308972226,
|
| 1149 |
+
"grad_norm": 0.20785097777843475,
|
| 1150 |
+
"learning_rate": 6.030449984974916e-06,
|
| 1151 |
+
"loss": 0.199,
|
| 1152 |
+
"step": 11300
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 4.377319851529502,
|
| 1156 |
+
"grad_norm": 0.20532238483428955,
|
| 1157 |
+
"learning_rate": 5.971346653085025e-06,
|
| 1158 |
+
"loss": 0.199,
|
| 1159 |
+
"step": 11400
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 4.415717394086778,
|
| 1163 |
+
"grad_norm": 0.21589842438697815,
|
| 1164 |
+
"learning_rate": 5.912101930730329e-06,
|
| 1165 |
+
"loss": 0.1992,
|
| 1166 |
+
"step": 11500
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 4.415717394086778,
|
| 1170 |
+
"eval_valid_loss": 0.18833594024181366,
|
| 1171 |
+
"eval_valid_runtime": 4.6904,
|
| 1172 |
+
"eval_valid_samples_per_second": 213.203,
|
| 1173 |
+
"eval_valid_steps_per_second": 6.823,
|
| 1174 |
+
"step": 11500
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 4.415717394086778,
|
| 1178 |
+
"eval_valid_target_loss": 0.211976557970047,
|
| 1179 |
+
"eval_valid_target_runtime": 4.658,
|
| 1180 |
+
"eval_valid_target_samples_per_second": 214.686,
|
| 1181 |
+
"eval_valid_target_steps_per_second": 6.87,
|
| 1182 |
+
"step": 11500
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 4.4541149366440544,
|
| 1186 |
+
"grad_norm": 0.2021540254354477,
|
| 1187 |
+
"learning_rate": 5.852724441648614e-06,
|
| 1188 |
+
"loss": 0.1987,
|
| 1189 |
+
"step": 11600
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 4.492512479201331,
|
| 1193 |
+
"grad_norm": 0.24406403303146362,
|
| 1194 |
+
"learning_rate": 5.7932228289033506e-06,
|
| 1195 |
+
"loss": 0.1984,
|
| 1196 |
+
"step": 11700
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 4.530910021758608,
|
| 1200 |
+
"grad_norm": 0.20519228279590607,
|
| 1201 |
+
"learning_rate": 5.7336057536256216e-06,
|
| 1202 |
+
"loss": 0.1984,
|
| 1203 |
+
"step": 11800
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 4.569307564315884,
|
| 1207 |
+
"grad_norm": 0.21227143704891205,
|
| 1208 |
+
"learning_rate": 5.67388189375337e-06,
|
| 1209 |
+
"loss": 0.1976,
|
| 1210 |
+
"step": 11900
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 4.60770510687316,
|
| 1214 |
+
"grad_norm": 0.2325662076473236,
|
| 1215 |
+
"learning_rate": 5.614059942768254e-06,
|
| 1216 |
+
"loss": 0.1977,
|
| 1217 |
+
"step": 12000
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 4.60770510687316,
|
| 1221 |
+
"eval_valid_loss": 0.18742187321186066,
|
| 1222 |
+
"eval_valid_runtime": 4.6831,
|
| 1223 |
+
"eval_valid_samples_per_second": 213.535,
|
| 1224 |
+
"eval_valid_steps_per_second": 6.833,
|
| 1225 |
+
"step": 12000
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 4.60770510687316,
|
| 1229 |
+
"eval_valid_target_loss": 0.21108593046665192,
|
| 1230 |
+
"eval_valid_target_runtime": 4.6502,
|
| 1231 |
+
"eval_valid_target_samples_per_second": 215.046,
|
| 1232 |
+
"eval_valid_target_steps_per_second": 6.881,
|
| 1233 |
+
"step": 12000
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 4.6461026494304365,
|
| 1237 |
+
"grad_norm": 0.2245544046163559,
|
| 1238 |
+
"learning_rate": 5.554148608430192e-06,
|
| 1239 |
+
"loss": 0.1965,
|
| 1240 |
+
"step": 12100
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 4.684500191987713,
|
| 1244 |
+
"grad_norm": 0.22662824392318726,
|
| 1245 |
+
"learning_rate": 5.4941566115098614e-06,
|
| 1246 |
+
"loss": 0.1971,
|
| 1247 |
+
"step": 12200
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 4.722897734544989,
|
| 1251 |
+
"grad_norm": 0.19245535135269165,
|
| 1252 |
+
"learning_rate": 5.4340926845192874e-06,
|
| 1253 |
+
"loss": 0.1974,
|
| 1254 |
+
"step": 12300
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 4.761295277102265,
|
| 1258 |
+
"grad_norm": 0.18942756950855255,
|
| 1259 |
+
"learning_rate": 5.373965570440729e-06,
|
| 1260 |
+
"loss": 0.1966,
|
| 1261 |
+
"step": 12400
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 4.799692819659541,
|
| 1265 |
+
"grad_norm": 0.1962059736251831,
|
| 1266 |
+
"learning_rate": 5.3137840214540395e-06,
|
| 1267 |
+
"loss": 0.1958,
|
| 1268 |
+
"step": 12500
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 4.799692819659541,
|
| 1272 |
+
"eval_valid_loss": 0.18663281202316284,
|
| 1273 |
+
"eval_valid_runtime": 4.6972,
|
| 1274 |
+
"eval_valid_samples_per_second": 212.895,
|
| 1275 |
+
"eval_valid_steps_per_second": 6.813,
|
| 1276 |
+
"step": 12500
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 4.799692819659541,
|
| 1280 |
+
"eval_valid_target_loss": 0.21009375154972076,
|
| 1281 |
+
"eval_valid_target_runtime": 4.6708,
|
| 1282 |
+
"eval_valid_target_samples_per_second": 214.096,
|
| 1283 |
+
"eval_valid_target_steps_per_second": 6.851,
|
| 1284 |
+
"step": 12500
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 4.838090362216818,
|
| 1288 |
+
"grad_norm": 0.2151457667350769,
|
| 1289 |
+
"learning_rate": 5.2535567976626846e-06,
|
| 1290 |
+
"loss": 0.1963,
|
| 1291 |
+
"step": 12600
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 4.876487904774095,
|
| 1295 |
+
"grad_norm": 0.18380814790725708,
|
| 1296 |
+
"learning_rate": 5.1932926658186166e-06,
|
| 1297 |
+
"loss": 0.1959,
|
| 1298 |
+
"step": 12700
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 4.914885447331371,
|
| 1302 |
+
"grad_norm": 0.19516663253307343,
|
| 1303 |
+
"learning_rate": 5.133000398046168e-06,
|
| 1304 |
+
"loss": 0.1953,
|
| 1305 |
+
"step": 12800
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 4.953282989888647,
|
| 1309 |
+
"grad_norm": 0.24182352423667908,
|
| 1310 |
+
"learning_rate": 5.072688770565177e-06,
|
| 1311 |
+
"loss": 0.1953,
|
| 1312 |
+
"step": 12900
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 4.9916805324459235,
|
| 1316 |
+
"grad_norm": 0.23720215260982513,
|
| 1317 |
+
"learning_rate": 5.012366562413501e-06,
|
| 1318 |
+
"loss": 0.1955,
|
| 1319 |
+
"step": 13000
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 4.9916805324459235,
|
| 1323 |
+
"eval_valid_loss": 0.18524999916553497,
|
| 1324 |
+
"eval_valid_runtime": 4.6908,
|
| 1325 |
+
"eval_valid_samples_per_second": 213.184,
|
| 1326 |
+
"eval_valid_steps_per_second": 6.822,
|
| 1327 |
+
"step": 13000
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 4.9916805324459235,
|
| 1331 |
+
"eval_valid_target_loss": 0.20893749594688416,
|
| 1332 |
+
"eval_valid_target_runtime": 4.6667,
|
| 1333 |
+
"eval_valid_target_samples_per_second": 214.285,
|
| 1334 |
+
"eval_valid_target_steps_per_second": 6.857,
|
| 1335 |
+
"step": 13000
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 5.0300780750032,
|
| 1339 |
+
"grad_norm": 0.20271484553813934,
|
| 1340 |
+
"learning_rate": 4.952042554169138e-06,
|
| 1341 |
+
"loss": 0.1948,
|
| 1342 |
+
"step": 13100
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 5.068475617560476,
|
| 1346 |
+
"grad_norm": 0.2053770273923874,
|
| 1347 |
+
"learning_rate": 4.891725526672107e-06,
|
| 1348 |
+
"loss": 0.1947,
|
| 1349 |
+
"step": 13200
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 5.106873160117752,
|
| 1353 |
+
"grad_norm": 0.20811979472637177,
|
| 1354 |
+
"learning_rate": 4.8314242597463e-06,
|
| 1355 |
+
"loss": 0.1939,
|
| 1356 |
+
"step": 13300
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 5.145270702675028,
|
| 1360 |
+
"grad_norm": 0.19889037311077118,
|
| 1361 |
+
"learning_rate": 4.771147530921483e-06,
|
| 1362 |
+
"loss": 0.1943,
|
| 1363 |
+
"step": 13400
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 5.1836682452323055,
|
| 1367 |
+
"grad_norm": 0.2038932591676712,
|
| 1368 |
+
"learning_rate": 4.710904114155621e-06,
|
| 1369 |
+
"loss": 0.1938,
|
| 1370 |
+
"step": 13500
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 5.1836682452323055,
|
| 1374 |
+
"eval_valid_loss": 0.1847265660762787,
|
| 1375 |
+
"eval_valid_runtime": 4.698,
|
| 1376 |
+
"eval_valid_samples_per_second": 212.854,
|
| 1377 |
+
"eval_valid_steps_per_second": 6.811,
|
| 1378 |
+
"step": 13500
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 5.1836682452323055,
|
| 1382 |
+
"eval_valid_target_loss": 0.20839843153953552,
|
| 1383 |
+
"eval_valid_target_runtime": 4.6593,
|
| 1384 |
+
"eval_valid_target_samples_per_second": 214.626,
|
| 1385 |
+
"eval_valid_target_steps_per_second": 6.868,
|
| 1386 |
+
"step": 13500
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 5.222065787789582,
|
| 1390 |
+
"grad_norm": 0.19585560262203217,
|
| 1391 |
+
"learning_rate": 4.650702778557736e-06,
|
| 1392 |
+
"loss": 0.1932,
|
| 1393 |
+
"step": 13600
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 5.260463330346858,
|
| 1397 |
+
"grad_norm": 0.23953603208065033,
|
| 1398 |
+
"learning_rate": 4.59055228711146e-06,
|
| 1399 |
+
"loss": 0.1933,
|
| 1400 |
+
"step": 13700
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 5.298860872904134,
|
| 1404 |
+
"grad_norm": 0.21477288007736206,
