Hugging Face's logo

Spaces:
Darwinkel
/
gemma-2b-sciq
Runtime error

App Files Community
Darwinkel commited on
Commit
ead8650
·
verified ·
1 Parent(s): 42b5a8f

Upload 10 files

Browse files
Files changed (11) hide show
  1. .gitattributes +1 -0
  2. adapter_config.json +34 -0
  3. adapter_model.safetensors +3 -0
  4. optimizer.pt +3 -0
  5. rng_state.pth +3 -0
  6. scheduler.pt +3 -0
  7. special_tokens_map.json +30 -0
  8. tokenizer.json +3 -0
  9. tokenizer_config.json +49 -0
  10. trainer_state.json +3959 -0
  11. training_args.bin +3 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
adapter_config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "google/gemma-2b",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 8,
14
+ "lora_dropout": 0.0,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 8,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "o_proj",
24
+ "v_proj",
25
+ "down_proj",
26
+ "gate_proj",
27
+ "k_proj",
28
+ "up_proj",
29
+ "q_proj"
30
+ ],
31
+ "task_type": "CAUSAL_LM",
32
+ "use_dora": false,
33
+ "use_rslora": false
34
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0738c2f76374d5c2940d4eee7edd0dbf6608429d22cc3a3fbecea3363892d4a3
3
+ size 39256456
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e32ebe7063731b420de607193a84ef7cfca5475808adf96c5301afab4bd7ae35
3
+ size 78658234
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7682299c566684ea51cf26f0c86b6ffaa3c0bc63cbdf84674b29a2c62ac72143
3
+ size 14244
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d442cf640e1171d85a5eff33c5234f9fb4933557b9265257a38ac8b0a913431
3
+ size 1064
special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<bos>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<eos>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "<pad>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "<unk>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cb7e592b7314175501c7fc56b904d581de569169ac90e5aa2ec11a860c2cbbaa
3
+ size 17477652
tokenizer_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "0": {
6
+ "content": "<pad>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "1": {
14
+ "content": "<eos>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "2": {
22
+ "content": "<bos>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "3": {
30
+ "content": "<unk>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ }
37
+ },
38
+ "bos_token": "<bos>",
39
+ "clean_up_tokenization_spaces": false,
40
+ "eos_token": "<eos>",
41
+ "legacy": null,
42
+ "model_max_length": 1000000000000000019884624838656,
43
+ "pad_token": "<pad>",
44
+ "sp_model_kwargs": {},
45
+ "spaces_between_special_tokens": false,
46
+ "tokenizer_class": "GemmaTokenizer",
47
+ "unk_token": "<unk>",
48
+ "use_default_system_prompt": false
49
+ }
trainer_state.json ADDED
@@ -0,0 +1,3959 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.8837229214830038,
5
+ "eval_steps": 500,
6
+ "global_step": 5500,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0,
13
+ "grad_norm": 0.5656424760818481,
14
+ "learning_rate": 0.00019972583961617548,
15
+ "loss": 2.6115,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.01,
20
+ "grad_norm": 0.9568225145339966,
21
+ "learning_rate": 0.0001993831391363948,
22
+ "loss": 2.3674,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.01,
27
+ "grad_norm": 0.8508038520812988,
28
+ "learning_rate": 0.00019904043865661412,
29
+ "loss": 2.1487,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.01,
34
+ "grad_norm": 1.0186710357666016,
35
+ "learning_rate": 0.00019869773817683345,
36
+ "loss": 2.1421,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.02,
41
+ "grad_norm": 1.7234398126602173,
42
+ "learning_rate": 0.0001983550376970528,
43
+ "loss": 2.0021,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.02,
48
+ "grad_norm": 1.197520136833191,
49
+ "learning_rate": 0.0001980123372172721,
50
+ "loss": 2.0381,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.02,
55
+ "grad_norm": 1.0739991664886475,
56
+ "learning_rate": 0.00019766963673749144,
57
+ "loss": 1.8883,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.03,
62
+ "grad_norm": 1.0150132179260254,
63
+ "learning_rate": 0.00019732693625771076,
64
+ "loss": 1.8494,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.03,
69
+ "grad_norm": 1.236234426498413,
70
+ "learning_rate": 0.0001969842357779301,
71
+ "loss": 1.9171,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.03,
76
+ "grad_norm": 1.0886958837509155,
77
+ "learning_rate": 0.00019664153529814942,
78
+ "loss": 1.9387,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.04,
83
+ "grad_norm": 1.1191097497940063,
84
+ "learning_rate": 0.00019629883481836875,
85
+ "loss": 1.976,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.04,
90
+ "grad_norm": 1.0738675594329834,
91
+ "learning_rate": 0.00019595613433858808,
92
+ "loss": 1.8378,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.04,
97
+ "grad_norm": 0.648668646812439,
98
+ "learning_rate": 0.0001956134338588074,
99
+ "loss": 1.768,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.05,
104
+ "grad_norm": 0.9386289119720459,
105
+ "learning_rate": 0.00019527073337902674,
106
+ "loss": 2.0067,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.05,
111
+ "grad_norm": 1.1613832712173462,
112
+ "learning_rate": 0.00019492803289924607,
113
+ "loss": 1.9355,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.05,
118
+ "grad_norm": 0.7319044470787048,
119
+ "learning_rate": 0.0001945853324194654,
120
+ "loss": 1.9042,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.06,
125
+ "grad_norm": 0.9041644930839539,
126
+ "learning_rate": 0.00019424263193968473,
127
+ "loss": 1.7163,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.06,
132
+ "grad_norm": 0.9293299317359924,
133
+ "learning_rate": 0.00019389993145990406,
134
+ "loss": 1.807,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.07,
139
+ "grad_norm": 0.9214122295379639,
140
+ "learning_rate": 0.00019355723098012336,
141
+ "loss": 1.9056,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.07,
146
+ "grad_norm": 0.7177646160125732,
147
+ "learning_rate": 0.0001932145305003427,
148
+ "loss": 1.9574,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.07,
153
+ "grad_norm": 0.813965916633606,
154
+ "learning_rate": 0.00019287183002056205,
155
+ "loss": 1.7995,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.08,
160
+ "grad_norm": 1.0333760976791382,
161
+ "learning_rate": 0.00019252912954078138,
162
+ "loss": 1.814,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.08,
167
+ "grad_norm": 0.6691217422485352,
168
+ "learning_rate": 0.0001921864290610007,
169
+ "loss": 1.8261,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.08,
174
+ "grad_norm": 1.1737751960754395,
175
+ "learning_rate": 0.00019184372858122,
176
+ "loss": 1.9473,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.09,
181
+ "grad_norm": 1.1508344411849976,
182
+ "learning_rate": 0.00019150102810143934,
183
+ "loss": 1.9176,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.09,
188
+ "grad_norm": 0.6660133600234985,
189
+ "learning_rate": 0.00019115832762165867,
190
+ "loss": 1.835,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.09,
195
+ "grad_norm": 0.6423531174659729,
196
+ "learning_rate": 0.00019081562714187803,
197
+ "loss": 1.7194,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.1,
202
+ "grad_norm": 0.8241636157035828,
203
+ "learning_rate": 0.00019047292666209733,
204
+ "loss": 1.8679,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.1,
209
+ "grad_norm": 0.7184795141220093,
210
+ "learning_rate": 0.00019013022618231666,
211
+ "loss": 1.8129,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.1,
216
+ "grad_norm": 0.8253782391548157,
217
+ "learning_rate": 0.000189787525702536,
218
+ "loss": 1.8567,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.11,
223
+ "grad_norm": 1.417243242263794,
224
+ "learning_rate": 0.00018944482522275532,
225
+ "loss": 1.7741,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.11,
230
+ "grad_norm": 0.9040454626083374,
231
+ "learning_rate": 0.00018910212474297465,
232
+ "loss": 1.8458,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.11,
237
+ "grad_norm": 0.6580069065093994,
238
+ "learning_rate": 0.00018875942426319398,
239
+ "loss": 1.7982,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.12,
244
+ "grad_norm": 0.8849833011627197,
245
+ "learning_rate": 0.0001884167237834133,
246
+ "loss": 1.8622,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.12,
251
+ "grad_norm": 1.0523239374160767,
252
+ "learning_rate": 0.00018807402330363264,
253
+ "loss": 1.8608,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.12,
258
+ "grad_norm": 1.0496423244476318,
259
+ "learning_rate": 0.00018773132282385194,
260
+ "loss": 1.8245,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.13,
265
+ "grad_norm": 0.9488272070884705,
266
+ "learning_rate": 0.00018738862234407127,
267
+ "loss": 1.8933,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.13,
272
+ "grad_norm": 0.9461072087287903,
273
+ "learning_rate": 0.00018704592186429063,
274
+ "loss": 1.7277,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.13,
279
+ "grad_norm": 0.6415026187896729,
280
+ "learning_rate": 0.00018670322138450996,
281
+ "loss": 1.7843,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.14,
286
+ "grad_norm": 1.0457078218460083,
287
+ "learning_rate": 0.0001863605209047293,
288
+ "loss": 1.8874,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.14,
293
+ "grad_norm": 1.0890721082687378,
294
+ "learning_rate": 0.0001860178204249486,
295
+ "loss": 1.8536,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.14,
300
+ "grad_norm": 0.8896569013595581,
301
+ "learning_rate": 0.00018567511994516792,
302
+ "loss": 1.8297,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.15,
307
+ "grad_norm": 0.9457584023475647,
308
+ "learning_rate": 0.00018533241946538728,
309
+ "loss": 1.8061,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.15,
314
+ "grad_norm": 0.8208130598068237,
315
+ "learning_rate": 0.0001849897189856066,
316
+ "loss": 1.8238,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.15,
321
+ "grad_norm": 0.7884149551391602,
322
+ "learning_rate": 0.0001846470185058259,
323
+ "loss": 1.7419,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.16,
328
+ "grad_norm": 1.5733205080032349,
329
+ "learning_rate": 0.00018430431802604524,
330
+ "loss": 1.8829,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.16,
335
+ "grad_norm": 0.963455319404602,
336
+ "learning_rate": 0.00018396161754626457,
337
+ "loss": 1.822,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.16,
342
+ "grad_norm": 0.616909384727478,
343
+ "learning_rate": 0.0001836189170664839,
344
+ "loss": 1.7923,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.17,
349
+ "grad_norm": 0.5382218360900879,
350
+ "learning_rate": 0.00018327621658670323,
351
+ "loss": 1.719,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.17,
356
+ "grad_norm": 1.171004056930542,
357
+ "learning_rate": 0.00018293351610692256,
358
+ "loss": 1.8522,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 0.17,
363
+ "eval_loss": 1.9394277334213257,
364
+ "eval_runtime": 33.4276,
365
+ "eval_samples_per_second": 29.915,
366
+ "eval_steps_per_second": 3.739,
367
+ "step": 500
368
+ },
369
+ {
370
+ "epoch": 0.17,
371
+ "grad_norm": 0.7731293439865112,
372
+ "learning_rate": 0.0001825908156271419,
373
+ "loss": 1.9151,
374
+ "step": 510
375
+ },
376
+ {
377
+ "epoch": 0.18,
378
+ "grad_norm": 0.8664043545722961,
379
+ "learning_rate": 0.00018224811514736122,
380
+ "loss": 1.6679,
381
+ "step": 520
382
+ },
383
+ {
384
+ "epoch": 0.18,
385
+ "grad_norm": 1.3886076211929321,
386
+ "learning_rate": 0.00018190541466758055,
387
+ "loss": 1.8509,
388
+ "step": 530
389
+ },
390
+ {
391
+ "epoch": 0.18,
392
+ "grad_norm": 0.7000617384910583,
393
+ "learning_rate": 0.00018156271418779988,
394
+ "loss": 1.8046,
395
+ "step": 540
396
+ },
397
+ {
398
+ "epoch": 0.19,
399
+ "grad_norm": 0.8490706086158752,
400
+ "learning_rate": 0.0001812200137080192,
401
+ "loss": 1.748,
402
+ "step": 550
403
+ },
404
+ {
405
+ "epoch": 0.19,
406
+ "grad_norm": 1.4293190240859985,
407
+ "learning_rate": 0.00018087731322823854,
408
+ "loss": 1.9725,
409
+ "step": 560
410
+ },
411
+ {
