Hugging Face's logo

manu
/
colpali-3b-mix-448-docmatix-only-mined-ib

ColPali
Safetensors
English
vidore
Model card Files Community
manu commited on
Commit
d44207d
•
1 Parent(s): c9e89b1

Upload folder using huggingface_hub

Browse files
Files changed (17) hide show
  1. .gitattributes +1 -0
  2. adapter_config.json +26 -0
  3. adapter_model.safetensors +3 -0
  4. checkpoint-3000/adapter_config.json +26 -0
  5. checkpoint-3000/adapter_model.safetensors +3 -0
  6. checkpoint-3000/optimizer.pt +3 -0
  7. checkpoint-3000/rng_state.pth +3 -0
  8. checkpoint-3000/scheduler.pt +3 -0
  9. checkpoint-3000/trainer_state.json +2613 -0
  10. checkpoint-3000/training_args.bin +3 -0
  11. git_hash.txt +1 -0
  12. preprocessor_config.json +40 -0
  13. results.json +1 -0
  14. special_tokens_map.json +39 -0
  15. tokenizer.json +3 -0
  16. tokenizer_config.json +0 -0
  17. training_config.yml +42 -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 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "./models/paligemma-3b-mix-448",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": false,
8
+ "init_lora_weights": "gaussian",
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.1,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 32,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": "(.*(language_model).*(down_proj|gate_proj|up_proj|k_proj|q_proj|v_proj|o_proj).*$|.*(custom_text_proj).*$)",
23
+ "task_type": "FEATURE_EXTRACTION",
24
+ "use_dora": false,
25
+ "use_rslora": false
26
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6c318f350a055c439f7dac58f6d8fc7730cea964091f278211e1acc440a5fbce
3
+ size 78625112
checkpoint-3000/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "./models/paligemma-3b-mix-448",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": false,
8
+ "init_lora_weights": "gaussian",
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.1,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 32,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": "(.*(language_model).*(down_proj|gate_proj|up_proj|k_proj|q_proj|v_proj|o_proj).*$|.*(custom_text_proj).*$)",
23
+ "task_type": "FEATURE_EXTRACTION",
24
+ "use_dora": false,
25
+ "use_rslora": false
26
+ }
checkpoint-3000/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b5a4ebc968b8b59a3adbc600f7493a8597dbaea90019daf5f0849732b4ba840c
3
+ size 78625112
checkpoint-3000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a71627ccc8037d6d4327a32cd7102973916904d380026cb86013dc164370cc48
3
+ size 157385722
checkpoint-3000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35c4698ebd78e6859374b1a01a0e10844d54821ab1ea77b6f5376b3dfe8a019f
3
+ size 14244
checkpoint-3000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:34b329b6477966fc78a7f8d0ff7401eaffcffd77f7d0b84cd752d9116ad97135
3
+ size 1064
checkpoint-3000/trainer_state.json ADDED
@@ -0,0 +1,2613 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.96,
5
+ "eval_steps": 50,
6
+ "global_step": 3000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0032,
13
+ "grad_norm": 0.2421875,
14
+ "learning_rate": 5e-06,
15
+ "loss": 0.7,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.0064,
20
+ "grad_norm": 0.30859375,
21
+ "learning_rate": 1e-05,
22
+ "loss": 0.6988,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.0096,
27
+ "grad_norm": 0.28515625,
28
+ "learning_rate": 1.5e-05,
29
+ "loss": 0.6977,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.0128,
34
+ "grad_norm": 0.244140625,
35
+ "learning_rate": 2e-05,
36
+ "loss": 0.6969,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.016,
41
+ "grad_norm": 0.259765625,
42
+ "learning_rate": 2.5e-05,
43
+ "loss": 0.6941,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.016,
48
+ "eval_loss": 0.6939687728881836,
49
+ "eval_runtime": 192.7405,
50
+ "eval_samples_per_second": 2.594,
51
+ "eval_steps_per_second": 0.083,
52
+ "step": 50
53
+ },
54
+ {
55
+ "epoch": 0.0192,
56
+ "grad_norm": 0.232421875,
57
+ "learning_rate": 3e-05,
58
+ "loss": 0.6871,
59
+ "step": 60
60
+ },
61
+ {
62
+ "epoch": 0.0224,
63
+ "grad_norm": 0.302734375,
64
+ "learning_rate": 3.5e-05,
65
+ "loss": 0.6828,
66
+ "step": 70
67
+ },
68
+ {
69
+ "epoch": 0.0256,
70
+ "grad_norm": 0.8203125,
71
+ "learning_rate": 4e-05,
72
+ "loss": 0.6687,
73
+ "step": 80
74
+ },
75
+ {
76
+ "epoch": 0.0288,
77
+ "grad_norm": 1.2421875,
78
+ "learning_rate": 4.5e-05,
79
+ "loss": 0.6309,
80
+ "step": 90
81
+ },
82
+ {
83
+ "epoch": 0.032,
84
+ "grad_norm": 2.03125,
85
+ "learning_rate": 5e-05,
86
+ "loss": 0.5781,
87
+ "step": 100
88
+ },
89
+ {
90
+ "epoch": 0.032,
91
+ "eval_loss": 0.5389062762260437,
92
+ "eval_runtime": 108.3608,
93
+ "eval_samples_per_second": 4.614,
94
+ "eval_steps_per_second": 0.148,
95
+ "step": 100
96
+ },
97
+ {
98
+ "epoch": 0.0352,
99
+ "grad_norm": 2.328125,
100
+ "learning_rate": 4.9834710743801654e-05,
101
+ "loss": 0.4818,
102
+ "step": 110
103
+ },
104
+ {
105
+ "epoch": 0.0384,
106
+ "grad_norm": 2.265625,
107
+ "learning_rate": 4.9669421487603305e-05,
108
+ "loss": 0.4291,
109
+ "step": 120
110
+ },
111
+ {
112
+ "epoch": 0.0416,
113
+ "grad_norm": 2.203125,
114
+ "learning_rate": 4.9504132231404956e-05,
115
+ "loss": 0.3615,
116
+ "step": 130
117
+ },
118
+ {
119
+ "epoch": 0.0448,
120
+ "grad_norm": 2.515625,
121
+ "learning_rate": 4.9338842975206614e-05,
122
+ "loss": 0.3408,
123
+ "step": 140
124
+ },
125
+ {
126
+ "epoch": 0.048,
127
+ "grad_norm": 1.515625,
128
+ "learning_rate": 4.917355371900827e-05,
129
+ "loss": 0.306,
130
+ "step": 150
131
+ },
132
+ {
133
+ "epoch": 0.048,
134
+ "eval_loss": 0.2862031161785126,
135
+ "eval_runtime": 110.0641,
136
+ "eval_samples_per_second": 4.543,
137
+ "eval_steps_per_second": 0.145,
138
+ "step": 150
139
+ },
140
+ {
141
+ "epoch": 0.0512,
142
+ "grad_norm": 1.7265625,
143
+ "learning_rate": 4.900826446280992e-05,
144
+ "loss": 0.2904,
145
+ "step": 160
146
+ },
147
+ {
148
+ "epoch": 0.0544,
149
+ "grad_norm": 1.5703125,
150
+ "learning_rate": 4.8842975206611575e-05,
151
+ "loss": 0.2648,
152
+ "step": 170
153
+ },
154
+ {
155
+ "epoch": 0.0576,
156
+ "grad_norm": 1.6875,
157
+ "learning_rate": 4.8677685950413226e-05,
158
+ "loss": 0.2373,
159
+ "step": 180
160
+ },
161
+ {
162
+ "epoch": 0.0608,
163
+ "grad_norm": 1.0859375,
164
+ "learning_rate": 4.851239669421488e-05,
165
+ "loss": 0.1749,
166
+ "step": 190
167
+ },
168
+ {
169
+ "epoch": 0.064,
170
+ "grad_norm": 0.8671875,
171
+ "learning_rate": 4.834710743801653e-05,
172
+ "loss": 0.1556,
173
+ "step": 200
174
+ },
175
+ {
176
+ "epoch": 0.064,
177
+ "eval_loss": 0.16703906655311584,
178
+ "eval_runtime": 109.9961,
179
+ "eval_samples_per_second": 4.546,
180
+ "eval_steps_per_second": 0.145,
181
+ "step": 200
182
+ },
183
+ {
184
+ "epoch": 0.0672,
185
+ "grad_norm": 2.5625,
186
+ "learning_rate": 4.8181818181818186e-05,
187
+ "loss": 0.1382,
188
+ "step": 210
189
+ },
190
+ {
191
+ "epoch": 0.0704,
192
+ "grad_norm": 1.859375,
193
+ "learning_rate": 4.801652892561984e-05,
194
+ "loss": 0.125,
195
+ "step": 220
196
+ },
197
+ {
198
+ "epoch": 0.0736,
199
+ "grad_norm": 1.3515625,
200
+ "learning_rate": 4.785123966942149e-05,
201
+ "loss": 0.1729,
202
+ "step": 230
203
+ },
204
+ {
205
+ "epoch": 0.0768,
206
+ "grad_norm": 1.1953125,
207
+ "learning_rate": 4.768595041322314e-05,
208
+ "loss": 0.1252,
209
+ "step": 240
210
+ },
211
+ {
212
+ "epoch": 0.08,
213
+ "grad_norm": 1.0078125,
214
+ "learning_rate": 4.75206611570248e-05,
215
+ "loss": 0.1108,
216
+ "step": 250
217
+ },
218
+ {
219
+ "epoch": 0.08,
220
+ "eval_loss": 0.13416016101837158,
221
+ "eval_runtime": 110.9679,
222
+ "eval_samples_per_second": 4.506,
223
+ "eval_steps_per_second": 0.144,
224
+ "step": 250
225
+ },
226
+ {
227
+ "epoch": 0.0832,
228
+ "grad_norm": 1.7890625,
229
+ "learning_rate": 4.735537190082645e-05,
230
+ "loss": 0.1502,
231
+ "step": 260
232
+ },
233
+ {
234
+ "epoch": 0.0864,
235
+ "grad_norm": 0.66796875,
236
+ "learning_rate": 4.71900826446281e-05,
237
+ "loss": 0.1291,
238
+ "step": 270
239
+ },
240
+ {
241
+ "epoch": 0.0896,
242
+ "grad_norm": 2.546875,
243
+ "learning_rate": 4.702479338842976e-05,
244
+ "loss": 0.1326,
245
+ "step": 280
246
+ },
247
+ {
248
+ "epoch": 0.0928,
249
+ "grad_norm": 1.7265625,
250
+ "learning_rate": 4.685950413223141e-05,
251
+ "loss": 0.1374,
252
+ "step": 290
253
+ },
254
+ {
255
+ "epoch": 0.096,
256
+ "grad_norm": 1.15625,
257
+ "learning_rate": 4.669421487603306e-05,
258
+ "loss": 0.1081,
259
+ "step": 300
260
+ },
261
+ {
262
+ "epoch": 0.096,
263
+ "eval_loss": 0.12271875143051147,
264
+ "eval_runtime": 110.6168,
265
+ "eval_samples_per_second": 4.52,
266
+ "eval_steps_per_second": 0.145,
267
+ "step": 300
268
+ },
269
+ {
270
+ "epoch": 0.0992,
271
+ "grad_norm": 1.140625,
272
+ "learning_rate": 4.652892561983471e-05,
273
+ "loss": 0.1266,
274
+ "step": 310
275
+ },
276
+ {
277
+ "epoch": 0.1024,
278
+ "grad_norm": 1.2265625,
279
+ "learning_rate": 4.636363636363636e-05,
280
+ "loss": 0.1368,
281
+ "step": 320
282
+ },
283
+ {
284
+ "epoch": 0.1056,
285
+ "grad_norm": 0.7109375,
286
+ "learning_rate": 4.6198347107438014e-05,
287
+ "loss": 0.1064,
288
+ "step": 330
289
+ },
290
+ {
291
+ "epoch": 0.1088,
292
+ "grad_norm": 1.4375,
293
+ "learning_rate": 4.603305785123967e-05,
294
+ "loss": 0.1233,
295
+ "step": 340
296
+ },
297
+ {
298
+ "epoch": 0.112,
299
+ "grad_norm": 0.6328125,
300
+ "learning_rate": 4.586776859504133e-05,
301
+ "loss": 0.0881,
302
+ "step": 350
303
+ },
304
+ {
305
+ "epoch": 0.112,
306
+ "eval_loss": 0.1180742159485817,
307
+ "eval_runtime": 109.5127,
308
+ "eval_samples_per_second": 4.566,
309
+ "eval_steps_per_second": 0.146,
310
+ "step": 350
311
+ },
312
+ {
313
+ "epoch": 0.1152,
314
+ "grad_norm": 1.109375,
315
+ "learning_rate": 4.570247933884298e-05,
316
+ "loss": 0.1136,
317
+ "step": 360
318
+ },
319
+ {
320
+ "epoch": 0.1184,
321
+ "grad_norm": 0.89453125,
322
+ "learning_rate": 4.553719008264463e-05,
323
+ "loss": 0.1338,
324
+ "step": 370
325
+ },
326
+ {
327
+ "epoch": 0.1216,
328
+ "grad_norm": 2.65625,
329
+ "learning_rate": 4.5371900826446284e-05,
330
+ "loss": 0.1636,
331
+ "step": 380
332
+ },
333
+ {
334
+ "epoch": 0.1248,
335
+ "grad_norm": 2.375,
336
+ "learning_rate": 4.5206611570247935e-05,
337
+ "loss": 0.1067,
338
+ "step": 390
339
+ },
340
+ {
341
+ "epoch": 0.128,
342
+ "grad_norm": 1.078125,
343
+ "learning_rate": 4.504132231404959e-05,
344
+ "loss": 0.1287,
345
+ "step": 400
346
+ },
347
+ {
348
+ "epoch": 0.128,
349
+ "eval_loss": 0.1135859340429306,
