Hugging Face's logo

jdorairaj
/
LoRA-large-tasks

Safetensors
Model card Files
xet
 Community
jdorairaj commited on
Commit
c91ff8d
·
1 Parent(s): 4ad5c11

using lfs

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +1 -0
  2. outputs/qqp/args.json +3 -41
  3. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/args.json +3 -32
  4. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/adapter_config.json +3 -32
  5. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/all_results.json +3 -1
  6. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/all_results_la_kron_all_homo_mc_corr_1000.json +3 -1
  7. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/all_results_val.json +3 -1
  8. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/eval_res.json +0 -0
  9. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/eval_res_la_kron_all_homo_mc_corr_1000.json +0 -0
  10. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/gpu_stats.json +3 -130
  11. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/gpu_stats_la.json +3 -130
  12. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/special_tokens_map.json +3 -7
  13. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/tokenizer.json +0 -0
  14. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/tokenizer_config.json +3 -55
  15. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/adapter_config.json +3 -32
  16. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/all_results.json +3 -1
  17. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/all_results_la_kron_all_homo_mc_corr_1000.json +3 -1
  18. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/all_results_val.json +3 -1
  19. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/eval_res.json +0 -0
  20. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/eval_res_la_kron_all_homo_mc_corr_1000.json +0 -0
  21. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/gpu_stats.json +3 -130
  22. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/gpu_stats_la.json +3 -130
  23. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/special_tokens_map.json +3 -7
  24. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/tokenizer.json +0 -0
  25. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/tokenizer_config.json +3 -55
  26. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/adapter_config.json +3 -32
  27. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/all_results.json +3 -1
  28. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/all_results_la_kron_all_homo_mc_corr_1000.json +3 -1
  29. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/all_results_val.json +3 -1
  30. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/eval_res.json +0 -0
  31. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/eval_res_la_kron_all_homo_mc_corr_1000.json +0 -0
  32. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/gpu_stats.json +3 -130
  33. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/gpu_stats_la.json +3 -130
  34. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/special_tokens_map.json +3 -7
  35. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/tokenizer.json +0 -0
  36. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/tokenizer_config.json +3 -55
  37. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/adapter_config.json +3 -32
  38. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/all_results.json +3 -1
  39. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/all_results_la_kron_all_homo_mc_corr_1000.json +3 -1
  40. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/all_results_val.json +3 -1
  41. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/eval_res.json +0 -0
  42. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/eval_res_la_kron_all_homo_mc_corr_1000.json +0 -0
  43. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/gpu_stats.json +3 -130
  44. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/gpu_stats_la.json +3 -130
  45. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/special_tokens_map.json +3 -7
  46. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/tokenizer.json +0 -0
  47. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/tokenizer_config.json +3 -55
  48. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_27290/adapter_config.json +3 -32
  49. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_27290/all_results.json +3 -1
  50. outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_27290/all_results_la_kron_all_homo_mc_corr_1000.json +3 -1
.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
+ outputs/**/*.json filter=lfs diff=lfs merge=lfs -text
outputs/qqp/args.json CHANGED
@@ -1,41 +1,3 @@
1
- {
2
- "task_name": "qqp",
3
- "train_file": null,
4
- "validation_file": null,
5
- "max_length": 300,
6
- "pad_to_max_length": false,
7
- "model_name_or_path": "bert-base-uncased",
8
- "use_slow_tokenizer": false,
9
- "per_device_train_batch_size": 32,
10
- "per_device_eval_batch_size": 8,
11
- "learning_rate": 0.0001,
12
- "weight_decay": 0.0,
13
- "num_train_epochs": 3,
14
- "max_train_steps": null,
15
- "peft_method": null,
16
- "gradient_accumulation_steps": 1,
17
- "lr_scheduler_type": "linear",
18
- "num_warmup_steps": 0,
19
- "output_dir": "./outputs",
20
- "seed": 65,
21
- "push_to_hub": false,
22
- "hub_model_id": null,
23
- "hub_token": null,
24
- "checkpointing_steps": "1000",
25
- "resume_from_checkpoint": null,
26
- "with_tracking": false,
27
- "report_to": "all",
28
- "ignore_mismatched_sizes": true,
29
- "save": false,
30
- "load_step": 999,
31
- "lora_r": 8,
32
- "lora_alpha": 16,
33
- "lora_dropout": 0.1,
34
- "laplace_hessian": "kron",
35
- "laplace_sub": "all",
36
- "laplace_prior": "homo",
37
- "laplace_optim_step": 1000,
38
- "testing_set": "train_val",
39
- "cache_dir": "/content/cache/huggingface/metrics/",
40
- "laplace_predict": "mc_corr"
41
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:36f4c55f7083c703dc8ed9be8cba0336bd837b325dbcaff4f3e323dcf587d140
3
+ size 1152
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/args.json CHANGED
@@ -1,32 +1,3 @@
1
- {
2
- "task_name": "qqp",
3
- "train_file": null,
4
- "validation_file": null,
5
- "max_length": 300,
6
- "pad_to_max_length": false,
7
- "model_name_or_path": "bert-base-uncased",
8
- "use_slow_tokenizer": false,
9
- "per_device_train_batch_size": 32,
10
- "per_device_eval_batch_size": 8,
11
- "learning_rate": 0.0001,
12
- "weight_decay": 0.0,
13
- "num_train_epochs": 3,
14
- "max_train_steps": null,
15
- "gradient_accumulation_steps": 1,
16
- "lr_scheduler_type": "linear",
17
- "num_warmup_steps": 0,
18
- "output_dir": "./outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65",
19
- "seed": 65,
20
- "push_to_hub": false,
21
- "hub_model_id": null,
22
- "hub_token": null,
23
- "checkpointing_steps": null,
24
- "resume_from_checkpoint": null,
25
- "with_tracking": false,
26
- "report_to": "all",
27
- "ignore_mismatched_sizes": true,
28
- "lora_r": 8,
29
- "lora_alpha": 16,
30
- "lora_dropout": 0.1,
31
- "testing_set": "train_val"
32
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eabb031b73f0d5eda72a08456a5ec4273c24d831df84a3447eb559f0d8d2df5d
3
+ size 931
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/adapter_config.json CHANGED
@@ -1,32 +1,3 @@
1
- {
2
- "alpha_pattern": {},
3
- "auto_mapping": null,
4
- "base_model_name_or_path": "bert-base-uncased",
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": 16,
14
- "lora_dropout": 0.1,
15
- "megatron_config": null,
16
- "megatron_core": "megatron.core",
17
- "modules_to_save": [
18
- "classifier",
19
- "score"
20
- ],
21
- "peft_type": "LORA",
22
- "r": 8,
23
- "rank_pattern": {},
24
- "revision": null,
25
- "target_modules": [
26
- "value",
27
- "query"
28
- ],
29
- "task_type": "SEQ_CLS",
30
- "use_dora": false,
31
- "use_rslora": false
32
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81a29a631d083a248a7fe8b5acc0d972f1e62158c7e9d95bdbbeffc2630802a8
3
+ size 668
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/all_results.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.3832797427652733, "eval_f1": 0.5440681685196022}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b4b1837d6e394da33a3778991c82b29ef9bfe1ee112aaa879f591170f16ea7c
3
+ size 68
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/all_results_la_kron_all_homo_mc_corr_1000.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.3832797427652733, "eval_f1": 0.5440681685196022}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b4b1837d6e394da33a3778991c82b29ef9bfe1ee112aaa879f591170f16ea7c
3
+ size 68