|
| 1405 |
+
"learning_rate": 4.530461395399485e-06,
|
| 1406 |
+
"loss": 0.1929,
|
| 1407 |
+
"step": 13800
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 5.33725841546141,
|
| 1411 |
+
"grad_norm": 0.22662727534770966,
|
| 1412 |
+
"learning_rate": 4.470438850329089e-06,
|
| 1413 |
+
"loss": 0.1935,
|
| 1414 |
+
"step": 13900
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 5.375655958018687,
|
| 1418 |
+
"grad_norm": 0.18912354111671448,
|
| 1419 |
+
"learning_rate": 4.410493388858925e-06,
|
| 1420 |
+
"loss": 0.1931,
|
| 1421 |
+
"step": 14000
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 5.375655958018687,
|
| 1425 |
+
"eval_valid_loss": 0.18379686772823334,
|
| 1426 |
+
"eval_valid_runtime": 4.6729,
|
| 1427 |
+
"eval_valid_samples_per_second": 214.001,
|
| 1428 |
+
"eval_valid_steps_per_second": 6.848,
|
| 1429 |
+
"step": 14000
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 5.375655958018687,
|
| 1433 |
+
"eval_valid_target_loss": 0.20746874809265137,
|
| 1434 |
+
"eval_valid_target_runtime": 4.6581,
|
| 1435 |
+
"eval_valid_target_samples_per_second": 214.682,
|
| 1436 |
+
"eval_valid_target_steps_per_second": 6.87,
|
| 1437 |
+
"step": 14000
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 5.414053500575963,
|
| 1441 |
+
"grad_norm": 0.21155835688114166,
|
| 1442 |
+
"learning_rate": 4.350633736727259e-06,
|
| 1443 |
+
"loss": 0.193,
|
| 1444 |
+
"step": 14100
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 5.452451043133239,
|
| 1448 |
+
"grad_norm": 0.2160138338804245,
|
| 1449 |
+
"learning_rate": 4.29086860718184e-06,
|
| 1450 |
+
"loss": 0.1931,
|
| 1451 |
+
"step": 14200
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 5.490848585690516,
|
| 1455 |
+
"grad_norm": 0.19270409643650055,
|
| 1456 |
+
"learning_rate": 4.231206699711587e-06,
|
| 1457 |
+
"loss": 0.1925,
|
| 1458 |
+
"step": 14300
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 5.5292461282477925,
|
| 1462 |
+
"grad_norm": 0.18501386046409607,
|
| 1463 |
+
"learning_rate": 4.171656698780281e-06,
|
| 1464 |
+
"loss": 0.1925,
|
| 1465 |
+
"step": 14400
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 5.567643670805069,
|
| 1469 |
+
"grad_norm": 0.20564299821853638,
|
| 1470 |
+
"learning_rate": 4.112227272562447e-06,
|
| 1471 |
+
"loss": 0.1918,
|
| 1472 |
+
"step": 14500
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 5.567643670805069,
|
| 1476 |
+
"eval_valid_loss": 0.18317969143390656,
|
| 1477 |
+
"eval_valid_runtime": 4.679,
|
| 1478 |
+
"eval_valid_samples_per_second": 213.72,
|
| 1479 |
+
"eval_valid_steps_per_second": 6.839,
|
| 1480 |
+
"step": 14500
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 5.567643670805069,
|
| 1484 |
+
"eval_valid_target_loss": 0.20700781047344208,
|
| 1485 |
+
"eval_valid_target_runtime": 4.674,
|
| 1486 |
+
"eval_valid_target_samples_per_second": 213.95,
|
| 1487 |
+
"eval_valid_target_steps_per_second": 6.846,
|
| 1488 |
+
"step": 14500
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 5.606041213362345,
|
| 1492 |
+
"grad_norm": 0.21509169042110443,
|
| 1493 |
+
"learning_rate": 4.052927071681593e-06,
|
| 1494 |
+
"loss": 0.1919,
|
| 1495 |
+
"step": 14600
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 5.644438755919621,
|
| 1499 |
+
"grad_norm": 0.18730491399765015,
|
| 1500 |
+
"learning_rate": 3.99376472795103e-06,
|
| 1501 |
+
"loss": 0.1921,
|
| 1502 |
+
"step": 14700
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 5.682836298476897,
|
| 1506 |
+
"grad_norm": 0.21269969642162323,
|
| 1507 |
+
"learning_rate": 3.934748853117398e-06,
|
| 1508 |
+
"loss": 0.1918,
|
| 1509 |
+
"step": 14800
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 5.721233841034174,
|
| 1513 |
+
"grad_norm": 0.18910899758338928,
|
| 1514 |
+
"learning_rate": 3.8758880376071415e-06,
|
| 1515 |
+
"loss": 0.1914,
|
| 1516 |
+
"step": 14900
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 5.75963138359145,
|
| 1520 |
+
"grad_norm": 0.22251802682876587,
|
| 1521 |
+
"learning_rate": 3.8171908492760665e-06,
|
| 1522 |
+
"loss": 0.1916,
|
| 1523 |
+
"step": 15000
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 5.75963138359145,
|
| 1527 |
+
"eval_valid_loss": 0.18259374797344208,
|
| 1528 |
+
"eval_valid_runtime": 4.67,
|
| 1529 |
+
"eval_valid_samples_per_second": 214.134,
|
| 1530 |
+
"eval_valid_steps_per_second": 6.852,
|
| 1531 |
+
"step": 15000
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 5.75963138359145,
|
| 1535 |
+
"eval_valid_target_loss": 0.20646093785762787,
|
| 1536 |
+
"eval_valid_target_runtime": 4.67,
|
| 1537 |
+
"eval_valid_target_samples_per_second": 214.131,
|
| 1538 |
+
"eval_valid_target_steps_per_second": 6.852,
|
| 1539 |
+
"step": 15000
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 5.798028926148726,
|
| 1543 |
+
"grad_norm": 0.17328619956970215,
|
| 1544 |
+
"learning_rate": 3.758665832162203e-06,
|
| 1545 |
+
"loss": 0.1911,
|
| 1546 |
+
"step": 15100
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 5.836426468706003,
|
| 1550 |
+
"grad_norm": 0.20850612223148346,
|
| 1551 |
+
"learning_rate": 3.7003215052421116e-06,
|
| 1552 |
+
"loss": 0.1915,
|
| 1553 |
+
"step": 15200
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 5.8748240112632795,
|
| 1557 |
+
"grad_norm": 0.1912785917520523,
|
| 1558 |
+
"learning_rate": 3.642166361190859e-06,
|
| 1559 |
+
"loss": 0.1908,
|
| 1560 |
+
"step": 15300
|
| 1561 |
+
},
|
| 1562 |
+
{
|
| 1563 |
+
"epoch": 5.913221553820556,
|
| 1564 |
+
"grad_norm": 0.2138790339231491,
|
| 1565 |
+
"learning_rate": 3.584208865145812e-06,
|
| 1566 |
+
"loss": 0.1907,
|
| 1567 |
+
"step": 15400
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 5.951619096377832,
|
| 1571 |
+
"grad_norm": 0.19723013043403625,
|
| 1572 |
+
"learning_rate": 3.5264574534744373e-06,
|
| 1573 |
+
"loss": 0.1913,
|
| 1574 |
+
"step": 15500
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 5.951619096377832,
|
| 1578 |
+
"eval_valid_loss": 0.1817968785762787,
|
| 1579 |
+
"eval_valid_runtime": 4.6726,
|
| 1580 |
+
"eval_valid_samples_per_second": 214.016,
|
| 1581 |
+
"eval_valid_steps_per_second": 6.849,
|
| 1582 |
+
"step": 15500
|
| 1583 |
+
},
|
| 1584 |
+
{
|
| 1585 |
+
"epoch": 5.951619096377832,
|
| 1586 |
+
"eval_valid_target_loss": 0.20574218034744263,
|
| 1587 |
+
"eval_valid_target_runtime": 4.6817,
|
| 1588 |
+
"eval_valid_target_samples_per_second": 213.599,
|
| 1589 |
+
"eval_valid_target_steps_per_second": 6.835,
|
| 1590 |
+
"step": 15500
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 5.990016638935108,
|
| 1594 |
+
"grad_norm": 0.19212548434734344,
|
| 1595 |
+
"learning_rate": 3.4689205325462997e-06,
|
| 1596 |
+
"loss": 0.1907,
|
| 1597 |
+
"step": 15600
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 6.028414181492384,
|
| 1601 |
+
"grad_norm": 0.19529464840888977,
|
| 1602 |
+
"learning_rate": 3.4116064775094126e-06,
|
| 1603 |
+
"loss": 0.1901,
|
| 1604 |
+
"step": 15700
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 6.066811724049661,
|
| 1608 |
+
"grad_norm": 0.2088070809841156,
|
| 1609 |
+
"learning_rate": 3.354523631071147e-06,
|
| 1610 |
+
"loss": 0.1902,
|
| 1611 |
+
"step": 15800
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 6.105209266606937,
|
| 1615 |
+
"grad_norm": 0.19294045865535736,
|
| 1616 |
+
"learning_rate": 3.2976803022838514e-06,
|
| 1617 |
+
"loss": 0.1903,
|
| 1618 |
+
"step": 15900
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 6.143606809164213,
|
| 1622 |
+
"grad_norm": 0.20844899117946625,
|
| 1623 |
+
"learning_rate": 3.2410847653353805e-06,
|
| 1624 |
+
"loss": 0.1897,
|
| 1625 |
+
"step": 16000
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 6.143606809164213,
|
| 1629 |
+
"eval_valid_loss": 0.1809999942779541,
|
| 1630 |
+
"eval_valid_runtime": 4.6789,
|
| 1631 |
+
"eval_valid_samples_per_second": 213.724,
|
| 1632 |
+
"eval_valid_steps_per_second": 6.839,
|
| 1633 |
+
"step": 16000
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 6.143606809164213,
|
| 1637 |
+
"eval_valid_target_loss": 0.20546874403953552,
|
| 1638 |
+
"eval_valid_target_runtime": 4.6614,
|
| 1639 |
+
"eval_valid_target_samples_per_second": 214.53,
|
| 1640 |
+
"eval_valid_target_steps_per_second": 6.865,
|
| 1641 |
+
"step": 16000
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 6.18200435172149,
|
| 1645 |
+
"grad_norm": 0.19932307302951813,
|
| 1646 |
+
"learning_rate": 3.184745258344688e-06,
|
| 1647 |
+
"loss": 0.1894,
|
| 1648 |
+
"step": 16100
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 6.220401894278766,
|
| 1652 |
+
"grad_norm": 0.19776058197021484,
|
| 1653 |
+
"learning_rate": 3.128669982162681e-06,
|
| 1654 |
+
"loss": 0.1899,
|
| 1655 |
+
"step": 16200
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 6.258799436836043,
|
| 1659 |
+
"grad_norm": 0.20467509329319,
|
| 1660 |
+
"learning_rate": 3.07286709917849e-06,
|
| 1661 |