350
+ "eval_runtime": 115.1853,
351
+ "eval_samples_per_second": 4.341,
352
+ "eval_steps_per_second": 0.139,
353
+ "step": 400
354
+ },
355
+ {
356
+ "epoch": 0.1312,
357
+ "grad_norm": 1.5859375,
358
+ "learning_rate": 4.487603305785124e-05,
359
+ "loss": 0.122,
360
+ "step": 410
361
+ },
362
+ {
363
+ "epoch": 0.1344,
364
+ "grad_norm": 1.6796875,
365
+ "learning_rate": 4.4710743801652896e-05,
366
+ "loss": 0.1232,
367
+ "step": 420
368
+ },
369
+ {
370
+ "epoch": 0.1376,
371
+ "grad_norm": 0.99609375,
372
+ "learning_rate": 4.454545454545455e-05,
373
+ "loss": 0.11,
374
+ "step": 430
375
+ },
376
+ {
377
+ "epoch": 0.1408,
378
+ "grad_norm": 1.6328125,
379
+ "learning_rate": 4.43801652892562e-05,
380
+ "loss": 0.0937,
381
+ "step": 440
382
+ },
383
+ {
384
+ "epoch": 0.144,
385
+ "grad_norm": 1.8671875,
386
+ "learning_rate": 4.4214876033057856e-05,
387
+ "loss": 0.1162,
388
+ "step": 450
389
+ },
390
+ {
391
+ "epoch": 0.144,
392
+ "eval_loss": 0.10819531232118607,
393
+ "eval_runtime": 110.4172,
394
+ "eval_samples_per_second": 4.528,
395
+ "eval_steps_per_second": 0.145,
396
+ "step": 450
397
+ },
398
+ {
399
+ "epoch": 0.1472,
400
+ "grad_norm": 3.78125,
401
+ "learning_rate": 4.404958677685951e-05,
402
+ "loss": 0.1171,
403
+ "step": 460
404
+ },
405
+ {
406
+ "epoch": 0.1504,
407
+ "grad_norm": 1.71875,
408
+ "learning_rate": 4.388429752066116e-05,
409
+ "loss": 0.1032,
410
+ "step": 470
411
+ },
412
+ {
413
+ "epoch": 0.1536,
414
+ "grad_norm": 2.796875,
415
+ "learning_rate": 4.371900826446281e-05,
416
+ "loss": 0.1217,
417
+ "step": 480
418
+ },
419
+ {
420
+ "epoch": 0.1568,
421
+ "grad_norm": 0.703125,
422
+ "learning_rate": 4.355371900826447e-05,
423
+ "loss": 0.1197,
424
+ "step": 490
425
+ },
426
+ {
427
+ "epoch": 0.16,
428
+ "grad_norm": 1.734375,
429
+ "learning_rate": 4.338842975206612e-05,
430
+ "loss": 0.0878,
431
+ "step": 500
432
+ },
433
+ {
434
+ "epoch": 0.16,
435
+ "eval_loss": 0.10691406577825546,
436
+ "eval_runtime": 110.9181,
437
+ "eval_samples_per_second": 4.508,
438
+ "eval_steps_per_second": 0.144,
439
+ "step": 500
440
+ },
441
+ {
442
+ "epoch": 0.1632,
443
+ "grad_norm": 1.3125,
444
+ "learning_rate": 4.322314049586777e-05,
445
+ "loss": 0.1102,
446
+ "step": 510
447
+ },
448
+ {
449
+ "epoch": 0.1664,
450
+ "grad_norm": 1.171875,
451
+ "learning_rate": 4.305785123966942e-05,
452
+ "loss": 0.0982,
453
+ "step": 520
454
+ },
455
+ {
456
+ "epoch": 0.1696,
457
+ "grad_norm": 0.90625,
458
+ "learning_rate": 4.289256198347107e-05,
459
+ "loss": 0.1096,
460
+ "step": 530
461
+ },
462
+ {
463
+ "epoch": 0.1728,
464
+ "grad_norm": 0.93359375,
465
+ "learning_rate": 4.2727272727272724e-05,
466
+ "loss": 0.1269,
467
+ "step": 540
468
+ },
469
+ {
470
+ "epoch": 0.176,
471
+ "grad_norm": 1.609375,
472
+ "learning_rate": 4.256198347107438e-05,
473
+ "loss": 0.0994,
474
+ "step": 550
475
+ },
476
+ {
477
+ "epoch": 0.176,
478
+ "eval_loss": 0.10447265952825546,
479
+ "eval_runtime": 110.3941,
480
+ "eval_samples_per_second": 4.529,
481
+ "eval_steps_per_second": 0.145,
482
+ "step": 550
483
+ },
484
+ {
485
+ "epoch": 0.1792,
486
+ "grad_norm": 0.703125,
487
+ "learning_rate": 4.239669421487604e-05,
488
+ "loss": 0.1118,
489
+ "step": 560
490
+ },
491
+ {
492
+ "epoch": 0.1824,
493
+ "grad_norm": 1.234375,
494
+ "learning_rate": 4.223140495867769e-05,
495
+ "loss": 0.098,
496
+ "step": 570
497
+ },
498
+ {
499
+ "epoch": 0.1856,
500
+ "grad_norm": 0.9375,
501
+ "learning_rate": 4.206611570247934e-05,
502
+ "loss": 0.0811,
503
+ "step": 580
504
+ },
505
+ {
506
+ "epoch": 0.1888,
507
+ "grad_norm": 1.3984375,
508
+ "learning_rate": 4.1900826446280994e-05,
509
+ "loss": 0.1208,
510
+ "step": 590
511
+ },
512
+ {
513
+ "epoch": 0.192,
514
+ "grad_norm": 1.34375,
515
+ "learning_rate": 4.1735537190082645e-05,
516
+ "loss": 0.1275,
517
+ "step": 600
518
+ },
519
+ {
520
+ "epoch": 0.192,
521
+ "eval_loss": 0.10120312869548798,
522
+ "eval_runtime": 109.2003,
523
+ "eval_samples_per_second": 4.579,
524
+ "eval_steps_per_second": 0.147,
525
+ "step": 600
526
+ },
527
+ {
528
+ "epoch": 0.1952,
529
+ "grad_norm": 0.5859375,
530
+ "learning_rate": 4.1570247933884296e-05,
531
+ "loss": 0.0872,
532
+ "step": 610
533
+ },
534
+ {
535
+ "epoch": 0.1984,
536
+ "grad_norm": 1.0390625,
537
+ "learning_rate": 4.1404958677685954e-05,
538
+ "loss": 0.1025,
539
+ "step": 620
540
+ },
541
+ {
542
+ "epoch": 0.2016,
543
+ "grad_norm": 1.0546875,
544
+ "learning_rate": 4.1239669421487606e-05,
545
+ "loss": 0.0932,
546
+ "step": 630
547
+ },
548
+ {
549
+ "epoch": 0.2048,
550
+ "grad_norm": 1.3046875,
551
+ "learning_rate": 4.107438016528926e-05,
552
+ "loss": 0.0838,
553
+ "step": 640
554
+ },
555
+ {
556
+ "epoch": 0.208,
557
+ "grad_norm": 1.0703125,
558
+ "learning_rate": 4.0909090909090915e-05,
559
+ "loss": 0.0921,
560
+ "step": 650
561
+ },
562
+ {
563
+ "epoch": 0.208,
564
+ "eval_loss": 0.10271484404802322,
565
+ "eval_runtime": 110.3072,
566
+ "eval_samples_per_second": 4.533,
567
+ "eval_steps_per_second": 0.145,
568
+ "step": 650
569
+ },
570
+ {
571
+ "epoch": 0.2112,
572
+ "grad_norm": 1.453125,
573
+ "learning_rate": 4.0743801652892566e-05,
574
+ "loss": 0.0842,
575
+ "step": 660
576
+ },
577
+ {
578
+ "epoch": 0.2144,
579
+ "grad_norm": 1.9609375,
580
+ "learning_rate": 4.057851239669422e-05,
581
+ "loss": 0.0969,
582
+ "step": 670
583
+ },
584
+ {
585
+ "epoch": 0.2176,
586
+ "grad_norm": 1.5859375,
587
+ "learning_rate": 4.041322314049587e-05,
588
+ "loss": 0.1037,
589
+ "step": 680
590
+ },
591
+ {
592
+ "epoch": 0.2208,
593
+ "grad_norm": 0.96875,
594
+ "learning_rate": 4.024793388429752e-05,
595
+ "loss": 0.0872,
596
+ "step": 690
597
+ },
598
+ {
599
+ "epoch": 0.224,
600
+ "grad_norm": 0.703125,
601
+ "learning_rate": 4.008264462809918e-05,
602
+ "loss": 0.0722,
603
+ "step": 700
604
+ },
605
+ {
606
+ "epoch": 0.224,
607
+ "eval_loss": 0.10465820133686066,
608
+ "eval_runtime": 109.5448,
609
+ "eval_samples_per_second": 4.564,
610
+ "eval_steps_per_second": 0.146,
611
+ "step": 700
612
+ },
613
+ {
614
+ "epoch": 0.2272,
615
+ "grad_norm": 0.66015625,
616
+ "learning_rate": 3.991735537190083e-05,
617
+ "loss": 0.0926,
618
+ "step": 710
619
+ },
620
+ {
621
+ "epoch": 0.2304,
622
+ "grad_norm": 3.234375,
623
+ "learning_rate": 3.975206611570248e-05,
624
+ "loss": 0.0945,
625
+ "step": 720
626
+ },
627
+ {
628
+ "epoch": 0.2336,
629
+ "grad_norm": 0.5390625,
630
+ "learning_rate": 3.958677685950413e-05,
631
+ "loss": 0.076,
632
+ "step": 730
633
+ },
634
+ {
635
+ "epoch": 0.2368,
636
+ "grad_norm": 1.0234375,
637
+ "learning_rate": 3.942148760330578e-05,
638
+ "loss": 0.0959,
639
+ "step": 740
640
+ },
641
+ {
642
+ "epoch": 0.24,
643
+ "grad_norm": 0.455078125,
644
+ "learning_rate": 3.925619834710744e-05,
645
+ "loss": 0.0916,
646
+ "step": 750
647
+ },
648
+ {
649
+ "epoch": 0.24,
650
+ "eval_loss": 0.10287109017372131,
651
+ "eval_runtime": 122.5858,
652
+ "eval_samples_per_second": 4.079,
653
+ "eval_steps_per_second": 0.131,
654
+ "step": 750
655
+ },
656
+ {
657
+ "epoch": 0.2432,
658
+ "grad_norm": 1.609375,
659
+ "learning_rate": 3.909090909090909e-05,
660
+ "loss": 0.0891,
661
+ "step": 760
662
+ },
663
+ {
664
+ "epoch": 0.2464,
665
+ "grad_norm": 0.8359375,
666
+ "learning_rate": 3.892561983471075e-05,
667
+ "loss": 0.0954,
668
+ "step": 770
669
+ },
670
+ {
671
+ "epoch": 0.2496,
672
+ "grad_norm": 1.203125,
673
+ "learning_rate": 3.87603305785124e-05,
674
+ "loss": 0.11,
675
+ "step": 780
676
+ },
677
+ {
678
+ "epoch": 0.2528,
679
+ "grad_norm": 0.365234375,
680
+ "learning_rate": 3.859504132231405e-05,
681
+ "loss": 0.067,
682
+ "step": 790
683
+ },
684
+ {
685
+ "epoch": 0.256,
686
+ "grad_norm": 1.1171875,
687
+ "learning_rate": 3.8429752066115703e-05,
688
+ "loss": 0.0909,
689
+ "step": 800
690
+ },
691
+ {
692
+ "epoch": 0.256,
693
+ "eval_loss": 0.09786719083786011,
694
+ "eval_runtime": 112.8442,
695
+ "eval_samples_per_second": 4.431,
696
+ "eval_steps_per_second": 0.142,
697
+ "step": 800
698
+ },
699
+ {
700
+ "epoch": 0.2592,
701
+ "grad_norm": 0.61328125,
702
+ "learning_rate": 3.8264462809917355e-05,
703
+ "loss": 0.0801,
704
+ "step": 810
705
+ },
706
+ {
707
+ "epoch": 0.2624,
708
+ "grad_norm": 1.9296875,
709
+ "learning_rate": 3.8099173553719006e-05,
710
+ "loss": 0.1088,
711
+ "step": 820
712
+ },
713
+ {
714
+ "epoch": 0.2656,
715
+ "grad_norm": 1.7421875,
716
+ "learning_rate": 3.7933884297520664e-05,
717
+ "loss": 0.08,
718
+ "step": 830
719
+ },
720
+ {
721
+ "epoch": 0.2688,
722
+ "grad_norm": 1.1875,
723
+ "learning_rate": 3.7768595041322315e-05,
724
+ "loss": 0.1055,
725
+ "step": 840
726
+ },
727
+ {
728
+ "epoch": 0.272,
729
+ "grad_norm": 0.85546875,
730
+ "learning_rate": 3.760330578512397e-05,
731
+ "loss": 0.0951,
732
+ "step": 850
733
+ },
734
+ {
735
+ "epoch": 0.272,
736
+ "eval_loss": 0.09604492038488388,
737
+ "eval_runtime": 109.2932,
738
+ "eval_samples_per_second": 4.575,
739
+ "eval_steps_per_second": 0.146,
740
+ "step": 850
741
+ },
742
+ {
743
+ "epoch": 0.2752,
744
+ "grad_norm": 1.7265625,
745
+ "learning_rate": 3.7438016528925624e-05,
746
+ "loss": 0.116,
747
+ "step": 860
748
+ },
749
+ {
750
+ "epoch": 0.2784,
751
+ "grad_norm": 1.640625,
752
+ "learning_rate": 3.7272727272727276e-05,
753
+ "loss": 0.0771,
754
+ "step": 870
755
+ },
756
+ {
757
+ "epoch": 0.2816,
758
+ "grad_norm": 0.92578125,
759
+ "learning_rate": 3.710743801652893e-05,
760
+ "loss": 0.08,
761
+ "step": 880
762
+ },
763
+ {
764
+ "epoch": 0.2848,
765
+ "grad_norm": 1.7890625,
766
+ "learning_rate": 3.694214876033058e-05,
767
+ "loss": 0.105,
768
+ "step": 890
769
+ },
770
+ {
771
+ "epoch": 0.288,
772
+ "grad_norm": 0.60546875,
773
+ "learning_rate": 3.6776859504132236e-05,
774
+ "loss": 0.0686,
775
+ "step": 900
776
+ },
777
+ {
778
+ "epoch": 0.288,
779
+ "eval_loss": 0.08956640958786011,
780
+ "eval_runtime": 109.9587,
781
+ "eval_samples_per_second": 4.547,
782
+ "eval_steps_per_second": 0.146,
783
+ "step": 900
784
+ },
785
+ {
786
+ "epoch": 0.2912,
787
+ "grad_norm": 0.484375,
788
+ "learning_rate": 3.661157024793389e-05,
789
+ "loss": 0.0867,
790
+ "step": 910
791
+ },
792
+ {
793
+ "epoch": 0.2944,
794
+ "grad_norm": 1.25,
795
+ "learning_rate": 3.644628099173554e-05,
796
+ "loss": 0.0975,
797
+ "step": 920
798
+ },
799
+ {
800
+ "epoch": 0.2976,
801
+ "grad_norm": 1.625,
802
+ "learning_rate": 3.628099173553719e-05,
803
+ "loss": 0.0905,
804
+ "step": 930
805
+ },
806
+ {
807
+ "epoch": 0.3008,
808
+ "grad_norm": 0.44140625,
809