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/all_results_val.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.38418304246255325, "eval_f1": 0.5459261736125888}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b267063a5554011a15c9dc7f6a42f59a6b94377288e3a9168f80dfa3105930d
3
+ size 69
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/eval_res.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/eval_res_la_kron_all_homo_mc_corr_1000.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/gpu_stats.json CHANGED
@@ -1,130 +1,3 @@
1
- {
2
- "memory_allocated": 448852992,
3
- "max_memory_allocated": 532900352,
4
- "memory_reserved": 648019968,
5
- "max_memory_reserved": 648019968,
6
- "memory_stats": {
7
- "active.all.allocated": 5880425,
8
- "active.all.current": 266,
9
- "active.all.freed": 5880159,
10
- "active.all.peak": 280,
11
- "active.large_pool.allocated": 2290614,
12
- "active.large_pool.current": 76,
13
- "active.large_pool.freed": 2290538,
14
- "active.large_pool.peak": 84,
15
- "active.small_pool.allocated": 3589811,
16
- "active.small_pool.current": 190,
17
- "active.small_pool.freed": 3589621,
18
- "active.small_pool.peak": 203,
19
- "active_bytes.all.allocated": 7740209286144,
20
- "active_bytes.all.current": 448852992,
21
- "active_bytes.all.freed": 7739760433152,
22
- "active_bytes.all.peak": 532900352,
23
- "active_bytes.large_pool.allocated": 5868623697920,
24
- "active_bytes.large_pool.current": 447086592,
25
- "active_bytes.large_pool.freed": 5868176611328,
26
- "active_bytes.large_pool.peak": 531067392,
27
- "active_bytes.small_pool.allocated": 1871585588224,
28
- "active_bytes.small_pool.current": 1766400,
29
- "active_bytes.small_pool.freed": 1871583821824,
30
- "active_bytes.small_pool.peak": 9081344,
31
- "allocated_bytes.all.allocated": 7740209286144,
32
- "allocated_bytes.all.current": 448852992,
33
- "allocated_bytes.all.freed": 7739760433152,
34
- "allocated_bytes.all.peak": 532900352,
35
- "allocated_bytes.large_pool.allocated": 5868623697920,
36
- "allocated_bytes.large_pool.current": 447086592,
37
- "allocated_bytes.large_pool.freed": 5868176611328,
38
- "allocated_bytes.large_pool.peak": 531067392,
39
- "allocated_bytes.small_pool.allocated": 1871585588224,
40
- "allocated_bytes.small_pool.current": 1766400,
41
- "allocated_bytes.small_pool.freed": 1871583821824,
42
- "allocated_bytes.small_pool.peak": 9081344,
43
- "allocation.all.allocated": 5880425,
44
- "allocation.all.current": 266,
45
- "allocation.all.freed": 5880159,
46
- "allocation.all.peak": 280,
47
- "allocation.large_pool.allocated": 2290614,
48
- "allocation.large_pool.current": 76,
49
- "allocation.large_pool.freed": 2290538,
50
- "allocation.large_pool.peak": 84,
51
- "allocation.small_pool.allocated": 3589811,
52
- "allocation.small_pool.current": 190,
53
- "allocation.small_pool.freed": 3589621,
54
- "allocation.small_pool.peak": 203,
55
- "inactive_split.all.allocated": 3773981,
56
- "inactive_split.all.current": 22,
57
- "inactive_split.all.freed": 3773959,
58
- "inactive_split.all.peak": 33,
59
- "inactive_split.large_pool.allocated": 2090315,
60
- "inactive_split.large_pool.current": 18,
61
- "inactive_split.large_pool.freed": 2090297,
62
- "inactive_split.large_pool.peak": 24,
63
- "inactive_split.small_pool.allocated": 1683666,
64
- "inactive_split.small_pool.current": 4,
65
- "inactive_split.small_pool.freed": 1683662,
66
- "inactive_split.small_pool.peak": 15,
67
- "inactive_split_bytes.all.allocated": 7808626842112,
68
- "inactive_split_bytes.all.current": 46074880,
69
- "inactive_split_bytes.all.freed": 7808580767232,
70
- "inactive_split_bytes.all.peak": 90030080,
71
- "inactive_split_bytes.large_pool.allocated": 5881975429120,
72
- "inactive_split_bytes.large_pool.current": 45744128,
73
- "inactive_split_bytes.large_pool.freed": 5881929684992,
74
- "inactive_split_bytes.large_pool.peak": 89755136,
75
- "inactive_split_bytes.small_pool.allocated": 1926651412992,
76
- "inactive_split_bytes.small_pool.current": 330752,
77
- "inactive_split_bytes.small_pool.freed": 1926651082240,
78
- "inactive_split_bytes.small_pool.peak": 6580736,
79
- "max_split_size": -1,
80
- "num_alloc_retries": 0,
81
- "num_device_alloc": 32,
82
- "num_device_free": 0,
83
- "num_ooms": 0,
84
- "num_sync_all_streams": 0,
85
- "oversize_allocations.allocated": 0,
86
- "oversize_allocations.current": 0,
87
- "oversize_allocations.freed": 0,
88
- "oversize_allocations.peak": 0,
89
- "oversize_segments.allocated": 0,
90
- "oversize_segments.current": 0,
91
- "oversize_segments.freed": 0,
92
- "oversize_segments.peak": 0,
93
- "requested_bytes.all.allocated": 6383968346031,
94
- "requested_bytes.all.current": 447715384,
95
- "requested_bytes.all.freed": 6383520630647,
96
- "requested_bytes.all.peak": 531762008,
97
- "requested_bytes.large_pool.allocated": 4512820212544,
98
- "requested_bytes.large_pool.current": 445954048,
99
- "requested_bytes.large_pool.freed": 4512374258496,
100
- "requested_bytes.large_pool.peak": 529934848,
101
- "requested_bytes.small_pool.allocated": 1871148133487,
102
- "requested_bytes.small_pool.current": 1761336,
103
- "requested_bytes.small_pool.freed": 1871146372151,
104
- "requested_bytes.small_pool.peak": 9073624,
105
- "reserved_bytes.all.allocated": 648019968,
106
- "reserved_bytes.all.current": 648019968,
107
- "reserved_bytes.all.freed": 0,
108
- "reserved_bytes.all.peak": 648019968,
109
- "reserved_bytes.large_pool.allocated": 637534208,
110
- "reserved_bytes.large_pool.current": 637534208,
111
- "reserved_bytes.large_pool.freed": 0,
112
- "reserved_bytes.large_pool.peak": 637534208,
113
- "reserved_bytes.small_pool.allocated": 10485760,
114
- "reserved_bytes.small_pool.current": 10485760,
115
- "reserved_bytes.small_pool.freed": 0,
116
- "reserved_bytes.small_pool.peak": 10485760,
117
- "segment.all.allocated": 32,
118
- "segment.all.current": 32,
119
- "segment.all.freed": 0,
120
- "segment.all.peak": 32,
121
- "segment.large_pool.allocated": 27,
122
- "segment.large_pool.current": 27,
123
- "segment.large_pool.freed": 0,
124
- "segment.large_pool.peak": 27,
125
- "segment.small_pool.allocated": 5,
126
- "segment.small_pool.current": 5,
127
- "segment.small_pool.freed": 0,
128
- "segment.small_pool.peak": 5
129
- }
130
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea40bb55d84163198cdbeefcabee8d1ba4814a01e59c062b2034dff472564800
3
+ size 6046
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/gpu_stats_la.json CHANGED
@@ -1,130 +1,3 @@
1
- {
2
- "memory_allocated": 808749568,
3
- "max_memory_allocated": 5014978048,
4
- "memory_reserved": 6358564864,
5
- "max_memory_reserved": 6727663616,
6
- "memory_stats": {
7
- "active.all.allocated": 24482704,
8
- "active.all.current": 10809,
9
- "active.all.freed": 24471895,
10
- "active.all.peak": 10985,
11
- "active.large_pool.allocated": 11722181,
12
- "active.large_pool.current": 222,
13
- "active.large_pool.freed": 11721959,
14
- "active.large_pool.peak": 360,
15
- "active.small_pool.allocated": 12760523,
16
- "active.small_pool.current": 10587,
17
- "active.small_pool.freed": 12749936,
18
- "active.small_pool.peak": 10750,
19
- "active_bytes.all.allocated": 78209693356544,
20
- "active_bytes.all.current": 808749568,
21
- "active_bytes.all.freed": 78208884606976,
22
- "active_bytes.all.peak": 5014978048,
23
- "active_bytes.large_pool.allocated": 74862721955840,
24
- "active_bytes.large_pool.current": 799473664,
25
- "active_bytes.large_pool.freed": 74861922482176,
26
- "active_bytes.large_pool.peak": 4997783552,
27
- "active_bytes.small_pool.allocated": 3346971400704,
28
- "active_bytes.small_pool.current": 9275904,
29
- "active_bytes.small_pool.freed": 3346962124800,
30
- "active_bytes.small_pool.peak": 124137472,
31
- "allocated_bytes.all.allocated": 78209693356544,
32
- "allocated_bytes.all.current": 808749568,
33
- "allocated_bytes.all.freed": 78208884606976,
34
- "allocated_bytes.all.peak": 5014978048,
35
- "allocated_bytes.large_pool.allocated": 74862721955840,
36
- "allocated_bytes.large_pool.current": 799473664,
37
- "allocated_bytes.large_pool.freed": 74861922482176,
38
- "allocated_bytes.large_pool.peak": 4997783552,
39
- "allocated_bytes.small_pool.allocated": 3346971400704,
40
- "allocated_bytes.small_pool.current": 9275904,
41
- "allocated_bytes.small_pool.freed": 3346962124800,
42
- "allocated_bytes.small_pool.peak": 124137472,
43
- "allocation.all.allocated": 24482704,
44
- "allocation.all.current": 10809,
45
- "allocation.all.freed": 24471895,
46
- "allocation.all.peak": 10985,
47
- "allocation.large_pool.allocated": 11722181,
48
- "allocation.large_pool.current": 222,
49
- "allocation.large_pool.freed": 11721959,
50
- "allocation.large_pool.peak": 360,
51
- "allocation.small_pool.allocated": 12760523,
52
- "allocation.small_pool.current": 10587,
53
- "allocation.small_pool.freed": 12749936,
54
- "allocation.small_pool.peak": 10750,