412
+ "epoch": 0.2,
413
+ "grad_norm": 0.7126957178115845,
414
+ "learning_rate": 0.00018053461274845787,
415
+ "loss": 1.6888,
416
+ "step": 570
417
+ },
418
+ {
419
+ "epoch": 0.2,
420
+ "grad_norm": 0.9974524974822998,
421
+ "learning_rate": 0.00018019191226867717,
422
+ "loss": 1.8405,
423
+ "step": 580
424
+ },
425
+ {
426
+ "epoch": 0.2,
427
+ "grad_norm": 0.9911081790924072,
428
+ "learning_rate": 0.0001798492117888965,
429
+ "loss": 1.7753,
430
+ "step": 590
431
+ },
432
+ {
433
+ "epoch": 0.21,
434
+ "grad_norm": 1.3659840822219849,
435
+ "learning_rate": 0.00017950651130911585,
436
+ "loss": 1.7435,
437
+ "step": 600
438
+ },
439
+ {
440
+ "epoch": 0.21,
441
+ "grad_norm": 0.4976978302001953,
442
+ "learning_rate": 0.00017916381082933518,
443
+ "loss": 1.759,
444
+ "step": 610
445
+ },
446
+ {
447
+ "epoch": 0.21,
448
+ "grad_norm": 0.7868736982345581,
449
+ "learning_rate": 0.0001788211103495545,
450
+ "loss": 1.7654,
451
+ "step": 620
452
+ },
453
+ {
454
+ "epoch": 0.22,
455
+ "grad_norm": 1.006628155708313,
456
+ "learning_rate": 0.00017847840986977382,
457
+ "loss": 1.7862,
458
+ "step": 630
459
+ },
460
+ {
461
+ "epoch": 0.22,
462
+ "grad_norm": 0.8664697408676147,
463
+ "learning_rate": 0.00017813570938999315,
464
+ "loss": 1.8815,
465
+ "step": 640
466
+ },
467
+ {
468
+ "epoch": 0.22,
469
+ "grad_norm": 0.44789645075798035,
470
+ "learning_rate": 0.00017779300891021248,
471
+ "loss": 1.779,
472
+ "step": 650
473
+ },
474
+ {
475
+ "epoch": 0.23,
476
+ "grad_norm": 0.9740760326385498,
477
+ "learning_rate": 0.00017745030843043183,
478
+ "loss": 1.7026,
479
+ "step": 660
480
+ },
481
+ {
482
+ "epoch": 0.23,
483
+ "grad_norm": 0.9802984595298767,
484
+ "learning_rate": 0.00017710760795065114,
485
+ "loss": 1.8359,
486
+ "step": 670
487
+ },
488
+ {
489
+ "epoch": 0.23,
490
+ "grad_norm": 1.0521053075790405,
491
+ "learning_rate": 0.00017676490747087047,
492
+ "loss": 1.7777,
493
+ "step": 680
494
+ },
495
+ {
496
+ "epoch": 0.24,
497
+ "grad_norm": 0.6399825215339661,
498
+ "learning_rate": 0.0001764222069910898,
499
+ "loss": 1.8129,
500
+ "step": 690
501
+ },
502
+ {
503
+ "epoch": 0.24,
504
+ "grad_norm": 1.1847810745239258,
505
+ "learning_rate": 0.00017607950651130912,
506
+ "loss": 1.8775,
507
+ "step": 700
508
+ },
509
+ {
510
+ "epoch": 0.24,
511
+ "grad_norm": 0.7050787806510925,
512
+ "learning_rate": 0.00017573680603152845,
513
+ "loss": 1.8454,
514
+ "step": 710
515
+ },
516
+ {
517
+ "epoch": 0.25,
518
+ "grad_norm": 0.8241177797317505,
519
+ "learning_rate": 0.00017539410555174778,
520
+ "loss": 1.7047,
521
+ "step": 720
522
+ },
523
+ {
524
+ "epoch": 0.25,
525
+ "grad_norm": 1.743680477142334,
526
+ "learning_rate": 0.00017505140507196711,
527
+ "loss": 1.8251,
528
+ "step": 730
529
+ },
530
+ {
531
+ "epoch": 0.25,
532
+ "grad_norm": 0.776196300983429,
533
+ "learning_rate": 0.00017470870459218644,
534
+ "loss": 1.8341,
535
+ "step": 740
536
+ },
537
+ {
538
+ "epoch": 0.26,
539
+ "grad_norm": 0.6896054744720459,
540
+ "learning_rate": 0.00017436600411240575,
541
+ "loss": 1.7569,
542
+ "step": 750
543
+ },
544
+ {
545
+ "epoch": 0.26,
546
+ "grad_norm": 0.703697919845581,
547
+ "learning_rate": 0.0001740233036326251,
548
+ "loss": 1.7696,
549
+ "step": 760
550
+ },
551
+ {
552
+ "epoch": 0.26,
553
+ "grad_norm": 0.6734452247619629,
554
+ "learning_rate": 0.00017368060315284443,
555
+ "loss": 1.6639,
556
+ "step": 770
557
+ },
558
+ {
559
+ "epoch": 0.27,
560
+ "grad_norm": 0.6856238842010498,
561
+ "learning_rate": 0.00017333790267306376,
562
+ "loss": 1.8419,
563
+ "step": 780
564
+ },
565
+ {
566
+ "epoch": 0.27,
567
+ "grad_norm": 1.1194758415222168,
568
+ "learning_rate": 0.00017299520219328306,
569
+ "loss": 1.7916,
570
+ "step": 790
571
+ },
572
+ {
573
+ "epoch": 0.27,
574
+ "grad_norm": 1.455841064453125,
575
+ "learning_rate": 0.0001726525017135024,
576
+ "loss": 1.7368,
577
+ "step": 800
578
+ },
579
+ {
580
+ "epoch": 0.28,
581
+ "grad_norm": 0.5988683700561523,
582
+ "learning_rate": 0.00017230980123372172,
583
+ "loss": 1.8434,
584
+ "step": 810
585
+ },
586
+ {
587
+ "epoch": 0.28,
588
+ "grad_norm": 0.9031710028648376,
589
+ "learning_rate": 0.00017196710075394108,
590
+ "loss": 1.7447,
591
+ "step": 820
592
+ },
593
+ {
594
+ "epoch": 0.28,
595
+ "grad_norm": 1.2125264406204224,
596
+ "learning_rate": 0.0001716244002741604,
597
+ "loss": 1.9449,
598
+ "step": 830
599
+ },
600
+ {
601
+ "epoch": 0.29,
602
+ "grad_norm": 0.9563066959381104,
603
+ "learning_rate": 0.0001712816997943797,
604
+ "loss": 1.7063,
605
+ "step": 840
606
+ },
607
+ {
608
+ "epoch": 0.29,
609
+ "grad_norm": 0.8778769969940186,
610
+ "learning_rate": 0.00017093899931459904,
611
+ "loss": 1.802,
612
+ "step": 850
613
+ },
614
+ {
615
+ "epoch": 0.29,
616
+ "grad_norm": 1.0570799112319946,
617
+ "learning_rate": 0.00017059629883481837,
618
+ "loss": 1.7331,
619
+ "step": 860
620
+ },
621
+ {
622
+ "epoch": 0.3,
623
+ "grad_norm": 0.8234407305717468,
624
+ "learning_rate": 0.0001702535983550377,
625
+ "loss": 1.7943,
626
+ "step": 870
627
+ },
628
+ {
629
+ "epoch": 0.3,
630
+ "grad_norm": 0.968658983707428,
631
+ "learning_rate": 0.00016991089787525703,
632
+ "loss": 1.8527,
633
+ "step": 880
634
+ },
635
+ {
636
+ "epoch": 0.3,
637
+ "grad_norm": 0.6607180237770081,
638
+ "learning_rate": 0.00016956819739547636,
639
+ "loss": 1.8521,
640
+ "step": 890
641
+ },
642
+ {
643
+ "epoch": 0.31,
644
+ "grad_norm": 0.8055354952812195,
645
+ "learning_rate": 0.0001692254969156957,
646
+ "loss": 1.6901,
647
+ "step": 900
648
+ },
649
+ {
650
+ "epoch": 0.31,
651
+ "grad_norm": 0.8606925010681152,
652
+ "learning_rate": 0.00016888279643591502,
653
+ "loss": 1.7248,
654
+ "step": 910
655
+ },
656
+ {
657
+ "epoch": 0.32,
658
+ "grad_norm": 0.9894892573356628,
659
+ "learning_rate": 0.00016854009595613432,
660
+ "loss": 1.7541,
661
+ "step": 920
662
+ },
663
+ {
664
+ "epoch": 0.32,
665
+ "grad_norm": 0.8559629321098328,
666
+ "learning_rate": 0.00016819739547635368,
667
+ "loss": 1.7803,
668
+ "step": 930
669
+ },
670
+ {
671
+ "epoch": 0.32,
672
+ "grad_norm": 0.8917673826217651,
673
+ "learning_rate": 0.000167854694996573,
674
+ "loss": 1.8224,
675
+ "step": 940
676
+ },
677
+ {
678
+ "epoch": 0.33,
679
+ "grad_norm": 1.2621186971664429,
680
+ "learning_rate": 0.00016751199451679234,
681
+ "loss": 1.8253,
682
+ "step": 950
683
+ },
684
+ {
685
+ "epoch": 0.33,
686
+ "grad_norm": 1.1135177612304688,
687
+ "learning_rate": 0.00016716929403701167,
688
+ "loss": 1.6519,
689
+ "step": 960
690
+ },
691
+ {
692
+ "epoch": 0.33,
693
+ "grad_norm": 0.7034028172492981,
694
+ "learning_rate": 0.00016682659355723097,
695
+ "loss": 1.7079,
696
+ "step": 970
697
+ },
698
+ {
699
+ "epoch": 0.34,
700
+ "grad_norm": 0.7942814826965332,
701
+ "learning_rate": 0.0001664838930774503,
702
+ "loss": 1.828,
703
+ "step": 980
704
+ },
705
+ {
706
+ "epoch": 0.34,
707
+ "grad_norm": 0.9687950611114502,
708
+ "learning_rate": 0.00016614119259766966,
709
+ "loss": 1.7203,
710
+ "step": 990
711
+ },
712
+ {
713
+ "epoch": 0.34,
714
+ "grad_norm": 1.1074302196502686,
715
+ "learning_rate": 0.000165798492117889,
716
+ "loss": 1.7146,
717
+ "step": 1000
718
+ },
719
+ {
720
+ "epoch": 0.34,
721
+ "eval_loss": 1.9078810214996338,
722
+ "eval_runtime": 33.2486,
723
+ "eval_samples_per_second": 30.076,
724
+ "eval_steps_per_second": 3.76,
725
+ "step": 1000
726
+ },
727
+ {
728
+ "epoch": 0.35,
729
+ "grad_norm": 0.9533829689025879,
730
+ "learning_rate": 0.0001654557916381083,
731
+ "loss": 1.7596,
732
+ "step": 1010
733
+ },
734
+ {
735
+ "epoch": 0.35,
736
+ "grad_norm": 1.0547090768814087,
737
+ "learning_rate": 0.00016511309115832762,
738
+ "loss": 1.9113,
739
+ "step": 1020
740
+ },
741
+ {
742
+ "epoch": 0.35,
743
+ "grad_norm": 1.0186220407485962,
744
+ "learning_rate": 0.00016477039067854695,
745
+ "loss": 1.7845,
746
+ "step": 1030
747
+ },
748
+ {
749
+ "epoch": 0.36,
750
+ "grad_norm": 0.9044001698493958,
751
+ "learning_rate": 0.00016442769019876628,
752
+ "loss": 1.8174,
753
+ "step": 1040
754
+ },
755
+ {
756
+ "epoch": 0.36,
757
+ "grad_norm": 0.6433171033859253,
758
+ "learning_rate": 0.0001640849897189856,
759
+ "loss": 1.7702,
760
+ "step": 1050
761
+ },
762
+ {
763
+ "epoch": 0.36,
764
+ "grad_norm": 1.2511520385742188,
765
+ "learning_rate": 0.00016374228923920494,
766
+ "loss": 1.9304,
767
+ "step": 1060
768
+ },
769
+ {
770
+ "epoch": 0.37,
771
+ "grad_norm": 0.7901211977005005,
772
+ "learning_rate": 0.00016339958875942427,
773
+ "loss": 1.8432,
774
+ "step": 1070
775
+ },
776
+ {
777
+ "epoch": 0.37,
778
+ "grad_norm": 1.515535831451416,
779
+ "learning_rate": 0.0001630568882796436,
780
+ "loss": 1.8818,
781
+ "step": 1080
782
+ },
783
+ {
784
+ "epoch": 0.37,
785
+ "grad_norm": 0.9449120759963989,
786
+ "learning_rate": 0.00016271418779986293,
787
+ "loss": 1.8594,
788
+ "step": 1090
789
+ },
790
+ {
791
+ "epoch": 0.38,
792
+ "grad_norm": 0.7776308059692383,
793
+ "learning_rate": 0.00016237148732008226,
794
+ "loss": 1.8896,
795
+ "step": 1100
796
+ },
797
+ {
798
+ "epoch": 0.38,
799
+ "grad_norm": 1.3541969060897827,
800
+ "learning_rate": 0.0001620287868403016,
801
+ "loss": 1.8208,
802
+ "step": 1110
803
+ },
804
+ {
805
+ "epoch": 0.38,
806
+ "grad_norm": 0.7614444494247437,
807
+ "learning_rate": 0.00016168608636052092,
808
+ "loss": 1.759,
809
+ "step": 1120
810
+ },
811
+ {
812
+ "epoch": 0.39,
813
+ "grad_norm": 1.170345425605774,
814
+ "learning_rate": 0.00016134338588074025,
815
+ "loss": 1.6713,
816
+ "step": 1130
817
+ },
818
+ {
819
+ "epoch": 0.39,
820
+ "grad_norm": 0.8094021081924438,
821
+ "learning_rate": 0.00016100068540095955,
822
+ "loss": 1.7394,
823
+ "step": 1140
824
+ },
825
+ {
826
+ "epoch": 0.39,
827
+ "grad_norm": 1.169124722480774,
828
+ "learning_rate": 0.0001606579849211789,
829
+ "loss": 1.7609,
830
+ "step": 1150
831
+ },
832
+ {
833
+ "epoch": 0.4,
834
+ "grad_norm": 0.6766496300697327,
835
+ "learning_rate": 0.00016031528444139824,
836
+ "loss": 1.7812,
837
+ "step": 1160
838
+ },
839
+ {
840
+ "epoch": 0.4,
841
+ "grad_norm": 1.0808138847351074,
842
+ "learning_rate": 0.00015997258396161757,
843
+ "loss": 1.7777,
844
+ "step": 1170
845
+ },
846
+ {
847
+ "epoch": 0.4,
848
+ "grad_norm": 0.6450923681259155,
849
+ "learning_rate": 0.00015962988348183687,
850
+ "loss": 1.8539,
851
+ "step": 1180
852
+ },
853
+ {
854
+ "epoch": 0.41,
855
+ "grad_norm": 1.0518946647644043,
856
+ "learning_rate": 0.0001592871830020562,
857
+ "loss": 1.7799,
858
+ "step": 1190
859
+ },
860
+ {
861
+ "epoch": 0.41,
862
+ "grad_norm": 0.7807414531707764,
863
+ "learning_rate": 0.00015894448252227553,
864
+ "loss": 1.774,
865
+ "step": 1200
866
+ },
867
+ {
868
+ "epoch": 0.41,
869
+ "grad_norm": 1.4259986877441406,
870
+ "learning_rate": 0.00015860178204249488,
871
+ "loss": 1.8153,
872
+ "step": 1210
873
+ },
874
+ {
875
+ "epoch": 0.42,
876
+ "grad_norm": 0.9342586994171143,
877
+ "learning_rate": 0.0001582590815627142,
878
+ "loss": 1.7495,
879
+ "step": 1220
880
+ },
881
+ {
882
+ "epoch": 0.42,
883
+ "grad_norm": 0.7621099948883057,
884
+ "learning_rate": 0.00015791638108293352,
885
+ "loss": 1.7964,
886
+ "step": 1230
887
+ },
888
+ {
889
+ "epoch": 0.42,
890
+ "grad_norm": 0.8253260254859924,
891
+ "learning_rate": 0.00015757368060315285,
892
+ "loss": 1.7669,
893
+ "step": 1240
894
+ },
895
+ {
896
+ "epoch": 0.43,
897
+ "grad_norm": 0.6914420127868652,
898
+ "learning_rate": 0.00015723098012337218,
899
+ "loss": 1.803,
900
+ "step": 1250
901
+ },
902
+ {
903
+ "epoch": 0.43,
904
+ "grad_norm": 0.7147281765937805,
905
+ "learning_rate": 0.0001568882796435915,
906
+ "loss": 1.8226,
907
+ "step": 1260
908
+ },
909
+ {
910
+ "epoch": 0.43,
911
+ "grad_norm": 2.0851213932037354,
912
+ "learning_rate": 0.00015654557916381084,
913
+ "loss": 1.6957,
914
+ "step": 1270
915
+ },
916
+ {
917
+ "epoch": 0.44,
918
+ "grad_norm": 0.6254770159721375,
919
+ "learning_rate": 0.00015620287868403017,
920
+ "loss": 1.75,
921
+ "step": 1280
922
+ },
923
+ {
924
+ "epoch": 0.44,
925
+ "grad_norm": 1.0984652042388916,
926
+ "learning_rate": 0.0001558601782042495,
927
+ "loss": 1.8425,
928
+ "step": 1290
929
+ },
930
+ {
931
+ "epoch": 0.45,
932
+ "grad_norm": 1.0353467464447021,
933
+ "learning_rate": 0.00015551747772446882,
934
+ "loss": 1.7995,
935
+ "step": 1300
936
+ },
937
+ {
938
+ "epoch": 0.45,