+ "learning_rate": 3.611570247933884e-05,
810
+ "loss": 0.0983,
811
+ "step": 940
812
+ },
813
+ {
814
+ "epoch": 0.304,
815
+ "grad_norm": 0.51953125,
816
+ "learning_rate": 3.59504132231405e-05,
817
+ "loss": 0.0772,
818
+ "step": 950
819
+ },
820
+ {
821
+ "epoch": 0.304,
822
+ "eval_loss": 0.09039648622274399,
823
+ "eval_runtime": 111.9644,
824
+ "eval_samples_per_second": 4.466,
825
+ "eval_steps_per_second": 0.143,
826
+ "step": 950
827
+ },
828
+ {
829
+ "epoch": 0.3072,
830
+ "grad_norm": 0.828125,
831
+ "learning_rate": 3.578512396694215e-05,
832
+ "loss": 0.0907,
833
+ "step": 960
834
+ },
835
+ {
836
+ "epoch": 0.3104,
837
+ "grad_norm": 0.84765625,
838
+ "learning_rate": 3.56198347107438e-05,
839
+ "loss": 0.0939,
840
+ "step": 970
841
+ },
842
+ {
843
+ "epoch": 0.3136,
844
+ "grad_norm": 2.203125,
845
+ "learning_rate": 3.545454545454546e-05,
846
+ "loss": 0.0984,
847
+ "step": 980
848
+ },
849
+ {
850
+ "epoch": 0.3168,
851
+ "grad_norm": 0.494140625,
852
+ "learning_rate": 3.528925619834711e-05,
853
+ "loss": 0.0924,
854
+ "step": 990
855
+ },
856
+ {
857
+ "epoch": 0.32,
858
+ "grad_norm": 0.91015625,
859
+ "learning_rate": 3.512396694214876e-05,
860
+ "loss": 0.0761,
861
+ "step": 1000
862
+ },
863
+ {
864
+ "epoch": 0.32,
865
+ "eval_loss": 0.09162402153015137,
866
+ "eval_runtime": 112.9796,
867
+ "eval_samples_per_second": 4.426,
868
+ "eval_steps_per_second": 0.142,
869
+ "step": 1000
870
+ },
871
+ {
872
+ "epoch": 0.3232,
873
+ "grad_norm": 0.7265625,
874
+ "learning_rate": 3.495867768595041e-05,
875
+ "loss": 0.0845,
876
+ "step": 1010
877
+ },
878
+ {
879
+ "epoch": 0.3264,
880
+ "grad_norm": 1.0859375,
881
+ "learning_rate": 3.4793388429752064e-05,
882
+ "loss": 0.0852,
883
+ "step": 1020
884
+ },
885
+ {
886
+ "epoch": 0.3296,
887
+ "grad_norm": 1.40625,
888
+ "learning_rate": 3.4628099173553716e-05,
889
+ "loss": 0.1208,
890
+ "step": 1030
891
+ },
892
+ {
893
+ "epoch": 0.3328,
894
+ "grad_norm": 1.3203125,
895
+ "learning_rate": 3.4462809917355374e-05,
896
+ "loss": 0.0807,
897
+ "step": 1040
898
+ },
899
+ {
900
+ "epoch": 0.336,
901
+ "grad_norm": 1.515625,
902
+ "learning_rate": 3.429752066115703e-05,
903
+ "loss": 0.0756,
904
+ "step": 1050
905
+ },
906
+ {
907
+ "epoch": 0.336,
908
+ "eval_loss": 0.09120605140924454,
909
+ "eval_runtime": 111.5564,
910
+ "eval_samples_per_second": 4.482,
911
+ "eval_steps_per_second": 0.143,
912
+ "step": 1050
913
+ },
914
+ {
915
+ "epoch": 0.3392,
916
+ "grad_norm": 0.84765625,
917
+ "learning_rate": 3.413223140495868e-05,
918
+ "loss": 0.1092,
919
+ "step": 1060
920
+ },
921
+ {
922
+ "epoch": 0.3424,
923
+ "grad_norm": 2.671875,
924
+ "learning_rate": 3.3966942148760334e-05,
925
+ "loss": 0.1217,
926
+ "step": 1070
927
+ },
928
+ {
929
+ "epoch": 0.3456,
930
+ "grad_norm": 0.9921875,
931
+ "learning_rate": 3.3801652892561985e-05,
932
+ "loss": 0.0921,
933
+ "step": 1080
934
+ },
935
+ {
936
+ "epoch": 0.3488,
937
+ "grad_norm": 1.0703125,
938
+ "learning_rate": 3.3636363636363636e-05,
939
+ "loss": 0.1023,
940
+ "step": 1090
941
+ },
942
+ {
943
+ "epoch": 0.352,
944
+ "grad_norm": 2.34375,
945
+ "learning_rate": 3.347107438016529e-05,
946
+ "loss": 0.107,
947
+ "step": 1100
948
+ },
949
+ {
950
+ "epoch": 0.352,
951
+ "eval_loss": 0.08991601318120956,
952
+ "eval_runtime": 109.5083,
953
+ "eval_samples_per_second": 4.566,
954
+ "eval_steps_per_second": 0.146,
955
+ "step": 1100
956
+ },
957
+ {
958
+ "epoch": 0.3552,
959
+ "grad_norm": 0.98828125,
960
+ "learning_rate": 3.3305785123966946e-05,
961
+ "loss": 0.0888,
962
+ "step": 1110
963
+ },
964
+ {
965
+ "epoch": 0.3584,
966
+ "grad_norm": 2.640625,
967
+ "learning_rate": 3.31404958677686e-05,
968
+ "loss": 0.0896,
969
+ "step": 1120
970
+ },
971
+ {
972
+ "epoch": 0.3616,
973
+ "grad_norm": 0.4375,
974
+ "learning_rate": 3.297520661157025e-05,
975
+ "loss": 0.072,
976
+ "step": 1130
977
+ },
978
+ {
979
+ "epoch": 0.3648,
980
+ "grad_norm": 1.515625,
981
+ "learning_rate": 3.28099173553719e-05,
982
+ "loss": 0.1046,
983
+ "step": 1140
984
+ },
985
+ {
986
+ "epoch": 0.368,
987
+ "grad_norm": 0.6640625,
988
+ "learning_rate": 3.264462809917356e-05,
989
+ "loss": 0.0791,
990
+ "step": 1150
991
+ },
992
+ {
993
+ "epoch": 0.368,
994
+ "eval_loss": 0.08706250041723251,
995
+ "eval_runtime": 108.4302,
996
+ "eval_samples_per_second": 4.611,
997
+ "eval_steps_per_second": 0.148,
998
+ "step": 1150
999
+ },
1000
+ {
1001
+ "epoch": 0.3712,
1002
+ "grad_norm": 1.4140625,
1003
+ "learning_rate": 3.247933884297521e-05,
1004
+ "loss": 0.0791,
1005
+ "step": 1160
1006
+ },
1007
+ {
1008
+ "epoch": 0.3744,
1009
+ "grad_norm": 1.734375,
1010
+ "learning_rate": 3.231404958677686e-05,
1011
+ "loss": 0.0854,
1012
+ "step": 1170
1013
+ },
1014
+ {
1015
+ "epoch": 0.3776,
1016
+ "grad_norm": 1.09375,
1017
+ "learning_rate": 3.214876033057852e-05,
1018
+ "loss": 0.0912,
1019
+ "step": 1180
1020
+ },
1021
+ {
1022
+ "epoch": 0.3808,
1023
+ "grad_norm": 0.48828125,
1024
+ "learning_rate": 3.198347107438017e-05,
1025
+ "loss": 0.1034,
1026
+ "step": 1190
1027
+ },
1028
+ {
1029
+ "epoch": 0.384,
1030
+ "grad_norm": 1.3515625,
1031
+ "learning_rate": 3.181818181818182e-05,
1032
+ "loss": 0.0863,
1033
+ "step": 1200
1034
+ },
1035
+ {
1036
+ "epoch": 0.384,
1037
+ "eval_loss": 0.08619336038827896,
1038
+ "eval_runtime": 108.0491,
1039
+ "eval_samples_per_second": 4.628,
1040
+ "eval_steps_per_second": 0.148,
1041
+ "step": 1200
1042
+ },
1043
+ {
1044
+ "epoch": 0.3872,
1045
+ "grad_norm": 0.80078125,
1046
+ "learning_rate": 3.165289256198347e-05,
1047
+ "loss": 0.0904,
1048
+ "step": 1210
1049
+ },
1050
+ {
1051
+ "epoch": 0.3904,
1052
+ "grad_norm": 0.7890625,
1053
+ "learning_rate": 3.148760330578512e-05,
1054
+ "loss": 0.1262,
1055
+ "step": 1220
1056
+ },
1057
+ {
1058
+ "epoch": 0.3936,
1059
+ "grad_norm": 1.21875,
1060
+ "learning_rate": 3.1322314049586774e-05,
1061
+ "loss": 0.08,
1062
+ "step": 1230
1063
+ },
1064
+ {
1065
+ "epoch": 0.3968,
1066
+ "grad_norm": 1.609375,
1067
+ "learning_rate": 3.1157024793388425e-05,
1068
+ "loss": 0.0861,
1069
+ "step": 1240
1070
+ },
1071
+ {
1072
+ "epoch": 0.4,
1073
+ "grad_norm": 1.28125,
1074
+ "learning_rate": 3.099173553719008e-05,
1075
+ "loss": 0.0803,
1076
+ "step": 1250
1077
+ },
1078
+ {
1079
+ "epoch": 0.4,
1080
+ "eval_loss": 0.08419531583786011,
1081
+ "eval_runtime": 110.4296,
1082
+ "eval_samples_per_second": 4.528,
1083
+ "eval_steps_per_second": 0.145,
1084
+ "step": 1250
1085
+ },
1086
+ {
1087
+ "epoch": 0.4032,
1088
+ "grad_norm": 1.140625,
1089
+ "learning_rate": 3.082644628099174e-05,
1090
+ "loss": 0.0889,
1091
+ "step": 1260
1092
+ },
1093
+ {
1094
+ "epoch": 0.4064,
1095
+ "grad_norm": 0.7890625,
1096
+ "learning_rate": 3.066115702479339e-05,
1097
+ "loss": 0.0879,
1098
+ "step": 1270
1099
+ },
1100
+ {
1101
+ "epoch": 0.4096,
1102
+ "grad_norm": 1.015625,
1103
+ "learning_rate": 3.0495867768595044e-05,
1104
+ "loss": 0.0776,
1105
+ "step": 1280
1106
+ },
1107
+ {
1108
+ "epoch": 0.4128,
1109
+ "grad_norm": 0.8984375,
1110
+ "learning_rate": 3.0330578512396695e-05,
1111
+ "loss": 0.0903,
1112
+ "step": 1290
1113
+ },
1114
+ {
1115
+ "epoch": 0.416,
1116
+ "grad_norm": 0.7578125,
1117
+ "learning_rate": 3.016528925619835e-05,
1118
+ "loss": 0.0877,
1119
+ "step": 1300
1120
+ },
1121
+ {
1122
+ "epoch": 0.416,
1123
+ "eval_loss": 0.08515820652246475,
1124
+ "eval_runtime": 107.6996,
1125
+ "eval_samples_per_second": 4.643,
1126
+ "eval_steps_per_second": 0.149,
1127
+ "step": 1300
1128
+ },
1129
+ {
1130
+ "epoch": 0.4192,
1131
+ "grad_norm": 0.84765625,
1132
+ "learning_rate": 3e-05,
1133
+ "loss": 0.0912,
1134
+ "step": 1310
1135
+ },
1136
+ {
1137
+ "epoch": 0.4224,
1138
+ "grad_norm": 0.671875,
1139
+ "learning_rate": 2.9834710743801652e-05,
1140
+ "loss": 0.0634,
1141
+ "step": 1320
1142
+ },
1143
+ {
1144
+ "epoch": 0.4256,
1145
+ "grad_norm": 1.03125,
1146
+ "learning_rate": 2.9669421487603307e-05,
1147
+ "loss": 0.076,
1148
+ "step": 1330
1149
+ },
1150
+ {
1151
+ "epoch": 0.4288,
1152
+ "grad_norm": 1.7421875,
1153
+ "learning_rate": 2.950413223140496e-05,
1154
+ "loss": 0.0911,
1155
+ "step": 1340
1156
+ },
1157
+ {
1158
+ "epoch": 0.432,
1159
+ "grad_norm": 0.859375,
1160
+ "learning_rate": 2.9338842975206616e-05,
1161
+ "loss": 0.0804,
1162
+ "step": 1350
1163
+ },
1164
+ {
1165
+ "epoch": 0.432,
1166
+ "eval_loss": 0.0840449184179306,
1167
+ "eval_runtime": 106.9192,
1168
+ "eval_samples_per_second": 4.676,
1169
+ "eval_steps_per_second": 0.15,
1170
+ "step": 1350
1171
+ },
1172
+ {
1173
+ "epoch": 0.4352,
1174
+ "grad_norm": 0.74609375,
1175
+ "learning_rate": 2.9173553719008267e-05,
1176
+ "loss": 0.0779,
1177
+ "step": 1360
1178
+ },
1179
+ {
1180
+ "epoch": 0.4384,
1181
+ "grad_norm": 0.93359375,
1182
+ "learning_rate": 2.9008264462809918e-05,
1183
+ "loss": 0.0749,
1184
+ "step": 1370
1185
+ },
1186
+ {
1187
+ "epoch": 0.4416,
1188
+ "grad_norm": 1.359375,
1189
+ "learning_rate": 2.8842975206611573e-05,
1190
+ "loss": 0.0772,
1191
+ "step": 1380
1192
+ },
1193
+ {
1194
+ "epoch": 0.4448,
1195
+ "grad_norm": 0.86328125,
1196
+ "learning_rate": 2.8677685950413224e-05,
1197
+ "loss": 0.0853,
1198
+ "step": 1390
1199
+ },
1200
+ {
1201
+ "epoch": 0.448,
1202
+ "grad_norm": 1.5390625,
1203
+ "learning_rate": 2.8512396694214875e-05,
1204
+ "loss": 0.1136,
1205
+ "step": 1400
1206
+ },
1207
+ {
1208
+ "epoch": 0.448,
1209
+ "eval_loss": 0.0832226574420929,
1210
+ "eval_runtime": 109.8641,
1211
+ "eval_samples_per_second": 4.551,
1212
+ "eval_steps_per_second": 0.146,
1213
+ "step": 1400
1214
+ },
1215
+ {
1216
+ "epoch": 0.4512,
1217
+ "grad_norm": 0.59765625,
1218
+ "learning_rate": 2.834710743801653e-05,
1219
+ "loss": 0.0655,
1220
+ "step": 1410
1221
+ },
1222
+ {
1223
+ "epoch": 0.4544,
1224
+ "grad_norm": 2.03125,
1225
+ "learning_rate": 2.818181818181818e-05,
1226
+ "loss": 0.0952,
1227
+ "step": 1420
1228
+ },
1229
+ {
1230
+ "epoch": 0.4576,
1231
+ "grad_norm": 1.1796875,
1232
+ "learning_rate": 2.8016528925619832e-05,
1233
+ "loss": 0.0812,
1234
+ "step": 1430
1235
+ },
1236
+ {
1237
+ "epoch": 0.4608,
1238
+ "grad_norm": 0.54296875,
1239
+ "learning_rate": 2.785123966942149e-05,
1240
+ "loss": 0.0649,
1241
+ "step": 1440
1242
+ },
1243
+ {
1244
+ "epoch": 0.464,
1245
+ "grad_norm": 0.671875,
1246
+ "learning_rate": 2.7685950413223145e-05,
1247
+ "loss": 0.0799,
1248
+ "step": 1450
1249
+ },
1250
+ {
1251
+ "epoch": 0.464,
1252
+ "eval_loss": 0.08352148532867432,
1253
+ "eval_runtime": 109.4479,
1254
+ "eval_samples_per_second": 4.568,