55
- "inactive_split.all.allocated": 12808394,
56
- "inactive_split.all.current": 115,
57
- "inactive_split.all.freed": 12808279,
58
- "inactive_split.all.peak": 246,
59
- "inactive_split.large_pool.allocated": 8088991,
60
- "inactive_split.large_pool.current": 38,
61
- "inactive_split.large_pool.freed": 8088953,
62
- "inactive_split.large_pool.peak": 135,
63
- "inactive_split.small_pool.allocated": 4719403,
64
- "inactive_split.small_pool.current": 77,
65
- "inactive_split.small_pool.freed": 4719326,
66
- "inactive_split.small_pool.peak": 156,
67
- "inactive_split_bytes.all.allocated": 71846063992832,
68
- "inactive_split_bytes.all.current": 116094464,
69
- "inactive_split_bytes.all.freed": 71845947898368,
70
- "inactive_split_bytes.all.peak": 1893863936,
71
- "inactive_split_bytes.large_pool.allocated": 68396959819776,
72
- "inactive_split_bytes.large_pool.current": 91815936,
73
- "inactive_split_bytes.large_pool.freed": 68396868003840,
74
- "inactive_split_bytes.large_pool.peak": 1887404032,
75
- "inactive_split_bytes.small_pool.allocated": 3449104173056,
76
- "inactive_split_bytes.small_pool.current": 24278528,
77
- "inactive_split_bytes.small_pool.freed": 3449079894528,
78
- "inactive_split_bytes.small_pool.peak": 84646400,
79
- "max_split_size": -1,
80
- "num_alloc_retries": 0,
81
- "num_device_alloc": 436,
82
- "num_device_free": 130,
83
- "num_ooms": 0,
84
- "num_sync_all_streams": 1,
85
- "oversize_allocations.allocated": 0,
86
- "oversize_allocations.current": 0,
87
- "oversize_allocations.freed": 0,
88
- "oversize_allocations.peak": 0,
89
- "oversize_segments.allocated": 0,
90
- "oversize_segments.current": 0,
91
- "oversize_segments.freed": 0,
92
- "oversize_segments.peak": 0,
93
- "requested_bytes.all.allocated": 77380438428195,
94
- "requested_bytes.all.current": 800396976,
95
- "requested_bytes.all.freed": 77379638031219,
96
- "requested_bytes.all.peak": 4992933420,
97
- "requested_bytes.large_pool.allocated": 74034645754624,
98
- "requested_bytes.large_pool.current": 795392000,
99
- "requested_bytes.large_pool.freed": 74033850362624,
100
- "requested_bytes.large_pool.peak": 4975769600,
101
- "requested_bytes.small_pool.allocated": 3345792673571,
102
- "requested_bytes.small_pool.current": 5004976,
103
- "requested_bytes.small_pool.freed": 3345787668595,
104
- "requested_bytes.small_pool.peak": 119884008,
105
- "reserved_bytes.all.allocated": 12182355968,
106
- "reserved_bytes.all.current": 6358564864,
107
- "reserved_bytes.all.freed": 5823791104,
108
- "reserved_bytes.all.peak": 6727663616,
109
- "reserved_bytes.large_pool.allocated": 12002000896,
110
- "reserved_bytes.large_pool.current": 6230638592,
111
- "reserved_bytes.large_pool.freed": 5771362304,
112
- "reserved_bytes.large_pool.peak": 6662651904,
113
- "reserved_bytes.small_pool.allocated": 180355072,
114
- "reserved_bytes.small_pool.current": 127926272,
115
- "reserved_bytes.small_pool.freed": 52428800,
116
- "reserved_bytes.small_pool.peak": 127926272,
117
- "segment.all.allocated": 436,
118
- "segment.all.current": 306,
119
- "segment.all.freed": 130,
120
- "segment.all.peak": 306,
121
- "segment.large_pool.allocated": 350,
122
- "segment.large_pool.current": 245,
123
- "segment.large_pool.freed": 105,
124
- "segment.large_pool.peak": 245,
125
- "segment.small_pool.allocated": 86,
126
- "segment.small_pool.current": 61,
127
- "segment.small_pool.freed": 25,
128
- "segment.small_pool.peak": 61
129
- }
130
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a08e54b4ea5ec71e13d699e69ca0fd0ff6279cc55eadba8a5e1b9e9bb0c3ee9f
3
+ size 6177
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/special_tokens_map.json CHANGED
@@ -1,7 +1,3 @@
1
- {
2
- "cls_token": "[CLS]",
3
- "mask_token": "[MASK]",
4
- "pad_token": "[PAD]",
5
- "sep_token": "[SEP]",
6
- "unk_token": "[UNK]"
7
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b6d346be366a7d1d48332dbc9fdf3bf8960b5d879522b7799ddba59e76237ee3
3
+ size 125
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_0/tokenizer_config.json CHANGED
@@ -1,55 +1,3 @@
1
- {
2
- "added_tokens_decoder": {
3
- "0": {
4
- "content": "[PAD]",
5
- "lstrip": false,
6
- "normalized": false,
7
- "rstrip": false,
8
- "single_word": false,
9
- "special": true
10
- },
11
- "100": {
12
- "content": "[UNK]",
13
- "lstrip": false,
14
- "normalized": false,
15
- "rstrip": false,
16
- "single_word": false,
17
- "special": true
18
- },
19
- "101": {
20
- "content": "[CLS]",
21
- "lstrip": false,
22
- "normalized": false,
23
- "rstrip": false,
24
- "single_word": false,
25
- "special": true
26
- },
27
- "102": {
28
- "content": "[SEP]",
29
- "lstrip": false,
30
- "normalized": false,
31
- "rstrip": false,
32
- "single_word": false,
33
- "special": true
34
- },
35
- "103": {
36
- "content": "[MASK]",
37
- "lstrip": false,
38
- "normalized": false,
39
- "rstrip": false,
40
- "single_word": false,
41
- "special": true
42
- }
43
- },
44
- "clean_up_tokenization_spaces": true,
45
- "cls_token": "[CLS]",
46
- "do_lower_case": true,
47
- "mask_token": "[MASK]",
48
- "model_max_length": 512,
49
- "pad_token": "[PAD]",
50
- "sep_token": "[SEP]",
51
- "strip_accents": null,
52
- "tokenize_chinese_chars": true,
53
- "tokenizer_class": "BertTokenizer",
54
- "unk_token": "[UNK]"
55
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f62a57a75856b93282501c92a86f62b169997c81e93cf6f75b7cc15d6285968e
3
+ size 1190
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/adapter_config.json CHANGED
@@ -1,32 +1,3 @@
1
- {
2
- "alpha_pattern": {},
3
- "auto_mapping": null,
4
- "base_model_name_or_path": "bert-base-uncased",
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": 16,
14
- "lora_dropout": 0.1,
15
- "megatron_config": null,
16
- "megatron_core": "megatron.core",
17
- "modules_to_save": [
18
- "classifier",
19
- "score"
20
- ],
21
- "peft_type": "LORA",
22
- "r": 8,
23
- "rank_pattern": {},
24
- "revision": null,
25
- "target_modules": [
26
- "value",
27
- "query"
28
- ],
29
- "task_type": "SEQ_CLS",
30
- "use_dora": false,
31
- "use_rslora": false
32
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81a29a631d083a248a7fe8b5acc0d972f1e62158c7e9d95bdbbeffc2630802a8
3
+ size 668
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/all_results.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.8533762057877814, "eval_f1": 0.8011138696906663}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8a2b532b49799110512f9cdc930cff4015cb4f79b91c0877f12cc036f3c2ce5
3
+ size 68
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/all_results_la_kron_all_homo_mc_corr_1000.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.8534504081127875, "eval_f1": 0.801221189653437}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a89c9e0a18de01aad9dd42f84630c267bec8a14c7bfa29b3c84586011cb0599
3
+ size 67
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/all_results_val.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.854541706747286, "eval_f1": 0.804354656858215}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d90a2c37f96886daac450e45e9fcb3f1bd8aee4676ca3e2107202575ba48506
3
+ size 66
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/eval_res.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/eval_res_la_kron_all_homo_mc_corr_1000.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/gpu_stats.json CHANGED
@@ -1,130 +1,3 @@
1
- {
2
- "memory_allocated": 459737600,
3
- "max_memory_allocated": 5480528896,
4
- "memory_reserved": 6840909824,
5
- "max_memory_reserved": 6840909824,
6
- "memory_stats": {
7
- "active.all.allocated": 27775346,
8
- "active.all.current": 367,
9
- "active.all.freed": 27774979,
10
- "active.all.peak": 627,
11
- "active.large_pool.allocated": 13181532,
12
- "active.large_pool.current": 77,
13
- "active.large_pool.freed": 13181455,
14
- "active.large_pool.peak": 250,
15
- "active.small_pool.allocated": 14593814,
16
- "active.small_pool.current": 290,
17
- "active.small_pool.freed": 14593524,
18
- "active.small_pool.peak": 538,
19
- "active_bytes.all.allocated": 75488457111552,
20
- "active_bytes.all.current": 459737600,
21
- "active_bytes.all.freed": 75487997373952,
22
- "active_bytes.all.peak": 5480528896,
23
- "active_bytes.large_pool.allocated": 68824605308416,
24
- "active_bytes.large_pool.current": 455606272,
25
- "active_bytes.large_pool.freed": 68824149702144,
26
- "active_bytes.large_pool.peak": 5461108736,
27
- "active_bytes.small_pool.allocated": 6663851803136,
28
- "active_bytes.small_pool.current": 4131328,
29
- "active_bytes.small_pool.freed": 6663847671808,
30
- "active_bytes.small_pool.peak": 98207744,
31
- "allocated_bytes.all.allocated": 75488457111552,
32
- "allocated_bytes.all.current": 459737600,
33
- "allocated_bytes.all.freed": 75487997373952,
34