+
"loss": 0.1898,
|
| 1662 |
+
"step": 16300
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 6.297196979393319,
|
| 1666 |
+
"grad_norm": 0.19593088328838348,
|
| 1667 |
+
"learning_rate": 3.017344732131342e-06,
|
| 1668 |
+
"loss": 0.1895,
|
| 1669 |
+
"step": 16400
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 6.335594521950595,
|
| 1673 |
+
"grad_norm": 0.20078891515731812,
|
| 1674 |
+
"learning_rate": 2.9621109629282064e-06,
|
| 1675 |
+
"loss": 0.1897,
|
| 1676 |
+
"step": 16500
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 6.335594521950595,
|
| 1680 |
+
"eval_valid_loss": 0.1807578057050705,
|
| 1681 |
+
"eval_valid_runtime": 4.7017,
|
| 1682 |
+
"eval_valid_samples_per_second": 212.687,
|
| 1683 |
+
"eval_valid_steps_per_second": 6.806,
|
| 1684 |
+
"step": 16500
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 6.335594521950595,
|
| 1688 |
+
"eval_valid_target_loss": 0.2052578181028366,
|
| 1689 |
+
"eval_valid_target_runtime": 4.6726,
|
| 1690 |
+
"eval_valid_target_samples_per_second": 214.013,
|
| 1691 |
+
"eval_valid_target_steps_per_second": 6.848,
|
| 1692 |
+
"step": 16500
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 6.373992064507871,
|
| 1696 |
+
"grad_norm": 0.17822235822677612,
|
| 1697 |
+
"learning_rate": 2.9071738314673758e-06,
|
| 1698 |
+
"loss": 0.1889,
|
| 1699 |
+
"step": 16600
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 6.412389607065148,
|
| 1703 |
+
"grad_norm": 0.21160703897476196,
|
| 1704 |
+
"learning_rate": 2.8525413344681797e-06,
|
| 1705 |
+
"loss": 0.1889,
|
| 1706 |
+
"step": 16700
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 6.450787149622424,
|
| 1710 |
+
"grad_norm": 0.19472962617874146,
|
| 1711 |
+
"learning_rate": 2.798221424306953e-06,
|
| 1712 |
+
"loss": 0.1894,
|
| 1713 |
+
"step": 16800
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 6.4891846921797,
|
| 1717 |
+
"grad_norm": 0.17923222482204437,
|
| 1718 |
+
"learning_rate": 2.744222007859506e-06,
|
| 1719 |
+
"loss": 0.1891,
|
| 1720 |
+
"step": 16900
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 6.527582234736977,
|
| 1724 |
+
"grad_norm": 0.18077126145362854,
|
| 1725 |
+
"learning_rate": 2.690550945350157e-06,
|
| 1726 |
+
"loss": 0.1886,
|
| 1727 |
+
"step": 17000
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 6.527582234736977,
|
| 1731 |
+
"eval_valid_loss": 0.18031249940395355,
|
| 1732 |
+
"eval_valid_runtime": 4.6828,
|
| 1733 |
+
"eval_valid_samples_per_second": 213.548,
|
| 1734 |
+
"eval_valid_steps_per_second": 6.834,
|
| 1735 |
+
"step": 17000
|
| 1736 |
+
},
|
| 1737 |
+
{
|
| 1738 |
+
"epoch": 6.527582234736977,
|
| 1739 |
+
"eval_valid_target_loss": 0.20450781285762787,
|
| 1740 |
+
"eval_valid_target_runtime": 4.6685,
|
| 1741 |
+
"eval_valid_target_samples_per_second": 214.203,
|
| 1742 |
+
"eval_valid_target_steps_per_second": 6.854,
|
| 1743 |
+
"step": 17000
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 6.565979777294253,
|
| 1747 |
+
"grad_norm": 0.19065329432487488,
|
| 1748 |
+
"learning_rate": 2.637216049207615e-06,
|
| 1749 |
+
"loss": 0.188,
|
| 1750 |
+
"step": 17100
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 6.60437731985153,
|
| 1754 |
+
"grad_norm": 0.20368430018424988,
|
| 1755 |
+
"learning_rate": 2.5842250829277724e-06,
|
| 1756 |
+
"loss": 0.189,
|
| 1757 |
+
"step": 17200
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 6.642774862408806,
|
| 1761 |
+
"grad_norm": 0.21131780743598938,
|
| 1762 |
+
"learning_rate": 2.5315857599436575e-06,
|
| 1763 |
+
"loss": 0.1887,
|
| 1764 |
+
"step": 17300
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 6.681172404966082,
|
| 1768 |
+
"grad_norm": 0.2033446729183197,
|
| 1769 |
+
"learning_rate": 2.4793057425026467e-06,
|
| 1770 |
+
"loss": 0.1887,
|
| 1771 |
+
"step": 17400
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 6.719569947523358,
|
| 1775 |
+
"grad_norm": 0.19689294695854187,
|
| 1776 |
+
"learning_rate": 2.427392640551137e-06,
|
| 1777 |
+
"loss": 0.1887,
|
| 1778 |
+
"step": 17500
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 6.719569947523358,
|
| 1782 |
+
"eval_valid_loss": 0.17996874451637268,
|
| 1783 |
+
"eval_valid_runtime": 4.7043,
|
| 1784 |
+
"eval_valid_samples_per_second": 212.57,
|
| 1785 |
+
"eval_valid_steps_per_second": 6.802,
|
| 1786 |
+
"step": 17500
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 6.719569947523358,
|
| 1790 |
+
"eval_valid_target_loss": 0.20432811975479126,
|
| 1791 |
+
"eval_valid_target_runtime": 4.6638,
|
| 1792 |
+
"eval_valid_target_samples_per_second": 214.416,
|
| 1793 |
+
"eval_valid_target_steps_per_second": 6.861,
|
| 1794 |
+
"step": 17500
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 6.757967490080635,
|
| 1798 |
+
"grad_norm": 0.1994999349117279,
|
| 1799 |
+
"learning_rate": 2.3758540106268406e-06,
|
| 1800 |
+
"loss": 0.1881,
|
| 1801 |
+
"step": 17600
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 6.796365032637911,
|
| 1805 |
+
"grad_norm": 0.19650602340698242,
|
| 1806 |
+
"learning_rate": 2.32469735475884e-06,
|
| 1807 |
+
"loss": 0.1881,
|
| 1808 |
+
"step": 17700
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 6.834762575195187,
|
| 1812 |
+
"grad_norm": 0.21248474717140198,
|
| 1813 |
+
"learning_rate": 2.273930119375586e-06,
|
| 1814 |
+
"loss": 0.1882,
|
| 1815 |
+
"step": 17800
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 6.873160117752464,
|
| 1819 |
+
"grad_norm": 0.19042810797691345,
|
| 1820 |
+
"learning_rate": 2.2235596942209776e-06,
|
| 1821 |
+
"loss": 0.188,
|
| 1822 |
+
"step": 17900
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 6.91155766030974,
|
| 1826 |
+
"grad_norm": 0.23096908628940582,
|
| 1827 |
+
"learning_rate": 2.173593411278714e-06,
|
| 1828 |
+
"loss": 0.1886,
|
| 1829 |
+
"step": 18000
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 6.91155766030974,
|
| 1833 |
+
"eval_valid_loss": 0.17952343821525574,
|
| 1834 |
+
"eval_valid_runtime": 4.6878,
|
| 1835 |
+
"eval_valid_samples_per_second": 213.321,
|
| 1836 |
+
"eval_valid_steps_per_second": 6.826,
|
| 1837 |
+
"step": 18000
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 6.91155766030974,
|
| 1841 |
+
"eval_valid_target_loss": 0.20391406118869781,
|
| 1842 |
+
"eval_valid_target_runtime": 4.6595,
|
| 1843 |
+
"eval_valid_target_samples_per_second": 214.617,
|
| 1844 |
+
"eval_valid_target_steps_per_second": 6.868,
|
| 1845 |
+
"step": 18000
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 6.949955202867017,
|
| 1849 |
+
"grad_norm": 0.21275204420089722,
|
| 1850 |
+
"learning_rate": 2.124038543705034e-06,
|
| 1851 |
+
"loss": 0.1878,
|
| 1852 |
+
"step": 18100
|
| 1853 |
+
},
|
| 1854 |
+
{
|
| 1855 |
+
"epoch": 6.988352745424293,
|
| 1856 |
+
"grad_norm": 0.20453621447086334,
|
| 1857 |
+
"learning_rate": 2.0749023047700285e-06,
|
| 1858 |
+
"loss": 0.188,
|
| 1859 |
+
"step": 18200
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 7.026750287981569,
|
| 1863 |
+
"grad_norm": 0.20724526047706604,
|
| 1864 |
+
"learning_rate": 2.026191846807663e-06,
|
| 1865 |
+
"loss": 0.1883,
|
| 1866 |
+
"step": 18300
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 7.065147830538845,
|
| 1870 |
+
"grad_norm": 0.1886543333530426,
|
| 1871 |
+
"learning_rate": 1.9779142601746825e-06,
|
| 1872 |
+
"loss": 0.1874,
|
| 1873 |
+
"step": 18400
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 7.1035453730961216,
|
| 1877 |
+
"grad_norm": 0.20411571860313416,
|
| 1878 |
+
"learning_rate": 1.9300765722185265e-06,
|
| 1879 |
+
"loss": 0.187,
|
| 1880 |
+
"step": 18500
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 7.1035453730961216,
|
| 1884 |
+
"eval_valid_loss": 0.17924219369888306,
|
| 1885 |
+
"eval_valid_runtime": 4.6825,
|
| 1886 |
+
"eval_valid_samples_per_second": 213.561,
|
| 1887 |
+
"eval_valid_steps_per_second": 6.834,
|
| 1888 |
+
"step": 18500
|
| 1889 |
+
},
|
| 1890 |
+
{
|
| 1891 |
+
"epoch": 7.1035453730961216,
|
| 1892 |
+
"eval_valid_target_loss": 0.20393750071525574,
|
| 1893 |
+
"eval_valid_target_runtime": 4.6736,
|
| 1894 |
+
"eval_valid_target_samples_per_second": 213.97,
|
| 1895 |
+
"eval_valid_target_steps_per_second": 6.847,
|
| 1896 |
+
"step": 18500
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 7.141942915653398,
|
| 1900 |
+
"grad_norm": 0.18996645510196686,
|
| 1901 |
+
"learning_rate": 1.8826857462544129e-06,
|
| 1902 |
+
"loss": 0.1871,
|
| 1903 |
+
"step": 18600
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 7.180340458210675,
|
| 1907 |
+
"grad_norm": 0.21018381416797638,
|
| 1908 |
+
"learning_rate": 1.8357486805517615e-06,
|
| 1909 |
+
"loss": 0.1874,
|
| 1910 |
+
"step": 18700
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 7.218738000767951,
|
| 1914 |
+
"grad_norm": 0.19617675244808197,
|
| 1915 |
+
"learning_rate": 1.7892722073300627e-06,
|
| 1916 |
+
"loss": 0.1869,
|
| 1917 |
+
"step": 18800