1255
+ "eval_steps_per_second": 0.146,
1256
+ "step": 1450
1257
+ },
1258
+ {
1259
+ "epoch": 0.4672,
1260
+ "grad_norm": 0.62890625,
1261
+ "learning_rate": 2.7520661157024796e-05,
1262
+ "loss": 0.0699,
1263
+ "step": 1460
1264
+ },
1265
+ {
1266
+ "epoch": 0.4704,
1267
+ "grad_norm": 1.734375,
1268
+ "learning_rate": 2.7355371900826447e-05,
1269
+ "loss": 0.0688,
1270
+ "step": 1470
1271
+ },
1272
+ {
1273
+ "epoch": 0.4736,
1274
+ "grad_norm": 0.93359375,
1275
+ "learning_rate": 2.7190082644628102e-05,
1276
+ "loss": 0.085,
1277
+ "step": 1480
1278
+ },
1279
+ {
1280
+ "epoch": 0.4768,
1281
+ "grad_norm": 0.490234375,
1282
+ "learning_rate": 2.7024793388429753e-05,
1283
+ "loss": 0.0874,
1284
+ "step": 1490
1285
+ },
1286
+ {
1287
+ "epoch": 0.48,
1288
+ "grad_norm": 1.703125,
1289
+ "learning_rate": 2.6859504132231405e-05,
1290
+ "loss": 0.0896,
1291
+ "step": 1500
1292
+ },
1293
+ {
1294
+ "epoch": 0.48,
1295
+ "eval_loss": 0.08368749916553497,
1296
+ "eval_runtime": 110.4491,
1297
+ "eval_samples_per_second": 4.527,
1298
+ "eval_steps_per_second": 0.145,
1299
+ "step": 1500
1300
+ },
1301
+ {
1302
+ "epoch": 0.4832,
1303
+ "grad_norm": 0.828125,
1304
+ "learning_rate": 2.669421487603306e-05,
1305
+ "loss": 0.1036,
1306
+ "step": 1510
1307
+ },
1308
+ {
1309
+ "epoch": 0.4864,
1310
+ "grad_norm": 1.78125,
1311
+ "learning_rate": 2.652892561983471e-05,
1312
+ "loss": 0.0917,
1313
+ "step": 1520
1314
+ },
1315
+ {
1316
+ "epoch": 0.4896,
1317
+ "grad_norm": 1.1328125,
1318
+ "learning_rate": 2.636363636363636e-05,
1319
+ "loss": 0.0985,
1320
+ "step": 1530
1321
+ },
1322
+ {
1323
+ "epoch": 0.4928,
1324
+ "grad_norm": 1.171875,
1325
+ "learning_rate": 2.619834710743802e-05,
1326
+ "loss": 0.0881,
1327
+ "step": 1540
1328
+ },
1329
+ {
1330
+ "epoch": 0.496,
1331
+ "grad_norm": 0.7734375,
1332
+ "learning_rate": 2.6033057851239674e-05,
1333
+ "loss": 0.078,
1334
+ "step": 1550
1335
+ },
1336
+ {
1337
+ "epoch": 0.496,
1338
+ "eval_loss": 0.08341991901397705,
1339
+ "eval_runtime": 109.2192,
1340
+ "eval_samples_per_second": 4.578,
1341
+ "eval_steps_per_second": 0.146,
1342
+ "step": 1550
1343
+ },
1344
+ {
1345
+ "epoch": 0.4992,
1346
+ "grad_norm": 1.2265625,
1347
+ "learning_rate": 2.5867768595041325e-05,
1348
+ "loss": 0.0784,
1349
+ "step": 1560
1350
+ },
1351
+ {
1352
+ "epoch": 0.5024,
1353
+ "grad_norm": 0.71484375,
1354
+ "learning_rate": 2.5702479338842977e-05,
1355
+ "loss": 0.0748,
1356
+ "step": 1570
1357
+ },
1358
+ {
1359
+ "epoch": 0.5056,
1360
+ "grad_norm": 1.203125,
1361
+ "learning_rate": 2.553719008264463e-05,
1362
+ "loss": 0.0776,
1363
+ "step": 1580
1364
+ },
1365
+ {
1366
+ "epoch": 0.5088,
1367
+ "grad_norm": 1.40625,
1368
+ "learning_rate": 2.5371900826446283e-05,
1369
+ "loss": 0.0927,
1370
+ "step": 1590
1371
+ },
1372
+ {
1373
+ "epoch": 0.512,
1374
+ "grad_norm": 0.6640625,
1375
+ "learning_rate": 2.5206611570247934e-05,
1376
+ "loss": 0.0878,
1377
+ "step": 1600
1378
+ },
1379
+ {
1380
+ "epoch": 0.512,
1381
+ "eval_loss": 0.08232617378234863,
1382
+ "eval_runtime": 110.112,
1383
+ "eval_samples_per_second": 4.541,
1384
+ "eval_steps_per_second": 0.145,
1385
+ "step": 1600
1386
+ },
1387
+ {
1388
+ "epoch": 0.5152,
1389
+ "grad_norm": 1.8125,
1390
+ "learning_rate": 2.504132231404959e-05,
1391
+ "loss": 0.0929,
1392
+ "step": 1610
1393
+ },
1394
+ {
1395
+ "epoch": 0.5184,
1396
+ "grad_norm": 0.75,
1397
+ "learning_rate": 2.4876033057851243e-05,
1398
+ "loss": 0.0796,
1399
+ "step": 1620
1400
+ },
1401
+ {
1402
+ "epoch": 0.5216,
1403
+ "grad_norm": 0.9921875,
1404
+ "learning_rate": 2.4710743801652894e-05,
1405
+ "loss": 0.0932,
1406
+ "step": 1630
1407
+ },
1408
+ {
1409
+ "epoch": 0.5248,
1410
+ "grad_norm": 1.0078125,
1411
+ "learning_rate": 2.4545454545454545e-05,
1412
+ "loss": 0.1084,
1413
+ "step": 1640
1414
+ },
1415
+ {
1416
+ "epoch": 0.528,
1417
+ "grad_norm": 1.03125,
1418
+ "learning_rate": 2.43801652892562e-05,
1419
+ "loss": 0.0763,
1420
+ "step": 1650
1421
+ },
1422
+ {
1423
+ "epoch": 0.528,
1424
+ "eval_loss": 0.08151757717132568,
1425
+ "eval_runtime": 111.3314,
1426
+ "eval_samples_per_second": 4.491,
1427
+ "eval_steps_per_second": 0.144,
1428
+ "step": 1650
1429
+ },
1430
+ {
1431
+ "epoch": 0.5312,
1432
+ "grad_norm": 0.859375,
1433
+ "learning_rate": 2.421487603305785e-05,
1434
+ "loss": 0.062,
1435
+ "step": 1660
1436
+ },
1437
+ {
1438
+ "epoch": 0.5344,
1439
+ "grad_norm": 1.0859375,
1440
+ "learning_rate": 2.4049586776859506e-05,
1441
+ "loss": 0.0628,
1442
+ "step": 1670
1443
+ },
1444
+ {
1445
+ "epoch": 0.5376,
1446
+ "grad_norm": 0.59375,
1447
+ "learning_rate": 2.3884297520661157e-05,
1448
+ "loss": 0.0908,
1449
+ "step": 1680
1450
+ },
1451
+ {
1452
+ "epoch": 0.5408,
1453
+ "grad_norm": 1.3515625,
1454
+ "learning_rate": 2.3719008264462812e-05,
1455
+ "loss": 0.0927,
1456
+ "step": 1690
1457
+ },
1458
+ {
1459
+ "epoch": 0.544,
1460
+ "grad_norm": 1.109375,
1461
+ "learning_rate": 2.3553719008264463e-05,
1462
+ "loss": 0.0958,
1463
+ "step": 1700
1464
+ },
1465
+ {
1466
+ "epoch": 0.544,
1467
+ "eval_loss": 0.08200781047344208,
1468
+ "eval_runtime": 114.049,
1469
+ "eval_samples_per_second": 4.384,
1470
+ "eval_steps_per_second": 0.14,
1471
+ "step": 1700
1472
+ },
1473
+ {
1474
+ "epoch": 0.5472,
1475
+ "grad_norm": 0.53125,
1476
+ "learning_rate": 2.3388429752066114e-05,
1477
+ "loss": 0.0716,
1478
+ "step": 1710
1479
+ },
1480
+ {
1481
+ "epoch": 0.5504,
1482
+ "grad_norm": 0.421875,
1483
+ "learning_rate": 2.3223140495867772e-05,
1484
+ "loss": 0.0715,
1485
+ "step": 1720
1486
+ },
1487
+ {
1488
+ "epoch": 0.5536,
1489
+ "grad_norm": 2.703125,
1490
+ "learning_rate": 2.3057851239669423e-05,
1491
+ "loss": 0.0941,
1492
+ "step": 1730
1493
+ },
1494
+ {
1495
+ "epoch": 0.5568,
1496
+ "grad_norm": 0.89453125,
1497
+ "learning_rate": 2.2892561983471075e-05,
1498
+ "loss": 0.0785,
1499
+ "step": 1740
1500
+ },
1501
+ {
1502
+ "epoch": 0.56,
1503
+ "grad_norm": 0.76953125,
1504
+ "learning_rate": 2.272727272727273e-05,
1505
+ "loss": 0.0857,
1506
+ "step": 1750
1507
+ },
1508
+ {
1509
+ "epoch": 0.56,
1510
+ "eval_loss": 0.08109960705041885,
1511
+ "eval_runtime": 112.2299,
1512
+ "eval_samples_per_second": 4.455,
1513
+ "eval_steps_per_second": 0.143,
1514
+ "step": 1750
1515
+ },
1516
+ {
1517
+ "epoch": 0.5632,
1518
+ "grad_norm": 0.953125,
1519
+ "learning_rate": 2.256198347107438e-05,
1520
+ "loss": 0.0601,
1521
+ "step": 1760
1522
+ },
1523
+ {
1524
+ "epoch": 0.5664,
1525
+ "grad_norm": 0.322265625,
1526
+ "learning_rate": 2.2396694214876035e-05,
1527
+ "loss": 0.0866,
1528
+ "step": 1770
1529
+ },
1530
+ {
1531
+ "epoch": 0.5696,
1532
+ "grad_norm": 1.015625,
1533
+ "learning_rate": 2.2231404958677686e-05,
1534
+ "loss": 0.0708,
1535
+ "step": 1780
1536
+ },
1537
+ {
1538
+ "epoch": 0.5728,
1539
+ "grad_norm": 0.72265625,
1540
+ "learning_rate": 2.206611570247934e-05,
1541
+ "loss": 0.0762,
1542
+ "step": 1790
1543
+ },
1544
+ {
1545
+ "epoch": 0.576,
1546
+ "grad_norm": 0.5703125,
1547
+ "learning_rate": 2.1900826446280992e-05,
1548
+ "loss": 0.0831,
1549
+ "step": 1800
1550
+ },
1551
+ {
1552
+ "epoch": 0.576,
1553
+ "eval_loss": 0.07973046600818634,
1554
+ "eval_runtime": 108.6444,
1555
+ "eval_samples_per_second": 4.602,
1556
+ "eval_steps_per_second": 0.147,
1557
+ "step": 1800
1558
+ },
1559
+ {
1560
+ "epoch": 0.5792,
1561
+ "grad_norm": 0.87109375,
1562
+ "learning_rate": 2.1735537190082643e-05,
1563
+ "loss": 0.0677,
1564
+ "step": 1810
1565
+ },
1566
+ {
1567
+ "epoch": 0.5824,
1568
+ "grad_norm": 0.8203125,
1569
+ "learning_rate": 2.1570247933884298e-05,
1570
+ "loss": 0.0738,
1571
+ "step": 1820
1572
+ },
1573
+ {
1574
+ "epoch": 0.5856,
1575
+ "grad_norm": 0.58203125,
1576
+ "learning_rate": 2.1404958677685953e-05,
1577
+ "loss": 0.064,
1578
+ "step": 1830
1579
+ },
1580
+ {
1581
+ "epoch": 0.5888,
1582
+ "grad_norm": 1.3359375,
1583
+ "learning_rate": 2.1239669421487604e-05,
1584
+ "loss": 0.0777,
1585
+ "step": 1840
1586
+ },
1587
+ {
1588
+ "epoch": 0.592,
1589
+ "grad_norm": 1.0078125,
1590
+ "learning_rate": 2.1074380165289255e-05,
1591
+ "loss": 0.0792,
1592
+ "step": 1850
1593
+ },
1594
+ {
1595
+ "epoch": 0.592,
1596
+ "eval_loss": 0.08046093583106995,
1597
+ "eval_runtime": 110.6488,
1598
+ "eval_samples_per_second": 4.519,
1599
+ "eval_steps_per_second": 0.145,
1600
+ "step": 1850
1601
+ },
1602
+ {
1603
+ "epoch": 0.5952,
1604
+ "grad_norm": 1.28125,
1605
+ "learning_rate": 2.090909090909091e-05,
1606
+ "loss": 0.0778,
1607
+ "step": 1860
1608
+ },
1609
+ {
1610
+ "epoch": 0.5984,
1611
+ "grad_norm": 0.8125,
1612
+ "learning_rate": 2.0743801652892564e-05,
1613
+ "loss": 0.0831,
1614
+ "step": 1870
1615
+ },
1616
+ {
1617
+ "epoch": 0.6016,
1618
+ "grad_norm": 0.90234375,
1619
+ "learning_rate": 2.0578512396694216e-05,
1620
+ "loss": 0.0721,
1621
+ "step": 1880
1622
+ },
1623
+ {
1624
+ "epoch": 0.6048,
1625
+ "grad_norm": 0.69140625,
1626
+ "learning_rate": 2.0413223140495867e-05,
1627
+ "loss": 0.0661,
1628
+ "step": 1890
1629
+ },
1630
+ {
1631
+ "epoch": 0.608,
1632
+ "grad_norm": 0.21484375,
1633
+ "learning_rate": 2.024793388429752e-05,
1634
+ "loss": 0.0725,
1635
+ "step": 1900
1636
+ },
1637
+ {
1638
+ "epoch": 0.608,
1639
+ "eval_loss": 0.08086328208446503,
1640
+ "eval_runtime": 108.5455,
1641
+ "eval_samples_per_second": 4.606,
1642
+ "eval_steps_per_second": 0.147,
1643
+ "step": 1900
1644
+ },
1645
+ {
1646
+ "epoch": 0.6112,
1647
+ "grad_norm": 0.5,
1648
+ "learning_rate": 2.0082644628099173e-05,
1649
+ "loss": 0.0813,
1650
+ "step": 1910
1651
+ },
1652
+ {
1653
+ "epoch": 0.6144,
1654
+ "grad_norm": 1.2109375,
1655
+ "learning_rate": 1.9917355371900827e-05,
1656
+ "loss": 0.0724,
1657
+ "step": 1920
1658
+ },
1659
+ {
1660
+ "epoch": 0.6176,
1661
+ "grad_norm": 0.921875,
1662
+ "learning_rate": 1.9752066115702482e-05,
1663
+ "loss": 0.0585,
1664
+ "step": 1930
1665
+ },
1666
+ {
1667
+ "epoch": 0.6208,
1668
+ "grad_norm": 1.890625,
1669
+ "learning_rate": 1.9586776859504133e-05,
1670
+ "loss": 0.0928,
1671
+ "step": 1940
1672
+ },
1673
+ {
1674
+ "epoch": 0.624,
1675
+ "grad_norm": 1.0546875,
1676
+ "learning_rate": 1.9421487603305784e-05,
1677
+ "loss": 0.0795,
1678
+ "step": 1950
1679
+ },
1680
+ {
1681
+ "epoch": 0.624,
1682
+ "eval_loss": 0.08059374988079071,
1683
+ "eval_runtime": 107.6528,
1684
+ "eval_samples_per_second": 4.645,
1685
+ "eval_steps_per_second": 0.149,
1686
+ "step": 1950
1687
+ },
1688
+ {
1689
+ "epoch": 0.6272,
1690
+ "grad_norm": 0.765625,
1691
+ "learning_rate": 1.925619834710744e-05,