- "allocated_bytes.all.peak": 5480528896,
35
- "allocated_bytes.large_pool.allocated": 68824605308416,
36
- "allocated_bytes.large_pool.current": 455606272,
37
- "allocated_bytes.large_pool.freed": 68824149702144,
38
- "allocated_bytes.large_pool.peak": 5461108736,
39
- "allocated_bytes.small_pool.allocated": 6663851803136,
40
- "allocated_bytes.small_pool.current": 4131328,
41
- "allocated_bytes.small_pool.freed": 6663847671808,
42
- "allocated_bytes.small_pool.peak": 98207744,
43
- "allocation.all.allocated": 27775346,
44
- "allocation.all.current": 367,
45
- "allocation.all.freed": 27774979,
46
- "allocation.all.peak": 627,
47
- "allocation.large_pool.allocated": 13181532,
48
- "allocation.large_pool.current": 77,
49
- "allocation.large_pool.freed": 13181455,
50
- "allocation.large_pool.peak": 250,
51
- "allocation.small_pool.allocated": 14593814,
52
- "allocation.small_pool.current": 290,
53
- "allocation.small_pool.freed": 14593524,
54
- "allocation.small_pool.peak": 538,
55
- "inactive_split.all.allocated": 14966847,
56
- "inactive_split.all.current": 32,
57
- "inactive_split.all.freed": 14966815,
58
- "inactive_split.all.peak": 143,
59
- "inactive_split.large_pool.allocated": 10529227,
60
- "inactive_split.large_pool.current": 19,
61
- "inactive_split.large_pool.freed": 10529208,
62
- "inactive_split.large_pool.peak": 123,
63
- "inactive_split.small_pool.allocated": 4437620,
64
- "inactive_split.small_pool.current": 13,
65
- "inactive_split.small_pool.freed": 4437607,
66
- "inactive_split.small_pool.peak": 70,
67
- "inactive_split_bytes.all.allocated": 68752278194176,
68
- "inactive_split_bytes.all.current": 64550400,
69
- "inactive_split_bytes.all.freed": 68752213643776,
70
- "inactive_split_bytes.all.peak": 1129547776,
71
- "inactive_split_bytes.large_pool.allocated": 61954103755264,
72
- "inactive_split_bytes.large_pool.current": 58195968,
73
- "inactive_split_bytes.large_pool.freed": 61954045559296,
74
- "inactive_split_bytes.large_pool.peak": 1127497728,
75
- "inactive_split_bytes.small_pool.allocated": 6798174438912,
76
- "inactive_split_bytes.small_pool.current": 6354432,
77
- "inactive_split_bytes.small_pool.freed": 6798168084480,
78
- "inactive_split_bytes.small_pool.peak": 43309568,
79
- "max_split_size": -1,
80
- "num_alloc_retries": 0,
81
- "num_device_alloc": 207,
82
- "num_device_free": 0,
83
- "num_ooms": 0,
84
- "num_sync_all_streams": 0,
85
- "oversize_allocations.allocated": 0,
86
- "oversize_allocations.current": 0,
87
- "oversize_allocations.freed": 0,
88
- "oversize_allocations.peak": 0,
89
- "oversize_segments.allocated": 0,
90
- "oversize_segments.current": 0,
91
- "oversize_segments.freed": 0,
92
- "oversize_segments.peak": 0,
93
- "requested_bytes.all.allocated": 70973866325084,
94
- "requested_bytes.all.current": 458599752,
95
- "requested_bytes.all.freed": 70973407725332,
96
- "requested_bytes.all.peak": 5476793900,
97
- "requested_bytes.large_pool.allocated": 64311530915008,
98
- "requested_bytes.large_pool.current": 454473728,
99
- "requested_bytes.large_pool.freed": 64311076441280,
100
- "requested_bytes.large_pool.peak": 5457379328,
101
- "requested_bytes.small_pool.allocated": 6662335410076,
102
- "requested_bytes.small_pool.current": 4126024,
103
- "requested_bytes.small_pool.freed": 6662331284052,
104
- "requested_bytes.small_pool.peak": 98173436,
105
- "reserved_bytes.all.allocated": 6840909824,
106
- "reserved_bytes.all.current": 6840909824,
107
- "reserved_bytes.all.freed": 0,
108
- "reserved_bytes.all.peak": 6840909824,
109
- "reserved_bytes.large_pool.allocated": 6729760768,
110
- "reserved_bytes.large_pool.current": 6729760768,
111
- "reserved_bytes.large_pool.freed": 0,
112
- "reserved_bytes.large_pool.peak": 6729760768,
113
- "reserved_bytes.small_pool.allocated": 111149056,
114
- "reserved_bytes.small_pool.current": 111149056,
115
- "reserved_bytes.small_pool.freed": 0,
116
- "reserved_bytes.small_pool.peak": 111149056,
117
- "segment.all.allocated": 207,
118
- "segment.all.current": 207,
119
- "segment.all.freed": 0,
120
- "segment.all.peak": 207,
121
- "segment.large_pool.allocated": 154,
122
- "segment.large_pool.current": 154,
123
- "segment.large_pool.freed": 0,
124
- "segment.large_pool.peak": 154,
125
- "segment.small_pool.allocated": 53,
126
- "segment.small_pool.current": 53,
127
- "segment.small_pool.freed": 0,
128
- "segment.small_pool.peak": 53
129
- }
130
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b3cacdee0e75f6018b4cb7944afd30f8210df2e43bd561b364bcd740367360c4
3
+ size 6120
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/gpu_stats_la.json CHANGED
@@ -1,130 +1,3 @@
1
- {
2
- "memory_allocated": 808356352,
3
- "max_memory_allocated": 5015770624,
4
- "memory_reserved": 6350176256,
5
- "max_memory_reserved": 6752829440,
6
- "memory_stats": {
7
- "active.all.allocated": 73448108,
8
- "active.all.current": 10809,
9
- "active.all.freed": 73437299,
10
- "active.all.peak": 10985,
11
- "active.large_pool.allocated": 35166539,
12
- "active.large_pool.current": 222,
13
- "active.large_pool.freed": 35166317,
14
- "active.large_pool.peak": 360,
15
- "active.small_pool.allocated": 38281569,
16
- "active.small_pool.current": 10587,
17
- "active.small_pool.freed": 38270982,
18
- "active.small_pool.peak": 10750,
19
- "active_bytes.all.allocated": 234626955736064,
20
- "active_bytes.all.current": 808356352,
21
- "active_bytes.all.freed": 234626147379712,
22
- "active_bytes.all.peak": 5015770624,
23
- "active_bytes.large_pool.allocated": 224586041533952,
24
- "active_bytes.large_pool.current": 799080448,
25
- "active_bytes.large_pool.freed": 224585242453504,
26
- "active_bytes.large_pool.peak": 4998576128,
27
- "active_bytes.small_pool.allocated": 10040914202112,
28
- "active_bytes.small_pool.current": 9275904,
29
- "active_bytes.small_pool.freed": 10040904926208,
30
- "active_bytes.small_pool.peak": 124137472,
31
- "allocated_bytes.all.allocated": 234626955736064,
32
- "allocated_bytes.all.current": 808356352,
33
- "allocated_bytes.all.freed": 234626147379712,
34
- "allocated_bytes.all.peak": 5015770624,
35
- "allocated_bytes.large_pool.allocated": 224586041533952,
36
- "allocated_bytes.large_pool.current": 799080448,
37
- "allocated_bytes.large_pool.freed": 224585242453504,
38
- "allocated_bytes.large_pool.peak": 4998576128,
39
- "allocated_bytes.small_pool.allocated": 10040914202112,
40
- "allocated_bytes.small_pool.current": 9275904,
41
- "allocated_bytes.small_pool.freed": 10040904926208,
42
- "allocated_bytes.small_pool.peak": 124137472,
43
- "allocation.all.allocated": 73448108,
44
- "allocation.all.current": 10809,
45
- "allocation.all.freed": 73437299,
46
- "allocation.all.peak": 10985,
47
- "allocation.large_pool.allocated": 35166539,
48
- "allocation.large_pool.current": 222,
49
- "allocation.large_pool.freed": 35166317,
50
- "allocation.large_pool.peak": 360,
51
- "allocation.small_pool.allocated": 38281569,
52
- "allocation.small_pool.current": 10587,
53
- "allocation.small_pool.freed": 38270982,
54
- "allocation.small_pool.peak": 10750,
55
- "inactive_split.all.allocated": 38491526,
56
- "inactive_split.all.current": 119,
57
- "inactive_split.all.freed": 38491407,
58
- "inactive_split.all.peak": 1966,
59
- "inactive_split.large_pool.allocated": 24502987,
60
- "inactive_split.large_pool.current": 38,
61
- "inactive_split.large_pool.freed": 24502949,
62
- "inactive_split.large_pool.peak": 141,
63
- "inactive_split.small_pool.allocated": 13988539,
64
- "inactive_split.small_pool.current": 81,
65
- "inactive_split.small_pool.freed": 13988458,
66
- "inactive_split.small_pool.peak": 1920,
67
- "inactive_split_bytes.all.allocated": 216296435532288,
68
- "inactive_split_bytes.all.current": 120681984,
69
- "inactive_split_bytes.all.freed": 216296314850304,
70
- "inactive_split_bytes.all.peak": 1893863936,
71
- "inactive_split_bytes.large_pool.allocated": 205937432882688,
72
- "inactive_split_bytes.large_pool.current": 92209152,
73
- "inactive_split_bytes.large_pool.freed": 205937340673536,
74
- "inactive_split_bytes.large_pool.peak": 1887404032,
75
- "inactive_split_bytes.small_pool.allocated": 10359002649600,
76
- "inactive_split_bytes.small_pool.current": 28472832,
77
- "inactive_split_bytes.small_pool.freed": 10358974176768,
78
- "inactive_split_bytes.small_pool.peak": 84646400,
79
- "max_split_size": -1,
80
- "num_alloc_retries": 0,
81
- "num_device_alloc": 1230,
82
- "num_device_free": 924,
83
- "num_ooms": 0,
84
- "num_sync_all_streams": 5,
85
- "oversize_allocations.allocated": 0,
86
- "oversize_allocations.current": 0,
87
- "oversize_allocations.freed": 0,
88
- "oversize_allocations.peak": 0,
89
- "oversize_segments.allocated": 0,
90
- "oversize_segments.current": 0,
91
- "oversize_segments.freed": 0,
92