939
+ "grad_norm": 0.6647160053253174,
940
+ "learning_rate": 0.00015517477724468813,
941
+ "loss": 1.866,
942
+ "step": 1310
943
+ },
944
+ {
945
+ "epoch": 0.45,
946
+ "grad_norm": 0.6671775579452515,
947
+ "learning_rate": 0.00015483207676490748,
948
+ "loss": 1.6871,
949
+ "step": 1320
950
+ },
951
+ {
952
+ "epoch": 0.46,
953
+ "grad_norm": 1.0024131536483765,
954
+ "learning_rate": 0.00015448937628512681,
955
+ "loss": 1.7424,
956
+ "step": 1330
957
+ },
958
+ {
959
+ "epoch": 0.46,
960
+ "grad_norm": 1.0090551376342773,
961
+ "learning_rate": 0.00015414667580534614,
962
+ "loss": 1.7001,
963
+ "step": 1340
964
+ },
965
+ {
966
+ "epoch": 0.46,
967
+ "grad_norm": 0.9725455045700073,
968
+ "learning_rate": 0.00015380397532556545,
969
+ "loss": 1.7114,
970
+ "step": 1350
971
+ },
972
+ {
973
+ "epoch": 0.47,
974
+ "grad_norm": 0.6556392312049866,
975
+ "learning_rate": 0.00015346127484578478,
976
+ "loss": 1.5969,
977
+ "step": 1360
978
+ },
979
+ {
980
+ "epoch": 0.47,
981
+ "grad_norm": 1.156596302986145,
982
+ "learning_rate": 0.00015311857436600413,
983
+ "loss": 1.7334,
984
+ "step": 1370
985
+ },
986
+ {
987
+ "epoch": 0.47,
988
+ "grad_norm": 0.9172496199607849,
989
+ "learning_rate": 0.00015277587388622346,
990
+ "loss": 1.7373,
991
+ "step": 1380
992
+ },
993
+ {
994
+ "epoch": 0.48,
995
+ "grad_norm": 0.9010474681854248,
996
+ "learning_rate": 0.0001524331734064428,
997
+ "loss": 1.8032,
998
+ "step": 1390
999
+ },
1000
+ {
1001
+ "epoch": 0.48,
1002
+ "grad_norm": 0.9486579298973083,
1003
+ "learning_rate": 0.0001520904729266621,
1004
+ "loss": 1.6388,
1005
+ "step": 1400
1006
+ },
1007
+ {
1008
+ "epoch": 0.48,
1009
+ "grad_norm": 0.8411978483200073,
1010
+ "learning_rate": 0.00015174777244688142,
1011
+ "loss": 1.7671,
1012
+ "step": 1410
1013
+ },
1014
+ {
1015
+ "epoch": 0.49,
1016
+ "grad_norm": 0.9575003385543823,
1017
+ "learning_rate": 0.00015140507196710075,
1018
+ "loss": 1.6523,
1019
+ "step": 1420
1020
+ },
1021
+ {
1022
+ "epoch": 0.49,
1023
+ "grad_norm": 0.7651090025901794,
1024
+ "learning_rate": 0.0001510623714873201,
1025
+ "loss": 1.812,
1026
+ "step": 1430
1027
+ },
1028
+ {
1029
+ "epoch": 0.49,
1030
+ "grad_norm": 0.8477165699005127,
1031
+ "learning_rate": 0.0001507196710075394,
1032
+ "loss": 1.7125,
1033
+ "step": 1440
1034
+ },
1035
+ {
1036
+ "epoch": 0.5,
1037
+ "grad_norm": 0.9737070202827454,
1038
+ "learning_rate": 0.00015037697052775874,
1039
+ "loss": 1.7506,
1040
+ "step": 1450
1041
+ },
1042
+ {
1043
+ "epoch": 0.5,
1044
+ "grad_norm": 1.0645496845245361,
1045
+ "learning_rate": 0.00015003427004797807,
1046
+ "loss": 1.7335,
1047
+ "step": 1460
1048
+ },
1049
+ {
1050
+ "epoch": 0.5,
1051
+ "grad_norm": 0.9303259253501892,
1052
+ "learning_rate": 0.0001496915695681974,
1053
+ "loss": 1.8838,
1054
+ "step": 1470
1055
+ },
1056
+ {
1057
+ "epoch": 0.51,
1058
+ "grad_norm": 0.6571500897407532,
1059
+ "learning_rate": 0.00014934886908841673,
1060
+ "loss": 1.8093,
1061
+ "step": 1480
1062
+ },
1063
+ {
1064
+ "epoch": 0.51,
1065
+ "grad_norm": 0.7994106411933899,
1066
+ "learning_rate": 0.00014900616860863606,
1067
+ "loss": 1.6691,
1068
+ "step": 1490
1069
+ },
1070
+ {
1071
+ "epoch": 0.51,
1072
+ "grad_norm": 0.8453437685966492,
1073
+ "learning_rate": 0.0001486634681288554,
1074
+ "loss": 1.6731,
1075
+ "step": 1500
1076
+ },
1077
+ {
1078
+ "epoch": 0.51,
1079
+ "eval_loss": 1.8940061330795288,
1080
+ "eval_runtime": 33.2126,
1081
+ "eval_samples_per_second": 30.109,
1082
+ "eval_steps_per_second": 3.764,
1083
+ "step": 1500
1084
+ },
1085
+ {
1086
+ "epoch": 0.52,
1087
+ "grad_norm": 1.0370814800262451,
1088
+ "learning_rate": 0.00014832076764907472,
1089
+ "loss": 1.7869,
1090
+ "step": 1510
1091
+ },
1092
+ {
1093
+ "epoch": 0.52,
1094
+ "grad_norm": 1.0886887311935425,
1095
+ "learning_rate": 0.00014797806716929405,
1096
+ "loss": 1.7887,
1097
+ "step": 1520
1098
+ },
1099
+ {
1100
+ "epoch": 0.52,
1101
+ "grad_norm": 0.9058669209480286,
1102
+ "learning_rate": 0.00014763536668951335,
1103
+ "loss": 1.6781,
1104
+ "step": 1530
1105
+ },
1106
+ {
1107
+ "epoch": 0.53,
1108
+ "grad_norm": 0.46401920914649963,
1109
+ "learning_rate": 0.0001472926662097327,
1110
+ "loss": 1.6465,
1111
+ "step": 1540
1112
+ },
1113
+ {
1114
+ "epoch": 0.53,
1115
+ "grad_norm": 0.6265978813171387,
1116
+ "learning_rate": 0.00014694996572995204,
1117
+ "loss": 1.8399,
1118
+ "step": 1550
1119
+ },
1120
+ {
1121
+ "epoch": 0.53,
1122
+ "grad_norm": 0.7882290482521057,
1123
+ "learning_rate": 0.00014660726525017137,
1124
+ "loss": 1.7707,
1125
+ "step": 1560
1126
+ },
1127
+ {
1128
+ "epoch": 0.54,
1129
+ "grad_norm": 0.7576068043708801,
1130
+ "learning_rate": 0.00014626456477039067,
1131
+ "loss": 1.8781,
1132
+ "step": 1570
1133
+ },
1134
+ {
1135
+ "epoch": 0.54,
1136
+ "grad_norm": 0.8988894820213318,
1137
+ "learning_rate": 0.00014592186429061,
1138
+ "loss": 1.7109,
1139
+ "step": 1580
1140
+ },
1141
+ {
1142
+ "epoch": 0.54,
1143
+ "grad_norm": 0.7934654951095581,
1144
+ "learning_rate": 0.00014557916381082933,
1145
+ "loss": 1.8261,
1146
+ "step": 1590
1147
+ },
1148
+ {
1149
+ "epoch": 0.55,
1150
+ "grad_norm": 0.9526162147521973,
1151
+ "learning_rate": 0.0001452364633310487,
1152
+ "loss": 1.7286,
1153
+ "step": 1600
1154
+ },
1155
+ {
1156
+ "epoch": 0.55,
1157
+ "grad_norm": 0.8650903701782227,
1158
+ "learning_rate": 0.000144893762851268,
1159
+ "loss": 1.8075,
1160
+ "step": 1610
1161
+ },
1162
+ {
1163
+ "epoch": 0.55,
1164
+ "grad_norm": 0.8737215399742126,
1165
+ "learning_rate": 0.00014455106237148732,
1166
+ "loss": 1.7683,
1167
+ "step": 1620
1168
+ },
1169
+ {
1170
+ "epoch": 0.56,
1171
+ "grad_norm": 1.0927869081497192,
1172
+ "learning_rate": 0.00014420836189170665,
1173
+ "loss": 1.8238,
1174
+ "step": 1630
1175
+ },
1176
+ {
1177
+ "epoch": 0.56,
1178
+ "grad_norm": 0.7490981817245483,
1179
+ "learning_rate": 0.00014386566141192598,
1180
+ "loss": 1.7528,
1181
+ "step": 1640
1182
+ },
1183
+ {
1184
+ "epoch": 0.57,
1185
+ "grad_norm": 0.6721557974815369,
1186
+ "learning_rate": 0.0001435229609321453,
1187
+ "loss": 1.7212,
1188
+ "step": 1650
1189
+ },
1190
+ {
1191
+ "epoch": 0.57,
1192
+ "grad_norm": 0.8125373125076294,
1193
+ "learning_rate": 0.00014318026045236464,
1194
+ "loss": 1.8369,
1195
+ "step": 1660
1196
+ },
1197
+ {
1198
+ "epoch": 0.57,
1199
+ "grad_norm": 0.598507821559906,
1200
+ "learning_rate": 0.00014283755997258397,
1201
+ "loss": 1.8455,
1202
+ "step": 1670
1203
+ },
1204
+ {
1205
+ "epoch": 0.58,
1206
+ "grad_norm": 1.2567535638809204,
1207
+ "learning_rate": 0.0001424948594928033,
1208
+ "loss": 1.7656,
1209
+ "step": 1680
1210
+ },
1211
+ {
1212
+ "epoch": 0.58,
1213
+ "grad_norm": 1.5279853343963623,
1214
+ "learning_rate": 0.00014215215901302263,
1215
+ "loss": 1.8297,
1216
+ "step": 1690
1217
+ },
1218
+ {
1219
+ "epoch": 0.58,
1220
+ "grad_norm": 1.1410638093948364,
1221
+ "learning_rate": 0.00014180945853324196,
1222
+ "loss": 1.7489,
1223
+ "step": 1700
1224
+ },
1225
+ {
1226
+ "epoch": 0.59,
1227
+ "grad_norm": 0.9007987976074219,
1228
+ "learning_rate": 0.0001414667580534613,
1229
+ "loss": 1.7473,
1230
+ "step": 1710
1231
+ },
1232
+ {
1233
+ "epoch": 0.59,
1234
+ "grad_norm": 0.5736974477767944,
1235
+ "learning_rate": 0.00014112405757368062,
1236
+ "loss": 1.8022,
1237
+ "step": 1720
1238
+ },
1239
+ {
1240
+ "epoch": 0.59,
1241
+ "grad_norm": 0.6310347318649292,
1242
+ "learning_rate": 0.00014078135709389995,
1243
+ "loss": 1.7676,
1244
+ "step": 1730
1245
+ },
1246
+ {
1247
+ "epoch": 0.6,
1248
+ "grad_norm": 0.9788106679916382,
1249
+ "learning_rate": 0.00014043865661411925,
1250
+ "loss": 1.7303,
1251
+ "step": 1740
1252
+ },
1253
+ {
1254
+ "epoch": 0.6,
1255
+ "grad_norm": 0.6612042784690857,
1256
+ "learning_rate": 0.00014009595613433858,
1257
+ "loss": 1.675,
1258
+ "step": 1750
1259
+ },
1260
+ {
1261
+ "epoch": 0.6,
1262
+ "grad_norm": 0.8740193247795105,
1263
+ "learning_rate": 0.00013975325565455794,
1264
+ "loss": 1.7945,
1265
+ "step": 1760
1266
+ },
1267
+ {
1268
+ "epoch": 0.61,
1269
+ "grad_norm": 0.9548364877700806,
1270
+ "learning_rate": 0.00013941055517477727,
1271
+ "loss": 1.7485,
1272
+ "step": 1770
1273
+ },
1274
+ {
1275
+ "epoch": 0.61,
1276
+ "grad_norm": 0.6676565408706665,
1277
+ "learning_rate": 0.00013906785469499657,
1278
+ "loss": 1.7479,
1279
+ "step": 1780
1280
+ },
1281
+ {
1282
+ "epoch": 0.61,
1283
+ "grad_norm": 0.6287640333175659,
1284
+ "learning_rate": 0.0001387251542152159,
1285
+ "loss": 1.7007,
1286
+ "step": 1790
1287
+ },
1288
+ {
1289
+ "epoch": 0.62,
1290
+ "grad_norm": 1.5443295240402222,
1291
+ "learning_rate": 0.00013838245373543523,
1292
+ "loss": 1.8916,
1293
+ "step": 1800
1294
+ },
1295
+ {
1296
+ "epoch": 0.62,
1297
+ "grad_norm": 0.9970656037330627,
1298
+ "learning_rate": 0.00013803975325565456,
1299
+ "loss": 1.6733,
1300
+ "step": 1810
1301
+ },
1302
+ {
1303
+ "epoch": 0.62,
1304
+ "grad_norm": 0.9320075511932373,
1305
+ "learning_rate": 0.00013769705277587391,
1306
+ "loss": 1.8622,
1307
+ "step": 1820
1308
+ },
1309
+ {
1310
+ "epoch": 0.63,
1311
+ "grad_norm": 0.8384440541267395,
1312
+ "learning_rate": 0.00013735435229609322,
1313
+ "loss": 1.6825,
1314
+ "step": 1830
1315
+ },
1316
+ {
1317
+ "epoch": 0.63,
1318
+ "grad_norm": 1.1807342767715454,
1319
+ "learning_rate": 0.00013701165181631255,
1320
+ "loss": 1.6548,
1321
+ "step": 1840
1322
+ },
1323
+ {
1324
+ "epoch": 0.63,
1325
+ "grad_norm": 0.7640541195869446,
1326
+ "learning_rate": 0.00013666895133653188,
1327
+ "loss": 1.8134,
1328
+ "step": 1850
1329
+ },
1330
+ {
1331
+ "epoch": 0.64,
1332
+ "grad_norm": 0.9137887358665466,
1333
+ "learning_rate": 0.0001363262508567512,
1334
+ "loss": 1.7685,
1335
+ "step": 1860
1336
+ },
1337
+ {
1338
+ "epoch": 0.64,
1339
+ "grad_norm": 0.8986667394638062,
1340
+ "learning_rate": 0.00013598355037697054,
1341
+ "loss": 1.7455,
1342
+ "step": 1870
1343
+ },
1344
+ {
1345
+ "epoch": 0.64,
1346
+ "grad_norm": 0.96836918592453,
1347
+ "learning_rate": 0.00013564084989718987,
1348
+ "loss": 1.8705,
1349
+ "step": 1880
1350
+ },
1351
+ {
1352
+ "epoch": 0.65,
1353
+ "grad_norm": 1.381028175354004,
1354
+ "learning_rate": 0.0001352981494174092,
1355
+ "loss": 1.7644,
1356
+ "step": 1890
1357
+ },
1358
+ {
1359
+ "epoch": 0.65,
1360
+ "grad_norm": 0.617438018321991,
1361
+ "learning_rate": 0.00013495544893762853,
1362
+ "loss": 1.6194,
1363
+ "step": 1900
1364
+ },
1365
+ {
1366
+ "epoch": 0.65,
1367
+ "grad_norm": 0.8686628937721252,
1368
+ "learning_rate": 0.00013461274845784783,
1369
+ "loss": 1.7171,
1370
+ "step": 1910
1371
+ },
1372
+ {
1373
+ "epoch": 0.66,
1374
+ "grad_norm": 0.7735409140586853,
1375
+ "learning_rate": 0.00013427004797806716,
1376
+ "loss": 1.725,
1377
+ "step": 1920
1378
+ },
1379
+ {
1380
+ "epoch": 0.66,
1381
+ "grad_norm": 1.0692516565322876,
1382
+ "learning_rate": 0.00013392734749828651,
1383
+ "loss": 1.762,
1384
+ "step": 1930
1385
+ },
1386
+ {
1387
+ "epoch": 0.66,
1388
+ "grad_norm": 0.763136625289917,
1389
+ "learning_rate": 0.00013358464701850584,
1390
+ "loss": 1.6546,
1391
+ "step": 1940
1392
+ },
1393
+ {
1394
+ "epoch": 0.67,
1395
+ "grad_norm": 0.9908429980278015,
1396
+ "learning_rate": 0.00013324194653872517,
1397
+ "loss": 1.6499,
1398
+ "step": 1950
1399
+ },
1400
+ {
1401
+ "epoch": 0.67,
1402
+ "grad_norm": 0.9493003487586975,
1403
+ "learning_rate": 0.00013289924605894448,
1404
+ "loss": 1.5616,
1405
+ "step": 1960
1406
+ },
1407
+ {
1408
+ "epoch": 0.67,
1409
+ "grad_norm": 0.8336248993873596,
1410
+ "learning_rate": 0.0001325565455791638,
1411
+ "loss": 1.7914,
1412
+ "step": 1970
1413
+ },
1414
+ {
1415
+ "epoch": 0.68,
1416
+ "grad_norm": 0.8938840627670288,
1417
+ "learning_rate": 0.00013221384509938314,
1418
+ "loss": 1.7274,
1419
+ "step": 1980
1420
+ },
1421
+ {
1422
+ "epoch": 0.68,
1423
+ "grad_norm": 1.0243479013442993,
1424
+ "learning_rate": 0.0001318711446196025,
1425
+ "loss": 1.6643,
1426
+ "step": 1990
1427
+ },
1428
+ {
1429
+ "epoch": 0.68,
1430
+ "grad_norm": 1.0226181745529175,
1431
+ "learning_rate": 0.0001315284441398218,
1432
+ "loss": 1.7626,
1433
+ "step": 2000
1434
+ },
1435
+ {
1436
+ "epoch": 0.68,
1437
+ "eval_loss": 1.8913378715515137,
1438
+ "eval_runtime": 33.1473,
1439
+ "eval_samples_per_second": 30.168,
1440
+ "eval_steps_per_second": 3.771,
1441
+ "step": 2000