1692
+ "loss": 0.0675,
1693
+ "step": 1960
1694
+ },
1695
+ {
1696
+ "epoch": 0.6304,
1697
+ "grad_norm": 1.6328125,
1698
+ "learning_rate": 1.9090909090909094e-05,
1699
+ "loss": 0.0896,
1700
+ "step": 1970
1701
+ },
1702
+ {
1703
+ "epoch": 0.6336,
1704
+ "grad_norm": 0.76953125,
1705
+ "learning_rate": 1.8925619834710745e-05,
1706
+ "loss": 0.075,
1707
+ "step": 1980
1708
+ },
1709
+ {
1710
+ "epoch": 0.6368,
1711
+ "grad_norm": 1.7890625,
1712
+ "learning_rate": 1.8760330578512396e-05,
1713
+ "loss": 0.0808,
1714
+ "step": 1990
1715
+ },
1716
+ {
1717
+ "epoch": 0.64,
1718
+ "grad_norm": 1.21875,
1719
+ "learning_rate": 1.859504132231405e-05,
1720
+ "loss": 0.0774,
1721
+ "step": 2000
1722
+ },
1723
+ {
1724
+ "epoch": 0.64,
1725
+ "eval_loss": 0.07988671958446503,
1726
+ "eval_runtime": 107.6118,
1727
+ "eval_samples_per_second": 4.646,
1728
+ "eval_steps_per_second": 0.149,
1729
+ "step": 2000
1730
+ },
1731
+ {
1732
+ "epoch": 0.6432,
1733
+ "grad_norm": 0.63671875,
1734
+ "learning_rate": 1.8429752066115705e-05,
1735
+ "loss": 0.0795,
1736
+ "step": 2010
1737
+ },
1738
+ {
1739
+ "epoch": 0.6464,
1740
+ "grad_norm": 1.4140625,
1741
+ "learning_rate": 1.8264462809917356e-05,
1742
+ "loss": 0.1115,
1743
+ "step": 2020
1744
+ },
1745
+ {
1746
+ "epoch": 0.6496,
1747
+ "grad_norm": 0.8203125,
1748
+ "learning_rate": 1.8099173553719008e-05,
1749
+ "loss": 0.079,
1750
+ "step": 2030
1751
+ },
1752
+ {
1753
+ "epoch": 0.6528,
1754
+ "grad_norm": 0.28515625,
1755
+ "learning_rate": 1.7933884297520662e-05,
1756
+ "loss": 0.0646,
1757
+ "step": 2040
1758
+ },
1759
+ {
1760
+ "epoch": 0.656,
1761
+ "grad_norm": 1.25,
1762
+ "learning_rate": 1.7768595041322314e-05,
1763
+ "loss": 0.0846,
1764
+ "step": 2050
1765
+ },
1766
+ {
1767
+ "epoch": 0.656,
1768
+ "eval_loss": 0.07978320121765137,
1769
+ "eval_runtime": 107.8393,
1770
+ "eval_samples_per_second": 4.637,
1771
+ "eval_steps_per_second": 0.148,
1772
+ "step": 2050
1773
+ },
1774
+ {
1775
+ "epoch": 0.6592,
1776
+ "grad_norm": 0.55859375,
1777
+ "learning_rate": 1.7603305785123968e-05,
1778
+ "loss": 0.0949,
1779
+ "step": 2060
1780
+ },
1781
+ {
1782
+ "epoch": 0.6624,
1783
+ "grad_norm": 0.6171875,
1784
+ "learning_rate": 1.7438016528925623e-05,
1785
+ "loss": 0.0822,
1786
+ "step": 2070
1787
+ },
1788
+ {
1789
+ "epoch": 0.6656,
1790
+ "grad_norm": 0.75390625,
1791
+ "learning_rate": 1.7272727272727274e-05,
1792
+ "loss": 0.0666,
1793
+ "step": 2080
1794
+ },
1795
+ {
1796
+ "epoch": 0.6688,
1797
+ "grad_norm": 1.1796875,
1798
+ "learning_rate": 1.7107438016528925e-05,
1799
+ "loss": 0.0745,
1800
+ "step": 2090
1801
+ },
1802
+ {
1803
+ "epoch": 0.672,
1804
+ "grad_norm": 1.2265625,
1805
+ "learning_rate": 1.694214876033058e-05,
1806
+ "loss": 0.1087,
1807
+ "step": 2100
1808
+ },
1809
+ {
1810
+ "epoch": 0.672,
1811
+ "eval_loss": 0.07949023693799973,
1812
+ "eval_runtime": 112.9895,
1813
+ "eval_samples_per_second": 4.425,
1814
+ "eval_steps_per_second": 0.142,
1815
+ "step": 2100
1816
+ },
1817
+ {
1818
+ "epoch": 0.6752,
1819
+ "grad_norm": 1.953125,
1820
+ "learning_rate": 1.6776859504132234e-05,
1821
+ "loss": 0.0917,
1822
+ "step": 2110
1823
+ },
1824
+ {
1825
+ "epoch": 0.6784,
1826
+ "grad_norm": 0.56640625,
1827
+ "learning_rate": 1.6611570247933886e-05,
1828
+ "loss": 0.075,
1829
+ "step": 2120
1830
+ },
1831
+ {
1832
+ "epoch": 0.6816,
1833
+ "grad_norm": 1.296875,
1834
+ "learning_rate": 1.6446280991735537e-05,
1835
+ "loss": 0.0788,
1836
+ "step": 2130
1837
+ },
1838
+ {
1839
+ "epoch": 0.6848,
1840
+ "grad_norm": 1.203125,
1841
+ "learning_rate": 1.628099173553719e-05,
1842
+ "loss": 0.0821,
1843
+ "step": 2140
1844
+ },
1845
+ {
1846
+ "epoch": 0.688,
1847
+ "grad_norm": 1.296875,
1848
+ "learning_rate": 1.6115702479338843e-05,
1849
+ "loss": 0.0836,
1850
+ "step": 2150
1851
+ },
1852
+ {
1853
+ "epoch": 0.688,
1854
+ "eval_loss": 0.0802304670214653,
1855
+ "eval_runtime": 109.4516,
1856
+ "eval_samples_per_second": 4.568,
1857
+ "eval_steps_per_second": 0.146,
1858
+ "step": 2150
1859
+ },
1860
+ {
1861
+ "epoch": 0.6912,
1862
+ "grad_norm": 1.21875,
1863
+ "learning_rate": 1.5950413223140497e-05,
1864
+ "loss": 0.0845,
1865
+ "step": 2160
1866
+ },
1867
+ {
1868
+ "epoch": 0.6944,
1869
+ "grad_norm": 0.69921875,
1870
+ "learning_rate": 1.578512396694215e-05,
1871
+ "loss": 0.0701,
1872
+ "step": 2170
1873
+ },
1874
+ {
1875
+ "epoch": 0.6976,
1876
+ "grad_norm": 1.5078125,
1877
+ "learning_rate": 1.5619834710743803e-05,
1878
+ "loss": 0.0789,
1879
+ "step": 2180
1880
+ },
1881
+ {
1882
+ "epoch": 0.7008,
1883
+ "grad_norm": 1.1015625,
1884
+ "learning_rate": 1.5454545454545454e-05,
1885
+ "loss": 0.0611,
1886
+ "step": 2190
1887
+ },
1888
+ {
1889
+ "epoch": 0.704,
1890
+ "grad_norm": 1.578125,
1891
+ "learning_rate": 1.5289256198347106e-05,
1892
+ "loss": 0.0782,
1893
+ "step": 2200
1894
+ },
1895
+ {
1896
+ "epoch": 0.704,
1897
+ "eval_loss": 0.07987890392541885,
1898
+ "eval_runtime": 108.1627,
1899
+ "eval_samples_per_second": 4.623,
1900
+ "eval_steps_per_second": 0.148,
1901
+ "step": 2200
1902
+ },
1903
+ {
1904
+ "epoch": 0.7072,
1905
+ "grad_norm": 0.69921875,
1906
+ "learning_rate": 1.5123966942148762e-05,
1907
+ "loss": 0.0692,
1908
+ "step": 2210
1909
+ },
1910
+ {
1911
+ "epoch": 0.7104,
1912
+ "grad_norm": 0.9140625,
1913
+ "learning_rate": 1.4958677685950415e-05,
1914
+ "loss": 0.0656,
1915
+ "step": 2220
1916
+ },
1917
+ {
1918
+ "epoch": 0.7136,
1919
+ "grad_norm": 1.2890625,
1920
+ "learning_rate": 1.4793388429752068e-05,
1921
+ "loss": 0.0854,
1922
+ "step": 2230
1923
+ },
1924
+ {
1925
+ "epoch": 0.7168,
1926
+ "grad_norm": 0.5703125,
1927
+ "learning_rate": 1.4628099173553719e-05,
1928
+ "loss": 0.0798,
1929
+ "step": 2240
1930
+ },
1931
+ {
1932
+ "epoch": 0.72,
1933
+ "grad_norm": 0.458984375,
1934
+ "learning_rate": 1.4462809917355372e-05,
1935
+ "loss": 0.0885,
1936
+ "step": 2250
1937
+ },
1938
+ {
1939
+ "epoch": 0.72,
1940
+ "eval_loss": 0.07889453321695328,
1941
+ "eval_runtime": 110.4069,
1942
+ "eval_samples_per_second": 4.529,
1943
+ "eval_steps_per_second": 0.145,
1944
+ "step": 2250
1945
+ },
1946
+ {
1947
+ "epoch": 0.7232,
1948
+ "grad_norm": 1.234375,
1949
+ "learning_rate": 1.4297520661157027e-05,
1950
+ "loss": 0.0813,
1951
+ "step": 2260
1952
+ },
1953
+ {
1954
+ "epoch": 0.7264,
1955
+ "grad_norm": 0.76171875,
1956
+ "learning_rate": 1.413223140495868e-05,
1957
+ "loss": 0.0615,
1958
+ "step": 2270
1959
+ },
1960
+ {
1961
+ "epoch": 0.7296,
1962
+ "grad_norm": 1.0390625,
1963
+ "learning_rate": 1.396694214876033e-05,
1964
+ "loss": 0.0773,
1965
+ "step": 2280
1966
+ },
1967
+ {
1968
+ "epoch": 0.7328,
1969
+ "grad_norm": 1.1484375,
1970
+ "learning_rate": 1.3801652892561984e-05,
1971
+ "loss": 0.0854,
1972
+ "step": 2290
1973
+ },
1974
+ {
1975
+ "epoch": 0.736,
1976
+ "grad_norm": 1.390625,
1977
+ "learning_rate": 1.3636363636363637e-05,
1978
+ "loss": 0.0926,
1979
+ "step": 2300
1980
+ },
1981
+ {
1982
+ "epoch": 0.736,
1983
+ "eval_loss": 0.07876367121934891,
1984
+ "eval_runtime": 120.6799,
1985
+ "eval_samples_per_second": 4.143,
1986
+ "eval_steps_per_second": 0.133,
1987
+ "step": 2300
1988
+ },
1989
+ {
1990
+ "epoch": 0.7392,
1991
+ "grad_norm": 1.046875,
1992
+ "learning_rate": 1.3471074380165291e-05,
1993
+ "loss": 0.0992,
1994
+ "step": 2310
1995
+ },
1996
+ {
1997
+ "epoch": 0.7424,
1998
+ "grad_norm": 0.44140625,
1999
+ "learning_rate": 1.3305785123966944e-05,
2000
+ "loss": 0.0629,
2001
+ "step": 2320
2002
+ },
2003
+ {
2004
+ "epoch": 0.7456,
2005
+ "grad_norm": 0.7421875,
2006
+ "learning_rate": 1.3140495867768595e-05,
2007
+ "loss": 0.0799,
2008
+ "step": 2330
2009
+ },
2010
+ {
2011
+ "epoch": 0.7488,
2012
+ "grad_norm": 1.203125,
2013
+ "learning_rate": 1.2975206611570248e-05,
2014
+ "loss": 0.092,
2015
+ "step": 2340
2016
+ },
2017
+ {
2018
+ "epoch": 0.752,
2019
+ "grad_norm": 1.34375,
2020
+ "learning_rate": 1.2809917355371901e-05,
2021
+ "loss": 0.064,
2022
+ "step": 2350
2023
+ },
2024
+ {
2025
+ "epoch": 0.752,
2026
+ "eval_loss": 0.07899804413318634,
2027
+ "eval_runtime": 110.2904,
2028
+ "eval_samples_per_second": 4.533,
2029
+ "eval_steps_per_second": 0.145,
2030
+ "step": 2350
2031
+ },
2032
+ {
2033
+ "epoch": 0.7552,
2034
+ "grad_norm": 0.6875,
2035
+ "learning_rate": 1.2644628099173556e-05,
2036
+ "loss": 0.0833,
2037
+ "step": 2360
2038
+ },
2039
+ {
2040
+ "epoch": 0.7584,
2041
+ "grad_norm": 2.015625,
2042
+ "learning_rate": 1.2479338842975207e-05,
2043
+ "loss": 0.0841,
2044
+ "step": 2370
2045
+ },
2046
+ {
2047
+ "epoch": 0.7616,
2048
+ "grad_norm": 1.1875,
2049
+ "learning_rate": 1.231404958677686e-05,
2050
+ "loss": 0.0798,
2051
+ "step": 2380
2052
+ },
2053
+ {
2054
+ "epoch": 0.7648,
2055
+ "grad_norm": 1.7421875,
2056
+ "learning_rate": 1.2148760330578513e-05,
2057
+ "loss": 0.0956,
2058
+ "step": 2390
2059
+ },
2060
+ {
2061
+ "epoch": 0.768,
2062
+ "grad_norm": 0.99609375,
2063
+ "learning_rate": 1.1983471074380166e-05,
2064
+ "loss": 0.0798,
2065
+ "step": 2400
2066
+ },
2067
+ {
2068
+ "epoch": 0.768,
2069
+ "eval_loss": 0.07969531416893005,
2070
+ "eval_runtime": 110.6153,
2071
+ "eval_samples_per_second": 4.52,
2072
+ "eval_steps_per_second": 0.145,
2073
+ "step": 2400
2074
+ },
2075
+ {
2076
+ "epoch": 0.7712,
2077
+ "grad_norm": 0.5078125,
2078
+ "learning_rate": 1.1818181818181819e-05,
2079
+ "loss": 0.0711,
2080
+ "step": 2410
2081
+ },
2082
+ {
2083
+ "epoch": 0.7744,
2084
+ "grad_norm": 1.4765625,
2085
+ "learning_rate": 1.1652892561983472e-05,
2086
+ "loss": 0.0708,
2087
+ "step": 2420
2088
+ },
2089
+ {
2090
+ "epoch": 0.7776,
2091
+ "grad_norm": 0.9609375,
2092
+ "learning_rate": 1.1487603305785125e-05,
2093
+ "loss": 0.0656,
2094
+ "step": 2430
2095
+ },
2096
+ {
2097
+ "epoch": 0.7808,
2098
+ "grad_norm": 0.71484375,
2099
+ "learning_rate": 1.1322314049586777e-05,
2100
+ "loss": 0.0723,
2101
+ "step": 2440
2102
+ },
2103
+ {
2104
+ "epoch": 0.784,
2105
+ "grad_norm": 1.09375,
2106
+ "learning_rate": 1.115702479338843e-05,
2107
+ "loss": 0.0789,
2108
+ "step": 2450
2109
+ },
2110
+ {
2111
+ "epoch": 0.784,
2112
+ "eval_loss": 0.0796191394329071,
2113
+ "eval_runtime": 110.6578,
2114
+ "eval_samples_per_second": 4.518,
2115
+ "eval_steps_per_second": 0.145,
2116
+ "step": 2450
2117
+ },
2118
+ {
2119
+ "epoch": 0.7872,
2120
+ "grad_norm": 1.046875,
2121
+ "learning_rate": 1.0991735537190083e-05,
2122
+ "loss": 0.0755,
2123
+ "step": 2460
2124
+ },
2125
+ {
2126
+ "epoch": 0.7904,
2127
+ "grad_norm": 1.140625,
2128