- "oversize_segments.peak": 0,
93
- "requested_bytes.all.allocated": 232141281205865,
94
- "requested_bytes.all.current": 800396976,
95
- "requested_bytes.all.freed": 232140480808889,
96
- "requested_bytes.all.peak": 4992933420,
97
- "requested_bytes.large_pool.allocated": 222103903185152,
98
- "requested_bytes.large_pool.current": 795392000,
99
- "requested_bytes.large_pool.freed": 222103107793152,
100
- "requested_bytes.large_pool.peak": 4975769600,
101
- "requested_bytes.small_pool.allocated": 10037378020713,
102
- "requested_bytes.small_pool.current": 5004976,
103
- "requested_bytes.small_pool.freed": 10037373015737,
104
- "requested_bytes.small_pool.peak": 119884008,
105
- "reserved_bytes.all.allocated": 34997272576,
106
- "reserved_bytes.all.current": 6350176256,
107
- "reserved_bytes.all.freed": 28647096320,
108
- "reserved_bytes.all.peak": 6752829440,
109
- "reserved_bytes.large_pool.allocated": 34483470336,
110
- "reserved_bytes.large_pool.current": 6222249984,
111
- "reserved_bytes.large_pool.freed": 28261220352,
112
- "reserved_bytes.large_pool.peak": 6687817728,
113
- "reserved_bytes.small_pool.allocated": 513802240,
114
- "reserved_bytes.small_pool.current": 127926272,
115
- "reserved_bytes.small_pool.freed": 385875968,
116
- "reserved_bytes.small_pool.peak": 127926272,
117
- "segment.all.allocated": 1230,
118
- "segment.all.current": 306,
119
- "segment.all.freed": 924,
120
- "segment.all.peak": 306,
121
- "segment.large_pool.allocated": 985,
122
- "segment.large_pool.current": 245,
123
- "segment.large_pool.freed": 740,
124
- "segment.large_pool.peak": 245,
125
- "segment.small_pool.allocated": 245,
126
- "segment.small_pool.current": 61,
127
- "segment.small_pool.freed": 184,
128
- "segment.small_pool.peak": 61
129
- }
130
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c7ea034a00859dede9a9b30a227018a2ab98782be37df0d1525dd5688ccd88a7
3
+ size 6214
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/special_tokens_map.json CHANGED
@@ -1,7 +1,3 @@
1
- {
2
- "cls_token": "[CLS]",
3
- "mask_token": "[MASK]",
4
- "pad_token": "[PAD]",
5
- "sep_token": "[SEP]",
6
- "unk_token": "[UNK]"
7
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b6d346be366a7d1d48332dbc9fdf3bf8960b5d879522b7799ddba59e76237ee3
3
+ size 125
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_10915/tokenizer_config.json CHANGED
@@ -1,55 +1,3 @@
1
- {
2
- "added_tokens_decoder": {
3
- "0": {
4
- "content": "[PAD]",
5
- "lstrip": false,
6
- "normalized": false,
7
- "rstrip": false,
8
- "single_word": false,
9
- "special": true
10
- },
11
- "100": {
12
- "content": "[UNK]",
13
- "lstrip": false,
14
- "normalized": false,
15
- "rstrip": false,
16
- "single_word": false,
17
- "special": true
18
- },
19
- "101": {
20
- "content": "[CLS]",
21
- "lstrip": false,
22
- "normalized": false,
23
- "rstrip": false,
24
- "single_word": false,
25
- "special": true
26
- },
27
- "102": {
28
- "content": "[SEP]",
29
- "lstrip": false,
30
- "normalized": false,
31
- "rstrip": false,
32
- "single_word": false,
33
- "special": true
34
- },
35
- "103": {
36
- "content": "[MASK]",
37
- "lstrip": false,
38
- "normalized": false,
39
- "rstrip": false,
40
- "single_word": false,
41
- "special": true
42
- }
43
- },
44
- "clean_up_tokenization_spaces": true,
45
- "cls_token": "[CLS]",
46
- "do_lower_case": true,
47
- "mask_token": "[MASK]",
48
- "model_max_length": 512,
49
- "pad_token": "[PAD]",
50
- "sep_token": "[SEP]",
51
- "strip_accents": null,
52
- "tokenize_chinese_chars": true,
53
- "tokenizer_class": "BertTokenizer",
54
- "unk_token": "[UNK]"
55
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f62a57a75856b93282501c92a86f62b169997c81e93cf6f75b7cc15d6285968e
3
+ size 1190
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/adapter_config.json CHANGED
@@ -1,32 +1,3 @@
1
- {
2
- "alpha_pattern": {},
3
- "auto_mapping": null,
4
- "base_model_name_or_path": "bert-base-uncased",
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": 16,
14
- "lora_dropout": 0.1,
15
- "megatron_config": null,
16
- "megatron_core": "megatron.core",
17
- "modules_to_save": [
18
- "classifier",
19
- "score"
20
- ],
21
- "peft_type": "LORA",
22
- "r": 8,
23
- "rank_pattern": {},
24
- "revision": null,
25
- "target_modules": [
26
- "value",
27
- "query"
28
- ],
29
- "task_type": "SEQ_CLS",
30
- "use_dora": false,
31
- "use_rslora": false
32
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81a29a631d083a248a7fe8b5acc0d972f1e62158c7e9d95bdbbeffc2630802a8
3
+ size 668
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/all_results.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.8614642592134554, "eval_f1": 0.8166672122025466}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b84741c4d971f9bf30ba410095ba549f669d191b2237e039c7cc4fbcd73b3e03
3
+ size 68
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/all_results_la_kron_all_homo_mc_corr_1000.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.8614889933217907, "eval_f1": 0.8166939443535187}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e28e43b174cffd344dc9e04521992272a0609ab9632bab865338c0f931f1fabb
3
+ size 68
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/all_results_val.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.8613164765700151, "eval_f1": 0.8177352356871953}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6d51821dbbe2e76ed118fdce5f3fc47b472feb5b556e10801ef544f10bd48acd
3
+ size 68
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/eval_res.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/eval_res_la_kron_all_homo_mc_corr_1000.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/gpu_stats.json CHANGED
@@ -1,130 +1,3 @@
1
- {
2
- "memory_allocated": 459742208,
3
- "max_memory_allocated": 5480528896,
4
- "memory_reserved": 6840909824,
5
- "max_memory_reserved": 6840909824,
6
- "memory_stats": {
7
- "active.all.allocated": 38723233,
8
- "active.all.current": 367,
9
- "active.all.freed": 38722866,
10
- "active.all.peak": 627,
11
- "active.large_pool.allocated": 18627179,
12
- "active.large_pool.current": 77,
13
- "active.large_pool.freed": 18627102,
14
- "active.large_pool.peak": 250,
15
- "active.small_pool.allocated": 20096054,
16
- "active.small_pool.current": 290,
17
- "active.small_pool.freed": 20095764,
18
- "active.small_pool.peak": 538,
19
- "active_bytes.all.allocated": 109319365175808,
20
- "active_bytes.all.current": 459742208,
21
- "active_bytes.all.freed": 109318905433600,
22
- "active_bytes.all.peak": 5480528896,
23
- "active_bytes.large_pool.allocated": 100259252704768,
24
- "active_bytes.large_pool.current": 455606272,
25
- "active_bytes.large_pool.freed": 100258797098496,
26
- "active_bytes.large_pool.peak": 5461108736,
27
- "active_bytes.small_pool.allocated": 9060112471040,
28
- "active_bytes.small_pool.current": 4135936,
29
- "active_bytes.small_pool.freed": 9060108335104,
30
- "active_bytes.small_pool.peak": 98207744,
31
- "allocated_bytes.all.allocated": 109319365175808,
32
- "allocated_bytes.all.current": 459742208,
33
- "allocated_bytes.all.freed": 109318905433600,
34
- "allocated_bytes.all.peak": 5480528896,
35
- "allocated_bytes.large_pool.allocated": 100259252704768,
36
- "allocated_bytes.large_pool.current": 455606272,
37
- "allocated_bytes.large_pool.freed": 100258797098496,
38
- "allocated_bytes.large_pool.peak": 5461108736,
39
- "allocated_bytes.small_pool.allocated": 9060112471040,
40
- "allocated_bytes.small_pool.current": 4135936,
41
- "allocated_bytes.small_pool.freed": 9060108335104,
42
- "allocated_bytes.small_pool.peak": 98207744,
43
- "allocation.all.allocated": 38723233,
44
- "allocation.all.current": 367,
45
- "allocation.all.freed": 38722866,
46
- "allocation.all.peak": 627,
47
- "allocation.large_pool.allocated": 18627179,
48
- "allocation.large_pool.current": 77,
49
- "allocation.large_pool.freed": 18627102,
50
- "allocation.large_pool.peak": 250,
51
- "allocation.small_pool.allocated": 20096054,
52
- "allocation.small_pool.current": 290,
53
- "allocation.small_pool.freed": 20095764,
54
- "allocation.small_pool.peak": 538,
55
- "inactive_split.all.allocated": 20542981,
56
- "inactive_split.all.current": 32,
57
- "inactive_split.all.freed": 20542949,
58
- "inactive_split.all.peak": 143,
59
- "inactive_split.large_pool.allocated": 14749135,
60
- "inactive_split.large_pool.current": 19,
61
- "inactive_split.large_pool.freed": 14749116,
62
- "inactive_split.large_pool.peak": 126,
63
- "inactive_split.small_pool.allocated": 5793846,
64
- "inactive_split.small_pool.current": 13,
65
- "inactive_split.small_pool.freed": 5793833,
66
- "inactive_split.small_pool.peak": 70,
67
- "inactive_split_bytes.all.allocated": 99116112968192,
68
- "inactive_split_bytes.all.current": 64545792,
69