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 7.257135543325227,
|
| 1921 |
+
"grad_norm": 0.2340448796749115,
|
| 1922 |
+
"learning_rate": 1.743263091764379e-06,
|
| 1923 |
+
"loss": 0.187,
|
| 1924 |
+
"step": 18900
|
| 1925 |
+
},
|
| 1926 |
+
{
|
| 1927 |
+
"epoch": 7.295533085882504,
|
| 1928 |
+
"grad_norm": 0.22970305383205414,
|
| 1929 |
+
"learning_rate": 1.6977280310005845e-06,
|
| 1930 |
+
"loss": 0.1873,
|
| 1931 |
+
"step": 19000
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 7.295533085882504,
|
| 1935 |
+
"eval_valid_loss": 0.1788671910762787,
|
| 1936 |
+
"eval_valid_runtime": 4.6706,
|
| 1937 |
+
"eval_valid_samples_per_second": 214.105,
|
| 1938 |
+
"eval_valid_steps_per_second": 6.851,
|
| 1939 |
+
"step": 19000
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 7.295533085882504,
|
| 1943 |
+
"eval_valid_target_loss": 0.20334374904632568,
|
| 1944 |
+
"eval_valid_target_runtime": 4.6842,
|
| 1945 |
+
"eval_valid_target_samples_per_second": 213.484,
|
| 1946 |
+
"eval_valid_target_steps_per_second": 6.831,
|
| 1947 |
+
"step": 19000
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 7.33393062843978,
|
| 1951 |
+
"grad_norm": 0.20527499914169312,
|
| 1952 |
+
"learning_rate": 1.6526736531805354e-06,
|
| 1953 |
+
"loss": 0.1873,
|
| 1954 |
+
"step": 19100
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 7.372328170997056,
|
| 1958 |
+
"grad_norm": 0.1835908442735672,
|
| 1959 |
+
"learning_rate": 1.6081065164772624e-06,
|
| 1960 |
+
"loss": 0.187,
|
| 1961 |
+
"step": 19200
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 7.410725713554332,
|
| 1965 |
+
"grad_norm": 0.18936371803283691,
|
| 1966 |
+
"learning_rate": 1.564033108140348e-06,
|
| 1967 |
+
"loss": 0.1865,
|
| 1968 |
+
"step": 19300
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 7.4491232561116085,
|
| 1972 |
+
"grad_norm": 0.19136998057365417,
|
| 1973 |
+
"learning_rate": 1.520459843551646e-06,
|
| 1974 |
+
"loss": 0.1872,
|
| 1975 |
+
"step": 19400
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 7.487520798668886,
|
| 1979 |
+
"grad_norm": 0.19691316783428192,
|
| 1980 |
+
"learning_rate": 1.4773930652914426e-06,
|
| 1981 |
+
"loss": 0.187,
|
| 1982 |
+
"step": 19500
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 7.487520798668886,
|
| 1986 |
+
"eval_valid_loss": 0.17878125607967377,
|
| 1987 |
+
"eval_valid_runtime": 4.6602,
|
| 1988 |
+
"eval_valid_samples_per_second": 214.581,
|
| 1989 |
+
"eval_valid_steps_per_second": 6.867,
|
| 1990 |
+
"step": 19500
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 7.487520798668886,
|
| 1994 |
+
"eval_valid_target_loss": 0.20325781404972076,
|
| 1995 |
+
"eval_valid_target_runtime": 4.6796,
|
| 1996 |
+
"eval_valid_target_samples_per_second": 213.695,
|
| 1997 |
+
"eval_valid_target_steps_per_second": 6.838,
|
| 1998 |
+
"step": 19500
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 7.525918341226162,
|
| 2002 |
+
"grad_norm": 0.18792080879211426,
|
| 2003 |
+
"learning_rate": 1.434839042215227e-06,
|
| 2004 |
+
"loss": 0.1868,
|
| 2005 |
+
"step": 19600
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 7.564315883783438,
|
| 2009 |
+
"grad_norm": 0.1945939064025879,
|
| 2010 |
+
"learning_rate": 1.3928039685411793e-06,
|
| 2011 |
+
"loss": 0.1869,
|
| 2012 |
+
"step": 19700
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 7.602713426340714,
|
| 2016 |
+
"grad_norm": 0.17974095046520233,
|
| 2017 |
+
"learning_rate": 1.3512939629485456e-06,
|
| 2018 |
+
"loss": 0.187,
|
| 2019 |
+
"step": 19800
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 7.641110968897991,
|
| 2023 |
+
"grad_norm": 0.22416825592517853,
|
| 2024 |
+
"learning_rate": 1.3103150676869864e-06,
|
| 2025 |
+
"loss": 0.1871,
|
| 2026 |
+
"step": 19900
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 7.679508511455267,
|
| 2030 |
+
"grad_norm": 0.19613422453403473,
|
| 2031 |
+
"learning_rate": 1.2698732476970627e-06,
|
| 2032 |
+
"loss": 0.1869,
|
| 2033 |
+
"step": 20000
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 7.679508511455267,
|
| 2037 |
+
"eval_valid_loss": 0.1783437430858612,
|
| 2038 |
+
"eval_valid_runtime": 4.6716,
|
| 2039 |
+
"eval_valid_samples_per_second": 214.058,
|
| 2040 |
+
"eval_valid_steps_per_second": 6.85,
|
| 2041 |
+
"step": 20000
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 7.679508511455267,
|
| 2045 |
+
"eval_valid_target_loss": 0.2031562477350235,
|
| 2046 |
+
"eval_valid_target_runtime": 4.6803,
|
| 2047 |
+
"eval_valid_target_samples_per_second": 213.661,
|
| 2048 |
+
"eval_valid_target_steps_per_second": 6.837,
|
| 2049 |
+
"step": 20000
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 7.717906054012543,
|
| 2053 |
+
"grad_norm": 0.20145875215530396,
|
| 2054 |
+
"learning_rate": 1.229974389741964e-06,
|
| 2055 |
+
"loss": 0.187,
|
| 2056 |
+
"step": 20100
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"epoch": 7.756303596569819,
|
| 2060 |
+
"grad_norm": 0.18396620452404022,
|
| 2061 |
+
"learning_rate": 1.1906243015506375e-06,
|
| 2062 |
+
"loss": 0.1867,
|
| 2063 |
+
"step": 20200
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 7.7947011391270955,
|
| 2067 |
+
"grad_norm": 0.18105918169021606,
|
| 2068 |
+
"learning_rate": 1.1518287109723958e-06,
|
| 2069 |
+
"loss": 0.1862,
|
| 2070 |
+
"step": 20300
|
| 2071 |
+
},
|
| 2072 |
+
{
|
| 2073 |
+
"epoch": 7.833098681684373,
|
| 2074 |
+
"grad_norm": 0.20986780524253845,
|
| 2075 |
+
"learning_rate": 1.1135932651431651e-06,
|
| 2076 |
+
"loss": 0.1863,
|
| 2077 |
+
"step": 20400
|
| 2078 |
+
},
|
| 2079 |
+
{
|
| 2080 |
+
"epoch": 7.871496224241649,
|
| 2081 |
+
"grad_norm": 0.21804456412792206,
|
| 2082 |
+
"learning_rate": 1.075923529663489e-06,
|
| 2083 |
+
"loss": 0.1869,
|
| 2084 |
+
"step": 20500
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 7.871496224241649,
|
| 2088 |
+
"eval_valid_loss": 0.17836718261241913,
|
| 2089 |
+
"eval_valid_runtime": 4.6832,
|
| 2090 |
+
"eval_valid_samples_per_second": 213.531,
|
| 2091 |
+
"eval_valid_steps_per_second": 6.833,
|
| 2092 |
+
"step": 20500
|
| 2093 |
+
},
|
| 2094 |
+
{
|
| 2095 |
+
"epoch": 7.871496224241649,
|
| 2096 |
+
"eval_valid_target_loss": 0.20322656631469727,
|
| 2097 |
+
"eval_valid_target_runtime": 4.6763,
|
| 2098 |
+
"eval_valid_target_samples_per_second": 213.843,
|
| 2099 |
+
"eval_valid_target_steps_per_second": 6.843,
|
| 2100 |
+
"step": 20500
|
| 2101 |
+
},
|
| 2102 |
+
{
|
| 2103 |
+
"epoch": 7.909893766798925,
|
| 2104 |
+
"grad_norm": 0.22019818425178528,
|
| 2105 |
+
"learning_rate": 1.0388249877883827e-06,
|
| 2106 |
+
"loss": 0.1858,
|
| 2107 |
+
"step": 20600
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 7.948291309356201,
|
| 2111 |
+
"grad_norm": 0.1965310275554657,
|
| 2112 |
+
"learning_rate": 1.0023030396291916e-06,
|
| 2113 |
+
"loss": 0.1866,
|
| 2114 |
+
"step": 20700
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 7.9866888519134775,
|
| 2118 |
+
"grad_norm": 0.18218408524990082,
|
| 2119 |
+
"learning_rate": 9.66363001367534e-07,
|
| 2120 |
+
"loss": 0.1869,
|
| 2121 |
+
"step": 20800
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 8.025086394470755,
|
| 2125 |
+
"grad_norm": 0.1850380003452301,
|
| 2126 |
+
"learning_rate": 9.310101044814835e-07,
|
| 2127 |
+
"loss": 0.1861,
|
| 2128 |
+
"step": 20900
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"epoch": 8.06348393702803,
|
| 2132 |
+
"grad_norm": 0.18823818862438202,
|
| 2133 |
+
"learning_rate": 8.962494949840577e-07,
|
| 2134 |
+
"loss": 0.186,
|
| 2135 |
+
"step": 21000
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 8.06348393702803,
|
| 2139 |
+
"eval_valid_loss": 0.17808593809604645,
|
| 2140 |
+
"eval_valid_runtime": 4.6916,
|
| 2141 |
+
"eval_valid_samples_per_second": 213.147,
|
| 2142 |
+
"eval_valid_steps_per_second": 6.821,
|
| 2143 |
+
"step": 21000
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 8.06348393702803,
|
| 2147 |
+
"eval_valid_target_loss": 0.20311719179153442,
|
| 2148 |
+
"eval_valid_target_runtime": 4.6653,
|
| 2149 |
+
"eval_valid_target_samples_per_second": 214.347,
|
| 2150 |
+
"eval_valid_target_steps_per_second": 6.859,
|
| 2151 |
+
"step": 21000
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 8.101881479585307,
|
| 2155 |
+
"grad_norm": 0.20501789450645447,
|
| 2156 |
+
"learning_rate": 8.620862326741658e-07,
|
| 2157 |
+
"loss": 0.1862,
|
| 2158 |
+
"step": 21100
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 8.140279022142582,
|
| 2162 |
+
"grad_norm": 0.19500133395195007,
|
| 2163 |
+
"learning_rate": 8.285252904000906e-07,
|
| 2164 |
+
"loss": 0.1862,
|
| 2165 |
+
"step": 21200
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 8.17867656469986,
|
| 2169 |
+
"grad_norm": 0.18742544949054718,
|
| 2170 |
+
"learning_rate": 7.955715533356367e-07,
|
| 2171 |
+
"loss": 0.1863,
|
| 2172 |
+
"step": 21300
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 8.217074107257135,