+ "learning_rate": 1.0826446280991736e-05,
2129
+ "loss": 0.0833,
2130
+ "step": 2470
2131
+ },
2132
+ {
2133
+ "epoch": 0.7936,
2134
+ "grad_norm": 0.5625,
2135
+ "learning_rate": 1.0661157024793389e-05,
2136
+ "loss": 0.1063,
2137
+ "step": 2480
2138
+ },
2139
+ {
2140
+ "epoch": 0.7968,
2141
+ "grad_norm": 0.48046875,
2142
+ "learning_rate": 1.0495867768595042e-05,
2143
+ "loss": 0.0747,
2144
+ "step": 2490
2145
+ },
2146
+ {
2147
+ "epoch": 0.8,
2148
+ "grad_norm": 1.390625,
2149
+ "learning_rate": 1.0330578512396695e-05,
2150
+ "loss": 0.0997,
2151
+ "step": 2500
2152
+ },
2153
+ {
2154
+ "epoch": 0.8,
2155
+ "eval_loss": 0.0799550786614418,
2156
+ "eval_runtime": 106.8244,
2157
+ "eval_samples_per_second": 4.681,
2158
+ "eval_steps_per_second": 0.15,
2159
+ "step": 2500
2160
+ },
2161
+ {
2162
+ "epoch": 0.8032,
2163
+ "grad_norm": 1.046875,
2164
+ "learning_rate": 1.0165289256198348e-05,
2165
+ "loss": 0.0737,
2166
+ "step": 2510
2167
+ },
2168
+ {
2169
+ "epoch": 0.8064,
2170
+ "grad_norm": 1.1328125,
2171
+ "learning_rate": 1e-05,
2172
+ "loss": 0.099,
2173
+ "step": 2520
2174
+ },
2175
+ {
2176
+ "epoch": 0.8096,
2177
+ "grad_norm": 0.58203125,
2178
+ "learning_rate": 9.834710743801654e-06,
2179
+ "loss": 0.0668,
2180
+ "step": 2530
2181
+ },
2182
+ {
2183
+ "epoch": 0.8128,
2184
+ "grad_norm": 1.140625,
2185
+ "learning_rate": 9.669421487603305e-06,
2186
+ "loss": 0.0772,
2187
+ "step": 2540
2188
+ },
2189
+ {
2190
+ "epoch": 0.816,
2191
+ "grad_norm": 0.7734375,
2192
+ "learning_rate": 9.50413223140496e-06,
2193
+ "loss": 0.0755,
2194
+ "step": 2550
2195
+ },
2196
+ {
2197
+ "epoch": 0.816,
2198
+ "eval_loss": 0.07922851294279099,
2199
+ "eval_runtime": 106.4516,
2200
+ "eval_samples_per_second": 4.697,
2201
+ "eval_steps_per_second": 0.15,
2202
+ "step": 2550
2203
+ },
2204
+ {
2205
+ "epoch": 0.8192,
2206
+ "grad_norm": 2.125,
2207
+ "learning_rate": 9.338842975206613e-06,
2208
+ "loss": 0.0735,
2209
+ "step": 2560
2210
+ },
2211
+ {
2212
+ "epoch": 0.8224,
2213
+ "grad_norm": 1.7265625,
2214
+ "learning_rate": 9.173553719008265e-06,
2215
+ "loss": 0.0879,
2216
+ "step": 2570
2217
+ },
2218
+ {
2219
+ "epoch": 0.8256,
2220
+ "grad_norm": 0.2578125,
2221
+ "learning_rate": 9.008264462809918e-06,
2222
+ "loss": 0.075,
2223
+ "step": 2580
2224
+ },
2225
+ {
2226
+ "epoch": 0.8288,
2227
+ "grad_norm": 0.9375,
2228
+ "learning_rate": 8.84297520661157e-06,
2229
+ "loss": 0.0718,
2230
+ "step": 2590
2231
+ },
2232
+ {
2233
+ "epoch": 0.832,
2234
+ "grad_norm": 0.828125,
2235
+ "learning_rate": 8.677685950413224e-06,
2236
+ "loss": 0.069,
2237
+ "step": 2600
2238
+ },
2239
+ {
2240
+ "epoch": 0.832,
2241
+ "eval_loss": 0.07956640422344208,
2242
+ "eval_runtime": 116.0737,
2243
+ "eval_samples_per_second": 4.308,
2244
+ "eval_steps_per_second": 0.138,
2245
+ "step": 2600
2246
+ },
2247
+ {
2248
+ "epoch": 0.8352,
2249
+ "grad_norm": 1.7109375,
2250
+ "learning_rate": 8.512396694214875e-06,
2251
+ "loss": 0.1006,
2252
+ "step": 2610
2253
+ },
2254
+ {
2255
+ "epoch": 0.8384,
2256
+ "grad_norm": 1.34375,
2257
+ "learning_rate": 8.34710743801653e-06,
2258
+ "loss": 0.0771,
2259
+ "step": 2620
2260
+ },
2261
+ {
2262
+ "epoch": 0.8416,
2263
+ "grad_norm": 0.9375,
2264
+ "learning_rate": 8.181818181818183e-06,
2265
+ "loss": 0.0668,
2266
+ "step": 2630
2267
+ },
2268
+ {
2269
+ "epoch": 0.8448,
2270
+ "grad_norm": 1.1875,
2271
+ "learning_rate": 8.016528925619834e-06,
2272
+ "loss": 0.0663,
2273
+ "step": 2640
2274
+ },
2275
+ {
2276
+ "epoch": 0.848,
2277
+ "grad_norm": 0.466796875,
2278
+ "learning_rate": 7.851239669421489e-06,
2279
+ "loss": 0.0687,
2280
+ "step": 2650
2281
+ },
2282
+ {
2283
+ "epoch": 0.848,
2284
+ "eval_loss": 0.07871288806200027,
2285
+ "eval_runtime": 108.1878,
2286
+ "eval_samples_per_second": 4.622,
2287
+ "eval_steps_per_second": 0.148,
2288
+ "step": 2650
2289
+ },
2290
+ {
2291
+ "epoch": 0.8512,
2292
+ "grad_norm": 1.0546875,
2293
+ "learning_rate": 7.68595041322314e-06,
2294
+ "loss": 0.0899,
2295
+ "step": 2660
2296
+ },
2297
+ {
2298
+ "epoch": 0.8544,
2299
+ "grad_norm": 0.9921875,
2300
+ "learning_rate": 7.520661157024795e-06,
2301
+ "loss": 0.0735,
2302
+ "step": 2670
2303
+ },
2304
+ {
2305
+ "epoch": 0.8576,
2306
+ "grad_norm": 0.416015625,
2307
+ "learning_rate": 7.355371900826447e-06,
2308
+ "loss": 0.0905,
2309
+ "step": 2680
2310
+ },
2311
+ {
2312
+ "epoch": 0.8608,
2313
+ "grad_norm": 1.2109375,
2314
+ "learning_rate": 7.190082644628099e-06,
2315
+ "loss": 0.0664,
2316
+ "step": 2690
2317
+ },
2318
+ {
2319
+ "epoch": 0.864,
2320
+ "grad_norm": 1.09375,
2321
+ "learning_rate": 7.0247933884297525e-06,
2322
+ "loss": 0.0791,
2323
+ "step": 2700
2324
+ },
2325
+ {
2326
+ "epoch": 0.864,
2327
+ "eval_loss": 0.07861328125,
2328
+ "eval_runtime": 109.1401,
2329
+ "eval_samples_per_second": 4.581,
2330
+ "eval_steps_per_second": 0.147,
2331
+ "step": 2700
2332
+ },
2333
+ {
2334
+ "epoch": 0.8672,
2335
+ "grad_norm": 0.470703125,
2336
+ "learning_rate": 6.859504132231405e-06,
2337
+ "loss": 0.0588,
2338
+ "step": 2710
2339
+ },
2340
+ {
2341
+ "epoch": 0.8704,
2342
+ "grad_norm": 0.4453125,
2343
+ "learning_rate": 6.694214876033058e-06,
2344
+ "loss": 0.1022,
2345
+ "step": 2720
2346
+ },
2347
+ {
2348
+ "epoch": 0.8736,
2349
+ "grad_norm": 0.77734375,
2350
+ "learning_rate": 6.528925619834711e-06,
2351
+ "loss": 0.0645,
2352
+ "step": 2730
2353
+ },
2354
+ {
2355
+ "epoch": 0.8768,
2356
+ "grad_norm": 0.79296875,
2357
+ "learning_rate": 6.363636363636363e-06,
2358
+ "loss": 0.0688,
2359
+ "step": 2740
2360
+ },
2361
+ {
2362
+ "epoch": 0.88,
2363
+ "grad_norm": 1.0,
2364
+ "learning_rate": 6.198347107438017e-06,
2365
+ "loss": 0.0763,
2366
+ "step": 2750
2367
+ },
2368
+ {
2369
+ "epoch": 0.88,
2370
+ "eval_loss": 0.07813085615634918,
2371
+ "eval_runtime": 107.2687,
2372
+ "eval_samples_per_second": 4.661,
2373
+ "eval_steps_per_second": 0.149,
2374
+ "step": 2750
2375
+ },
2376
+ {
2377
+ "epoch": 0.8832,
2378
+ "grad_norm": 1.0625,
2379
+ "learning_rate": 6.03305785123967e-06,
2380
+ "loss": 0.0619,
2381
+ "step": 2760
2382
+ },
2383
+ {
2384
+ "epoch": 0.8864,
2385
+ "grad_norm": 0.81640625,
2386
+ "learning_rate": 5.867768595041322e-06,
2387
+ "loss": 0.0861,
2388
+ "step": 2770
2389
+ },
2390
+ {
2391
+ "epoch": 0.8896,
2392
+ "grad_norm": 1.140625,
2393
+ "learning_rate": 5.702479338842975e-06,
2394
+ "loss": 0.0573,
2395
+ "step": 2780
2396
+ },
2397
+ {
2398
+ "epoch": 0.8928,
2399
+ "grad_norm": 0.6015625,
2400
+ "learning_rate": 5.537190082644628e-06,
2401
+ "loss": 0.0546,
2402
+ "step": 2790
2403
+ },
2404
+ {
2405
+ "epoch": 0.896,
2406
+ "grad_norm": 0.408203125,
2407
+ "learning_rate": 5.371900826446282e-06,
2408
+ "loss": 0.0656,
2409
+ "step": 2800
2410
+ },
2411
+ {
2412
+ "epoch": 0.896,
2413
+ "eval_loss": 0.0781230479478836,
2414
+ "eval_runtime": 120.1101,
2415
+ "eval_samples_per_second": 4.163,
2416
+ "eval_steps_per_second": 0.133,
2417
+ "step": 2800
2418
+ },
2419
+ {
2420
+ "epoch": 0.8992,
2421
+ "grad_norm": 1.4609375,
2422
+ "learning_rate": 5.206611570247935e-06,
2423
+ "loss": 0.0634,
2424
+ "step": 2810
2425
+ },
2426
+ {
2427
+ "epoch": 0.9024,
2428
+ "grad_norm": 1.515625,
2429
+ "learning_rate": 5.041322314049587e-06,
2430
+ "loss": 0.093,
2431
+ "step": 2820
2432
+ },
2433
+ {
2434
+ "epoch": 0.9056,
2435
+ "grad_norm": 1.8359375,
2436
+ "learning_rate": 4.87603305785124e-06,
2437
+ "loss": 0.0624,
2438
+ "step": 2830
2439
+ },
2440
+ {
2441
+ "epoch": 0.9088,
2442
+ "grad_norm": 1.0859375,
2443
+ "learning_rate": 4.710743801652893e-06,
2444
+ "loss": 0.0911,
2445
+ "step": 2840
2446
+ },
2447
+ {
2448
+ "epoch": 0.912,
2449
+ "grad_norm": 1.171875,
2450
+ "learning_rate": 4.5454545454545455e-06,
2451
+ "loss": 0.074,
2452
+ "step": 2850
2453
+ },
2454
+ {
2455
+ "epoch": 0.912,
2456
+ "eval_loss": 0.07803711295127869,
2457
+ "eval_runtime": 109.0375,
2458
+ "eval_samples_per_second": 4.586,
2459
+ "eval_steps_per_second": 0.147,
2460
+ "step": 2850
2461
+ },
2462
+ {
2463
+ "epoch": 0.9152,
2464
+ "grad_norm": 0.62890625,
2465
+ "learning_rate": 4.3801652892561984e-06,
2466
+ "loss": 0.0734,
2467
+ "step": 2860
2468
+ },
2469
+ {
2470
+ "epoch": 0.9184,
2471
+ "grad_norm": 0.5703125,
2472
+ "learning_rate": 4.214876033057851e-06,
2473
+ "loss": 0.0521,
2474
+ "step": 2870
2475
+ },
2476
+ {
2477
+ "epoch": 0.9216,
2478
+ "grad_norm": 1.453125,
2479
+ "learning_rate": 4.049586776859504e-06,
2480
+ "loss": 0.0843,
2481
+ "step": 2880
2482
+ },
2483
+ {
2484
+ "epoch": 0.9248,
2485
+ "grad_norm": 0.302734375,
2486
+ "learning_rate": 3.884297520661157e-06,
2487
+ "loss": 0.0775,
2488
+ "step": 2890
2489
+ },
2490
+ {
2491
+ "epoch": 0.928,
2492
+ "grad_norm": 1.1328125,
2493
+ "learning_rate": 3.71900826446281e-06,
2494
+ "loss": 0.0974,
2495
+ "step": 2900
2496
+ },
2497
+ {
2498
+ "epoch": 0.928,
2499
+ "eval_loss": 0.07862695306539536,
2500
+ "eval_runtime": 109.8662,
2501
+ "eval_samples_per_second": 4.551,
2502
+ "eval_steps_per_second": 0.146,
2503
+ "step": 2900
2504
+ },
2505
+ {
2506
+ "epoch": 0.9312,
2507
+ "grad_norm": 1.3515625,
2508
+ "learning_rate": 3.553719008264463e-06,
2509
+ "loss": 0.0611,
2510
+ "step": 2910
2511
+ },
2512
+ {
2513
+ "epoch": 0.9344,
2514
+ "grad_norm": 1.0078125,
2515
+ "learning_rate": 3.3884297520661155e-06,
2516
+ "loss": 0.0697,
2517
+ "step": 2920
2518
+ },
2519
+ {
2520
+ "epoch": 0.9376,
2521
+ "grad_norm": 0.56640625,
2522
+ "learning_rate": 3.2231404958677685e-06,
2523
+ "loss": 0.076,
2524
+ "step": 2930
2525
+ },
2526
+ {
2527
+ "epoch": 0.9408,
2528
+ "grad_norm": 1.5,
2529
+ "learning_rate": 3.0578512396694214e-06,
2530
+ "loss": 0.084,
2531
+ "step": 2940
2532
+ },
2533
+ {
2534
+ "epoch": 0.944,
2535
+ "grad_norm": 1.015625,
2536
+ "learning_rate": 2.8925619834710747e-06,
2537
+ "loss": 0.0769,
2538
+ "step": 2950
2539
+ },
2540
+ {
2541
+ "epoch": 0.944,
2542
+ "eval_loss": 0.07859765738248825,
2543
+ "eval_runtime": 109.8176,
2544
+ "eval_samples_per_second": 4.553,
2545
+ "eval_steps_per_second": 0.146,
2546
+ "step": 2950
2547
+ },
2548
+ {
2549
+ "epoch": 0.9472,
2550
+ "grad_norm": 1.1640625,
2551
+ "learning_rate": 2.7272727272727272e-06,
2552
+ "loss": 0.0722,
2553
+ "step": 2960
2554
+ },
2555
+ {
2556
+ "epoch": 0.9504,
2557
+ "grad_norm": 1.1015625,
2558
+ "learning_rate": 2.56198347107438e-06,
2559
+ "loss": 0.0839,
2560
+ "step": 2970
2561
+ },
2562
+ {
2563
+ "epoch": 0.9536,
2564
+ "grad_norm": 1.171875,
2565
+ "learning_rate": 2.3966942148760335e-06,
2566
+ "loss": 0.0818,
2567