- "inactive_split_bytes.all.freed": 99116048422400,
70
- "inactive_split_bytes.all.peak": 1129547776,
71
- "inactive_split_bytes.large_pool.allocated": 89881653104128,
72
- "inactive_split_bytes.large_pool.current": 58195968,
73
- "inactive_split_bytes.large_pool.freed": 89881594908160,
74
- "inactive_split_bytes.large_pool.peak": 1127497728,
75
- "inactive_split_bytes.small_pool.allocated": 9234459864064,
76
- "inactive_split_bytes.small_pool.current": 6349824,
77
- "inactive_split_bytes.small_pool.freed": 9234453514240,
78
- "inactive_split_bytes.small_pool.peak": 43309568,
79
- "max_split_size": -1,
80
- "num_alloc_retries": 0,
81
- "num_device_alloc": 207,
82
- "num_device_free": 0,
83
- "num_ooms": 0,
84
- "num_sync_all_streams": 0,
85
- "oversize_allocations.allocated": 0,
86
- "oversize_allocations.current": 0,
87
- "oversize_allocations.freed": 0,
88
- "oversize_allocations.peak": 0,
89
- "oversize_segments.allocated": 0,
90
- "oversize_segments.current": 0,
91
- "oversize_segments.freed": 0,
92
- "oversize_segments.peak": 0,
93
- "requested_bytes.all.allocated": 103223002053941,
94
- "requested_bytes.all.current": 458604360,
95
- "requested_bytes.all.freed": 103222543449581,
96
- "requested_bytes.all.peak": 5476793900,
97
- "requested_bytes.large_pool.allocated": 94164945048224,
98
- "requested_bytes.large_pool.current": 454473728,
99
- "requested_bytes.large_pool.freed": 94164490574496,
100
- "requested_bytes.large_pool.peak": 5457379328,
101
- "requested_bytes.small_pool.allocated": 9058057005717,
102
- "requested_bytes.small_pool.current": 4130632,
103
- "requested_bytes.small_pool.freed": 9058052875085,
104
- "requested_bytes.small_pool.peak": 98173436,
105
- "reserved_bytes.all.allocated": 6840909824,
106
- "reserved_bytes.all.current": 6840909824,
107
- "reserved_bytes.all.freed": 0,
108
- "reserved_bytes.all.peak": 6840909824,
109
- "reserved_bytes.large_pool.allocated": 6729760768,
110
- "reserved_bytes.large_pool.current": 6729760768,
111
- "reserved_bytes.large_pool.freed": 0,
112
- "reserved_bytes.large_pool.peak": 6729760768,
113
- "reserved_bytes.small_pool.allocated": 111149056,
114
- "reserved_bytes.small_pool.current": 111149056,
115
- "reserved_bytes.small_pool.freed": 0,
116
- "reserved_bytes.small_pool.peak": 111149056,
117
- "segment.all.allocated": 207,
118
- "segment.all.current": 207,
119
- "segment.all.freed": 0,
120
- "segment.all.peak": 207,
121
- "segment.large_pool.allocated": 154,
122
- "segment.large_pool.current": 154,
123
- "segment.large_pool.freed": 0,
124
- "segment.large_pool.peak": 154,
125
- "segment.small_pool.allocated": 53,
126
- "segment.small_pool.current": 53,
127
- "segment.small_pool.freed": 0,
128
- "segment.small_pool.peak": 53
129
- }
130
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e3c9aeec49df5b4a818d145ea556790311547ed51460e64ce052e591607b75ae
3
+ size 6130
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/gpu_stats_la.json CHANGED
@@ -1,130 +1,3 @@
1
- {
2
- "memory_allocated": 808552960,
3
- "max_memory_allocated": 5015770624,
4
- "memory_reserved": 6345981952,
5
- "max_memory_reserved": 6752829440,
6
- "memory_stats": {
7
- "active.all.allocated": 97930810,
8
- "active.all.current": 10809,
9
- "active.all.freed": 97920001,
10
- "active.all.peak": 10985,
11
- "active.large_pool.allocated": 46888718,
12
- "active.large_pool.current": 222,
13
- "active.large_pool.freed": 46888496,
14
- "active.large_pool.peak": 360,
15
- "active.small_pool.allocated": 51042092,
16
- "active.small_pool.current": 10587,
17
- "active.small_pool.freed": 51031505,
18
- "active.small_pool.peak": 10750,
19
- "active_bytes.all.allocated": 312846567023616,
20
- "active_bytes.all.current": 808552960,
21
- "active_bytes.all.freed": 312845758470656,
22
- "active_bytes.all.peak": 5015770624,
23
- "active_bytes.large_pool.allocated": 299458681420800,
24
- "active_bytes.large_pool.current": 799277056,
25
- "active_bytes.large_pool.freed": 299457882143744,
26
- "active_bytes.large_pool.peak": 4998576128,
27
- "active_bytes.small_pool.allocated": 13387885602816,
28
- "active_bytes.small_pool.current": 9275904,
29
- "active_bytes.small_pool.freed": 13387876326912,
30
- "active_bytes.small_pool.peak": 124137472,
31
- "allocated_bytes.all.allocated": 312846567023616,
32
- "allocated_bytes.all.current": 808552960,
33
- "allocated_bytes.all.freed": 312845758470656,
34
- "allocated_bytes.all.peak": 5015770624,
35
- "allocated_bytes.large_pool.allocated": 299458681420800,
36
- "allocated_bytes.large_pool.current": 799277056,
37
- "allocated_bytes.large_pool.freed": 299457882143744,
38
- "allocated_bytes.large_pool.peak": 4998576128,
39
- "allocated_bytes.small_pool.allocated": 13387885602816,
40
- "allocated_bytes.small_pool.current": 9275904,
41
- "allocated_bytes.small_pool.freed": 13387876326912,
42
- "allocated_bytes.small_pool.peak": 124137472,
43
- "allocation.all.allocated": 97930810,
44
- "allocation.all.current": 10809,
45
- "allocation.all.freed": 97920001,
46
- "allocation.all.peak": 10985,
47
- "allocation.large_pool.allocated": 46888718,
48
- "allocation.large_pool.current": 222,
49
- "allocation.large_pool.freed": 46888496,
50
- "allocation.large_pool.peak": 360,
51
- "allocation.small_pool.allocated": 51042092,
52
- "allocation.small_pool.current": 10587,
53
- "allocation.small_pool.freed": 51031505,
54
- "allocation.small_pool.peak": 10750,
55
- "inactive_split.all.allocated": 51328722,
56
- "inactive_split.all.current": 125,
57
- "inactive_split.all.freed": 51328597,
58
- "inactive_split.all.peak": 1966,
59
- "inactive_split.large_pool.allocated": 32685617,
60
- "inactive_split.large_pool.current": 43,
61
- "inactive_split.large_pool.freed": 32685574,
62
- "inactive_split.large_pool.peak": 141,
63
- "inactive_split.small_pool.allocated": 18643105,
64
- "inactive_split.small_pool.current": 82,
65
- "inactive_split.small_pool.freed": 18643023,
66
- "inactive_split.small_pool.peak": 1920,
67
- "inactive_split_bytes.all.allocated": 288630388610048,
68
- "inactive_split_bytes.all.current": 143554048,
69
- "inactive_split_bytes.all.freed": 288630245056000,
70
- "inactive_split_bytes.all.peak": 1893863936,
71
- "inactive_split_bytes.large_pool.allocated": 274820238239232,
72
- "inactive_split_bytes.large_pool.current": 121372672,
73
- "inactive_split_bytes.large_pool.freed": 274820116866560,
74
- "inactive_split_bytes.large_pool.peak": 1887404032,
75
- "inactive_split_bytes.small_pool.allocated": 13810150370816,
76
- "inactive_split_bytes.small_pool.current": 22181376,
77
- "inactive_split_bytes.small_pool.freed": 13810128189440,
78
- "inactive_split_bytes.small_pool.peak": 85829632,
79
- "max_split_size": -1,
80
- "num_alloc_retries": 0,
81
- "num_device_alloc": 1627,
82
- "num_device_free": 1322,
83
- "num_ooms": 0,
84
- "num_sync_all_streams": 7,
85
- "oversize_allocations.allocated": 0,
86
- "oversize_allocations.current": 0,
87
- "oversize_allocations.freed": 0,
88
- "oversize_allocations.peak": 0,
89
- "oversize_segments.allocated": 0,
90
- "oversize_segments.current": 0,
91
- "oversize_segments.freed": 0,
92
- "oversize_segments.peak": 0,
93
- "requested_bytes.all.allocated": 309521702594700,
94
- "requested_bytes.all.current": 800396976,
95
- "requested_bytes.all.freed": 309520902197724,
96
- "requested_bytes.all.peak": 4992933420,
97
- "requested_bytes.large_pool.allocated": 296138531900416,
98
- "requested_bytes.large_pool.current": 795392000,
99
- "requested_bytes.large_pool.freed": 296137736508416,
100
- "requested_bytes.large_pool.peak": 4975769600,
101
- "requested_bytes.small_pool.allocated": 13383170694284,
102
- "requested_bytes.small_pool.current": 5004976,
103
- "requested_bytes.small_pool.freed": 13383165689308,
104
- "requested_bytes.small_pool.peak": 119884008,
105
- "reserved_bytes.all.allocated": 46401585152,
106
- "reserved_bytes.all.current": 6345981952,
107
- "reserved_bytes.all.freed": 40055603200,
108
- "reserved_bytes.all.peak": 6752829440,
109
- "reserved_bytes.large_pool.allocated": 45720010752,
110
- "reserved_bytes.large_pool.current": 6218055680,
111
- "reserved_bytes.large_pool.freed": 39501955072,
112
- "reserved_bytes.large_pool.peak": 6687817728,
113
- "reserved_bytes.small_pool.allocated": 681574400,
114
- "reserved_bytes.small_pool.current": 127926272,
115
- "reserved_bytes.small_pool.freed": 553648128,
116
- "reserved_bytes.small_pool.peak": 127926272,
117
- "segment.all.allocated": 1627,
118
- "segment.all.current": 305,
119
- "segment.all.freed": 1322,
120
- "segment.all.peak": 306,
121
- "segment.large_pool.allocated": 1302,
122
- "segment.large_pool.current": 244,
123
- "segment.large_pool.freed": 1058,
124
- "segment.large_pool.peak": 245,
125
- "segment.small_pool.allocated": 325,
126
- "segment.small_pool.current": 61,