|
| 2176 |
+
"grad_norm": 0.20386624336242676,
|
| 2177 |
+
"learning_rate": 7.632298182690473e-07,
|
| 2178 |
+
"loss": 0.186,
|
| 2179 |
+
"step": 21400
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 8.255471649814412,
|
| 2183 |
+
"grad_norm": 0.17727358639240265,
|
| 2184 |
+
"learning_rate": 7.315047929047608e-07,
|
| 2185 |
+
"loss": 0.1861,
|
| 2186 |
+
"step": 21500
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 8.255471649814412,
|
| 2190 |
+
"eval_valid_loss": 0.17788280546665192,
|
| 2191 |
+
"eval_valid_runtime": 4.679,
|
| 2192 |
+
"eval_valid_samples_per_second": 213.72,
|
| 2193 |
+
"eval_valid_steps_per_second": 6.839,
|
| 2194 |
+
"step": 21500
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 8.255471649814412,
|
| 2198 |
+
"eval_valid_target_loss": 0.2026640623807907,
|
| 2199 |
+
"eval_valid_target_runtime": 4.6709,
|
| 2200 |
+
"eval_valid_target_samples_per_second": 214.093,
|
| 2201 |
+
"eval_valid_target_steps_per_second": 6.851,
|
| 2202 |
+
"step": 21500
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 8.293869192371687,
|
| 2206 |
+
"grad_norm": 0.19971401989459991,
|
| 2207 |
+
"learning_rate": 7.004010951781648e-07,
|
| 2208 |
+
"loss": 0.1858,
|
| 2209 |
+
"step": 21600
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 8.332266734928965,
|
| 2213 |
+
"grad_norm": 0.17827193439006805,
|
| 2214 |
+
"learning_rate": 6.699232525833987e-07,
|
| 2215 |
+
"loss": 0.1868,
|
| 2216 |
+
"step": 21700
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 8.370664277486242,
|
| 2220 |
+
"grad_norm": 0.18275295197963715,
|
| 2221 |
+
"learning_rate": 6.400757015143266e-07,
|
| 2222 |
+
"loss": 0.1858,
|
| 2223 |
+
"step": 21800
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 8.409061820043517,
|
| 2227 |
+
"grad_norm": 0.19496768712997437,
|
| 2228 |
+
"learning_rate": 6.108627866187661e-07,
|
| 2229 |
+
"loss": 0.1854,
|
| 2230 |
+
"step": 21900
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"epoch": 8.447459362600794,
|
| 2234 |
+
"grad_norm": 0.19046269357204437,
|
| 2235 |
+
"learning_rate": 5.822887601660832e-07,
|
| 2236 |
+
"loss": 0.1862,
|
| 2237 |
+
"step": 22000
|
| 2238 |
+
},
|
| 2239 |
+
{
|
| 2240 |
+
"epoch": 8.447459362600794,
|
| 2241 |
+
"eval_valid_loss": 0.17781250178813934,
|
| 2242 |
+
"eval_valid_runtime": 4.6746,
|
| 2243 |
+
"eval_valid_samples_per_second": 213.921,
|
| 2244 |
+
"eval_valid_steps_per_second": 6.845,
|
| 2245 |
+
"step": 22000
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 8.447459362600794,
|
| 2249 |
+
"eval_valid_target_loss": 0.2026640623807907,
|
| 2250 |
+
"eval_valid_target_runtime": 4.6755,
|
| 2251 |
+
"eval_valid_target_samples_per_second": 213.88,
|
| 2252 |
+
"eval_valid_target_steps_per_second": 6.844,
|
| 2253 |
+
"step": 22000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 8.48585690515807,
|
| 2257 |
+
"grad_norm": 0.20896296203136444,
|
| 2258 |
+
"learning_rate": 5.543577814282219e-07,
|
| 2259 |
+
"loss": 0.1856,
|
| 2260 |
+
"step": 22100
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 8.524254447715347,
|
| 2264 |
+
"grad_norm": 0.19562530517578125,
|
| 2265 |
+
"learning_rate": 5.270739160742738e-07,
|
| 2266 |
+
"loss": 0.1857,
|
| 2267 |
+
"step": 22200
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 8.562651990272622,
|
| 2271 |
+
"grad_norm": 0.1972120851278305,
|
| 2272 |
+
"learning_rate": 5.004411355786792e-07,
|
| 2273 |
+
"loss": 0.1863,
|
| 2274 |
+
"step": 22300
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 8.601049532829899,
|
| 2278 |
+
"grad_norm": 0.19712330400943756,
|
| 2279 |
+
"learning_rate": 4.7446331664312786e-07,
|
| 2280 |
+
"loss": 0.1855,
|
| 2281 |
+
"step": 22400
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 8.639447075387174,
|
| 2285 |
+
"grad_norm": 0.20409992337226868,
|
| 2286 |
+
"learning_rate": 4.4914424063226937e-07,
|
| 2287 |
+
"loss": 0.1857,
|
| 2288 |
+
"step": 22500
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 8.639447075387174,
|
| 2292 |
+
"eval_valid_loss": 0.17765624821186066,
|
| 2293 |
+
"eval_valid_runtime": 4.6769,
|
| 2294 |
+
"eval_valid_samples_per_second": 213.818,
|
| 2295 |
+
"eval_valid_steps_per_second": 6.842,
|
| 2296 |
+
"step": 22500
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 8.639447075387174,
|
| 2300 |
+
"eval_valid_target_loss": 0.2025781273841858,
|
| 2301 |
+
"eval_valid_target_runtime": 4.6696,
|
| 2302 |
+
"eval_valid_target_samples_per_second": 214.151,
|
| 2303 |
+
"eval_valid_target_steps_per_second": 6.853,
|
| 2304 |
+
"step": 22500
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 8.677844617944451,
|
| 2308 |
+
"grad_norm": 0.21083636581897736,
|
| 2309 |
+
"learning_rate": 4.2448759302328336e-07,
|
| 2310 |
+
"loss": 0.1861,
|
| 2311 |
+
"step": 22600
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 8.716242160501729,
|
| 2315 |
+
"grad_norm": 0.18778979778289795,
|
| 2316 |
+
"learning_rate": 4.0049696286942496e-07,
|
| 2317 |
+
"loss": 0.1862,
|
| 2318 |
+
"step": 22700
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 8.754639703059004,
|
| 2322 |
+
"grad_norm": 0.18586015701293945,
|
| 2323 |
+
"learning_rate": 3.7717584227759117e-07,
|
| 2324 |
+
"loss": 0.1857,
|
| 2325 |
+
"step": 22800
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 8.793037245616281,
|
| 2329 |
+
"grad_norm": 0.1977422684431076,
|
| 2330 |
+
"learning_rate": 3.54527625900013e-07,
|
| 2331 |
+
"loss": 0.1856,
|
| 2332 |
+
"step": 22900
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 8.831434788173556,
|
| 2336 |
+
"grad_norm": 0.18881608545780182,
|
| 2337 |
+
"learning_rate": 3.3255561044011564e-07,
|
| 2338 |
+
"loss": 0.1857,
|
| 2339 |
+
"step": 23000
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 8.831434788173556,
|
| 2343 |
+
"eval_valid_loss": 0.17771874368190765,
|
| 2344 |
+
"eval_valid_runtime": 4.6727,
|
| 2345 |
+
"eval_valid_samples_per_second": 214.01,
|
| 2346 |
+
"eval_valid_steps_per_second": 6.848,
|
| 2347 |
+
"step": 23000
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 8.831434788173556,
|
| 2351 |
+
"eval_valid_target_loss": 0.20250000059604645,
|
| 2352 |
+
"eval_valid_target_runtime": 4.6666,
|
| 2353 |
+
"eval_valid_target_samples_per_second": 214.287,
|
| 2354 |
+
"eval_valid_target_steps_per_second": 6.857,
|
| 2355 |
+
"step": 23000
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 8.869832330730834,
|
| 2359 |
+
"grad_norm": 0.2037239372730255,
|
| 2360 |
+
"learning_rate": 3.112629941726547e-07,
|
| 2361 |
+
"loss": 0.1856,
|
| 2362 |
+
"step": 23100
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 8.908229873288109,
|
| 2366 |
+
"grad_norm": 0.18967826664447784,
|
| 2367 |
+
"learning_rate": 2.9065287647816744e-07,
|
| 2368 |
+
"loss": 0.1855,
|
| 2369 |
+
"step": 23200
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 8.946627415845386,
|
| 2373 |
+
"grad_norm": 0.17752571403980255,
|
| 2374 |
+
"learning_rate": 2.707282573918213e-07,
|
| 2375 |
+
"loss": 0.1858,
|
| 2376 |
+
"step": 23300
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 8.985024958402661,
|
| 2380 |
+
"grad_norm": 0.18709731101989746,
|
| 2381 |
+
"learning_rate": 2.514920371667301e-07,
|
| 2382 |
+
"loss": 0.1854,
|
| 2383 |
+
"step": 23400
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 9.023422500959938,
|
| 2387 |
+
"grad_norm": 0.21643956005573273,
|
| 2388 |
+
"learning_rate": 2.3294701585178213e-07,
|
| 2389 |
+
"loss": 0.1858,
|
| 2390 |
+
"step": 23500
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 9.023422500959938,
|
| 2394 |
+
"eval_valid_loss": 0.17762500047683716,
|
| 2395 |
+
"eval_valid_runtime": 4.6791,
|
| 2396 |
+
"eval_valid_samples_per_second": 213.717,
|
| 2397 |
+
"eval_valid_steps_per_second": 6.839,
|
| 2398 |
+
"step": 23500
|
| 2399 |
+
},
|
| 2400 |
+
{
|
| 2401 |
+
"epoch": 9.023422500959938,
|
| 2402 |
+
"eval_valid_target_loss": 0.20255468785762787,
|
| 2403 |
+
"eval_valid_target_runtime": 4.7103,
|
| 2404 |
+
"eval_valid_target_samples_per_second": 212.301,
|
| 2405 |
+
"eval_valid_target_steps_per_second": 6.794,
|
| 2406 |
+
"step": 23500
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 9.061820043517216,
|
| 2410 |
+
"grad_norm": 0.18775244057178497,
|
| 2411 |
+
"learning_rate": 2.1509589288407183e-07,
|
| 2412 |
+
"loss": 0.1855,
|
| 2413 |
+
"step": 23600
|
| 2414 |
+
},
|
| 2415 |
+
{
|
| 2416 |
+
"epoch": 9.100217586074491,
|
| 2417 |
+
"grad_norm": 0.17277489602565765,
|
| 2418 |
+
"learning_rate": 1.9794126669595403e-07,
|
| 2419 |
+
"loss": 0.1859,
|
| 2420 |
+
"step": 23700
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 9.138615128631768,
|
| 2424 |
+
"grad_norm": 0.18996348977088928,
|
| 2425 |
+
"learning_rate": 1.8148563433682264e-07,
|
| 2426 |
+
"loss": 0.1852,
|
| 2427 |
+
"step": 23800
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 9.177012671189043,
|
| 2431 |
+