1442
+ },
1443
+ {
1444
+ "epoch": 0.69,
1445
+ "grad_norm": 1.1059471368789673,
1446
+ "learning_rate": 0.00013118574366004112,
1447
+ "loss": 1.6362,
1448
+ "step": 2010
1449
+ },
1450
+ {
1451
+ "epoch": 0.69,
1452
+ "grad_norm": 1.3754314184188843,
1453
+ "learning_rate": 0.00013084304318026045,
1454
+ "loss": 1.8308,
1455
+ "step": 2020
1456
+ },
1457
+ {
1458
+ "epoch": 0.7,
1459
+ "grad_norm": 1.3899627923965454,
1460
+ "learning_rate": 0.00013050034270047978,
1461
+ "loss": 1.6982,
1462
+ "step": 2030
1463
+ },
1464
+ {
1465
+ "epoch": 0.7,
1466
+ "grad_norm": 0.8804599046707153,
1467
+ "learning_rate": 0.00013015764222069911,
1468
+ "loss": 1.8138,
1469
+ "step": 2040
1470
+ },
1471
+ {
1472
+ "epoch": 0.7,
1473
+ "grad_norm": 0.6578095555305481,
1474
+ "learning_rate": 0.00012981494174091844,
1475
+ "loss": 1.7211,
1476
+ "step": 2050
1477
+ },
1478
+ {
1479
+ "epoch": 0.71,
1480
+ "grad_norm": 1.5725558996200562,
1481
+ "learning_rate": 0.00012947224126113777,
1482
+ "loss": 1.8684,
1483
+ "step": 2060
1484
+ },
1485
+ {
1486
+ "epoch": 0.71,
1487
+ "grad_norm": 1.097717523574829,
1488
+ "learning_rate": 0.0001291295407813571,
1489
+ "loss": 1.7705,
1490
+ "step": 2070
1491
+ },
1492
+ {
1493
+ "epoch": 0.71,
1494
+ "grad_norm": 0.7564202547073364,
1495
+ "learning_rate": 0.00012878684030157643,
1496
+ "loss": 1.5935,
1497
+ "step": 2080
1498
+ },
1499
+ {
1500
+ "epoch": 0.72,
1501
+ "grad_norm": 0.732243537902832,
1502
+ "learning_rate": 0.00012844413982179576,
1503
+ "loss": 1.7694,
1504
+ "step": 2090
1505
+ },
1506
+ {
1507
+ "epoch": 0.72,
1508
+ "grad_norm": 0.6464608907699585,
1509
+ "learning_rate": 0.0001281014393420151,
1510
+ "loss": 1.8418,
1511
+ "step": 2100
1512
+ },
1513
+ {
1514
+ "epoch": 0.72,
1515
+ "grad_norm": 0.7090341448783875,
1516
+ "learning_rate": 0.00012775873886223442,
1517
+ "loss": 1.8122,
1518
+ "step": 2110
1519
+ },
1520
+ {
1521
+ "epoch": 0.73,
1522
+ "grad_norm": 1.1480237245559692,
1523
+ "learning_rate": 0.00012741603838245375,
1524
+ "loss": 1.7766,
1525
+ "step": 2120
1526
+ },
1527
+ {
1528
+ "epoch": 0.73,
1529
+ "grad_norm": 0.6737000346183777,
1530
+ "learning_rate": 0.00012707333790267305,
1531
+ "loss": 1.7876,
1532
+ "step": 2130
1533
+ },
1534
+ {
1535
+ "epoch": 0.73,
1536
+ "grad_norm": 0.7794924378395081,
1537
+ "learning_rate": 0.00012673063742289238,
1538
+ "loss": 1.8529,
1539
+ "step": 2140
1540
+ },
1541
+ {
1542
+ "epoch": 0.74,
1543
+ "grad_norm": 1.3136320114135742,
1544
+ "learning_rate": 0.00012638793694311174,
1545
+ "loss": 1.6699,
1546
+ "step": 2150
1547
+ },
1548
+ {
1549
+ "epoch": 0.74,
1550
+ "grad_norm": 0.884027361869812,
1551
+ "learning_rate": 0.00012604523646333107,
1552
+ "loss": 1.7689,
1553
+ "step": 2160
1554
+ },
1555
+ {
1556
+ "epoch": 0.74,
1557
+ "grad_norm": 1.103605031967163,
1558
+ "learning_rate": 0.00012570253598355037,
1559
+ "loss": 1.8594,
1560
+ "step": 2170
1561
+ },
1562
+ {
1563
+ "epoch": 0.75,
1564
+ "grad_norm": 1.3322539329528809,
1565
+ "learning_rate": 0.0001253598355037697,
1566
+ "loss": 1.6765,
1567
+ "step": 2180
1568
+ },
1569
+ {
1570
+ "epoch": 0.75,
1571
+ "grad_norm": 0.7840645909309387,
1572
+ "learning_rate": 0.00012501713502398903,
1573
+ "loss": 1.65,
1574
+ "step": 2190
1575
+ },
1576
+ {
1577
+ "epoch": 0.75,
1578
+ "grad_norm": 0.9259356260299683,
1579
+ "learning_rate": 0.00012467443454420836,
1580
+ "loss": 1.7805,
1581
+ "step": 2200
1582
+ },
1583
+ {
1584
+ "epoch": 0.76,
1585
+ "grad_norm": 1.3709288835525513,
1586
+ "learning_rate": 0.0001243317340644277,
1587
+ "loss": 1.7086,
1588
+ "step": 2210
1589
+ },
1590
+ {
1591
+ "epoch": 0.76,
1592
+ "grad_norm": 0.6325123310089111,
1593
+ "learning_rate": 0.00012398903358464702,
1594
+ "loss": 1.7124,
1595
+ "step": 2220
1596
+ },
1597
+ {
1598
+ "epoch": 0.76,
1599
+ "grad_norm": 0.854541003704071,
1600
+ "learning_rate": 0.00012364633310486635,
1601
+ "loss": 1.7089,
1602
+ "step": 2230
1603
+ },
1604
+ {
1605
+ "epoch": 0.77,
1606
+ "grad_norm": 0.8861531019210815,
1607
+ "learning_rate": 0.00012330363262508568,
1608
+ "loss": 1.8369,
1609
+ "step": 2240
1610
+ },
1611
+ {
1612
+ "epoch": 0.77,
1613
+ "grad_norm": 1.269750714302063,
1614
+ "learning_rate": 0.000122960932145305,
1615
+ "loss": 1.7598,
1616
+ "step": 2250
1617
+ },
1618
+ {
1619
+ "epoch": 0.77,
1620
+ "grad_norm": 0.999598503112793,
1621
+ "learning_rate": 0.00012261823166552434,
1622
+ "loss": 1.8376,
1623
+ "step": 2260
1624
+ },
1625
+ {
1626
+ "epoch": 0.78,
1627
+ "grad_norm": 0.7654330134391785,
1628
+ "learning_rate": 0.00012227553118574367,
1629
+ "loss": 1.7236,
1630
+ "step": 2270
1631
+ },
1632
+ {
1633
+ "epoch": 0.78,
1634
+ "grad_norm": 1.11728835105896,
1635
+ "learning_rate": 0.000121932830705963,
1636
+ "loss": 1.7375,
1637
+ "step": 2280
1638
+ },
1639
+ {
1640
+ "epoch": 0.78,
1641
+ "grad_norm": 0.7219797968864441,
1642
+ "learning_rate": 0.00012159013022618233,
1643
+ "loss": 1.7786,
1644
+ "step": 2290
1645
+ },
1646
+ {
1647
+ "epoch": 0.79,
1648
+ "grad_norm": 1.0127757787704468,
1649
+ "learning_rate": 0.00012124742974640165,
1650
+ "loss": 1.7003,
1651
+ "step": 2300
1652
+ },
1653
+ {
1654
+ "epoch": 0.79,
1655
+ "grad_norm": 1.0450137853622437,
1656
+ "learning_rate": 0.00012090472926662097,
1657
+ "loss": 1.7425,
1658
+ "step": 2310
1659
+ },
1660
+ {
1661
+ "epoch": 0.79,
1662
+ "grad_norm": 0.9303760528564453,
1663
+ "learning_rate": 0.0001205620287868403,
1664
+ "loss": 1.632,
1665
+ "step": 2320
1666
+ },
1667
+ {
1668
+ "epoch": 0.8,
1669
+ "grad_norm": 0.7303478717803955,
1670
+ "learning_rate": 0.00012021932830705965,
1671
+ "loss": 1.6918,
1672
+ "step": 2330
1673
+ },
1674
+ {
1675
+ "epoch": 0.8,
1676
+ "grad_norm": 0.6323578953742981,
1677
+ "learning_rate": 0.00011987662782727895,
1678
+ "loss": 1.672,
1679
+ "step": 2340
1680
+ },
1681
+ {
1682
+ "epoch": 0.8,
1683
+ "grad_norm": 0.715811014175415,
1684
+ "learning_rate": 0.00011953392734749828,
1685
+ "loss": 1.7613,
1686
+ "step": 2350
1687
+ },
1688
+ {
1689
+ "epoch": 0.81,
1690
+ "grad_norm": 0.7297527194023132,
1691
+ "learning_rate": 0.00011919122686771762,
1692
+ "loss": 1.7277,
1693
+ "step": 2360
1694
+ },
1695
+ {
1696
+ "epoch": 0.81,
1697
+ "grad_norm": 1.0844471454620361,
1698
+ "learning_rate": 0.00011884852638793695,
1699
+ "loss": 1.8143,
1700
+ "step": 2370
1701
+ },
1702
+ {
1703
+ "epoch": 0.82,
1704
+ "grad_norm": 0.9260643720626831,
1705
+ "learning_rate": 0.00011850582590815628,
1706
+ "loss": 1.7228,
1707
+ "step": 2380
1708
+ },
1709
+ {
1710
+ "epoch": 0.82,
1711
+ "grad_norm": 0.9541537761688232,
1712
+ "learning_rate": 0.0001181631254283756,
1713
+ "loss": 1.7143,
1714
+ "step": 2390
1715
+ },
1716
+ {
1717
+ "epoch": 0.82,
1718
+ "grad_norm": 1.0506033897399902,
1719
+ "learning_rate": 0.00011782042494859493,
1720
+ "loss": 1.7659,
1721
+ "step": 2400
1722
+ },
1723
+ {
1724
+ "epoch": 0.83,
1725
+ "grad_norm": 0.7201717495918274,
1726
+ "learning_rate": 0.00011747772446881427,
1727
+ "loss": 1.7257,
1728
+ "step": 2410
1729
+ },
1730
+ {
1731
+ "epoch": 0.83,
1732
+ "grad_norm": 0.8612362742424011,
1733
+ "learning_rate": 0.0001171350239890336,
1734
+ "loss": 1.7009,
1735
+ "step": 2420
1736
+ },
1737
+ {
1738
+ "epoch": 0.83,
1739
+ "grad_norm": 0.8745547533035278,
1740
+ "learning_rate": 0.0001167923235092529,
1741
+ "loss": 1.733,
1742
+ "step": 2430
1743
+ },
1744
+ {
1745
+ "epoch": 0.84,
1746
+ "grad_norm": 0.5927043557167053,
1747
+ "learning_rate": 0.00011644962302947225,
1748
+ "loss": 1.7724,
1749
+ "step": 2440
1750
+ },
1751
+ {
1752
+ "epoch": 0.84,
1753
+ "grad_norm": 0.6471837162971497,
1754
+ "learning_rate": 0.00011610692254969158,
1755
+ "loss": 1.7103,
1756
+ "step": 2450
1757
+ },
1758
+ {
1759
+ "epoch": 0.84,
1760
+ "grad_norm": 1.1340347528457642,
1761
+ "learning_rate": 0.0001157642220699109,
1762
+ "loss": 1.7053,
1763
+ "step": 2460
1764
+ },
1765
+ {
1766
+ "epoch": 0.85,
1767
+ "grad_norm": 0.8819349408149719,
1768
+ "learning_rate": 0.00011542152159013022,
1769
+ "loss": 1.7552,
1770
+ "step": 2470
1771
+ },
1772
+ {
1773
+ "epoch": 0.85,
1774
+ "grad_norm": 0.6587919592857361,
1775
+ "learning_rate": 0.00011507882111034955,
1776
+ "loss": 1.6482,
1777
+ "step": 2480
1778
+ },
1779
+ {
1780
+ "epoch": 0.85,
1781
+ "grad_norm": 1.0057884454727173,
1782
+ "learning_rate": 0.00011473612063056888,
1783
+ "loss": 1.7711,
1784
+ "step": 2490
1785
+ },
1786
+ {
1787
+ "epoch": 0.86,
1788
+ "grad_norm": 0.6465263962745667,
1789
+ "learning_rate": 0.00011439342015078823,
1790
+ "loss": 1.7565,
1791
+ "step": 2500
1792
+ },
1793
+ {
1794
+ "epoch": 0.86,
1795
+ "eval_loss": 1.8792312145233154,
1796
+ "eval_runtime": 33.1087,
1797
+ "eval_samples_per_second": 30.204,
1798
+ "eval_steps_per_second": 3.775,
1799
+ "step": 2500
1800
+ },
1801
+ {
1802
+ "epoch": 0.86,
1803
+ "grad_norm": 0.5970360040664673,
1804
+ "learning_rate": 0.00011405071967100756,
1805
+ "loss": 1.7179,
1806
+ "step": 2510
1807
+ },
1808
+ {
1809
+ "epoch": 0.86,
1810
+ "grad_norm": 1.3015583753585815,
1811
+ "learning_rate": 0.00011370801919122687,
1812
+ "loss": 1.7225,
1813
+ "step": 2520
1814
+ },
1815
+ {
1816
+ "epoch": 0.87,
1817
+ "grad_norm": 0.9235218167304993,
1818
+ "learning_rate": 0.0001133653187114462,
1819
+ "loss": 1.7657,
1820
+ "step": 2530
1821
+ },
1822
+ {
1823
+ "epoch": 0.87,
1824
+ "grad_norm": 1.025038480758667,
1825
+ "learning_rate": 0.00011302261823166553,
1826
+ "loss": 1.7755,
1827
+ "step": 2540
1828
+ },
1829
+ {
1830
+ "epoch": 0.87,
1831
+ "grad_norm": 0.8988834619522095,
1832
+ "learning_rate": 0.00011267991775188486,
1833
+ "loss": 1.8187,
1834
+ "step": 2550
1835
+ },
1836
+ {
1837
+ "epoch": 0.88,
1838
+ "grad_norm": 0.7810622453689575,
1839
+ "learning_rate": 0.00011233721727210418,
1840
+ "loss": 1.6565,
1841
+ "step": 2560
1842
+ },
1843
+ {
1844
+ "epoch": 0.88,
1845
+ "grad_norm": 1.6817054748535156,
1846
+ "learning_rate": 0.0001119945167923235,
1847
+ "loss": 1.7764,
1848
+ "step": 2570
1849
+ },
1850
+ {
1851
+ "epoch": 0.88,
1852
+ "grad_norm": 0.9688411355018616,
1853
+ "learning_rate": 0.00011165181631254285,
1854
+ "loss": 1.6599,
1855
+ "step": 2580
1856
+ },
1857
+ {
1858
+ "epoch": 0.89,
1859
+ "grad_norm": 0.742932915687561,
1860
+ "learning_rate": 0.00011130911583276218,
1861
+ "loss": 1.7552,
1862
+ "step": 2590
1863
+ },
1864
+ {
1865
+ "epoch": 0.89,
1866
+ "grad_norm": 0.5261206030845642,
1867
+ "learning_rate": 0.0001109664153529815,
1868
+ "loss": 1.6432,
1869
+ "step": 2600
1870
+ },
1871
+ {
1872
+ "epoch": 0.89,
1873
+ "grad_norm": 0.8997339606285095,
1874
+ "learning_rate": 0.00011062371487320082,
1875
+ "loss": 1.8438,
1876
+ "step": 2610
1877
+ },
1878
+ {
1879
+ "epoch": 0.9,
1880
+ "grad_norm": 0.8077126741409302,
1881
+ "learning_rate": 0.00011028101439342015,
1882
+ "loss": 1.8144,
1883
+ "step": 2620
1884
+ },
1885
+ {
1886
+ "epoch": 0.9,
1887
+ "grad_norm": 0.9872453212738037,
1888
+ "learning_rate": 0.00010993831391363948,
1889
+ "loss": 1.7427,
1890
+ "step": 2630
1891
+ },
1892
+ {
1893
+ "epoch": 0.9,
1894
+ "grad_norm": 1.1201390027999878,
1895
+ "learning_rate": 0.00010959561343385883,
1896
+ "loss": 1.7696,
1897
+ "step": 2640
1898
+ },
1899
+ {
1900
+ "epoch": 0.91,
1901
+ "grad_norm": 1.1584488153457642,
1902
+ "learning_rate": 0.00010925291295407813,
1903
+ "loss": 1.6236,
1904
+ "step": 2650
1905
+ },
1906
+ {
1907
+ "epoch": 0.91,
1908
+ "grad_norm": 0.8254250884056091,
1909
+ "learning_rate": 0.00010891021247429747,
1910
+ "loss": 1.6214,
1911
+ "step": 2660
1912
+ },
1913
+ {
1914
+ "epoch": 0.91,
1915
+ "grad_norm": 0.9825947284698486,
1916
+ "learning_rate": 0.0001085675119945168,
1917
+ "loss": 1.7889,
1918
+ "step": 2670
1919
+ },
1920
+ {
1921
+ "epoch": 0.92,
1922
+ "grad_norm": 1.0265246629714966,
1923
+ "learning_rate": 0.00010822481151473613,
1924
+ "loss": 1.7283,
1925
+ "step": 2680
1926
+ },
1927
+ {
1928
+ "epoch": 0.92,
1929
+ "grad_norm": 0.891777515411377,
1930
+ "learning_rate": 0.00010788211103495545,
1931
+ "loss": 1.8176,
1932
+ "step": 2690
1933
+ },
1934
+ {
1935
+ "epoch": 0.92,
1936
+ "grad_norm": 0.8920706510543823,
1937
+ "learning_rate": 0.00010753941055517478,
1938
+ "loss": 1.7676,
1939
+ "step": 2700
1940
+ },
1941
+ {
1942
+ "epoch": 0.93,
1943
+ "grad_norm": 1.072204828262329,
1944
+ "learning_rate": 0.00010719671007539411,