127
- "segment.small_pool.freed": 264,
128
- "segment.small_pool.peak": 61
129
- }
130
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:60850e6b893473f88e3421e4e2d6cf85cab0ecb9d65f19e813668eb10e67e80e
3
+ size 6219
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/special_tokens_map.json CHANGED
@@ -1,7 +1,3 @@
1
- {
2
- "cls_token": "[CLS]",
3
- "mask_token": "[MASK]",
4
- "pad_token": "[PAD]",
5
- "sep_token": "[SEP]",
6
- "unk_token": "[UNK]"
7
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b6d346be366a7d1d48332dbc9fdf3bf8960b5d879522b7799ddba59e76237ee3
3
+ size 125
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_16373/tokenizer_config.json CHANGED
@@ -1,55 +1,3 @@
1
- {
2
- "added_tokens_decoder": {
3
- "0": {
4
- "content": "[PAD]",
5
- "lstrip": false,
6
- "normalized": false,
7
- "rstrip": false,
8
- "single_word": false,
9
- "special": true
10
- },
11
- "100": {
12
- "content": "[UNK]",
13
- "lstrip": false,
14
- "normalized": false,
15
- "rstrip": false,
16
- "single_word": false,
17
- "special": true
18
- },
19
- "101": {
20
- "content": "[CLS]",
21
- "lstrip": false,
22
- "normalized": false,
23
- "rstrip": false,
24
- "single_word": false,
25
- "special": true
26
- },
27
- "102": {
28
- "content": "[SEP]",
29
- "lstrip": false,
30
- "normalized": false,
31
- "rstrip": false,
32
- "single_word": false,
33
- "special": true
34
- },
35
- "103": {
36
- "content": "[MASK]",
37
- "lstrip": false,
38
- "normalized": false,
39
- "rstrip": false,
40
- "single_word": false,
41
- "special": true
42
- }
43
- },
44
- "clean_up_tokenization_spaces": true,
45
- "cls_token": "[CLS]",
46
- "do_lower_case": true,
47
- "mask_token": "[MASK]",
48
- "model_max_length": 512,
49
- "pad_token": "[PAD]",
50
- "sep_token": "[SEP]",
51
- "strip_accents": null,
52
- "tokenize_chinese_chars": true,
53
- "tokenizer_class": "BertTokenizer",
54
- "unk_token": "[UNK]"
55
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f62a57a75856b93282501c92a86f62b169997c81e93cf6f75b7cc15d6285968e
3
+ size 1190
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/adapter_config.json CHANGED
@@ -1,32 +1,3 @@
1
- {
2
- "alpha_pattern": {},
3
- "auto_mapping": null,
4
- "base_model_name_or_path": "bert-base-uncased",
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": 16,
14
- "lora_dropout": 0.1,
15
- "megatron_config": null,
16
- "megatron_core": "megatron.core",
17
- "modules_to_save": [
18
- "classifier",
19
- "score"
20
- ],
21
- "peft_type": "LORA",
22
- "r": 8,
23
- "rank_pattern": {},
24
- "revision": null,
25
- "target_modules": [
26
- "value",
27
- "query"
28
- ],
29
- "task_type": "SEQ_CLS",
30
- "use_dora": false,
31
- "use_rslora": false
32
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81a29a631d083a248a7fe8b5acc0d972f1e62158c7e9d95bdbbeffc2630802a8
3
+ size 668
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/all_results.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.8638139995053178, "eval_f1": 0.8181638044914136}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ccf67cb852c7608079eea2c8cd7bc80f05b2eca0923e5509838443f6254c8e5
3
+ size 68
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/all_results_la_kron_all_homo_mc_corr_1000.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.8638139995053178, "eval_f1": 0.8181638044914136}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ccf67cb852c7608079eea2c8cd7bc80f05b2eca0923e5509838443f6254c8e5
3
+ size 68
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/all_results_val.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.8645320873986533, "eval_f1": 0.8204240746138152}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a6d52a14bd194a36b84555b9e5d162c67d7c2643e03a76d01b432cd519b7741a
3
+ size 68
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/eval_res.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/eval_res_la_kron_all_homo_mc_corr_1000.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/gpu_stats.json CHANGED
@@ -1,130 +1,3 @@
1
- {
2
- "memory_allocated": 459742208,
3
- "max_memory_allocated": 5480528896,
4
- "memory_reserved": 6840909824,
5
- "max_memory_reserved": 6840909824,
6
- "memory_stats": {
7
- "active.all.allocated": 49671300,
8
- "active.all.current": 367,
9
- "active.all.freed": 49670933,
10
- "active.all.peak": 627,
11
- "active.large_pool.allocated": 24072103,
12
- "active.large_pool.current": 77,
13
- "active.large_pool.freed": 24072026,
14
- "active.large_pool.peak": 250,
15
- "active.small_pool.allocated": 25599197,
16
- "active.small_pool.current": 290,
17
- "active.small_pool.freed": 25598907,
18
- "active.small_pool.peak": 538,
19
- "active_bytes.all.allocated": 143211789438464,
20
- "active_bytes.all.current": 459742208,
21
- "active_bytes.all.freed": 143211329696256,
22
- "active_bytes.all.peak": 5480528896,
23
- "active_bytes.large_pool.allocated": 131754566969344,
24
- "active_bytes.large_pool.current": 455606272,
25
- "active_bytes.large_pool.freed": 131754111363072,
26
- "active_bytes.large_pool.peak": 5461108736,
27
- "active_bytes.small_pool.allocated": 11457222469120,
28
- "active_bytes.small_pool.current": 4135936,
29
- "active_bytes.small_pool.freed": 11457218333184,
30
- "active_bytes.small_pool.peak": 98207744,
31
- "allocated_bytes.all.allocated": 143211789438464,
32
- "allocated_bytes.all.current": 459742208,
33
- "allocated_bytes.all.freed": 143211329696256,
34
- "allocated_bytes.all.peak": 5480528896,
35
- "allocated_bytes.large_pool.allocated": 131754566969344,
36
- "allocated_bytes.large_pool.current": 455606272,
37
- "allocated_bytes.large_pool.freed": 131754111363072,
38
- "allocated_bytes.large_pool.peak": 5461108736,
39
- "allocated_bytes.small_pool.allocated": 11457222469120,
40
- "allocated_bytes.small_pool.current": 4135936,
41
- "allocated_bytes.small_pool.freed": 11457218333184,
42
- "allocated_bytes.small_pool.peak": 98207744,
43
- "allocation.all.allocated": 49671300,
44
- "allocation.all.current": 367,
45
- "allocation.all.freed": 49670933,
46
- "allocation.all.peak": 627,
47
- "allocation.large_pool.allocated": 24072103,
48
- "allocation.large_pool.current": 77,
49
- "allocation.large_pool.freed": 24072026,
50
- "allocation.large_pool.peak": 250,
51
- "allocation.small_pool.allocated": 25599197,
52
- "allocation.small_pool.current": 290,
53
- "allocation.small_pool.freed": 25598907,
54
- "allocation.small_pool.peak": 538,
55
- "inactive_split.all.allocated": 26126377,
56
- "inactive_split.all.current": 33,
57
- "inactive_split.all.freed": 26126344,
58
- "inactive_split.all.peak": 143,
59
- "inactive_split.large_pool.allocated": 18969295,
60
- "inactive_split.large_pool.current": 19,
61
- "inactive_split.large_pool.freed": 18969276,
62
- "inactive_split.large_pool.peak": 127,
63
- "inactive_split.small_pool.allocated": 7157082,
64
- "inactive_split.small_pool.current": 14,
65
- "inactive_split.small_pool.freed": 7157068,
66
- "inactive_split.small_pool.peak": 70,
67
- "inactive_split_bytes.all.allocated": 129491747222528,
68
- "inactive_split_bytes.all.current": 64545792,
69
- "inactive_split_bytes.all.freed": 129491682676736,
70
- "inactive_split_bytes.all.peak": 1129547776,
71
- "inactive_split_bytes.large_pool.allocated": 117820220749824,
72
- "inactive_split_bytes.large_pool.current": 58195968,
73
- "inactive_split_bytes.large_pool.freed": 117820162553856,
74
- "inactive_split_bytes.large_pool.peak": 1127497728,
75
- "inactive_split_bytes.small_pool.allocated": 11671526472704,
76
- "inactive_split_bytes.small_pool.current": 6349824,
77
- "inactive_split_bytes.small_pool.freed": 11671520122880,
78
- "inactive_split_bytes.small_pool.peak": 43309568,
79
- "max_split_size": -1,
80
- "num_alloc_retries": 0,
81
- "num_device_alloc": 207,
82
- "num_device_free": 0,
83
- "num_ooms": 0,
84
- "num_sync_all_streams": 0,
85
- "oversize_allocations.allocated": 0,
86
- "oversize_allocations.current": 0,
87
- "oversize_allocations.freed": 0,
88
- "oversize_allocations.peak": 0,
89
- "oversize_segments.allocated": 0,
90
- "oversize_segments.current": 0,
91
- "oversize_segments.freed": 0,
92
- "oversize_segments.peak": 0,
93
- "requested_bytes.all.allocated": 135538096736142,
94
- "requested_bytes.all.current": 458603592,
95
- "requested_bytes.all.freed": 135537638132550,
96
- "requested_bytes.all.peak": 5476793900,
97
- "requested_bytes.large_pool.allocated": 124083469228256,
98
- "requested_bytes.large_pool.current": 454473728,
99
- "requested_bytes.large_pool.freed": 124083014754528,
100
- "requested_bytes.large_pool.peak": 5457379328,
101
- "requested_bytes.small_pool.allocated": 11454627507886,
102
- "requested_bytes.small_pool.current": 4129864,
103
- "requested_bytes.small_pool.freed": 11454623378022,
104
- "requested_bytes.small_pool.peak": 98173436,