"grad_norm": 0.1894453912973404,
|
| 2432 |
+
"learning_rate": 1.6573139110963087e-07,
|
| 2433 |
+
"loss": 0.1854,
|
| 2434 |
+
"step": 23900
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 9.21541021374632,
|
| 2438 |
+
"grad_norm": 0.2011975795030594,
|
| 2439 |
+
"learning_rate": 1.5068083022223346e-07,
|
| 2440 |
+
"loss": 0.1855,
|
| 2441 |
+
"step": 24000
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 9.21541021374632,
|
| 2445 |
+
"eval_valid_loss": 0.17754687368869781,
|
| 2446 |
+
"eval_valid_runtime": 4.6668,
|
| 2447 |
+
"eval_valid_samples_per_second": 214.279,
|
| 2448 |
+
"eval_valid_steps_per_second": 6.857,
|
| 2449 |
+
"step": 24000
|
| 2450 |
+
},
|
| 2451 |
+
{
|
| 2452 |
+
"epoch": 9.21541021374632,
|
| 2453 |
+
"eval_valid_target_loss": 0.20250000059604645,
|
| 2454 |
+
"eval_valid_target_runtime": 4.6766,
|
| 2455 |
+
"eval_valid_target_samples_per_second": 213.829,
|
| 2456 |
+
"eval_valid_target_steps_per_second": 6.843,
|
| 2457 |
+
"step": 24000
|
| 2458 |
+
},
|
| 2459 |
+
{
|
| 2460 |
+
"epoch": 9.253807756303596,
|
| 2461 |
+
"grad_norm": 0.2087700515985489,
|
| 2462 |
+
"learning_rate": 1.3633614245357807e-07,
|
| 2463 |
+
"loss": 0.1858,
|
| 2464 |
+
"step": 24100
|
| 2465 |
+
},
|
| 2466 |
+
{
|
| 2467 |
+
"epoch": 9.292205298860873,
|
| 2468 |
+
"grad_norm": 0.18402153253555298,
|
| 2469 |
+
"learning_rate": 1.2269941583481548e-07,
|
| 2470 |
+
"loss": 0.1859,
|
| 2471 |
+
"step": 24200
|
| 2472 |
+
},
|
| 2473 |
+
{
|
| 2474 |
+
"epoch": 9.330602841418148,
|
| 2475 |
+
"grad_norm": 0.17724697291851044,
|
| 2476 |
+
"learning_rate": 1.0977263534536597e-07,
|
| 2477 |
+
"loss": 0.1856,
|
| 2478 |
+
"step": 24300
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 9.369000383975425,
|
| 2482 |
+
"grad_norm": 0.1847800761461258,
|
| 2483 |
+
"learning_rate": 9.755768262397936e-08,
|
| 2484 |
+
"loss": 0.1858,
|
| 2485 |
+
"step": 24400
|
| 2486 |
+
},
|
| 2487 |
+
{
|
| 2488 |
+
"epoch": 9.407397926532703,
|
| 2489 |
+
"grad_norm": 0.1905263364315033,
|
| 2490 |
+
"learning_rate": 8.605633569484184e-08,
|
| 2491 |
+
"loss": 0.1856,
|
| 2492 |
+
"step": 24500
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 9.407397926532703,
|
| 2496 |
+
"eval_valid_loss": 0.1775546818971634,
|
| 2497 |
+
"eval_valid_runtime": 4.6591,
|
| 2498 |
+
"eval_valid_samples_per_second": 214.636,
|
| 2499 |
+
"eval_valid_steps_per_second": 6.868,
|
| 2500 |
+
"step": 24500
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 9.407397926532703,
|
| 2504 |
+
"eval_valid_target_loss": 0.2024531215429306,
|
| 2505 |
+
"eval_valid_target_runtime": 4.6763,
|
| 2506 |
+
"eval_valid_target_samples_per_second": 213.844,
|
| 2507 |
+
"eval_valid_target_steps_per_second": 6.843,
|
| 2508 |
+
"step": 24500
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 9.445795469089978,
|
| 2512 |
+
"grad_norm": 0.17600856721401215,
|
| 2513 |
+
"learning_rate": 7.52702687087653e-08,
|
| 2514 |
+
"loss": 0.1855,
|
| 2515 |
+
"step": 24600
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 9.484193011647255,
|
| 2519 |
+
"grad_norm": 0.19071801006793976,
|
| 2520 |
+
"learning_rate": 6.520105169949609e-08,
|
| 2521 |
+
"loss": 0.1856,
|
| 2522 |
+
"step": 24700
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 9.52259055420453,
|
| 2526 |
+
"grad_norm": 0.20268982648849487,
|
| 2527 |
+
"learning_rate": 5.5850150355178936e-08,
|
| 2528 |
+
"loss": 0.1855,
|
| 2529 |
+
"step": 24800
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 9.560988096761807,
|
| 2533 |
+
"grad_norm": 0.18069659173488617,
|
| 2534 |
+
"learning_rate": 4.721892580500709e-08,
|
| 2535 |
+
"loss": 0.1852,
|
| 2536 |
+
"step": 24900
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 9.599385639319083,
|
| 2540 |
+
"grad_norm": 0.19809788465499878,
|
| 2541 |
+
"learning_rate": 3.9308634421098e-08,
|
| 2542 |
+
"loss": 0.1853,
|
| 2543 |
+
"step": 25000
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 9.599385639319083,
|
| 2547 |
+
"eval_valid_loss": 0.17754687368869781,
|
| 2548 |
+
"eval_valid_runtime": 4.6689,
|
| 2549 |
+
"eval_valid_samples_per_second": 214.182,
|
| 2550 |
+
"eval_valid_steps_per_second": 6.854,
|
| 2551 |
+
"step": 25000
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 9.599385639319083,
|
| 2555 |
+
"eval_valid_target_loss": 0.20237499475479126,
|
| 2556 |
+
"eval_valid_target_runtime": 4.688,
|
| 2557 |
+
"eval_valid_target_samples_per_second": 213.313,
|
| 2558 |
+
"eval_valid_target_steps_per_second": 6.826,
|
| 2559 |
+
"step": 25000
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 9.63778318187636,
|
| 2563 |
+
"grad_norm": 0.1990041732788086,
|
| 2564 |
+
"learning_rate": 3.2120427635613517e-08,
|
| 2565 |
+
"loss": 0.1852,
|
| 2566 |
+
"step": 25100
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 9.676180724433635,
|
| 2570 |
+
"grad_norm": 0.20578785240650177,
|
| 2571 |
+
"learning_rate": 2.565535177315226e-08,
|
| 2572 |
+
"loss": 0.185,
|
| 2573 |
+
"step": 25200
|
| 2574 |
+
},
|
| 2575 |
+
{
|
| 2576 |
+
"epoch": 9.714578266990912,
|
| 2577 |
+
"grad_norm": 0.19831426441669464,
|
| 2578 |
+
"learning_rate": 1.991434789845037e-08,
|
| 2579 |
+
"loss": 0.1858,
|
| 2580 |
+
"step": 25300
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 9.75297580954819,
|
| 2584 |
+
"grad_norm": 0.18692290782928467,
|
| 2585 |
+
"learning_rate": 1.489825167939607e-08,
|
| 2586 |
+
"loss": 0.1848,
|
| 2587 |
+
"step": 25400
|
| 2588 |
+
},
|
| 2589 |
+
{
|
| 2590 |
+
"epoch": 9.791373352105465,
|
| 2591 |
+
"grad_norm": 0.20175856351852417,
|
| 2592 |
+
"learning_rate": 1.0607793265389742e-08,
|
| 2593 |
+
"loss": 0.1854,
|
| 2594 |
+
"step": 25500
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 9.791373352105465,
|
| 2598 |
+
"eval_valid_loss": 0.17751562595367432,
|
| 2599 |
+
"eval_valid_runtime": 4.667,
|
| 2600 |
+
"eval_valid_samples_per_second": 214.272,
|
| 2601 |
+
"eval_valid_steps_per_second": 6.857,
|
| 2602 |
+
"step": 25500
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 9.791373352105465,
|
| 2606 |
+
"eval_valid_target_loss": 0.20240625739097595,
|
| 2607 |
+
"eval_valid_target_runtime": 4.6781,
|
| 2608 |
+
"eval_valid_target_samples_per_second": 213.763,
|
| 2609 |
+
"eval_valid_target_steps_per_second": 6.84,
|
| 2610 |
+
"step": 25500
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 9.829770894662742,
|
| 2614 |
+
"grad_norm": 0.20650416612625122,
|
| 2615 |
+
"learning_rate": 7.0435971810606244e-09,
|
| 2616 |
+
"loss": 0.1859,
|
| 2617 |
+
"step": 25600
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 9.868168437220017,
|
| 2621 |
+
"grad_norm": 0.1880464404821396,
|
| 2622 |
+
"learning_rate": 4.206182235363399e-09,
|
| 2623 |
+
"loss": 0.1857,
|
| 2624 |
+
"step": 25700
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 9.906565979777294,
|
| 2628 |
+
"grad_norm": 0.19517436623573303,
|
| 2629 |
+
"learning_rate": 2.095961446056949e-09,
|
| 2630 |
+
"loss": 0.1851,
|
| 2631 |
+
"step": 25800
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 9.94496352233457,
|
| 2635 |
+
"grad_norm": 0.21848323941230774,
|
| 2636 |
+
"learning_rate": 7.132419795868872e-10,
|
| 2637 |
+
"loss": 0.1858,
|
| 2638 |
+
"step": 25900
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 9.983361064891847,
|
| 2642 |
+
"grad_norm": 0.20499403774738312,
|
| 2643 |
+
"learning_rate": 5.82251063713235e-11,
|
| 2644 |
+
"loss": 0.1851,
|
| 2645 |
+
"step": 26000
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 9.983361064891847,
|
| 2649 |
+
"eval_valid_loss": 0.1775234341621399,
|
| 2650 |
+
"eval_valid_runtime": 4.6706,
|
| 2651 |
+
"eval_valid_samples_per_second": 214.106,
|
| 2652 |
+
"eval_valid_steps_per_second": 6.851,
|
| 2653 |
+
"step": 26000
|
| 2654 |
+
},
|
| 2655 |
+
{
|
| 2656 |
+
"epoch": 9.983361064891847,
|
| 2657 |
+
"eval_valid_target_loss": 0.20242968201637268,
|
| 2658 |
+
"eval_valid_target_runtime": 4.679,
|
| 2659 |
+
"eval_valid_target_samples_per_second": 213.721,
|
| 2660 |
+
"eval_valid_target_steps_per_second": 6.839,
|
| 2661 |
+
"step": 26000
|
| 2662 |
+
}
|
| 2663 |
+
],
|
| 2664 |
+
"logging_steps": 100,
|
| 2665 |
+
"max_steps": 26040,
|
| 2666 |
+
"num_input_tokens_seen": 0,
|
| 2667 |
+
"num_train_epochs": 10,
|
| 2668 |
+
"save_steps": 5000,
|
| 2669 |
+
"stateful_callbacks": {
|
| 2670 |
+
"TrainerControl": {
|
| 2671 |
+
"args": {
|
| 2672 |
+
"should_epoch_stop": false,
|
| 2673 |
+
"should_evaluate": false,
|
| 2674 |
+
"should_log": false,
|
| 2675 |
+
"should_save": true,
|
| 2676 |
+
"should_training_stop": true
|
| 2677 |
+
},
|
| 2678 |
+
"attributes": {}
|
| 2679 |
+
}
|
| 2680 |
+
},
|
| 2681 |
+
"total_flos": 2.475781022436819e+19,
|
| 2682 |
+
"train_batch_size": 16,
|
| 2683 |
+
"trial_name": null,
|
| 2684 |
+
"trial_params": null
|
| 2685 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4e1dfeeeadad4bc43c97dcedacc619cfac42a0c0627e5ea03b98f5461236498c