1945
+ "loss": 1.5836,
1946
+ "step": 2710
1947
+ },
1948
+ {
1949
+ "epoch": 0.93,
1950
+ "grad_norm": 0.9175311923027039,
1951
+ "learning_rate": 0.00010685400959561345,
1952
+ "loss": 1.8073,
1953
+ "step": 2720
1954
+ },
1955
+ {
1956
+ "epoch": 0.94,
1957
+ "grad_norm": 0.6199253797531128,
1958
+ "learning_rate": 0.00010651130911583275,
1959
+ "loss": 1.828,
1960
+ "step": 2730
1961
+ },
1962
+ {
1963
+ "epoch": 0.94,
1964
+ "grad_norm": 0.653229296207428,
1965
+ "learning_rate": 0.0001061686086360521,
1966
+ "loss": 1.7308,
1967
+ "step": 2740
1968
+ },
1969
+ {
1970
+ "epoch": 0.94,
1971
+ "grad_norm": 0.790413498878479,
1972
+ "learning_rate": 0.00010582590815627143,
1973
+ "loss": 1.8169,
1974
+ "step": 2750
1975
+ },
1976
+ {
1977
+ "epoch": 0.95,
1978
+ "grad_norm": 0.8657679557800293,
1979
+ "learning_rate": 0.00010548320767649076,
1980
+ "loss": 1.7453,
1981
+ "step": 2760
1982
+ },
1983
+ {
1984
+ "epoch": 0.95,
1985
+ "grad_norm": 0.6758552193641663,
1986
+ "learning_rate": 0.00010514050719671007,
1987
+ "loss": 1.7171,
1988
+ "step": 2770
1989
+ },
1990
+ {
1991
+ "epoch": 0.95,
1992
+ "grad_norm": 1.0935484170913696,
1993
+ "learning_rate": 0.0001047978067169294,
1994
+ "loss": 1.6754,
1995
+ "step": 2780
1996
+ },
1997
+ {
1998
+ "epoch": 0.96,
1999
+ "grad_norm": 0.8095535635948181,
2000
+ "learning_rate": 0.00010445510623714873,
2001
+ "loss": 1.8387,
2002
+ "step": 2790
2003
+ },
2004
+ {
2005
+ "epoch": 0.96,
2006
+ "grad_norm": 0.8804395198822021,
2007
+ "learning_rate": 0.00010411240575736808,
2008
+ "loss": 1.7839,
2009
+ "step": 2800
2010
+ },
2011
+ {
2012
+ "epoch": 0.96,
2013
+ "grad_norm": 0.945090115070343,
2014
+ "learning_rate": 0.0001037697052775874,
2015
+ "loss": 1.7196,
2016
+ "step": 2810
2017
+ },
2018
+ {
2019
+ "epoch": 0.97,
2020
+ "grad_norm": 0.6158414483070374,
2021
+ "learning_rate": 0.00010342700479780672,
2022
+ "loss": 1.8011,
2023
+ "step": 2820
2024
+ },
2025
+ {
2026
+ "epoch": 0.97,
2027
+ "grad_norm": 0.7917384505271912,
2028
+ "learning_rate": 0.00010308430431802605,
2029
+ "loss": 1.744,
2030
+ "step": 2830
2031
+ },
2032
+ {
2033
+ "epoch": 0.97,
2034
+ "grad_norm": 0.6415919065475464,
2035
+ "learning_rate": 0.00010274160383824538,
2036
+ "loss": 1.6379,
2037
+ "step": 2840
2038
+ },
2039
+ {
2040
+ "epoch": 0.98,
2041
+ "grad_norm": 0.6077090501785278,
2042
+ "learning_rate": 0.00010239890335846471,
2043
+ "loss": 1.657,
2044
+ "step": 2850
2045
+ },
2046
+ {
2047
+ "epoch": 0.98,
2048
+ "grad_norm": 1.036901593208313,
2049
+ "learning_rate": 0.00010205620287868403,
2050
+ "loss": 1.7059,
2051
+ "step": 2860
2052
+ },
2053
+ {
2054
+ "epoch": 0.98,
2055
+ "grad_norm": 0.7633301019668579,
2056
+ "learning_rate": 0.00010171350239890336,
2057
+ "loss": 1.8085,
2058
+ "step": 2870
2059
+ },
2060
+ {
2061
+ "epoch": 0.99,
2062
+ "grad_norm": 1.04219651222229,
2063
+ "learning_rate": 0.0001013708019191227,
2064
+ "loss": 1.6641,
2065
+ "step": 2880
2066
+ },
2067
+ {
2068
+ "epoch": 0.99,
2069
+ "grad_norm": 0.9899976849555969,
2070
+ "learning_rate": 0.00010102810143934203,
2071
+ "loss": 1.6819,
2072
+ "step": 2890
2073
+ },
2074
+ {
2075
+ "epoch": 0.99,
2076
+ "grad_norm": 0.755636990070343,
2077
+ "learning_rate": 0.00010068540095956133,
2078
+ "loss": 1.7573,
2079
+ "step": 2900
2080
+ },
2081
+ {
2082
+ "epoch": 1.0,
2083
+ "grad_norm": 1.1326630115509033,
2084
+ "learning_rate": 0.00010034270047978068,
2085
+ "loss": 1.6942,
2086
+ "step": 2910
2087
+ },
2088
+ {
2089
+ "epoch": 1.0,
2090
+ "grad_norm": 0.7579949498176575,
2091
+ "learning_rate": 0.0001,
2092
+ "loss": 1.6903,
2093
+ "step": 2920
2094
+ },
2095
+ {
2096
+ "epoch": 1.0,
2097
+ "grad_norm": 0.7203909754753113,
2098
+ "learning_rate": 9.965729952021933e-05,
2099
+ "loss": 1.5946,
2100
+ "step": 2930
2101
+ },
2102
+ {
2103
+ "epoch": 1.01,
2104
+ "grad_norm": 0.8731165528297424,
2105
+ "learning_rate": 9.931459904043866e-05,
2106
+ "loss": 1.6224,
2107
+ "step": 2940
2108
+ },
2109
+ {
2110
+ "epoch": 1.01,
2111
+ "grad_norm": 0.6287246942520142,
2112
+ "learning_rate": 9.8971898560658e-05,
2113
+ "loss": 1.6279,
2114
+ "step": 2950
2115
+ },
2116
+ {
2117
+ "epoch": 1.01,
2118
+ "grad_norm": 0.8794381618499756,
2119
+ "learning_rate": 9.862919808087731e-05,
2120
+ "loss": 1.7103,
2121
+ "step": 2960
2122
+ },
2123
+ {
2124
+ "epoch": 1.02,
2125
+ "grad_norm": 1.1305402517318726,
2126
+ "learning_rate": 9.828649760109665e-05,
2127
+ "loss": 1.5876,
2128
+ "step": 2970
2129
+ },
2130
+ {
2131
+ "epoch": 1.02,
2132
+ "grad_norm": 1.349693775177002,
2133
+ "learning_rate": 9.794379712131597e-05,
2134
+ "loss": 1.549,
2135
+ "step": 2980
2136
+ },
2137
+ {
2138
+ "epoch": 1.02,
2139
+ "grad_norm": 1.1124284267425537,
2140
+ "learning_rate": 9.76010966415353e-05,
2141
+ "loss": 1.5688,
2142
+ "step": 2990
2143
+ },
2144
+ {
2145
+ "epoch": 1.03,
2146
+ "grad_norm": 0.5864982604980469,
2147
+ "learning_rate": 9.725839616175463e-05,
2148
+ "loss": 1.6437,
2149
+ "step": 3000
2150
+ },
2151
+ {
2152
+ "epoch": 1.03,
2153
+ "eval_loss": 1.8886157274246216,
2154
+ "eval_runtime": 33.1481,
2155
+ "eval_samples_per_second": 30.168,
2156
+ "eval_steps_per_second": 3.771,
2157
+ "step": 3000
2158
+ },
2159
+ {
2160
+ "epoch": 1.03,
2161
+ "grad_norm": 0.8807237148284912,
2162
+ "learning_rate": 9.691569568197396e-05,
2163
+ "loss": 1.5888,
2164
+ "step": 3010
2165
+ },
2166
+ {
2167
+ "epoch": 1.03,
2168
+ "grad_norm": 0.8454139232635498,
2169
+ "learning_rate": 9.657299520219329e-05,
2170
+ "loss": 1.5414,
2171
+ "step": 3020
2172
+ },
2173
+ {
2174
+ "epoch": 1.04,
2175
+ "grad_norm": 0.9541159272193909,
2176
+ "learning_rate": 9.623029472241262e-05,
2177
+ "loss": 1.7525,
2178
+ "step": 3030
2179
+ },
2180
+ {
2181
+ "epoch": 1.04,
2182
+ "grad_norm": 1.38509202003479,
2183
+ "learning_rate": 9.588759424263193e-05,
2184
+ "loss": 1.5302,
2185
+ "step": 3040
2186
+ },
2187
+ {
2188
+ "epoch": 1.04,
2189
+ "grad_norm": 1.242966651916504,
2190
+ "learning_rate": 9.554489376285128e-05,
2191
+ "loss": 1.6085,
2192
+ "step": 3050
2193
+ },
2194
+ {
2195
+ "epoch": 1.05,
2196
+ "grad_norm": 1.1269468069076538,
2197
+ "learning_rate": 9.52021932830706e-05,
2198
+ "loss": 1.603,
2199
+ "step": 3060
2200
+ },
2201
+ {
2202
+ "epoch": 1.05,
2203
+ "grad_norm": 1.1521382331848145,
2204
+ "learning_rate": 9.485949280328992e-05,
2205
+ "loss": 1.6984,
2206
+ "step": 3070
2207
+ },
2208
+ {
2209
+ "epoch": 1.05,
2210
+ "grad_norm": 1.3359086513519287,
2211
+ "learning_rate": 9.451679232350927e-05,
2212
+ "loss": 1.4839,
2213
+ "step": 3080
2214
+ },
2215
+ {
2216
+ "epoch": 1.06,
2217
+ "grad_norm": 1.057581901550293,
2218
+ "learning_rate": 9.417409184372858e-05,
2219
+ "loss": 1.5541,
2220
+ "step": 3090
2221
+ },
2222
+ {
2223
+ "epoch": 1.06,
2224
+ "grad_norm": 1.090909719467163,
2225
+ "learning_rate": 9.383139136394791e-05,
2226
+ "loss": 1.5811,
2227
+ "step": 3100
2228
+ },
2229
+ {
2230
+ "epoch": 1.07,
2231
+ "grad_norm": 1.3244885206222534,
2232
+ "learning_rate": 9.348869088416724e-05,
2233
+ "loss": 1.6006,
2234
+ "step": 3110
2235
+ },
2236
+ {
2237
+ "epoch": 1.07,
2238
+ "grad_norm": 0.8855965733528137,
2239
+ "learning_rate": 9.314599040438657e-05,
2240
+ "loss": 1.5577,
2241
+ "step": 3120
2242
+ },
2243
+ {
2244
+ "epoch": 1.07,
2245
+ "grad_norm": 0.9480008482933044,
2246
+ "learning_rate": 9.28032899246059e-05,
2247
+ "loss": 1.6064,
2248
+ "step": 3130
2249
+ },
2250
+ {
2251
+ "epoch": 1.08,
2252
+ "grad_norm": 1.397888422012329,
2253
+ "learning_rate": 9.246058944482523e-05,
2254
+ "loss": 1.5708,
2255
+ "step": 3140
2256
+ },
2257
+ {
2258
+ "epoch": 1.08,
2259
+ "grad_norm": 0.8178092241287231,
2260
+ "learning_rate": 9.211788896504455e-05,
2261
+ "loss": 1.4722,
2262
+ "step": 3150
2263
+ },
2264
+ {
2265
+ "epoch": 1.08,
2266
+ "grad_norm": 1.3776417970657349,
2267
+ "learning_rate": 9.177518848526389e-05,
2268
+ "loss": 1.6941,
2269
+ "step": 3160
2270
+ },
2271
+ {
2272
+ "epoch": 1.09,
2273
+ "grad_norm": 1.3224530220031738,
2274
+ "learning_rate": 9.14324880054832e-05,
2275
+ "loss": 1.5414,
2276
+ "step": 3170
2277
+ },
2278
+ {
2279
+ "epoch": 1.09,
2280
+ "grad_norm": 1.3367009162902832,
2281
+ "learning_rate": 9.108978752570254e-05,
2282
+ "loss": 1.6275,
2283
+ "step": 3180
2284
+ },
2285
+ {
2286
+ "epoch": 1.09,
2287
+ "grad_norm": 1.0063951015472412,
2288
+ "learning_rate": 9.074708704592187e-05,
2289
+ "loss": 1.6761,
2290
+ "step": 3190
2291
+ },
2292
+ {
2293
+ "epoch": 1.1,
2294
+ "grad_norm": 1.320760726928711,
2295
+ "learning_rate": 9.04043865661412e-05,
2296
+ "loss": 1.5896,
2297
+ "step": 3200
2298
+ },
2299
+ {
2300
+ "epoch": 1.1,
2301
+ "grad_norm": 1.0159672498703003,
2302
+ "learning_rate": 9.006168608636053e-05,
2303
+ "loss": 1.5728,
2304
+ "step": 3210
2305
+ },
2306
+ {
2307
+ "epoch": 1.1,
2308
+ "grad_norm": 1.095314383506775,
2309
+ "learning_rate": 8.971898560657985e-05,
2310
+ "loss": 1.5329,
2311
+ "step": 3220
2312
+ },
2313
+ {
2314
+ "epoch": 1.11,
2315
+ "grad_norm": 1.212713360786438,
2316
+ "learning_rate": 8.937628512679918e-05,
2317
+ "loss": 1.5746,
2318
+ "step": 3230
2319
+ },
2320
+ {
2321
+ "epoch": 1.11,
2322
+ "grad_norm": 0.8203460574150085,
2323
+ "learning_rate": 8.903358464701851e-05,
2324
+ "loss": 1.6119,
2325
+ "step": 3240
2326
+ },
2327
+ {
2328
+ "epoch": 1.11,
2329
+ "grad_norm": 0.9643343091011047,
2330
+ "learning_rate": 8.869088416723784e-05,
2331
+ "loss": 1.5893,
2332
+ "step": 3250
2333
+ },
2334
+ {
2335
+ "epoch": 1.12,
2336
+ "grad_norm": 1.2415894269943237,
2337
+ "learning_rate": 8.834818368745716e-05,
2338
+ "loss": 1.6291,
2339
+ "step": 3260
2340
+ },
2341
+ {
2342
+ "epoch": 1.12,
2343
+ "grad_norm": 1.826658844947815,
2344
+ "learning_rate": 8.80054832076765e-05,
2345
+ "loss": 1.6394,
2346
+ "step": 3270
2347
+ },
2348
+ {
2349
+ "epoch": 1.12,
2350
+ "grad_norm": 1.3455665111541748,
2351
+ "learning_rate": 8.766278272789582e-05,
2352
+ "loss": 1.568,
2353
+ "step": 3280
2354
+ },
2355
+ {
2356
+ "epoch": 1.13,
2357
+ "grad_norm": 1.8909701108932495,
2358
+ "learning_rate": 8.732008224811515e-05,
2359
+ "loss": 1.5733,
2360
+ "step": 3290
2361
+ },
2362
+ {
2363
+ "epoch": 1.13,
2364
+ "grad_norm": 1.4277849197387695,
2365
+ "learning_rate": 8.697738176833448e-05,
2366
+ "loss": 1.6339,
2367
+ "step": 3300
2368
+ },
2369
+ {
2370
+ "epoch": 1.13,
2371
+ "grad_norm": 0.9563093185424805,
2372
+ "learning_rate": 8.663468128855381e-05,
2373
+ "loss": 1.5775,
2374
+ "step": 3310
2375
+ },
2376
+ {
2377
+ "epoch": 1.14,
2378
+ "grad_norm": 0.8461637496948242,
2379
+ "learning_rate": 8.629198080877314e-05,
2380
+ "loss": 1.653,
2381
+ "step": 3320
2382
+ },
2383
+ {
2384
+ "epoch": 1.14,
2385
+ "grad_norm": 1.0858458280563354,
2386
+ "learning_rate": 8.594928032899247e-05,
2387
+ "loss": 1.4778,
2388
+ "step": 3330
2389
+ },
2390
+ {
2391
+ "epoch": 1.14,
2392
+ "grad_norm": 1.1627178192138672,
2393
+ "learning_rate": 8.560657984921178e-05,
2394
+ "loss": 1.5374,
2395
+ "step": 3340
2396
+ },
2397
+ {
2398
+ "epoch": 1.15,
2399
+ "grad_norm": 1.196664571762085,
2400
+ "learning_rate": 8.526387936943113e-05,
2401
+ "loss": 1.6483,
2402
+ "step": 3350
2403
+ },
2404
+ {
2405
+ "epoch": 1.15,
2406
+ "grad_norm": 1.1990993022918701,
2407
+ "learning_rate": 8.492117888965046e-05,
2408
+ "loss": 1.5993,
2409
+ "step": 3360
2410
+ },
2411
+ {
2412
+ "epoch": 1.15,
2413
+ "grad_norm": 1.0623687505722046,
2414
+ "learning_rate": 8.457847840986977e-05,
2415
+ "loss": 1.5743,
2416
+ "step": 3370
2417
+ },
2418
+ {
2419
+ "epoch": 1.16,
2420
+ "grad_norm": 1.1684637069702148,
2421
+ "learning_rate": 8.423577793008912e-05,
2422
+ "loss": 1.5546,
2423
+ "step": 3380
2424
+ },
2425
+ {
2426
+ "epoch": 1.16,
2427
+ "grad_norm": 1.2448011636734009,
2428
+ "learning_rate": 8.389307745030843e-05,
2429
+ "loss": 1.496,
2430
+ "step": 3390
2431
+ },
2432
+ {
2433
+ "epoch": 1.16,
2434
+ "grad_norm": 0.9411953091621399,
2435
+ "learning_rate": 8.355037697052776e-05,
2436
+ "loss": 1.5966,
2437
+ "step": 3400
2438
+ },
2439
+ {
2440
+ "epoch": 1.17,
2441
+ "grad_norm": 1.0667563676834106,
2442
+ "learning_rate": 8.320767649074709e-05,
2443
+ "loss": 1.5128,
2444
+ "step": 3410
2445
+ },
2446
+ {
2447
+ "epoch": 1.17,
2448
+ "grad_norm": 1.50753653049469,
2449
+ "learning_rate": 8.286497601096642e-05,
2450
+ "loss": 1.5772,