+ "step": 2980
2568
+ },
2569
+ {
2570
+ "epoch": 0.9568,
2571
+ "grad_norm": 1.5,
2572
+ "learning_rate": 2.231404958677686e-06,
2573
+ "loss": 0.0757,
2574
+ "step": 2990
2575
+ },
2576
+ {
2577
+ "epoch": 0.96,
2578
+ "grad_norm": 1.015625,
2579
+ "learning_rate": 2.066115702479339e-06,
2580
+ "loss": 0.0816,
2581
+ "step": 3000
2582
+ },
2583
+ {
2584
+ "epoch": 0.96,
2585
+ "eval_loss": 0.07787304371595383,
2586
+ "eval_runtime": 108.0913,
2587
+ "eval_samples_per_second": 4.626,
2588
+ "eval_steps_per_second": 0.148,
2589
+ "step": 3000
2590
+ }
2591
+ ],
2592
+ "logging_steps": 10,
2593
+ "max_steps": 3125,
2594
+ "num_input_tokens_seen": 0,
2595
+ "num_train_epochs": 1,
2596
+ "save_steps": 500,
2597
+ "stateful_callbacks": {
2598
+ "TrainerControl": {
2599
+ "args": {
2600
+ "should_epoch_stop": false,
2601
+ "should_evaluate": false,
2602
+ "should_log": false,
2603
+ "should_save": true,
2604
+ "should_training_stop": false
2605
+ },
2606
+ "attributes": {}
2607
+ }
2608
+ },
2609
+ "total_flos": 1.75772653691904e+18,
2610
+ "train_batch_size": 32,
2611
+ "trial_name": null,
2612
+ "trial_params": null
2613
+ }
checkpoint-3000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d56c7f1023bbc45d161ffc81a87d8cd8a02d487bf247b61119c50adeda71cc7
3
+ size 5048
git_hash.txt ADDED
@@ -0,0 +1 @@
 
 
1
+ fe7feca3073c9ddb6fde2516490b91bab7031ba3
preprocessor_config.json ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_valid_processor_keys": [
3
+ "images",
4
+ "do_resize",
5
+ "size",
6
+ "resample",
7
+ "do_rescale",
8
+ "rescale_factor",
9
+ "do_normalize",
10
+ "image_mean",
11
+ "image_std",
12
+ "return_tensors",
13
+ "data_format",
14
+ "input_data_format",
15
+ "do_convert_rgb"
16
+ ],
17
+ "do_convert_rgb": null,
18
+ "do_normalize": true,
19
+ "do_rescale": true,
20
+ "do_resize": true,
21
+ "image_mean": [
22
+ 0.5,
23
+ 0.5,
24
+ 0.5
25
+ ],
26
+ "image_processor_type": "SiglipImageProcessor",
27
+ "image_seq_length": 1024,
28
+ "image_std": [
29
+ 0.5,
30
+ 0.5,
31
+ 0.5
32
+ ],
33
+ "processor_class": "PaliGemmaProcessor",
34
+ "resample": 3,
35
+ "rescale_factor": 0.00392156862745098,
36
+ "size": {
37
+ "height": 448,
38
+ "width": 448
39
+ }
40
+ }
results.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"syntheticDocQA_energy": {"ndcg_at_1": 0.89, "ndcg_at_3": 0.93786, "ndcg_at_5": 0.93786, "ndcg_at_10": 0.93786, "ndcg_at_20": 0.94021, "ndcg_at_100": 0.94396, "ndcg_at_1000": 0.94396, "map_at_1": 0.89, "map_at_3": 0.92667, "map_at_5": 0.92667, "map_at_10": 0.92667, "map_at_20": 0.92722, "map_at_100": 0.92774, "map_at_1000": 0.92774, "recall_at_1": 0.89, "recall_at_3": 0.97, "recall_at_5": 0.97, "recall_at_10": 0.97, "recall_at_20": 0.98, "recall_at_100": 1.0, "recall_at_1000": 1.0, "precision_at_1": 0.89, "precision_at_3": 0.32333, "precision_at_5": 0.194, "precision_at_10": 0.097, "precision_at_20": 0.049, "precision_at_100": 0.01, "precision_at_1000": 0.001, "mrr_at_1": 0.9, "mrr_at_3": 0.9316666666666668, "mrr_at_5": 0.9316666666666668, "mrr_at_10": 0.9316666666666668, "mrr_at_20": 0.9322549019607842, "mrr_at_100": 0.9328175800234624, "mrr_at_1000": 0.9328175800234624, "naucs_at_1_max": 0.1416115453998798, "naucs_at_1_std": -0.3451163989347998, "naucs_at_1_diff1": 0.921398505283052, "naucs_at_3_max": 0.5659819483348902, "naucs_at_3_std": -0.4197012138188588, "naucs_at_3_diff1": 0.9564270152505466, "naucs_at_5_max": 0.5659819483348872, "naucs_at_5_std": -0.41970121381885955, "naucs_at_5_diff1": 0.9564270152505424, "naucs_at_10_max": 0.5659819483348872, "naucs_at_10_std": -0.41970121381885955, "naucs_at_10_diff1": 0.9564270152505424, "naucs_at_20_max": 0.9346405228758136, "naucs_at_20_std": 0.24042950513538955, "naucs_at_20_diff1": 0.9346405228758136, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "syntheticDocQA_healthcare_industry": {"ndcg_at_1": 0.88, "ndcg_at_3": 0.92286, "ndcg_at_5": 0.93147, "ndcg_at_10": 0.9348, "ndcg_at_20": 0.9372, "ndcg_at_100": 0.93938, "ndcg_at_1000": 0.93938, "map_at_1": 0.88, "map_at_3": 0.91333, "map_at_5": 0.91833, "map_at_10": 0.91976, "map_at_20": 0.92035, "map_at_100": 0.92078, "map_at_1000": 0.92078, "recall_at_1": 0.88, "recall_at_3": 0.95, "recall_at_5": 0.97, "recall_at_10": 0.98, "recall_at_20": 0.99, "recall_at_100": 1.0, "recall_at_1000": 1.0, "precision_at_1": 0.88, "precision_at_3": 0.31667, "precision_at_5": 0.194, "precision_at_10": 0.098, "precision_at_20": 0.0495, "precision_at_100": 0.01, "precision_at_1000": 0.001, "mrr_at_1": 0.88, "mrr_at_3": 0.9133333333333334, "mrr_at_5": 0.9183333333333334, "mrr_at_10": 0.92, "mrr_at_20": 0.9205882352941177, "mrr_at_100": 0.9210427807486632, "mrr_at_1000": 0.9210427807486632, "naucs_at_1_max": 0.41314031180400845, "naucs_at_1_std": 0.09091632198536481, "naucs_at_1_diff1": 0.8812838052815782, "naucs_at_3_max": 0.23874883286648252, "naucs_at_3_std": -0.37497665732959407, "naucs_at_3_diff1": 0.8627450980392143, "naucs_at_5_max": 0.03781512605041818, "naucs_at_5_std": -0.7268907563025172, "naucs_at_5_diff1": 0.8638344226579531, "naucs_at_10_max": -0.004668534080301362, "naucs_at_10_std": -0.5144724556489195, "naucs_at_10_diff1": 0.7957516339869297, "naucs_at_20_max": -0.5634920634920767, "naucs_at_20_std": -1.1517273576097316, "naucs_at_20_diff1": 0.7222222222222276, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "syntheticDocQA_artificial_intelligence_test": {"ndcg_at_1": 0.91, "ndcg_at_3": 0.95286, "ndcg_at_5": 0.95716, "ndcg_at_10": 0.96032, "ndcg_at_20": 0.96032, "ndcg_at_100": 0.96032, "ndcg_at_1000": 0.96032, "map_at_1": 0.91, "map_at_3": 0.94333, "map_at_5": 0.94583, "map_at_10": 0.94708, "map_at_20": 0.94708, "map_at_100": 0.94708, "map_at_1000": 0.94708, "recall_at_1": 0.91, "recall_at_3": 0.98, "recall_at_5": 0.99, "recall_at_10": 1.0, "recall_at_20": 1.0, "recall_at_100": 1.0, "recall_at_1000": 1.0, "precision_at_1": 0.91, "precision_at_3": 0.32667, "precision_at_5": 0.198, "precision_at_10": 0.1, "precision_at_20": 0.05, "precision_at_100": 0.01, "precision_at_1000": 0.001, "mrr_at_1": 0.92, "mrr_at_3": 0.9483333333333333, "mrr_at_5": 0.9508333333333333, "mrr_at_10": 0.9520833333333333, "mrr_at_20": 0.9520833333333333, "mrr_at_100": 0.9520833333333333, "mrr_at_1000": 0.9520833333333333, "naucs_at_1_max": 0.24592800082996116, "naucs_at_1_std": -0.49019607843137364, "naucs_at_1_diff1": 0.9273783587509066, "naucs_at_3_max": 0.7222222222222141, "naucs_at_3_std": -0.5144724556489392, "naucs_at_3_diff1": 0.9346405228758099, "naucs_at_5_max": 0.7222222222222276, "naucs_at_5_std": -1.1517273576097316, "naucs_at_5_diff1": 0.8692810457516413, "naucs_at_10_max": 1.0, "naucs_at_10_std": 1.0, "naucs_at_10_diff1": 1.0, "naucs_at_20_max": 1.0, "naucs_at_20_std": 1.0, "naucs_at_20_diff1": 1.0, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "syntheticDocQA_government_reports": {"ndcg_at_1": 0.83, "ndcg_at_3": 0.92464, "ndcg_at_5": 0.92464, "ndcg_at_10": 0.92765, "ndcg_at_20": 0.93028, "ndcg_at_100": 0.93028, "ndcg_at_1000": 0.93028, "map_at_1": 0.83, "map_at_3": 0.905, "map_at_5": 0.905, "map_at_10": 0.90611, "map_at_20": 0.90688, "map_at_100": 0.90688, "map_at_1000": 0.90688, "recall_at_1": 0.83, "recall_at_3": 0.98, "recall_at_5": 0.98, "recall_at_10": 0.99, "recall_at_20": 1.0, "recall_at_100": 1.0, "recall_at_1000": 1.0, "precision_at_1": 0.83, "precision_at_3": 0.32667, "precision_at_5": 0.196, "precision_at_10": 0.099, "precision_at_20": 0.05, "precision_at_100": 0.01, "precision_at_1000": 0.001, "mrr_at_1": 0.86, "mrr_at_3": 0.92, "mrr_at_5": 0.92, "mrr_at_10": 0.9211111111111111, "mrr_at_20": 0.9218803418803418, "mrr_at_100": 0.9218803418803418, "mrr_at_1000": 0.9218803418803418, "naucs_at_1_max": 0.35191850002902464, "naucs_at_1_std": -0.0016543797527126534, "naucs_at_1_diff1": 0.9017820862599414, "naucs_at_3_max": 1.0, "naucs_at_3_std": 0.8611111111111119, "naucs_at_3_diff1": 0.8611111111111119, "naucs_at_5_max": 1.0, "naucs_at_5_std": 0.861111111111116, "naucs_at_5_diff1": 0.861111111111116, "naucs_at_10_max": 1.0, "naucs_at_10_std": 0.7222222222222276, "naucs_at_10_diff1": 0.7222222222222276, "naucs_at_20_max": 1.0, "naucs_at_20_std": 1.0, "naucs_at_20_diff1": 1.0, "naucs_at_100_max": NaN, "naucs_at_100_std": NaN, "naucs_at_100_diff1": NaN, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "infovqa_subsampled": {"ndcg_at_1": 0.716, "ndcg_at_3": 0.78674, "ndcg_at_5": 0.7999, "ndcg_at_10": 0.81139, "ndcg_at_20": 0.81628, "ndcg_at_100": 0.82424, "ndcg_at_1000": 0.82843, "map_at_1": 0.716, "map_at_3": 0.76967, "map_at_5": 0.77697, "map_at_10": 0.78161, "map_at_20": 0.78286, "map_at_100": 0.78402, "map_at_1000": 0.78415, "recall_at_1": 0.716, "recall_at_3": 0.836, "recall_at_5": 0.868, "recall_at_10": 0.904, "recall_at_20": 0.924, "recall_at_100": 0.966, "recall_at_1000": 1.0, "precision_at_1": 0.716, "precision_at_3": 0.27867, "precision_at_5": 0.1736, "precision_at_10": 0.0904, "precision_at_20": 0.0462, "precision_at_100": 0.00966, "precision_at_1000": 0.001, "mrr_at_1": 0.716, "mrr_at_3": 0.7706666666666663, "mrr_at_5": 0.7774666666666662, "mrr_at_10": 0.7822595238095235, "mrr_at_20": 0.7834290391754783, "mrr_at_100": 0.7846837692508419, "mrr_at_1000": 0.7848128972573761, "naucs_at_1_max": 0.5144395755239958, "naucs_at_1_std": 0.056706162745971116, "naucs_at_1_diff1": 0.8757608211966486, "naucs_at_3_max": 0.6206474526649232, "naucs_at_3_std": 0.2310119747546263, "naucs_at_3_diff1": 0.8256041371697351, "naucs_at_5_max": 0.6981123805627789, "naucs_at_5_std": 0.3026984776319997, "naucs_at_5_diff1": 0.8120489144727236, "naucs_at_10_max": 0.7312188764394637, "naucs_at_10_std": 0.39961679116090754, "naucs_at_10_diff1": 0.7691118113912242, "naucs_at_20_max": 0.7535259717922257, "naucs_at_20_std": 0.4052901862499385, "naucs_at_20_diff1": 0.758477075040541, "naucs_at_100_max": 0.8885044213763893, "naucs_at_100_std": 0.6311858076563935, "naucs_at_100_diff1": 0.7411984401603742, "naucs_at_1000_max": 1.0, "naucs_at_1000_std": 1.0, "naucs_at_1000_diff1": 1.0}, "docvqa_subsampled": {"ndcg_at_1": 0.394, "ndcg_at_3": 0.46421, "ndcg_at_5": 0.48048, "ndcg_at_10": 0.49561, "ndcg_at_20": 0.50938, "ndcg_at_100": 0.53597, "ndcg_at_1000": 0.56127, "map_at_1": 0.394, "map_at_3": 0.447, "map_at_5": 0.4559, "map_at_10": 0.46192, "map_at_20": 0.46576, "map_at_100": 0.46928, "map_at_1000": 0.47021, "recall_at_1": 0.394, "recall_at_3": 0.514, "recall_at_5": 0.554, "recall_at_10": 0.602, "recall_at_20": 0.656, "recall_at_100": 0.802, "recall_at_1000": 1.0, "precision_at_1": 0.394, "precision_at_3": 0.17133, "precision_at_5": 