|
| 3 |
+
size 6200
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
zero_to_fp32.py
ADDED
|
@@ -0,0 +1,604 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python
|
| 2 |
+
|
| 3 |
+
# Copyright (c) Microsoft Corporation.
|
| 4 |
+
# SPDX-License-Identifier: Apache-2.0
|
| 5 |
+
|
| 6 |
+
# DeepSpeed Team
|
| 7 |
+
|
| 8 |
+
# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
|
| 9 |
+
# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
|
| 10 |
+
# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
|
| 11 |
+
# application.
|
| 12 |
+
#
|
| 13 |
+
# example: python zero_to_fp32.py . pytorch_model.bin
|
| 14 |
+
|
| 15 |
+
import argparse
|
| 16 |
+
import torch
|
| 17 |
+
import glob
|
| 18 |
+
import math
|
| 19 |
+
import os
|
| 20 |
+
import re
|
| 21 |
+
from collections import OrderedDict
|
| 22 |
+
from dataclasses import dataclass
|
| 23 |
+
|
| 24 |
+
# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
|
| 25 |
+
# DeepSpeed data structures it has to be available in the current python environment.
|
| 26 |
+
from deepspeed.utils import logger
|
| 27 |
+
from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
|
| 28 |
+
FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
|
| 29 |
+
FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
@dataclass
|
| 33 |
+
class zero_model_state:
|
| 34 |
+
buffers: dict()
|
| 35 |
+
param_shapes: dict()
|
| 36 |
+
shared_params: list
|
| 37 |
+
ds_version: int
|
| 38 |
+
frozen_param_shapes: dict()
|
| 39 |
+
frozen_param_fragments: dict()
|
| 40 |
+
|
| 41 |
+
|
| 42 |
+
debug = 0
|
| 43 |
+
|
| 44 |
+
# load to cpu
|
| 45 |
+
device = torch.device('cpu')
|
| 46 |
+
|
| 47 |
+
|
| 48 |
+
def atoi(text):
|
| 49 |
+
return int(text) if text.isdigit() else text
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
def natural_keys(text):
|
| 53 |
+
'''
|
| 54 |
+
alist.sort(key=natural_keys) sorts in human order
|
| 55 |
+
http://nedbatchelder.com/blog/200712/human_sorting.html
|
| 56 |
+
(See Toothy's implementation in the comments)
|
| 57 |
+
'''
|
| 58 |
+
return [atoi(c) for c in re.split(r'(\d+)', text)]
|
| 59 |
+
|
| 60 |
+
|
| 61 |
+
def get_model_state_file(checkpoint_dir, zero_stage):
|
| 62 |
+
if not os.path.isdir(checkpoint_dir):
|
| 63 |
+
raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
|
| 64 |
+
|
| 65 |
+
# there should be only one file
|
| 66 |
+
if zero_stage <= 2:
|
| 67 |
+
file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
|
| 68 |
+
elif zero_stage == 3:
|
| 69 |
+
file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
|
| 70 |
+
|
| 71 |
+
if not os.path.exists(file):
|
| 72 |
+
raise FileNotFoundError(f"can't find model states file at '{file}'")
|
| 73 |
+
|
| 74 |
+
return file
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
def get_checkpoint_files(checkpoint_dir, glob_pattern):
|
| 78 |
+
# XXX: need to test that this simple glob rule works for multi-node setup too
|
| 79 |
+
ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
|
| 80 |
+
|
| 81 |
+
if len(ckpt_files) == 0:
|
| 82 |
+
raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
|
| 83 |
+
|
| 84 |
+
return ckpt_files
|
| 85 |
+
|
| 86 |
+
|
| 87 |
+
def get_optim_files(checkpoint_dir):
|
| 88 |
+
return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
|
| 89 |
+
|
| 90 |
+
|
| 91 |
+
def get_model_state_files(checkpoint_dir):
|
| 92 |
+
return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
|
| 93 |
+
|
| 94 |
+
|
| 95 |
+
def parse_model_states(files):
|
| 96 |
+
zero_model_states = []
|
| 97 |
+
for file in files:
|
| 98 |
+
state_dict = torch.load(file, map_location=device)
|
| 99 |
+
|
| 100 |
+
if BUFFER_NAMES not in state_dict:
|
| 101 |
+
raise ValueError(f"{file} is not a model state checkpoint")
|
| 102 |
+
buffer_names = state_dict[BUFFER_NAMES]
|
| 103 |
+
if debug:
|
| 104 |
+
print("Found buffers:", buffer_names)
|
| 105 |
+
|
| 106 |
+
# recover just the buffers while restoring them to fp32 if they were saved in fp16
|
| 107 |
+
buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
|
| 108 |
+
param_shapes = state_dict[PARAM_SHAPES]
|
| 109 |
+
|
| 110 |
+
# collect parameters that are included in param_shapes
|
| 111 |
+
param_names = []
|
| 112 |
+
for s in param_shapes:
|
| 113 |
+
for name in s.keys():
|
| 114 |
+
param_names.append(name)
|
| 115 |
+
|
| 116 |
+
# update with frozen parameters
|
| 117 |
+
frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
|
| 118 |
+
if frozen_param_shapes is not None:
|
| 119 |
+
if debug:
|
| 120 |
+
print(f"Found frozen_param_shapes: {frozen_param_shapes}")
|
| 121 |
+
param_names += list(frozen_param_shapes.keys())
|
| 122 |
+
|
| 123 |
+
# handle shared params
|
| 124 |
+
shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
|
| 125 |
+
|
| 126 |
+
ds_version = state_dict.get(DS_VERSION, None)
|
| 127 |
+
|
| 128 |
+
frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
|
| 129 |
+
|
| 130 |
+
z_model_state = zero_model_state(buffers=buffers,
|
| 131 |
+
param_shapes=param_shapes,
|
| 132 |
+
shared_params=shared_params,
|
| 133 |
+
ds_version=ds_version,
|
| 134 |
+
frozen_param_shapes=frozen_param_shapes,
|
| 135 |
+
frozen_param_fragments=frozen_param_fragments)
|
| 136 |
+
zero_model_states.append(z_model_state)
|
| 137 |
+
|
| 138 |
+
return zero_model_states
|
| 139 |
+
|
| 140 |
+
|
| 141 |
+
def parse_optim_states(files, ds_checkpoint_dir):
|
| 142 |
+
|
| 143 |
+
total_files = len(files)
|
| 144 |
+
state_dicts = []
|
| 145 |
+
for f in files:
|
| 146 |
+
state_dict = torch.load(f, map_location=device)
|
| 147 |
+
# immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
|
| 148 |
+
# and also handle the case where it was already removed by another helper script
|
| 149 |
+
state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
|
| 150 |
+
state_dicts.append(state_dict)
|
| 151 |
+
|
| 152 |
+
if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
|
| 153 |
+
raise ValueError(f"{files[0]} is not a zero checkpoint")
|
| 154 |
+
zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
|
| 155 |
+
world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
|
| 156 |
+
|
| 157 |
+
# For ZeRO-2 each param group can have different partition_count as data parallelism for expert
|
| 158 |
+
# parameters can be different from data parallelism for non-expert parameters. So we can just
|
| 159 |
+
# use the max of the partition_count to get the dp world_size.
|
| 160 |
+
|
| 161 |
+
if type(world_size) is list:
|
| 162 |
+
world_size = max(world_size)
|
| 163 |
+
|
| 164 |
+
if world_size != total_files:
|
| 165 |
+
raise ValueError(
|
| 166 |
+
f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
|
| 167 |
+
"Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
|
| 168 |
+
)
|
| 169 |
+
|
| 170 |
+
# the groups are named differently in each stage
|
| 171 |
+
if zero_stage <= 2:
|
| 172 |
+
fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
|
| 173 |
+
elif zero_stage == 3:
|
| 174 |
+
fp32_groups_key = FP32_FLAT_GROUPS
|
| 175 |
+
else:
|
| 176 |
+
raise ValueError(f"unknown zero stage {zero_stage}")
|
| 177 |
+
|
| 178 |
+
if zero_stage <= 2:
|
| 179 |
+
fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
|
| 180 |
+
elif zero_stage == 3:
|
| 181 |
+
# if there is more than one param group, there will be multiple flattened tensors - one
|
| 182 |
+
# flattened tensor per group - for simplicity merge them into a single tensor
|
| 183 |
+
#
|
| 184 |
+
# XXX: could make the script more memory efficient for when there are multiple groups - it
|
| 185 |
+
# will require matching the sub-lists of param_shapes for each param group flattened tensor
|
| 186 |
+
|
| 187 |
+
fp32_flat_groups = [
|
| 188 |
+
torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
|
| 189 |
+
]
|
| 190 |
+
|
| 191 |
+
return zero_stage, world_size, fp32_flat_groups
|
| 192 |
+
|
| 193 |
+
|
| 194 |
+
def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
|
| 195 |
+
"""
|
| 196 |
+
Returns fp32 state_dict reconstructed from ds checkpoint
|
| 197 |
+
|
| 198 |
+
Args:
|
| 199 |
+
- ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
|
| 200 |
+
|
| 201 |
+
"""
|
| 202 |
+
print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
|
| 203 |
+
|
| 204 |
+
optim_files = get_optim_files(ds_checkpoint_dir)
|
| 205 |
+
zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
|
| 206 |
+
print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
|
| 207 |
+
|
| 208 |
+
model_files = get_model_state_files(ds_checkpoint_dir)
|
| 209 |
+
|
| 210 |
+
zero_model_states = parse_model_states(model_files)
|
| 211 |
+
print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
|
| 212 |
+
|
| 213 |
+
if zero_stage <= 2:
|
| 214 |
+
return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 215 |
+
exclude_frozen_parameters)
|
| 216 |
+
elif zero_stage == 3:
|
| 217 |
+
return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 218 |
+
exclude_frozen_parameters)
|
| 219 |
+
|
| 220 |
+
|
| 221 |
+
def _zero2_merge_frozen_params(state_dict, zero_model_states):
|
| 222 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 223 |
+
return
|
| 224 |
+
|
| 225 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 226 |
+
frozen_param_fragments = zero_model_states[0].frozen_param_fragments
|
| 227 |
+
|
| 228 |
+
if debug:
|
| 229 |
+
num_elem = sum(s.numel() for s in frozen_param_shapes.values())
|
| 230 |
+
print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 231 |
+
|
| 232 |
+
wanted_params = len(frozen_param_shapes)
|
| 233 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 234 |
+
avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
|
| 235 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 236 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 237 |
+
|
| 238 |
+
total_params = 0
|
| 239 |
+
total_numel = 0
|
| 240 |
+
for name, shape in frozen_param_shapes.items():
|
| 241 |
+
total_params += 1
|
| 242 |
+
unpartitioned_numel = shape.numel()
|
| 243 |
+
total_numel += unpartitioned_numel
|
| 244 |
+
|
| 245 |
+
state_dict[name] = frozen_param_fragments[name]
|
| 246 |
+
|
| 247 |
+
if debug:
|
| 248 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 249 |
+
|
| 250 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 251 |
+
|
| 252 |
+
|
| 253 |
+
def _has_callable(obj, fn):
|
| 254 |
+
attr = getattr(obj, fn, None)
|
| 255 |
+
return callable(attr)
|
| 256 |
+
|
| 257 |
+
|
| 258 |
+
def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 259 |
+
param_shapes = zero_model_states[0].param_shapes
|
| 260 |
+
|
| 261 |
+
# Reconstruction protocol:
|
| 262 |
+
#
|
| 263 |
+
# XXX: document this
|
| 264 |
+
|
| 265 |
+
if debug:
|
| 266 |
+
for i in range(world_size):
|
| 267 |
+
for j in range(len(fp32_flat_groups[0])):
|
| 268 |
+
print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
|
| 269 |
+
|
| 270 |
+
# XXX: memory usage doubles here (zero2)
|
| 271 |
+
num_param_groups = len(fp32_flat_groups[0])
|
| 272 |
+
merged_single_partition_of_fp32_groups = []
|
| 273 |
+
for i in range(num_param_groups):
|
| 274 |
+
merged_partitions = [sd[i] for sd in fp32_flat_groups]
|
| 275 |
+
full_single_fp32_vector = torch.cat(merged_partitions, 0)
|
| 276 |
+
merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
|
| 277 |
+
avail_numel = sum(
|
| 278 |
+
[full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
|
| 279 |
+
|
| 280 |
+
if debug:
|
| 281 |
+
wanted_params = sum([len(shapes) for shapes in param_shapes])
|
| 282 |
+
wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
|
| 283 |
+
# not asserting if there is a mismatch due to possible padding
|
| 284 |
+
print(f"Have {avail_numel} numels to process.")
|
| 285 |
+
print(f"Need {wanted_numel} numels in {wanted_params} params.")