2451
+ "step": 3420
2452
+ },
2453
+ {
2454
+ "epoch": 1.17,
2455
+ "grad_norm": 0.9346134662628174,
2456
+ "learning_rate": 8.252227553118574e-05,
2457
+ "loss": 1.6321,
2458
+ "step": 3430
2459
+ },
2460
+ {
2461
+ "epoch": 1.18,
2462
+ "grad_norm": 1.304190754890442,
2463
+ "learning_rate": 8.217957505140508e-05,
2464
+ "loss": 1.5656,
2465
+ "step": 3440
2466
+ },
2467
+ {
2468
+ "epoch": 1.18,
2469
+ "grad_norm": 1.058018684387207,
2470
+ "learning_rate": 8.18368745716244e-05,
2471
+ "loss": 1.5413,
2472
+ "step": 3450
2473
+ },
2474
+ {
2475
+ "epoch": 1.19,
2476
+ "grad_norm": 1.15809166431427,
2477
+ "learning_rate": 8.149417409184373e-05,
2478
+ "loss": 1.5673,
2479
+ "step": 3460
2480
+ },
2481
+ {
2482
+ "epoch": 1.19,
2483
+ "grad_norm": 1.092393159866333,
2484
+ "learning_rate": 8.115147361206306e-05,
2485
+ "loss": 1.5962,
2486
+ "step": 3470
2487
+ },
2488
+ {
2489
+ "epoch": 1.19,
2490
+ "grad_norm": 0.9390305876731873,
2491
+ "learning_rate": 8.080877313228239e-05,
2492
+ "loss": 1.5565,
2493
+ "step": 3480
2494
+ },
2495
+ {
2496
+ "epoch": 1.2,
2497
+ "grad_norm": 1.002120852470398,
2498
+ "learning_rate": 8.046607265250173e-05,
2499
+ "loss": 1.5803,
2500
+ "step": 3490
2501
+ },
2502
+ {
2503
+ "epoch": 1.2,
2504
+ "grad_norm": 1.0857172012329102,
2505
+ "learning_rate": 8.012337217272105e-05,
2506
+ "loss": 1.6345,
2507
+ "step": 3500
2508
+ },
2509
+ {
2510
+ "epoch": 1.2,
2511
+ "eval_loss": 1.8998303413391113,
2512
+ "eval_runtime": 33.1629,
2513
+ "eval_samples_per_second": 30.154,
2514
+ "eval_steps_per_second": 3.769,
2515
+ "step": 3500
2516
+ },
2517
+ {
2518
+ "epoch": 1.2,
2519
+ "grad_norm": 0.9931670427322388,
2520
+ "learning_rate": 7.978067169294038e-05,
2521
+ "loss": 1.605,
2522
+ "step": 3510
2523
+ },
2524
+ {
2525
+ "epoch": 1.21,
2526
+ "grad_norm": 1.3759890794754028,
2527
+ "learning_rate": 7.94379712131597e-05,
2528
+ "loss": 1.5059,
2529
+ "step": 3520
2530
+ },
2531
+ {
2532
+ "epoch": 1.21,
2533
+ "grad_norm": 1.2301968336105347,
2534
+ "learning_rate": 7.909527073337903e-05,
2535
+ "loss": 1.582,
2536
+ "step": 3530
2537
+ },
2538
+ {
2539
+ "epoch": 1.21,
2540
+ "grad_norm": 1.1518924236297607,
2541
+ "learning_rate": 7.875257025359835e-05,
2542
+ "loss": 1.5839,
2543
+ "step": 3540
2544
+ },
2545
+ {
2546
+ "epoch": 1.22,
2547
+ "grad_norm": 0.9161165952682495,
2548
+ "learning_rate": 7.84098697738177e-05,
2549
+ "loss": 1.5494,
2550
+ "step": 3550
2551
+ },
2552
+ {
2553
+ "epoch": 1.22,
2554
+ "grad_norm": 1.250705599784851,
2555
+ "learning_rate": 7.806716929403701e-05,
2556
+ "loss": 1.5178,
2557
+ "step": 3560
2558
+ },
2559
+ {
2560
+ "epoch": 1.22,
2561
+ "grad_norm": 0.7702249884605408,
2562
+ "learning_rate": 7.772446881425634e-05,
2563
+ "loss": 1.644,
2564
+ "step": 3570
2565
+ },
2566
+ {
2567
+ "epoch": 1.23,
2568
+ "grad_norm": 1.4425973892211914,
2569
+ "learning_rate": 7.738176833447567e-05,
2570
+ "loss": 1.5009,
2571
+ "step": 3580
2572
+ },
2573
+ {
2574
+ "epoch": 1.23,
2575
+ "grad_norm": 1.2036337852478027,
2576
+ "learning_rate": 7.7039067854695e-05,
2577
+ "loss": 1.5456,
2578
+ "step": 3590
2579
+ },
2580
+ {
2581
+ "epoch": 1.23,
2582
+ "grad_norm": 1.4006402492523193,
2583
+ "learning_rate": 7.669636737491433e-05,
2584
+ "loss": 1.5511,
2585
+ "step": 3600
2586
+ },
2587
+ {
2588
+ "epoch": 1.24,
2589
+ "grad_norm": 1.1983481645584106,
2590
+ "learning_rate": 7.635366689513366e-05,
2591
+ "loss": 1.5645,
2592
+ "step": 3610
2593
+ },
2594
+ {
2595
+ "epoch": 1.24,
2596
+ "grad_norm": 1.2755049467086792,
2597
+ "learning_rate": 7.601096641535297e-05,
2598
+ "loss": 1.6512,
2599
+ "step": 3620
2600
+ },
2601
+ {
2602
+ "epoch": 1.24,
2603
+ "grad_norm": 1.3783161640167236,
2604
+ "learning_rate": 7.566826593557232e-05,
2605
+ "loss": 1.6747,
2606
+ "step": 3630
2607
+ },
2608
+ {
2609
+ "epoch": 1.25,
2610
+ "grad_norm": 1.1947081089019775,
2611
+ "learning_rate": 7.532556545579165e-05,
2612
+ "loss": 1.6605,
2613
+ "step": 3640
2614
+ },
2615
+ {
2616
+ "epoch": 1.25,
2617
+ "grad_norm": 1.2230151891708374,
2618
+ "learning_rate": 7.498286497601096e-05,
2619
+ "loss": 1.6187,
2620
+ "step": 3650
2621
+ },
2622
+ {
2623
+ "epoch": 1.25,
2624
+ "grad_norm": 1.372226595878601,
2625
+ "learning_rate": 7.464016449623031e-05,
2626
+ "loss": 1.6354,
2627
+ "step": 3660
2628
+ },
2629
+ {
2630
+ "epoch": 1.26,
2631
+ "grad_norm": 1.2375085353851318,
2632
+ "learning_rate": 7.429746401644962e-05,
2633
+ "loss": 1.656,
2634
+ "step": 3670
2635
+ },
2636
+ {
2637
+ "epoch": 1.26,
2638
+ "grad_norm": 0.9703730940818787,
2639
+ "learning_rate": 7.395476353666895e-05,
2640
+ "loss": 1.5571,
2641
+ "step": 3680
2642
+ },
2643
+ {
2644
+ "epoch": 1.26,
2645
+ "grad_norm": 1.3475947380065918,
2646
+ "learning_rate": 7.361206305688828e-05,
2647
+ "loss": 1.5487,
2648
+ "step": 3690
2649
+ },
2650
+ {
2651
+ "epoch": 1.27,
2652
+ "grad_norm": 1.3879302740097046,
2653
+ "learning_rate": 7.326936257710761e-05,
2654
+ "loss": 1.6702,
2655
+ "step": 3700
2656
+ },
2657
+ {
2658
+ "epoch": 1.27,
2659
+ "grad_norm": 1.4043548107147217,
2660
+ "learning_rate": 7.292666209732694e-05,
2661
+ "loss": 1.5555,
2662
+ "step": 3710
2663
+ },
2664
+ {
2665
+ "epoch": 1.27,
2666
+ "grad_norm": 1.2937321662902832,
2667
+ "learning_rate": 7.258396161754627e-05,
2668
+ "loss": 1.5959,
2669
+ "step": 3720
2670
+ },
2671
+ {
2672
+ "epoch": 1.28,
2673
+ "grad_norm": 1.4525338411331177,
2674
+ "learning_rate": 7.224126113776559e-05,
2675
+ "loss": 1.6252,
2676
+ "step": 3730
2677
+ },
2678
+ {
2679
+ "epoch": 1.28,
2680
+ "grad_norm": 1.1089144945144653,
2681
+ "learning_rate": 7.189856065798493e-05,
2682
+ "loss": 1.5027,
2683
+ "step": 3740
2684
+ },
2685
+ {
2686
+ "epoch": 1.28,
2687
+ "grad_norm": 1.2625998258590698,
2688
+ "learning_rate": 7.155586017820425e-05,
2689
+ "loss": 1.5907,
2690
+ "step": 3750
2691
+ },
2692
+ {
2693
+ "epoch": 1.29,
2694
+ "grad_norm": 1.2458665370941162,
2695
+ "learning_rate": 7.121315969842358e-05,
2696
+ "loss": 1.54,
2697
+ "step": 3760
2698
+ },
2699
+ {
2700
+ "epoch": 1.29,
2701
+ "grad_norm": 1.2830859422683716,
2702
+ "learning_rate": 7.087045921864292e-05,
2703
+ "loss": 1.5867,
2704
+ "step": 3770
2705
+ },
2706
+ {
2707
+ "epoch": 1.29,
2708
+ "grad_norm": 1.0032719373703003,
2709
+ "learning_rate": 7.052775873886224e-05,
2710
+ "loss": 1.5374,
2711
+ "step": 3780
2712
+ },
2713
+ {
2714
+ "epoch": 1.3,
2715
+ "grad_norm": 0.9105421304702759,
2716
+ "learning_rate": 7.018505825908157e-05,
2717
+ "loss": 1.528,
2718
+ "step": 3790
2719
+ },
2720
+ {
2721
+ "epoch": 1.3,
2722
+ "grad_norm": 1.3588030338287354,
2723
+ "learning_rate": 6.98423577793009e-05,
2724
+ "loss": 1.6368,
2725
+ "step": 3800
2726
+ },
2727
+ {
2728
+ "epoch": 1.3,
2729
+ "grad_norm": 1.4903500080108643,
2730
+ "learning_rate": 6.949965729952023e-05,
2731
+ "loss": 1.675,
2732
+ "step": 3810
2733
+ },
2734
+ {
2735
+ "epoch": 1.31,
2736
+ "grad_norm": 1.229722261428833,
2737
+ "learning_rate": 6.915695681973956e-05,
2738
+ "loss": 1.555,
2739
+ "step": 3820
2740
+ },
2741
+ {
2742
+ "epoch": 1.31,
2743
+ "grad_norm": 0.9523776769638062,
2744
+ "learning_rate": 6.881425633995888e-05,
2745
+ "loss": 1.6608,
2746
+ "step": 3830
2747
+ },
2748
+ {
2749
+ "epoch": 1.32,
2750
+ "grad_norm": 1.986708164215088,
2751
+ "learning_rate": 6.84715558601782e-05,
2752
+ "loss": 1.7199,
2753
+ "step": 3840
2754
+ },
2755
+ {
2756
+ "epoch": 1.32,
2757
+ "grad_norm": 0.79183429479599,
2758
+ "learning_rate": 6.812885538039754e-05,
2759
+ "loss": 1.5034,
2760
+ "step": 3850
2761
+ },
2762
+ {
2763
+ "epoch": 1.32,
2764
+ "grad_norm": 1.1760715246200562,
2765
+ "learning_rate": 6.778615490061686e-05,
2766
+ "loss": 1.6812,
2767
+ "step": 3860
2768
+ },
2769
+ {
2770
+ "epoch": 1.33,
2771
+ "grad_norm": 1.7899055480957031,
2772
+ "learning_rate": 6.744345442083619e-05,
2773
+ "loss": 1.7389,
2774
+ "step": 3870
2775
+ },
2776
+ {
2777
+ "epoch": 1.33,
2778
+ "grad_norm": 1.2628593444824219,
2779
+ "learning_rate": 6.710075394105552e-05,
2780
+ "loss": 1.5317,
2781
+ "step": 3880
2782
+ },
2783
+ {
2784
+ "epoch": 1.33,
2785
+ "grad_norm": 1.037351131439209,
2786
+ "learning_rate": 6.675805346127485e-05,
2787
+ "loss": 1.5858,
2788
+ "step": 3890
2789
+ },
2790
+ {
2791
+ "epoch": 1.34,
2792
+ "grad_norm": 1.2006704807281494,
2793
+ "learning_rate": 6.641535298149417e-05,
2794
+ "loss": 1.4587,
2795
+ "step": 3900
2796
+ },
2797
+ {
2798
+ "epoch": 1.34,
2799
+ "grad_norm": 1.0877715349197388,
2800
+ "learning_rate": 6.607265250171351e-05,
2801
+ "loss": 1.5306,
2802
+ "step": 3910
2803
+ },
2804
+ {
2805
+ "epoch": 1.34,
2806
+ "grad_norm": 1.4047476053237915,
2807
+ "learning_rate": 6.572995202193284e-05,
2808
+ "loss": 1.5603,
2809
+ "step": 3920
2810
+ },
2811
+ {
2812
+ "epoch": 1.35,
2813
+ "grad_norm": 1.2444441318511963,
2814
+ "learning_rate": 6.538725154215215e-05,
2815
+ "loss": 1.5809,
2816
+ "step": 3930
2817
+ },
2818
+ {
2819
+ "epoch": 1.35,
2820
+ "grad_norm": 1.5738134384155273,
2821
+ "learning_rate": 6.50445510623715e-05,
2822
+ "loss": 1.5606,
2823
+ "step": 3940
2824
+ },
2825
+ {
2826
+ "epoch": 1.35,
2827
+ "grad_norm": 1.4850690364837646,
2828
+ "learning_rate": 6.470185058259081e-05,
2829
+ "loss": 1.4945,
2830
+ "step": 3950
2831
+ },
2832
+ {
2833
+ "epoch": 1.36,
2834
+ "grad_norm": 1.3746342658996582,
2835
+ "learning_rate": 6.435915010281016e-05,
2836
+ "loss": 1.5152,
2837
+ "step": 3960
2838
+ },
2839
+ {
2840
+ "epoch": 1.36,
2841
+ "grad_norm": 1.139249324798584,
2842
+ "learning_rate": 6.401644962302947e-05,
2843
+ "loss": 1.6004,
2844
+ "step": 3970
2845
+ },
2846
+ {
2847
+ "epoch": 1.36,
2848
+ "grad_norm": 1.3590480089187622,
2849
+ "learning_rate": 6.36737491432488e-05,
2850
+ "loss": 1.4926,
2851
+ "step": 3980
2852
+ },
2853
+ {
2854
+ "epoch": 1.37,
2855
+ "grad_norm": 1.6366995573043823,
2856
+ "learning_rate": 6.333104866346813e-05,
2857
+ "loss": 1.6734,
2858
+ "step": 3990
2859
+ },
2860
+ {
2861
+ "epoch": 1.37,
2862
+ "grad_norm": 1.1154892444610596,
2863
+ "learning_rate": 6.298834818368746e-05,
2864
+ "loss": 1.5628,
2865
+ "step": 4000
2866
+ },
2867
+ {
2868
+ "epoch": 1.37,
2869
+ "eval_loss": 1.9075069427490234,
2870
+ "eval_runtime": 33.1019,
2871
+ "eval_samples_per_second": 30.21,
2872
+ "eval_steps_per_second": 3.776,
2873
+ "step": 4000
2874
+ },
2875
+ {
2876
+ "epoch": 1.37,
2877
+ "grad_norm": 1.123923897743225,
2878
+ "learning_rate": 6.264564770390678e-05,
2879
+ "loss": 1.6206,
2880
+ "step": 4010
2881
+ },
2882
+ {
2883
+ "epoch": 1.38,
2884
+ "grad_norm": 1.3015213012695312,
2885
+ "learning_rate": 6.230294722412612e-05,
2886
+ "loss": 1.6292,
2887
+ "step": 4020
2888
+ },
2889
+ {
2890
+ "epoch": 1.38,
2891
+ "grad_norm": 1.8867294788360596,
2892
+ "learning_rate": 6.196024674434544e-05,
2893
+ "loss": 1.6625,
2894
+ "step": 4030
2895
+ },
2896
+ {
2897
+ "epoch": 1.38,
2898
+ "grad_norm": 1.5840169191360474,
2899
+ "learning_rate": 6.161754626456477e-05,
2900
+ "loss": 1.6224,
2901
+ "step": 4040
2902
+ },
2903
+ {
2904
+ "epoch": 1.39,
2905
+ "grad_norm": 0.9141889810562134,
2906
+ "learning_rate": 6.12748457847841e-05,
2907
+ "loss": 1.5051,
2908
+ "step": 4050
2909
+ },
2910
+ {
2911
+ "epoch": 1.39,
2912
+ "grad_norm": 1.5261061191558838,
2913
+ "learning_rate": 6.093214530500343e-05,
2914
+ "loss": 1.4289,
2915
+ "step": 4060
2916
+ },
2917
+ {
2918
+ "epoch": 1.39,
2919
+ "grad_norm": 1.2253016233444214,
2920
+ "learning_rate": 6.0589444825222764e-05,
2921
+ "loss": 1.6065,
2922
+ "step": 4070
2923
+ },
2924
+ {
2925
+ "epoch": 1.4,
2926
+ "grad_norm": 1.7163646221160889,
2927
+ "learning_rate": 6.0246744345442087e-05,
2928
+ "loss": 1.5978,
2929
+ "step": 4080
2930
+ },
2931
+ {
2932
+ "epoch": 1.4,
2933
+ "grad_norm": 1.0204969644546509,
2934
+ "learning_rate": 5.9904043865661416e-05,
2935
+ "loss": 1.6267,
2936
+ "step": 4090
2937
+ },
2938
+ {
2939
+ "epoch": 1.4,
2940
+ "grad_norm": 1.9314994812011719,
2941
+ "learning_rate": 5.956134338588074e-05,
2942
+ "loss": 1.6486,
2943
+ "step": 4100
2944
+ },
2945
+ {
2946
+ "epoch": 1.41,
2947
+ "grad_norm": 1.1685149669647217,
2948
+ "learning_rate": 5.9218642906100076e-05,
2949
+ "loss": 1.6397,
2950
+ "step": 4110
2951
+ },
2952
+ {
2953
+ "epoch": 1.41,
2954
+ "grad_norm": 1.422166347503662,
2955
+ "learning_rate": 5.88759424263194e-05,
2956