0.1108, "precision_at_10": 0.0602, "precision_at_20": 0.0328, "precision_at_100": 0.00802, "precision_at_1000": 0.001, "mrr_at_1": 0.394, "mrr_at_3": 0.44466666666666693, "mrr_at_5": 0.45446666666666674, "mrr_at_10": 0.4606452380952381, "mrr_at_20": 0.46431832950331403, "mrr_at_100": 0.46793416222345513, "mrr_at_1000": 0.4688637410389999, "naucs_at_1_max": 0.59049210041854, "naucs_at_1_std": 0.05796555116331319, "naucs_at_1_diff1": 0.8098445903552552, "naucs_at_3_max": 0.5727328519172382, "naucs_at_3_std": 0.06011122496154157, "naucs_at_3_diff1": 0.6923280730829647, "naucs_at_5_max": 0.5486981759514414, "naucs_at_5_std": 0.061820776722579115, "naucs_at_5_diff1": 0.6590126388767453, "naucs_at_10_max": 0.5135874988457401, "naucs_at_10_std": 0.08675863013943434, "naucs_at_10_diff1": 0.6397577692850738, "naucs_at_20_max": 0.512128536803836, "naucs_at_20_std": 0.14237775962906352, "naucs_at_20_diff1": 0.605205095105116, "naucs_at_100_max": 0.48413818834514916, "naucs_at_100_std": 0.37696760076233576, "naucs_at_100_diff1": 0.5448073490707792, "naucs_at_1000_max": 1.0, "naucs_at_1000_std": 1.0, "naucs_at_1000_diff1": 1.0}, "arxivqa_subsampled": {"ndcg_at_1": 0.64, "ndcg_at_3": 0.69686, "ndcg_at_5": 0.71312, "ndcg_at_10": 0.72568, "ndcg_at_20": 0.73619, "ndcg_at_100": 0.7531, "ndcg_at_1000": 0.7598, "map_at_1": 0.64, "map_at_3": 0.68267, "map_at_5": 0.69157, "map_at_10": 0.69692, "map_at_20": 0.69975, "map_at_100": 0.70204, "map_at_1000": 0.70234, "recall_at_1": 0.64, "recall_at_3": 0.738, "recall_at_5": 0.778, "recall_at_10": 0.816, "recall_at_20": 0.858, "recall_at_100": 0.95, "recall_at_1000": 1.0, "precision_at_1": 0.64, "precision_at_3": 0.246, "precision_at_5": 0.1556, "precision_at_10": 0.0816, "precision_at_20": 0.0429, "precision_at_100": 0.0095, "precision_at_1000": 0.001, "mrr_at_1": 0.646, "mrr_at_3": 0.6843333333333333, "mrr_at_5": 0.6939333333333332, "mrr_at_10": 0.6995880952380951, "mrr_at_20": 0.7024205660696371, "mrr_at_100": 0.7046153644586061, "mrr_at_1000": 0.7049344272544787, "naucs_at_1_max": 0.6400498147911611, "naucs_at_1_std": 0.30982245497509275, "naucs_at_1_diff1": 0.8982085834717077, "naucs_at_3_max": 0.6633502178056638, "naucs_at_3_std": 0.3770435651623774, "naucs_at_3_diff1": 0.8408064249648411, "naucs_at_5_max": 0.6786025034872963, "naucs_at_5_std": 0.3739256135569494, "naucs_at_5_diff1": 0.8232090490155005, "naucs_at_10_max": 0.7271186806544918, "naucs_at_10_std": 0.45356819064887993, "naucs_at_10_diff1": 0.8012077451107367, "naucs_at_20_max": 0.761528570060601, "naucs_at_20_std": 0.5156837799756484, "naucs_at_20_diff1": 0.7929713683807331, "naucs_at_100_max": 0.7033986928104585, "naucs_at_100_std": 0.42941176470588494, "naucs_at_100_diff1": 0.7332773109243655, "naucs_at_1000_max": 1.0, "naucs_at_1000_std": 1.0, "naucs_at_1000_diff1": 1.0}, "tabfquad_subsampled": {"ndcg_at_1": 0.77857, "ndcg_at_3": 0.8321, "ndcg_at_5": 0.83932, "ndcg_at_10": 0.85306, "ndcg_at_20": 0.86309, "ndcg_at_100": 0.8696, "ndcg_at_1000": 0.8696, "map_at_1": 0.77857, "map_at_3": 0.81964, "map_at_5": 0.82357, "map_at_10": 0.82918, "map_at_20": 0.83197, "map_at_100": 0.83306, "map_at_1000": 0.83306, "recall_at_1": 0.77857, "recall_at_3": 0.86786, "recall_at_5": 0.88571, "recall_at_10": 0.92857, "recall_at_20": 0.96786, "recall_at_100": 1.0, "recall_at_1000": 1.0, "precision_at_1": 0.77857, "precision_at_3": 0.28929, "precision_at_5": 0.17714, "precision_at_10": 0.09286, "precision_at_20": 0.04839, "precision_at_100": 0.01, "precision_at_1000": 0.001, "mrr_at_1": 0.775, "mrr_at_3": 0.8172619047619047, "mrr_at_5": 0.8213690476190477, "mrr_at_10": 0.8272264739229026, "mrr_at_20": 0.8296797314496894, "mrr_at_100": 0.8307939591209218, "mrr_at_1000": 0.8307939591209218, "naucs_at_1_max": 0.3645402573438554, "naucs_at_1_std": 0.19164868115392844, "naucs_at_1_diff1": 0.7632584806497849, "naucs_at_3_max": 0.37951885735136837, "naucs_at_3_std": 0.2538639977063613, "naucs_at_3_diff1": 0.7619308259702339, "naucs_at_5_max": 0.3969654722354073, "naucs_at_5_std": 0.30882178452776354, "naucs_at_5_diff1": 0.7626364499288074, "naucs_at_10_max": 0.40688608776844226, "naucs_at_10_std": 0.3362978524743217, "naucs_at_10_diff1": 0.7623949579831951, "naucs_at_20_max": 0.47883597883597717, "naucs_at_20_std": 0.5260400456478881, "naucs_at_20_diff1": 0.9255628177196795, "naucs_at_100_max": 1.0, "naucs_at_100_std": 1.0, "naucs_at_100_diff1": 1.0, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "tatdqa": {"ndcg_at_1": 0.46723, "ndcg_at_3": 0.56002, "ndcg_at_5": 0.59103, "ndcg_at_10": 0.62058, "ndcg_at_20": 0.63842, "ndcg_at_100": 0.65687, "ndcg_at_1000": 0.66202, "map_at_1": 0.46723, "map_at_3": 0.53668, "map_at_5": 0.55406, "map_at_10": 0.56648, "map_at_20": 0.57141, "map_at_100": 0.57398, "map_at_1000": 0.57422, "recall_at_1": 0.46723, "recall_at_3": 0.62778, "recall_at_5": 0.70235, "recall_at_10": 0.79254, "recall_at_20": 0.8629, "recall_at_100": 0.96212, "recall_at_1000": 1.0, "precision_at_1": 0.46723, "precision_at_3": 0.20926, "precision_at_5": 0.14047, "precision_at_10": 0.07925, "precision_at_20": 0.04314, "precision_at_100": 0.00962, "precision_at_1000": 0.001, "mrr_at_1": 0.4660252555622369, "mrr_at_3": 0.5375826819001811, "mrr_at_5": 0.5542092603728216, "mrr_at_10": 0.5662937032900962, "mrr_at_20": 0.571217019105013, "mrr_at_100": 0.5738015340316196, "mrr_at_1000": 0.5740512063675741, "naucs_at_1_max": 0.12316388049685562, "naucs_at_1_std": -0.03302911730144414, "naucs_at_1_diff1": 0.6367700909505911, "naucs_at_3_max": 0.1491828892683818, "naucs_at_3_std": -0.01776625584052866, "naucs_at_3_diff1": 0.5441671908451038, "naucs_at_5_max": 0.13509434719203903, "naucs_at_5_std": -0.008750994497691225, "naucs_at_5_diff1": 0.4852071271216003, "naucs_at_10_max": 0.16644379506149076, "naucs_at_10_std": 0.05915036291157827, "naucs_at_10_diff1": 0.46415728517159416, "naucs_at_20_max": 0.21050100386062404, "naucs_at_20_std": 0.12189428322749622, "naucs_at_20_diff1": 0.39296395537492884, "naucs_at_100_max": 0.34701914032223563, "naucs_at_100_std": 0.3284671046283788, "naucs_at_100_diff1": 0.3933540743785471, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}, "shift_project": {"ndcg_at_1": 0.59, "ndcg_at_3": 0.70964, "ndcg_at_5": 0.73847, "ndcg_at_10": 0.7544, "ndcg_at_20": 0.76229, "ndcg_at_100": 0.77165, "ndcg_at_1000": 0.77302, "map_at_1": 0.59, "map_at_3": 0.68167, "map_at_5": 0.69767, "map_at_10": 0.7041, "map_at_20": 0.70641, "map_at_100": 0.70774, "map_at_1000": 0.7078, "recall_at_1": 0.59, "recall_at_3": 0.79, "recall_at_5": 0.86, "recall_at_10": 0.91, "recall_at_20": 0.94, "recall_at_100": 0.99, "recall_at_1000": 1.0, "precision_at_1": 0.59, "precision_at_3": 0.26333, "precision_at_5": 0.172, "precision_at_10": 0.091, "precision_at_20": 0.047, "precision_at_100": 0.0099, "precision_at_1000": 0.001, "mrr_at_1": 0.59, "mrr_at_3": 0.6866666666666668, "mrr_at_5": 0.7001666666666668, "mrr_at_10": 0.7071507936507938, "mrr_at_20": 0.7095317460317462, "mrr_at_100": 0.7108961223961225, "mrr_at_1000": 0.7109623475616854, "naucs_at_1_max": -0.09331298516442392, "naucs_at_1_std": -0.20234199597928013, "naucs_at_1_diff1": 0.6018313352018425, "naucs_at_3_max": -0.061100824776973234, "naucs_at_3_std": -0.285545963882944, "naucs_at_3_diff1": 0.6008367999615265, "naucs_at_5_max": -0.25419149230982163, "naucs_at_5_std": -0.42237079118747267, "naucs_at_5_diff1": 0.5672717195510587, "naucs_at_10_max": -0.05612615416536863, "naucs_at_10_std": -0.4241622574955902, "naucs_at_10_diff1": 0.7744060587197855, "naucs_at_20_max": -0.05757858699035113, "naucs_at_20_std": -0.027699968876440207, "naucs_at_20_diff1": 0.8295985060690946, "naucs_at_100_max": -1.1517273576096694, "naucs_at_100_std": -1.7399626517273008, "naucs_at_100_diff1": 0.554154995331464, "naucs_at_1000_max": NaN, "naucs_at_1000_std": NaN, "naucs_at_1000_diff1": NaN}}
special_tokens_map.json ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ {
4
+ "content": "<image>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false
9
+ }
10
+ ],
11
+ "bos_token": {
12
+ "content": "<bos>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false
17
+ },
18
+ "eos_token": {
19
+ "content": "<eos>",
20
+ "lstrip": false,
21
+ "normalized": false,
22
+ "rstrip": false,
23
+ "single_word": false
24
+ },
25
+ "pad_token": {
26
+ "content": "<pad>",
27
+ "lstrip": false,
28
+ "normalized": false,
29
+ "rstrip": false,
30
+ "single_word": false
31
+ },
32
+ "unk_token": {
33
+ "content": "<unk>",
34
+ "lstrip": false,
35
+ "normalized": false,
36
+ "rstrip": false,
37
+ "single_word": false
38
+ }
39
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ffd310e50986db7a039948ab83441d612689e7f989198e31b5c8984ca458adf6
3
+ size 17763459
tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff
 
training_config.yml ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ config:
2
+ (): colpali_engine.utils.train_colpali_engine_models.ColModelTrainingConfig
3
+ output_dir: !path ../../../models/train_colpali_docmatix_hardneg_ib_3b-mix-448
4
+ processor:
5
+ () : colpali_engine.utils.wrapper.AutoProcessorWrapper
6
+ pretrained_model_name_or_path: "./models/paligemma-3b-mix-448"
7
+ max_length: 50
8
+ model:
9
+ (): colpali_engine.utils.wrapper.AutoColModelWrapper
10
+ pretrained_model_name_or_path: "./models/paligemma-3b-mix-448"
11
+ training_objective: "colbertv1"
12
+ # attn_implementation: "eager"
13
+ torch_dtype: !ext torch.bfloat16
14
+ # device_map: "auto"
15
+ # quantization_config:
16
+ # (): transformers.BitsAndBytesConfig
17
+ # load_in_4bit: true
18
+ # bnb_4bit_quant_type: "nf4"
19
+ # bnb_4bit_compute_dtype: "bfloat16"
20
+ # bnb_4bit_use_double_quant: true
21
+
22
+ dataset_loading_func: !ext colpali_engine.utils.dataset_transformation.load_docmatix_ir_negs
23
+ eval_dataset_loader: !import ../data/test_data.yaml
24
+
25
+ max_length: 50
26
+ run_eval: true
27
+ add_suffix: true
28
+ loss_func:
29
+ (): colpali_engine.loss.colbert_loss.ColbertPairwiseNegativeCELoss
30
+ in_batch_term: true
31
+ tr_args: !import ../tr_args/default_neg_tr_args.yaml
32
+ peft_config:
33
+ (): peft.LoraConfig
34
+ r: 32
35
+ lora_alpha: 32
36
+ lora_dropout: 0.1
37
+ init_lora_weights: "gaussian"
38
+ bias: "none"
39
+ task_type: "FEATURE_EXTRACTION"
40
+ target_modules: '(.*(language_model).*(down_proj|gate_proj|up_proj|k_proj|q_proj|v_proj|o_proj).*$|.*(custom_text_proj).*$)'
41
+ # target_modules: '(.*(language_model).*(down_proj|gate_proj|up_proj|k_proj|q_proj|v_proj|o_proj).*$|.*(custom_text_proj).*$)'
42
+