|
| 286 |
+
|
| 287 |
+
# params
|
| 288 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 289 |
+
# out-of-core computing solution
|
| 290 |
+
total_numel = 0
|
| 291 |
+
total_params = 0
|
| 292 |
+
for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
|
| 293 |
+
offset = 0
|
| 294 |
+
avail_numel = full_single_fp32_vector.numel()
|
| 295 |
+
for name, shape in shapes.items():
|
| 296 |
+
|
| 297 |
+
unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
|
| 298 |
+
total_numel += unpartitioned_numel
|
| 299 |
+
total_params += 1
|
| 300 |
+
|
| 301 |
+
if debug:
|
| 302 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 303 |
+
state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
|
| 304 |
+
offset += unpartitioned_numel
|
| 305 |
+
|
| 306 |
+
# Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
|
| 307 |
+
# avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
|
| 308 |
+
# paddings performed in the code it's almost impossible to predict the exact numbers w/o the
|
| 309 |
+
# live optimizer object, so we are checking that the numbers are within the right range
|
| 310 |
+
align_to = 2 * world_size
|
| 311 |
+
|
| 312 |
+
def zero2_align(x):
|
| 313 |
+
return align_to * math.ceil(x / align_to)
|
| 314 |
+
|
| 315 |
+
if debug:
|
| 316 |
+
print(f"original offset={offset}, avail_numel={avail_numel}")
|
| 317 |
+
|
| 318 |
+
offset = zero2_align(offset)
|
| 319 |
+
avail_numel = zero2_align(avail_numel)
|
| 320 |
+
|
| 321 |
+
if debug:
|
| 322 |
+
print(f"aligned offset={offset}, avail_numel={avail_numel}")
|
| 323 |
+
|
| 324 |
+
# Sanity check
|
| 325 |
+
if offset != avail_numel:
|
| 326 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 327 |
+
|
| 328 |
+
print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
|
| 329 |
+
|
| 330 |
+
|
| 331 |
+
def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 332 |
+
exclude_frozen_parameters):
|
| 333 |
+
state_dict = OrderedDict()
|
| 334 |
+
|
| 335 |
+
# buffers
|
| 336 |
+
buffers = zero_model_states[0].buffers
|
| 337 |
+
state_dict.update(buffers)
|
| 338 |
+
if debug:
|
| 339 |
+
print(f"added {len(buffers)} buffers")
|
| 340 |
+
|
| 341 |
+
if not exclude_frozen_parameters:
|
| 342 |
+
_zero2_merge_frozen_params(state_dict, zero_model_states)
|
| 343 |
+
|
| 344 |
+
_zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 345 |
+
|
| 346 |
+
# recover shared parameters
|
| 347 |
+
for pair in zero_model_states[0].shared_params:
|
| 348 |
+
if pair[1] in state_dict:
|
| 349 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
| 350 |
+
|
| 351 |
+
return state_dict
|
| 352 |
+
|
| 353 |
+
|
| 354 |
+
def zero3_partitioned_param_info(unpartitioned_numel, world_size):
|
| 355 |
+
remainder = unpartitioned_numel % world_size
|
| 356 |
+
padding_numel = (world_size - remainder) if remainder else 0
|
| 357 |
+
partitioned_numel = math.ceil(unpartitioned_numel / world_size)
|
| 358 |
+
return partitioned_numel, padding_numel
|
| 359 |
+
|
| 360 |
+
|
| 361 |
+
def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
|
| 362 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 363 |
+
return
|
| 364 |
+
|
| 365 |
+
if debug:
|
| 366 |
+
for i in range(world_size):
|
| 367 |
+
num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
|
| 368 |
+
print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 369 |
+
|
| 370 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 371 |
+
wanted_params = len(frozen_param_shapes)
|
| 372 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 373 |
+
avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
|
| 374 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 375 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 376 |
+
|
| 377 |
+
total_params = 0
|
| 378 |
+
total_numel = 0
|
| 379 |
+
for name, shape in zero_model_states[0].frozen_param_shapes.items():
|
| 380 |
+
total_params += 1
|
| 381 |
+
unpartitioned_numel = shape.numel()
|
| 382 |
+
total_numel += unpartitioned_numel
|
| 383 |
+
|
| 384 |
+
param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
|
| 385 |
+
state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
|
| 386 |
+
|
| 387 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 388 |
+
|
| 389 |
+
if debug:
|
| 390 |
+
print(
|
| 391 |
+
f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 392 |
+
)
|
| 393 |
+
|
| 394 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 395 |
+
|
| 396 |
+
|
| 397 |
+
def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 398 |
+
param_shapes = zero_model_states[0].param_shapes
|
| 399 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
| 400 |
+
# Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
|
| 401 |
+
# param, re-consolidating each param, while dealing with padding if any
|
| 402 |
+
|
| 403 |
+
# merge list of dicts, preserving order
|
| 404 |
+
param_shapes = {k: v for d in param_shapes for k, v in d.items()}
|
| 405 |
+
|
| 406 |
+
if debug:
|
| 407 |
+
for i in range(world_size):
|
| 408 |
+
print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
|
| 409 |
+
|
| 410 |
+
wanted_params = len(param_shapes)
|
| 411 |
+
wanted_numel = sum(shape.numel() for shape in param_shapes.values())
|
| 412 |
+
# not asserting if there is a mismatch due to possible padding
|
| 413 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
| 414 |
+
print(f"Trainable params: Have {avail_numel} numels to process.")
|
| 415 |
+
print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
|
| 416 |
+
|
| 417 |
+
# params
|
| 418 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 419 |
+
# out-of-core computing solution
|
| 420 |
+
offset = 0
|
| 421 |
+
total_numel = 0
|
| 422 |
+
total_params = 0
|
| 423 |
+
for name, shape in param_shapes.items():
|
| 424 |
+
|
| 425 |
+
unpartitioned_numel = shape.numel()
|
| 426 |
+
total_numel += unpartitioned_numel
|
| 427 |
+
total_params += 1
|
| 428 |
+
|
| 429 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 430 |
+
|
| 431 |
+
if debug:
|
| 432 |
+
print(
|
| 433 |
+
f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 434 |
+
)
|
| 435 |
+
|
| 436 |
+
# XXX: memory usage doubles here
|
| 437 |
+
state_dict[name] = torch.cat(
|
| 438 |
+
tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
|
| 439 |
+
0).narrow(0, 0, unpartitioned_numel).view(shape)
|
| 440 |
+
offset += partitioned_numel
|
| 441 |
+
|
| 442 |
+
offset *= world_size
|
| 443 |
+
|
| 444 |
+
# Sanity check
|
| 445 |
+
if offset != avail_numel:
|
| 446 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 447 |
+
|
| 448 |
+
print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
|
| 449 |
+
|
| 450 |
+
|
| 451 |
+
def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 452 |
+
exclude_frozen_parameters):
|
| 453 |
+
state_dict = OrderedDict()
|
| 454 |
+
|
| 455 |
+
# buffers
|
| 456 |
+
buffers = zero_model_states[0].buffers
|
| 457 |
+
state_dict.update(buffers)
|
| 458 |
+
if debug:
|
| 459 |
+
print(f"added {len(buffers)} buffers")
|
| 460 |
+
|
| 461 |
+
if not exclude_frozen_parameters:
|
| 462 |
+
_zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
|
| 463 |
+
|
| 464 |
+
_zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 465 |
+
|
| 466 |
+
# recover shared parameters
|
| 467 |
+
for pair in zero_model_states[0].shared_params:
|
| 468 |
+
if pair[1] in state_dict:
|
| 469 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
| 470 |
+
|
| 471 |
+
return state_dict
|
| 472 |
+
|
| 473 |
+
|
| 474 |
+
def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
|
| 475 |
+
"""
|
| 476 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
|
| 477 |
+
``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
|
| 478 |
+
via a model hub.
|
| 479 |
+
|
| 480 |
+
Args:
|
| 481 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder
|
| 482 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
|
| 483 |
+
- ``exclude_frozen_parameters``: exclude frozen parameters
|
| 484 |
+
|
| 485 |
+
Returns:
|
| 486 |
+
- pytorch ``state_dict``
|
| 487 |
+
|
| 488 |
+
Note: this approach may not work if your application doesn't have sufficient free CPU memory and
|
| 489 |
+
you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
|
| 490 |
+
the checkpoint.
|
| 491 |
+
|
| 492 |
+
A typical usage might be ::
|
| 493 |
+
|
| 494 |
+
from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
|
| 495 |
+
# do the training and checkpoint saving
|
| 496 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
|
| 497 |
+
model = model.cpu() # move to cpu
|
| 498 |
+
model.load_state_dict(state_dict)
|
| 499 |
+
# submit to model hub or save the model to share with others
|
| 500 |
+
|
| 501 |
+
In this example the ``model`` will no longer be usable in the deepspeed context of the same
|
| 502 |
+
application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 503 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 504 |
+
|
| 505 |
+
If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
|
| 506 |
+
|
| 507 |
+
"""
|
| 508 |
+
if tag is None:
|
| 509 |
+
latest_path = os.path.join(checkpoint_dir, 'latest')
|
| 510 |
+
if os.path.isfile(latest_path):
|
| 511 |
+
with open(latest_path, 'r') as fd:
|
| 512 |
+
tag = fd.read().strip()
|
| 513 |
+
else:
|
| 514 |
+
raise ValueError(f"Unable to find 'latest' file at {latest_path}")
|
| 515 |
+
|
| 516 |
+
ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
|
| 517 |
+
|
| 518 |
+
if not os.path.isdir(ds_checkpoint_dir):
|
| 519 |
+
raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
|
| 520 |
+
|
| 521 |
+
return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
|
| 522 |
+
|
| 523 |
+
|
| 524 |
+
def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False):
|
| 525 |
+
"""
|
| 526 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
|
| 527 |
+
loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
|
| 528 |
+
|
| 529 |
+
Args:
|
| 530 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 531 |
+
- ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
|
| 532 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
|
| 533 |
+
- ``exclude_frozen_parameters``: exclude frozen parameters
|
| 534 |
+
"""
|
| 535 |
+
|
| 536 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
|
| 537 |
+
print(f"Saving fp32 state dict to {output_file}")
|
| 538 |
+
torch.save(state_dict, output_file)
|
| 539 |
+
|
| 540 |
+
|
| 541 |
+
def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
|
| 542 |
+
"""
|
| 543 |
+
1. Put the provided model to cpu
|
| 544 |
+
2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
|
| 545 |
+
3. Load it into the provided model
|
| 546 |
+
|
| 547 |
+
Args:
|
| 548 |
+
- ``model``: the model object to update
|
| 549 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 550 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
|
| 551 |
+
|
| 552 |
+
Returns:
|
| 553 |
+
- ``model`: modified model
|
| 554 |
+
|
| 555 |
+
Make sure you have plenty of CPU memory available before you call this function. If you don't
|
| 556 |
+
have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
|
| 557 |
+
conveniently placed for you in the checkpoint folder.
|
| 558 |
+
|
| 559 |
+
A typical usage might be ::
|
| 560 |
+
|
| 561 |
+
from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
|
| 562 |
+
model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
|
| 563 |
+
# submit to model hub or save the model to share with others
|
| 564 |
+
|
| 565 |
+
Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
|
| 566 |
+
of the same application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 567 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 568 |
+
|
| 569 |
+
"""
|
| 570 |
+
logger.info(f"Extracting fp32 weights")
|
| 571 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
|
| 572 |
+
|
| 573 |
+
logger.info(f"Overwriting model with fp32 weights")
|
| 574 |
+
model = model.cpu()
|
| 575 |
+
model.load_state_dict(state_dict, strict=False)
|
| 576 |
+
|
| 577 |
+
return model
|
| 578 |
+
|
| 579 |
+
|
| 580 |
+
if __name__ == "__main__":
|
| 581 |
+
|
| 582 |
+
parser = argparse.ArgumentParser()
|
| 583 |
+
parser.add_argument("checkpoint_dir",
|
| 584 |
+
type=str,
|
| 585 |
+
help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
|
| 586 |
+
parser.add_argument(
|
| 587 |
+
"output_file",
|
| 588 |
+
type=str,
|
| 589 |
+
help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
|
| 590 |
+
parser.add_argument("-t",
|
| 591 |
+
"--tag",
|
| 592 |
+
type=str,
|
| 593 |
+
default=None,
|
| 594 |
+
help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
|
| 595 |
+
parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
|
| 596 |
+
parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
|
| 597 |
+
args = parser.parse_args()
|
| 598 |
+
|
| 599 |
+
debug = args.debug
|
| 600 |
+
|
| 601 |
+
convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
|
| 602 |
+
args.output_file,
|
| 603 |
+
tag=args.tag,
|
| 604 |
+
exclude_frozen_parameters=args.exclude_frozen_parameters)
|