+ "loss": 1.6419,
2957
+ "step": 4120
2958
+ },
2959
+ {
2960
+ "epoch": 1.41,
2961
+ "grad_norm": 1.3074285984039307,
2962
+ "learning_rate": 5.853324194653873e-05,
2963
+ "loss": 1.565,
2964
+ "step": 4130
2965
+ },
2966
+ {
2967
+ "epoch": 1.42,
2968
+ "grad_norm": 0.965584933757782,
2969
+ "learning_rate": 5.819054146675805e-05,
2970
+ "loss": 1.5841,
2971
+ "step": 4140
2972
+ },
2973
+ {
2974
+ "epoch": 1.42,
2975
+ "grad_norm": 0.9101732969284058,
2976
+ "learning_rate": 5.784784098697739e-05,
2977
+ "loss": 1.6144,
2978
+ "step": 4150
2979
+ },
2980
+ {
2981
+ "epoch": 1.42,
2982
+ "grad_norm": 1.183640718460083,
2983
+ "learning_rate": 5.750514050719671e-05,
2984
+ "loss": 1.5998,
2985
+ "step": 4160
2986
+ },
2987
+ {
2988
+ "epoch": 1.43,
2989
+ "grad_norm": 1.1072790622711182,
2990
+ "learning_rate": 5.716244002741604e-05,
2991
+ "loss": 1.4634,
2992
+ "step": 4170
2993
+ },
2994
+ {
2995
+ "epoch": 1.43,
2996
+ "grad_norm": 1.608017086982727,
2997
+ "learning_rate": 5.681973954763536e-05,
2998
+ "loss": 1.5629,
2999
+ "step": 4180
3000
+ },
3001
+ {
3002
+ "epoch": 1.44,
3003
+ "grad_norm": 1.4969751834869385,
3004
+ "learning_rate": 5.64770390678547e-05,
3005
+ "loss": 1.5966,
3006
+ "step": 4190
3007
+ },
3008
+ {
3009
+ "epoch": 1.44,
3010
+ "grad_norm": 1.727695107460022,
3011
+ "learning_rate": 5.613433858807403e-05,
3012
+ "loss": 1.5456,
3013
+ "step": 4200
3014
+ },
3015
+ {
3016
+ "epoch": 1.44,
3017
+ "grad_norm": 1.4587767124176025,
3018
+ "learning_rate": 5.579163810829335e-05,
3019
+ "loss": 1.5238,
3020
+ "step": 4210
3021
+ },
3022
+ {
3023
+ "epoch": 1.45,
3024
+ "grad_norm": 1.5338579416275024,
3025
+ "learning_rate": 5.544893762851269e-05,
3026
+ "loss": 1.5485,
3027
+ "step": 4220
3028
+ },
3029
+ {
3030
+ "epoch": 1.45,
3031
+ "grad_norm": 0.8002244234085083,
3032
+ "learning_rate": 5.510623714873201e-05,
3033
+ "loss": 1.634,
3034
+ "step": 4230
3035
+ },
3036
+ {
3037
+ "epoch": 1.45,
3038
+ "grad_norm": 1.281417727470398,
3039
+ "learning_rate": 5.476353666895134e-05,
3040
+ "loss": 1.589,
3041
+ "step": 4240
3042
+ },
3043
+ {
3044
+ "epoch": 1.46,
3045
+ "grad_norm": 0.906808078289032,
3046
+ "learning_rate": 5.4420836189170664e-05,
3047
+ "loss": 1.57,
3048
+ "step": 4250
3049
+ },
3050
+ {
3051
+ "epoch": 1.46,
3052
+ "grad_norm": 1.799028992652893,
3053
+ "learning_rate": 5.407813570939e-05,
3054
+ "loss": 1.623,
3055
+ "step": 4260
3056
+ },
3057
+ {
3058
+ "epoch": 1.46,
3059
+ "grad_norm": 1.2560220956802368,
3060
+ "learning_rate": 5.3735435229609324e-05,
3061
+ "loss": 1.4231,
3062
+ "step": 4270
3063
+ },
3064
+ {
3065
+ "epoch": 1.47,
3066
+ "grad_norm": 1.315132737159729,
3067
+ "learning_rate": 5.339273474982865e-05,
3068
+ "loss": 1.553,
3069
+ "step": 4280
3070
+ },
3071
+ {
3072
+ "epoch": 1.47,
3073
+ "grad_norm": 1.1687719821929932,
3074
+ "learning_rate": 5.3050034270047976e-05,
3075
+ "loss": 1.691,
3076
+ "step": 4290
3077
+ },
3078
+ {
3079
+ "epoch": 1.47,
3080
+ "grad_norm": 1.182626724243164,
3081
+ "learning_rate": 5.270733379026731e-05,
3082
+ "loss": 1.58,
3083
+ "step": 4300
3084
+ },
3085
+ {
3086
+ "epoch": 1.48,
3087
+ "grad_norm": 0.819560170173645,
3088
+ "learning_rate": 5.2364633310486636e-05,
3089
+ "loss": 1.574,
3090
+ "step": 4310
3091
+ },
3092
+ {
3093
+ "epoch": 1.48,
3094
+ "grad_norm": 1.4093881845474243,
3095
+ "learning_rate": 5.2021932830705965e-05,
3096
+ "loss": 1.5805,
3097
+ "step": 4320
3098
+ },
3099
+ {
3100
+ "epoch": 1.48,
3101
+ "grad_norm": 2.079927921295166,
3102
+ "learning_rate": 5.167923235092529e-05,
3103
+ "loss": 1.6296,
3104
+ "step": 4330
3105
+ },
3106
+ {
3107
+ "epoch": 1.49,
3108
+ "grad_norm": 1.1056098937988281,
3109
+ "learning_rate": 5.1336531871144625e-05,
3110
+ "loss": 1.4964,
3111
+ "step": 4340
3112
+ },
3113
+ {
3114
+ "epoch": 1.49,
3115
+ "grad_norm": 1.924827218055725,
3116
+ "learning_rate": 5.0993831391363954e-05,
3117
+ "loss": 1.5223,
3118
+ "step": 4350
3119
+ },
3120
+ {
3121
+ "epoch": 1.49,
3122
+ "grad_norm": 1.461719274520874,
3123
+ "learning_rate": 5.065113091158328e-05,
3124
+ "loss": 1.5323,
3125
+ "step": 4360
3126
+ },
3127
+ {
3128
+ "epoch": 1.5,
3129
+ "grad_norm": 1.6647108793258667,
3130
+ "learning_rate": 5.0308430431802614e-05,
3131
+ "loss": 1.6025,
3132
+ "step": 4370
3133
+ },
3134
+ {
3135
+ "epoch": 1.5,
3136
+ "grad_norm": 1.33492910861969,
3137
+ "learning_rate": 4.996572995202194e-05,
3138
+ "loss": 1.5554,
3139
+ "step": 4380
3140
+ },
3141
+ {
3142
+ "epoch": 1.5,
3143
+ "grad_norm": 1.134582757949829,
3144
+ "learning_rate": 4.962302947224126e-05,
3145
+ "loss": 1.621,
3146
+ "step": 4390
3147
+ },
3148
+ {
3149
+ "epoch": 1.51,
3150
+ "grad_norm": 1.315508246421814,
3151
+ "learning_rate": 4.928032899246059e-05,
3152
+ "loss": 1.5828,
3153
+ "step": 4400
3154
+ },
3155
+ {
3156
+ "epoch": 1.51,
3157
+ "grad_norm": 1.3290214538574219,
3158
+ "learning_rate": 4.8937628512679926e-05,
3159
+ "loss": 1.578,
3160
+ "step": 4410
3161
+ },
3162
+ {
3163
+ "epoch": 1.51,
3164
+ "grad_norm": 1.2616337537765503,
3165
+ "learning_rate": 4.859492803289925e-05,
3166
+ "loss": 1.6177,
3167
+ "step": 4420
3168
+ },
3169
+ {
3170
+ "epoch": 1.52,
3171
+ "grad_norm": 1.4099230766296387,
3172
+ "learning_rate": 4.825222755311858e-05,
3173
+ "loss": 1.4926,
3174
+ "step": 4430
3175
+ },
3176
+ {
3177
+ "epoch": 1.52,
3178
+ "grad_norm": 0.9720429182052612,
3179
+ "learning_rate": 4.790952707333791e-05,
3180
+ "loss": 1.6552,
3181
+ "step": 4440
3182
+ },
3183
+ {
3184
+ "epoch": 1.52,
3185
+ "grad_norm": 1.1491189002990723,
3186
+ "learning_rate": 4.756682659355723e-05,
3187
+ "loss": 1.7001,
3188
+ "step": 4450
3189
+ },
3190
+ {
3191
+ "epoch": 1.53,
3192
+ "grad_norm": 1.1790263652801514,
3193
+ "learning_rate": 4.722412611377656e-05,
3194
+ "loss": 1.544,
3195
+ "step": 4460
3196
+ },
3197
+ {
3198
+ "epoch": 1.53,
3199
+ "grad_norm": 1.1880890130996704,
3200
+ "learning_rate": 4.688142563399589e-05,
3201
+ "loss": 1.6053,
3202
+ "step": 4470
3203
+ },
3204
+ {
3205
+ "epoch": 1.53,
3206
+ "grad_norm": 1.0895016193389893,
3207
+ "learning_rate": 4.653872515421522e-05,
3208
+ "loss": 1.455,
3209
+ "step": 4480
3210
+ },
3211
+ {
3212
+ "epoch": 1.54,
3213
+ "grad_norm": 1.230600118637085,
3214
+ "learning_rate": 4.619602467443454e-05,
3215
+ "loss": 1.5752,
3216
+ "step": 4490
3217
+ },
3218
+ {
3219
+ "epoch": 1.54,
3220
+ "grad_norm": 1.4027129411697388,
3221
+ "learning_rate": 4.585332419465387e-05,
3222
+ "loss": 1.5461,
3223
+ "step": 4500
3224
+ },
3225
+ {
3226
+ "epoch": 1.54,
3227
+ "eval_loss": 1.9048413038253784,
3228
+ "eval_runtime": 33.129,
3229
+ "eval_samples_per_second": 30.185,
3230
+ "eval_steps_per_second": 3.773,
3231
+ "step": 4500
3232
+ },
3233
+ {
3234
+ "epoch": 1.54,
3235
+ "grad_norm": 0.8590341806411743,
3236
+ "learning_rate": 4.55106237148732e-05,
3237
+ "loss": 1.6303,
3238
+ "step": 4510
3239
+ },
3240
+ {
3241
+ "epoch": 1.55,
3242
+ "grad_norm": 1.0827935934066772,
3243
+ "learning_rate": 4.516792323509253e-05,
3244
+ "loss": 1.5712,
3245
+ "step": 4520
3246
+ },
3247
+ {
3248
+ "epoch": 1.55,
3249
+ "grad_norm": 0.8795824646949768,
3250
+ "learning_rate": 4.4825222755311855e-05,
3251
+ "loss": 1.4882,
3252
+ "step": 4530
3253
+ },
3254
+ {
3255
+ "epoch": 1.55,
3256
+ "grad_norm": 1.509653091430664,
3257
+ "learning_rate": 4.4482522275531185e-05,
3258
+ "loss": 1.5534,
3259
+ "step": 4540
3260
+ },
3261
+ {
3262
+ "epoch": 1.56,
3263
+ "grad_norm": 1.0400638580322266,
3264
+ "learning_rate": 4.413982179575052e-05,
3265
+ "loss": 1.5681,
3266
+ "step": 4550
3267
+ },
3268
+ {
3269
+ "epoch": 1.56,
3270
+ "grad_norm": 1.1006004810333252,
3271
+ "learning_rate": 4.3797121315969844e-05,
3272
+ "loss": 1.5715,
3273
+ "step": 4560
3274
+ },
3275
+ {
3276
+ "epoch": 1.57,
3277
+ "grad_norm": 1.1621884107589722,
3278
+ "learning_rate": 4.3454420836189174e-05,
3279
+ "loss": 1.6373,
3280
+ "step": 4570
3281
+ },
3282
+ {
3283
+ "epoch": 1.57,
3284
+ "grad_norm": 1.0296626091003418,
3285
+ "learning_rate": 4.3111720356408503e-05,
3286
+ "loss": 1.6076,
3287
+ "step": 4580
3288
+ },
3289
+ {
3290
+ "epoch": 1.57,
3291
+ "grad_norm": 1.6784312725067139,
3292
+ "learning_rate": 4.276901987662783e-05,
3293
+ "loss": 1.6046,
3294
+ "step": 4590
3295
+ },
3296
+ {
3297
+ "epoch": 1.58,
3298
+ "grad_norm": 1.0730016231536865,
3299
+ "learning_rate": 4.2426319396847156e-05,
3300
+ "loss": 1.6317,
3301
+ "step": 4600
3302
+ },
3303
+ {
3304
+ "epoch": 1.58,
3305
+ "grad_norm": 1.0070710182189941,
3306
+ "learning_rate": 4.2083618917066486e-05,
3307
+ "loss": 1.5472,
3308
+ "step": 4610
3309
+ },
3310
+ {
3311
+ "epoch": 1.58,
3312
+ "grad_norm": 1.143546462059021,
3313
+ "learning_rate": 4.1740918437285815e-05,
3314
+ "loss": 1.4993,
3315
+ "step": 4620
3316
+ },
3317
+ {
3318
+ "epoch": 1.59,
3319
+ "grad_norm": 1.8565304279327393,
3320
+ "learning_rate": 4.1398217957505145e-05,
3321
+ "loss": 1.6021,
3322
+ "step": 4630
3323
+ },
3324
+ {
3325
+ "epoch": 1.59,
3326
+ "grad_norm": 1.1914728879928589,
3327
+ "learning_rate": 4.105551747772447e-05,
3328
+ "loss": 1.7231,
3329
+ "step": 4640
3330
+ },
3331
+ {
3332
+ "epoch": 1.59,
3333
+ "grad_norm": 1.6387224197387695,
3334
+ "learning_rate": 4.07128169979438e-05,
3335
+ "loss": 1.5804,
3336
+ "step": 4650
3337
+ },
3338
+ {
3339
+ "epoch": 1.6,
3340
+ "grad_norm": 1.65473210811615,
3341
+ "learning_rate": 4.037011651816313e-05,
3342
+ "loss": 1.6404,
3343
+ "step": 4660
3344
+ },
3345
+ {
3346
+ "epoch": 1.6,
3347
+ "grad_norm": 1.6097077131271362,
3348
+ "learning_rate": 4.002741603838245e-05,
3349
+ "loss": 1.4651,
3350
+ "step": 4670
3351
+ },
3352
+ {
3353
+ "epoch": 1.6,
3354
+ "grad_norm": 1.4290515184402466,
3355
+ "learning_rate": 3.968471555860178e-05,
3356
+ "loss": 1.5668,
3357
+ "step": 4680
3358
+ },
3359
+ {
3360
+ "epoch": 1.61,
3361
+ "grad_norm": 1.047481894493103,
3362
+ "learning_rate": 3.934201507882111e-05,
3363
+ "loss": 1.5275,
3364
+ "step": 4690
3365
+ },
3366
+ {
3367
+ "epoch": 1.61,
3368
+ "grad_norm": 1.3638914823532104,
3369
+ "learning_rate": 3.8999314599040446e-05,
3370
+ "loss": 1.6588,
3371
+ "step": 4700
3372
+ },
3373
+ {
3374
+ "epoch": 1.61,
3375
+ "grad_norm": 1.7712153196334839,
3376
+ "learning_rate": 3.865661411925977e-05,
3377
+ "loss": 1.6079,
3378
+ "step": 4710
3379
+ },
3380
+ {
3381
+ "epoch": 1.62,
3382
+ "grad_norm": 1.0898468494415283,
3383
+ "learning_rate": 3.83139136394791e-05,
3384
+ "loss": 1.574,
3385
+ "step": 4720
3386
+ },
3387
+ {
3388
+ "epoch": 1.62,
3389
+ "grad_norm": 1.4913599491119385,
3390
+ "learning_rate": 3.797121315969843e-05,
3391
+ "loss": 1.5376,
3392
+ "step": 4730
3393
+ },
3394
+ {
3395
+ "epoch": 1.62,
3396
+ "grad_norm": 1.225707769393921,
3397
+ "learning_rate": 3.762851267991775e-05,
3398
+ "loss": 1.5925,
3399
+ "step": 4740
3400
+ },
3401
+ {
3402
+ "epoch": 1.63,
3403
+ "grad_norm": 1.5699125528335571,
3404
+ "learning_rate": 3.728581220013708e-05,
3405
+ "loss": 1.593,
3406
+ "step": 4750
3407
+ },
3408
+ {
3409
+ "epoch": 1.63,
3410
+ "grad_norm": 1.318574070930481,
3411
+ "learning_rate": 3.694311172035641e-05,
3412
+ "loss": 1.5525,
3413
+ "step": 4760
3414
+ },
3415
+ {
3416
+ "epoch": 1.63,
3417
+ "grad_norm": 1.4544116258621216,
3418
+ "learning_rate": 3.660041124057574e-05,
3419
+ "loss": 1.5678,
3420
+ "step": 4770
3421
+ },
3422
+ {
3423
+ "epoch": 1.64,
3424
+ "grad_norm": 1.7460687160491943,
3425
+ "learning_rate": 3.6257710760795063e-05,
3426
+ "loss": 1.6081,
3427
+ "step": 4780
3428
+ },
3429
+ {
3430
+ "epoch": 1.64,
3431
+ "grad_norm": 1.4106998443603516,
3432
+ "learning_rate": 3.591501028101439e-05,
3433
+ "loss": 1.5687,
3434
+ "step": 4790
3435
+ },
3436
+ {
3437
+ "epoch": 1.64,
3438
+ "grad_norm": 1.0583499670028687,
3439
+ "learning_rate": 3.557230980123372e-05,
3440
+ "loss": 1.5467,
3441
+ "step": 4800
3442
+ },
3443
+ {
3444
+ "epoch": 1.65,
3445
+ "grad_norm": 1.2292665243148804,
3446
+ "learning_rate": 3.522960932145305e-05,
3447
+ "loss": 1.5491,
3448
+ "step": 4810
3449
+ },
3450
+ {
3451
+ "epoch": 1.65,
3452
+ "grad_norm": 1.3556251525878906,
3453
+ "learning_rate": 3.4886908841672375e-05,
3454
+ "loss": 1.5568,
3455
+ "step": 4820
3456
+ },
3457
+ {
3458
+ "epoch": 1.65,
3459
+ "grad_norm": 1.6374377012252808,
3460
+ "learning_rate": 3.4544208361891705e-05,
3461
+ "loss": 1.6016,
3462
+ "step": 4830
3463
+ },
3464
+ {
3465
+ "epoch": 1.66,
3466
+ "grad_norm": 1.0343750715255737,
3467
+ "learning_rate": 3.420150788211104e-05,
3468
+ "loss": 1.4693,
3469
+ "step": 4840
3470
+ },
3471
+ {
3472
+ "epoch": 1.66,
3473
+ "grad_norm": 1.378056526184082,
3474
+ "learning_rate": 3.3858807402330365e-05,
3475
+ "loss": 1.6081,
3476
+ "step": 4850
3477
+ },
3478
+ {
3479
+ "epoch": 1.66,
3480
+ "grad_norm": 1.370970368385315,
3481
+ "learning_rate": 3.3516106922549694e-05,
3482
+ "loss": 1.515,
3483
+ "step": 4860
3484
+ },
3485
+ {
3486
+ "epoch": 1.67,
3487
+ "grad_norm": 1.3780639171600342,
3488
+ "learning_rate": 3.3173406442769024e-05,
3489
+ "loss": 1.5644,
3490
+ "step": 4870
3491
+ },
3492
+ {
3493
+ "epoch": 1.67,
3494
+ "grad_norm": 1.0907922983169556,
3495
+ "learning_rate": 3.2830705962988354e-05,
3496
+ "loss": 1.5701,
3497
+ "step": 4880
3498
+ },
3499
+ {
3500
+ "epoch": 1.67,
3501
+ "grad_norm": 1.4807682037353516,
3502
+ "learning_rate": 3.2488005483207677e-05,
3503
+ "loss": 1.5535,
3504
+ "step": 4890
3505
+ },
3506
+ {
3507
+ "epoch": 1.68,
3508
+ "grad_norm": 1.7207825183868408,
3509
+ "learning_rate": 3.2145305003427006e-05,
3510
+ "loss": 1.6049,
3511
+ "step": 4900
3512
+ },
3513
+ {
3514
+ "epoch": 1.68,
3515
+ "grad_norm": 0.8784241676330566,
3516
+ "learning_rate": 3.1802604523646336e-05,
3517
+ "loss": 1.5213,
3518
+ "step": 4910
3519
+ },
3520
+ {
3521
+ "epoch": 1.69,
3522
+ "grad_norm": 1.6283917427062988,
3523
+ "learning_rate": 3.1459904043865666e-05,
3524
+ "loss": 1.4902,
3525
+ "step": 4920
3526
+ },
3527
+ {
3528
+ "epoch": 1.69,
3529
+ "grad_norm": 1.0017669200897217,
3530
+ "learning_rate": 3.111720356408499e-05,
3531
+ "loss": 1.5147,
3532
+ "step": 4930
3533
+ },
3534
+ {
3535
+ "epoch": 1.69,
3536
+ "grad_norm": 1.4256327152252197,
3537
+ "learning_rate": 3.077450308430432e-05,
3538
+ "loss": 1.518,
3539
+ "step": 4940
3540
+ },
3541
+ {
3542
+ "epoch": 1.7,
3543
+ "grad_norm": 1.4298090934753418,
3544
+ "learning_rate": 3.0431802604523645e-05,
3545
+ "loss": 1.5843,
3546
+ "step": 4950
3547
+ },
3548
+ {
3549
+ "epoch": 1.7,
3550
+ "grad_norm": 1.0894560813903809,
3551
+ "learning_rate": 3.0089102124742974e-05,
3552
+ "loss": 1.5931,
3553
+ "step": 4960
3554
+ },
3555
+ {
3556
+ "epoch": 1.7,
3557
+ "grad_norm": 1.8101505041122437,
3558
+ "learning_rate": 2.97464016449623e-05,
3559
+ "loss": 1.5629,
3560
+ "step": 4970
3561
+ },
3562
+ {
3563
+ "epoch": 1.71,
3564
+ "grad_norm": 0.966204047203064,
3565
+ "learning_rate": 2.9403701165181634e-05,
3566
+ "loss": 1.5048,
3567
+ "step": 4980
3568
+ },
3569
+ {
3570
+ "epoch": 1.71,
3571
+ "grad_norm": 1.2718944549560547,
3572
+ "learning_rate": 2.9061000685400963e-05,
3573
+ "loss": 1.6685,
3574
+ "step": 4990
3575
+ },
3576
+ {
3577
+ "epoch": 1.71,
3578
+ "grad_norm": 0.9012284874916077,
3579
+ "learning_rate": 2.871830020562029e-05,
3580
+ "loss": 1.5769,
3581
+ "step": 5000
3582
+ },
3583
+ {
3584
+ "epoch": 1.71,
3585
+ "eval_loss": 1.9053254127502441,
3586
+ "eval_runtime": 33.1389,
3587
+ "eval_samples_per_second": 30.176,
3588
+ "eval_steps_per_second": 3.772,
3589
+ "step": 5000
3590
+ },
3591
+ {
3592
+ "epoch": 1.72,
3593
+ "grad_norm": 1.4876313209533691,
3594
+ "learning_rate": 2.837559972583962e-05,
3595
+ "loss": 1.4846,
3596
+ "step": 5010
3597
+ },
3598
+ {
3599
+ "epoch": 1.72,
3600
+ "grad_norm": 0.9953039288520813,
3601
+ "learning_rate": 2.8032899246058946e-05,
3602
+ "loss": 1.6145,
3603
+ "step": 5020
3604
+ },
3605
+ {
3606
+ "epoch": 1.72,
3607
+ "grad_norm": 1.4575115442276,
3608
+ "learning_rate": 2.7690198766278275e-05,
3609
+ "loss": 1.5442,
3610
+ "step": 5030
3611
+ },
3612
+ {
3613
+ "epoch": 1.73,
3614
+ "grad_norm": 1.3410977125167847,
3615
+ "learning_rate": 2.73474982864976e-05,
3616
+ "loss": 1.5617,
3617
+ "step": 5040
3618
+ },
3619
+ {
3620
+ "epoch": 1.73,
3621
+ "grad_norm": 1.5489014387130737,
3622
+ "learning_rate": 2.700479780671693e-05,
3623
+ "loss": 1.6061,
3624
+ "step": 5050
3625
+ },
3626
+ {
3627
+ "epoch": 1.73,
3628
+ "grad_norm": 2.2693567276000977,
3629
+ "learning_rate": 2.6662097326936258e-05,
3630
+ "loss": 1.576,
3631
+ "step": 5060
3632
+ },
3633
+ {
3634
+ "epoch": 1.74,
3635
+ "grad_norm": 1.776106595993042,
3636
+ "learning_rate": 2.6319396847155587e-05,
3637
+ "loss": 1.57,
3638
+ "step": 5070
3639
+ },
3640
+ {
3641
+ "epoch": 1.74,
3642
+ "grad_norm": 1.0588148832321167,
3643
+ "learning_rate": 2.5976696367374914e-05,
3644
+ "loss": 1.476,
3645
+ "step": 5080
3646
+ },
3647
+ {
3648
+ "epoch": 1.74,
3649
+ "grad_norm": 1.133484125137329,
3650
+ "learning_rate": 2.5633995887594243e-05,
3651
+ "loss": 1.5103,
3652
+ "step": 5090
3653
+ },
3654
+ {
3655
+ "epoch": 1.75,
3656
+ "grad_norm": 1.3961825370788574,
3657
+ "learning_rate": 2.529129540781357e-05,
3658
+ "loss": 1.59,
3659
+ "step": 5100
3660
+ },
3661
+ {
3662
+ "epoch": 1.75,
3663
+ "grad_norm": 1.7427486181259155,
3664
+ "learning_rate": 2.49485949280329e-05,
3665
+ "loss": 1.5608,
3666
+ "step": 5110
3667
+ },
3668
+ {
3669
+ "epoch": 1.75,
3670
+ "grad_norm": 1.961029291152954,
3671
+ "learning_rate": 2.460589444825223e-05,
3672
+ "loss": 1.6278,
3673
+ "step": 5120
3674
+ },
3675
+ {
3676
+ "epoch": 1.76,
3677
+ "grad_norm": 1.2870323657989502,
3678
+ "learning_rate": 2.4263193968471555e-05,
3679
+ "loss": 1.5877,
3680
+ "step": 5130
3681
+ },
3682
+ {
3683
+ "epoch": 1.76,
3684
+ "grad_norm": 1.204353928565979,
3685
+ "learning_rate": 2.3920493488690885e-05,
3686
+ "loss": 1.5,
3687
+ "step": 5140
3688
+ },
3689
+ {
3690
+ "epoch": 1.76,
3691
+ "grad_norm": 0.9764713644981384,
3692
+ "learning_rate": 2.357779300891021e-05,
3693
+ "loss": 1.6451,
3694
+ "step": 5150
3695
+ },
3696
+ {
3697
+ "epoch": 1.77,
3698
+ "grad_norm": 1.2140144109725952,
3699
+ "learning_rate": 2.3235092529129544e-05,
3700
+ "loss": 1.4958,
3701
+ "step": 5160
3702
+ },
3703
+ {
3704
+ "epoch": 1.77,
3705
+ "grad_norm": 1.9167425632476807,
3706
+ "learning_rate": 2.289239204934887e-05,
3707
+ "loss": 1.569,
3708
+ "step": 5170
3709
+ },
3710
+ {
3711
+ "epoch": 1.77,
3712
+ "grad_norm": 1.864986538887024,
3713
+ "learning_rate": 2.25496915695682e-05,
3714
+ "loss": 1.5232,
3715
+ "step": 5180
3716
+ },
3717
+ {
3718
+ "epoch": 1.78,
3719
+ "grad_norm": 1.2421759366989136,
3720
+ "learning_rate": 2.2206991089787527e-05,
3721
+ "loss": 1.5894,
3722
+ "step": 5190
3723
+ },
3724
+ {
3725
+ "epoch": 1.78,
3726
+ "grad_norm": 1.412864089012146,
3727
+ "learning_rate": 2.1898560657984922e-05,
3728
+ "loss": 1.5737,
3729
+ "step": 5200
3730
+ },
3731
+ {
3732
+ "epoch": 1.78,
3733
+ "grad_norm": 1.105542778968811,
3734
+ "learning_rate": 2.1555860178204252e-05,
3735
+ "loss": 1.5747,
3736
+ "step": 5210
3737
+ },
3738
+ {
3739
+ "epoch": 1.79,
3740
+ "grad_norm": 1.7511961460113525,
3741
+ "learning_rate": 2.1213159698423578e-05,
3742
+ "loss": 1.5455,
3743
+ "step": 5220
3744
+ },
3745
+ {
3746
+ "epoch": 1.79,
3747
+ "grad_norm": 1.4287422895431519,
3748
+ "learning_rate": 2.0870459218642908e-05,
3749
+ "loss": 1.4871,
3750
+ "step": 5230
3751
+ },
3752
+ {
3753
+ "epoch": 1.79,
3754
+ "grad_norm": 0.835995614528656,
3755
+ "learning_rate": 2.0527758738862234e-05,
3756
+ "loss": 1.6128,
3757
+ "step": 5240
3758
+ },
3759
+ {
3760
+ "epoch": 1.8,
3761
+ "grad_norm": 1.8323885202407837,
3762
+ "learning_rate": 2.0185058259081564e-05,
3763
+ "loss": 1.6333,
3764
+ "step": 5250
3765
+ },
3766
+ {
3767
+ "epoch": 1.8,
3768
+ "grad_norm": 1.5953247547149658,
3769
+ "learning_rate": 1.984235777930089e-05,
3770
+ "loss": 1.5631,
3771
+ "step": 5260
3772
+ },
3773
+ {
3774
+ "epoch": 1.8,
3775
+ "grad_norm": 1.4622983932495117,
3776
+ "learning_rate": 1.9499657299520223e-05,
3777
+ "loss": 1.5065,
3778
+ "step": 5270
3779
+ },
3780
+ {
3781
+ "epoch": 1.81,
3782
+ "grad_norm": 1.6321667432785034,
3783
+ "learning_rate": 1.915695681973955e-05,
3784
+ "loss": 1.59,
3785
+ "step": 5280
3786
+ },
3787
+ {
3788
+ "epoch": 1.81,
3789
+ "grad_norm": 1.3693170547485352,
3790
+ "learning_rate": 1.8814256339958876e-05,
3791
+ "loss": 1.5847,
3792
+ "step": 5290
3793
+ },
3794
+ {
3795
+ "epoch": 1.82,
3796
+ "grad_norm": 1.5187667608261108,
3797
+ "learning_rate": 1.8471555860178205e-05,
3798
+ "loss": 1.541,
3799
+ "step": 5300
3800
+ },
3801
+ {
3802
+ "epoch": 1.82,
3803
+ "grad_norm": 1.1000255346298218,
3804
+ "learning_rate": 1.8128855380397532e-05,
3805
+ "loss": 1.4775,
3806
+ "step": 5310
3807
+ },
3808
+ {
3809
+ "epoch": 1.82,
3810
+ "grad_norm": 1.4071645736694336,
3811
+ "learning_rate": 1.778615490061686e-05,
3812
+ "loss": 1.6336,
3813
+ "step": 5320
3814
+ },
3815
+ {
3816
+ "epoch": 1.83,
3817
+ "grad_norm": 1.5703157186508179,
3818
+ "learning_rate": 1.7443454420836188e-05,
3819
+ "loss": 1.6725,
3820
+ "step": 5330
3821
+ },
3822
+ {
3823
+ "epoch": 1.83,
3824
+ "grad_norm": 1.0555702447891235,
3825
+ "learning_rate": 1.710075394105552e-05,
3826
+ "loss": 1.4712,
3827
+ "step": 5340
3828
+ },
3829
+ {
3830
+ "epoch": 1.83,
3831
+ "grad_norm": 1.4873102903366089,
3832
+ "learning_rate": 1.6758053461274847e-05,
3833
+ "loss": 1.5741,
3834
+ "step": 5350
3835
+ },
3836
+ {
3837
+ "epoch": 1.84,
3838
+ "grad_norm": 1.1715468168258667,
3839
+ "learning_rate": 1.6415352981494177e-05,
3840
+ "loss": 1.4884,
3841
+ "step": 5360
3842
+ },
3843
+ {
3844
+ "epoch": 1.84,
3845
+ "grad_norm": 1.82741379737854,
3846
+ "learning_rate": 1.6072652501713503e-05,
3847
+ "loss": 1.5778,
3848
+ "step": 5370
3849
+ },
3850
+ {
3851
+ "epoch": 1.84,
3852
+ "grad_norm": 1.6479945182800293,
3853
+ "learning_rate": 1.5729952021932833e-05,
3854
+ "loss": 1.6802,
3855
+ "step": 5380
3856
+ },
3857
+ {
3858
+ "epoch": 1.85,
3859
+ "grad_norm": 1.0871607065200806,
3860
+ "learning_rate": 1.538725154215216e-05,
3861
+ "loss": 1.5509,
3862
+ "step": 5390
3863
+ },
3864
+ {
3865
+ "epoch": 1.85,
3866
+ "grad_norm": 1.7326961755752563,
3867
+ "learning_rate": 1.5044551062371487e-05,
3868
+ "loss": 1.5746,
3869
+ "step": 5400
3870
+ },
3871
+ {
3872
+ "epoch": 1.85,
3873
+ "grad_norm": 1.3573588132858276,
3874
+ "learning_rate": 1.4701850582590817e-05,
3875
+ "loss": 1.6147,
3876
+ "step": 5410
3877
+ },
3878
+ {
3879
+ "epoch": 1.86,
3880
+ "grad_norm": 1.807897925376892,
3881
+ "learning_rate": 1.4359150102810145e-05,
3882
+ "loss": 1.6446,
3883
+ "step": 5420
3884
+ },
3885
+ {
3886
+ "epoch": 1.86,
3887
+ "grad_norm": 1.0243467092514038,
3888
+ "learning_rate": 1.4016449623029473e-05,
3889
+ "loss": 1.5844,
3890
+ "step": 5430
3891
+ },
3892
+ {
3893
+ "epoch": 1.86,
3894
+ "grad_norm": 1.709069013595581,
3895
+ "learning_rate": 1.36737491432488e-05,
3896
+ "loss": 1.5774,
3897
+ "step": 5440
3898
+ },
3899
+ {
3900
+ "epoch": 1.87,
3901
+ "grad_norm": 1.717564582824707,
3902
+ "learning_rate": 1.3331048663468129e-05,
3903
+ "loss": 1.5898,
3904
+ "step": 5450
3905
+ },
3906
+ {
3907
+ "epoch": 1.87,
3908
+ "grad_norm": 1.1066781282424927,
3909
+ "learning_rate": 1.2988348183687457e-05,
3910
+ "loss": 1.5828,
3911
+ "step": 5460
3912
+ },
3913
+ {
3914
+ "epoch": 1.87,
3915
+ "grad_norm": 1.119360089302063,
3916
+ "learning_rate": 1.2645647703906785e-05,
3917
+ "loss": 1.5321,
3918
+ "step": 5470
3919
+ },
3920
+ {
3921
+ "epoch": 1.88,
3922
+ "grad_norm": 1.0519651174545288,
3923
+ "learning_rate": 1.2302947224126114e-05,
3924
+ "loss": 1.5691,
3925
+ "step": 5480
3926
+ },
3927
+ {
3928
+ "epoch": 1.88,
3929
+ "grad_norm": 1.7377208471298218,
3930
+ "learning_rate": 1.1960246744345442e-05,
3931
+ "loss": 1.5568,
3932
+ "step": 5490
3933
+ },
3934
+ {
3935
+ "epoch": 1.88,
3936
+ "grad_norm": 1.4080170392990112,
3937
+ "learning_rate": 1.1617546264564772e-05,
3938
+ "loss": 1.6109,
3939
+ "step": 5500
3940
+ },
3941
+ {
3942
+ "epoch": 1.88,
3943
+ "eval_loss": 1.9016900062561035,
3944
+ "eval_runtime": 33.1979,
3945
+ "eval_samples_per_second": 30.122,
3946
+ "eval_steps_per_second": 3.765,
3947
+ "step": 5500
3948
+ }
3949
+ ],
3950
+ "logging_steps": 10,
3951
+ "max_steps": 5838,
3952
+ "num_input_tokens_seen": 0,
3953
+ "num_train_epochs": 2,
3954
+ "save_steps": 500,
3955
+ "total_flos": 2.980420245786624e+16,
3956
+ "train_batch_size": 1,
3957
+ "trial_name": null,
3958
+ "trial_params": null
3959
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5883e40264ebb017d148f5789d4245b7dbea2027127c87a02eba3a7db9f12f93
3
+ size 4920