105
- "reserved_bytes.all.allocated": 6840909824,
106
- "reserved_bytes.all.current": 6840909824,
107
- "reserved_bytes.all.freed": 0,
108
- "reserved_bytes.all.peak": 6840909824,
109
- "reserved_bytes.large_pool.allocated": 6729760768,
110
- "reserved_bytes.large_pool.current": 6729760768,
111
- "reserved_bytes.large_pool.freed": 0,
112
- "reserved_bytes.large_pool.peak": 6729760768,
113
- "reserved_bytes.small_pool.allocated": 111149056,
114
- "reserved_bytes.small_pool.current": 111149056,
115
- "reserved_bytes.small_pool.freed": 0,
116
- "reserved_bytes.small_pool.peak": 111149056,
117
- "segment.all.allocated": 207,
118
- "segment.all.current": 207,
119
- "segment.all.freed": 0,
120
- "segment.all.peak": 207,
121
- "segment.large_pool.allocated": 154,
122
- "segment.large_pool.current": 154,
123
- "segment.large_pool.freed": 0,
124
- "segment.large_pool.peak": 154,
125
- "segment.small_pool.allocated": 53,
126
- "segment.small_pool.current": 53,
127
- "segment.small_pool.freed": 0,
128
- "segment.small_pool.peak": 53
129
- }
130
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26b30c08de5574b5ef27faf351180ea0a5644d8627266c3e5329f25addb63944
3
+ size 6144
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/gpu_stats_la.json CHANGED
@@ -1,130 +1,3 @@
1
- {
2
- "memory_allocated": 808356352,
3
- "max_memory_allocated": 5015770624,
4
- "memory_reserved": 6350176256,
5
- "max_memory_reserved": 6752829440,
6
- "memory_stats": {
7
- "active.all.allocated": 122413512,
8
- "active.all.current": 10809,
9
- "active.all.freed": 122402703,
10
- "active.all.peak": 10985,
11
- "active.large_pool.allocated": 58610897,
12
- "active.large_pool.current": 222,
13
- "active.large_pool.freed": 58610675,
14
- "active.large_pool.peak": 360,
15
- "active.small_pool.allocated": 63802615,
16
- "active.small_pool.current": 10587,
17
- "active.small_pool.freed": 63792028,
18
- "active.small_pool.peak": 10750,
19
- "active_bytes.all.allocated": 391045242961408,
20
- "active_bytes.all.current": 808356352,
21
- "active_bytes.all.freed": 391044434605056,
22
- "active_bytes.all.peak": 5015770624,
23
- "active_bytes.large_pool.allocated": 374310385957888,
24
- "active_bytes.large_pool.current": 799080448,
25
- "active_bytes.large_pool.freed": 374309586877440,
26
- "active_bytes.large_pool.peak": 4998576128,
27
- "active_bytes.small_pool.allocated": 16734857003520,
28
- "active_bytes.small_pool.current": 9275904,
29
- "active_bytes.small_pool.freed": 16734847727616,
30
- "active_bytes.small_pool.peak": 124137472,
31
- "allocated_bytes.all.allocated": 391045242961408,
32
- "allocated_bytes.all.current": 808356352,
33
- "allocated_bytes.all.freed": 391044434605056,
34
- "allocated_bytes.all.peak": 5015770624,
35
- "allocated_bytes.large_pool.allocated": 374310385957888,
36
- "allocated_bytes.large_pool.current": 799080448,
37
- "allocated_bytes.large_pool.freed": 374309586877440,
38
- "allocated_bytes.large_pool.peak": 4998576128,
39
- "allocated_bytes.small_pool.allocated": 16734857003520,
40
- "allocated_bytes.small_pool.current": 9275904,
41
- "allocated_bytes.small_pool.freed": 16734847727616,
42
- "allocated_bytes.small_pool.peak": 124137472,
43
- "allocation.all.allocated": 122413512,
44
- "allocation.all.current": 10809,
45
- "allocation.all.freed": 122402703,
46
- "allocation.all.peak": 10985,
47
- "allocation.large_pool.allocated": 58610897,
48
- "allocation.large_pool.current": 222,
49
- "allocation.large_pool.freed": 58610675,
50
- "allocation.large_pool.peak": 360,
51
- "allocation.small_pool.allocated": 63802615,
52
- "allocation.small_pool.current": 10587,
53
- "allocation.small_pool.freed": 63792028,
54
- "allocation.small_pool.peak": 10750,
55
- "inactive_split.all.allocated": 64144648,
56
- "inactive_split.all.current": 110,
57
- "inactive_split.all.freed": 64144538,
58
- "inactive_split.all.peak": 1980,
59
- "inactive_split.large_pool.allocated": 40914502,
60
- "inactive_split.large_pool.current": 38,
61
- "inactive_split.large_pool.freed": 40914464,
62
- "inactive_split.large_pool.peak": 141,
63
- "inactive_split.small_pool.allocated": 23230146,
64
- "inactive_split.small_pool.current": 72,
65
- "inactive_split.small_pool.freed": 23230074,
66
- "inactive_split.small_pool.peak": 1934,
67
- "inactive_split_bytes.all.allocated": 360737834738688,
68
- "inactive_split_bytes.all.current": 120681984,
69
- "inactive_split_bytes.all.freed": 360737714056704,
70
- "inactive_split_bytes.all.peak": 1893863936,
71
- "inactive_split_bytes.large_pool.allocated": 343478534163456,
72
- "inactive_split_bytes.large_pool.current": 92209152,
73
- "inactive_split_bytes.large_pool.freed": 343478441954304,
74
- "inactive_split_bytes.large_pool.peak": 1887404032,
75
- "inactive_split_bytes.small_pool.allocated": 17259300575232,
76
- "inactive_split_bytes.small_pool.current": 28472832,
77
- "inactive_split_bytes.small_pool.freed": 17259272102400,
78
- "inactive_split_bytes.small_pool.peak": 85829632,
79
- "max_split_size": -1,
80
- "num_alloc_retries": 0,
81
- "num_device_alloc": 2025,
82
- "num_device_free": 1719,
83
- "num_ooms": 0,
84
- "num_sync_all_streams": 9,
85
- "oversize_allocations.allocated": 0,
86
- "oversize_allocations.current": 0,
87
- "oversize_allocations.freed": 0,
88
- "oversize_allocations.peak": 0,
89
- "oversize_segments.allocated": 0,
90
- "oversize_segments.current": 0,
91
- "oversize_segments.freed": 0,
92
- "oversize_segments.peak": 0,
93
- "requested_bytes.all.allocated": 386902123983535,
94
- "requested_bytes.all.current": 800396976,
95
- "requested_bytes.all.freed": 386901323586559,
96
- "requested_bytes.all.peak": 4992933420,
97
- "requested_bytes.large_pool.allocated": 370173160615680,
98
- "requested_bytes.large_pool.current": 795392000,
99
- "requested_bytes.large_pool.freed": 370172365223680,
100
- "requested_bytes.large_pool.peak": 4975769600,
101
- "requested_bytes.small_pool.allocated": 16728963367855,
102
- "requested_bytes.small_pool.current": 5004976,
103
- "requested_bytes.small_pool.freed": 16728958362879,
104
- "requested_bytes.small_pool.peak": 119884008,
105
- "reserved_bytes.all.allocated": 57814286336,
106
- "reserved_bytes.all.current": 6350176256,
107
- "reserved_bytes.all.freed": 51464110080,
108
- "reserved_bytes.all.peak": 6752829440,
109
- "reserved_bytes.large_pool.allocated": 56964939776,
110
- "reserved_bytes.large_pool.current": 6222249984,
111
- "reserved_bytes.large_pool.freed": 50742689792,
112
- "reserved_bytes.large_pool.peak": 6687817728,
113
- "reserved_bytes.small_pool.allocated": 849346560,
114
- "reserved_bytes.small_pool.current": 127926272,
115
- "reserved_bytes.small_pool.freed": 721420288,
116
- "reserved_bytes.small_pool.peak": 127926272,
117
- "segment.all.allocated": 2025,
118
- "segment.all.current": 306,
119
- "segment.all.freed": 1719,
120
- "segment.all.peak": 306,
121
- "segment.large_pool.allocated": 1620,
122
- "segment.large_pool.current": 245,
123
- "segment.large_pool.freed": 1375,
124
- "segment.large_pool.peak": 245,
125
- "segment.small_pool.allocated": 405,
126
- "segment.small_pool.current": 61,
127
- "segment.small_pool.freed": 344,
128
- "segment.small_pool.peak": 61
129
- }
130
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5fcd4a2ef2bb96bfd342772ad610ebbc7979c09468fa057fb1f25e0939ceb08b
3
+ size 6222
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/special_tokens_map.json CHANGED
@@ -1,7 +1,3 @@
1
- {
2
- "cls_token": "[CLS]",
3
- "mask_token": "[MASK]",
4
- "pad_token": "[PAD]",
5
- "sep_token": "[SEP]",
6
- "unk_token": "[UNK]"
7
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b6d346be366a7d1d48332dbc9fdf3bf8960b5d879522b7799ddba59e76237ee3
3
+ size 125
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_21831/tokenizer_config.json CHANGED
@@ -1,55 +1,3 @@
1
- {
2
- "added_tokens_decoder": {
3
- "0": {
4
- "content": "[PAD]",
5
- "lstrip": false,
6
- "normalized": false,
7
- "rstrip": false,
8
- "single_word": false,
9
- "special": true
10
- },
11
- "100": {
12
- "content": "[UNK]",
13
- "lstrip": false,
14
- "normalized": false,
15
- "rstrip": false,
16
- "single_word": false,
17
- "special": true
18
- },
19
- "101": {
20
- "content": "[CLS]",
21
- "lstrip": false,
22
- "normalized": false,
23
- "rstrip": false,
24
- "single_word": false,
25
- "special": true
26
- },
27
- "102": {
28
- "content": "[SEP]",
29
- "lstrip": false,
30
- "normalized": false,
31
- "rstrip": false,
32
- "single_word": false,
33
- "special": true
34
- },
35
- "103": {
36
- "content": "[MASK]",
37
- "lstrip": false,
38
- "normalized": false,
39
- "rstrip": false,
40
- "single_word": false,
41
- "special": true
42
- }
43
- },
44
- "clean_up_tokenization_spaces": true,
45
- "cls_token": "[CLS]",
46
- "do_lower_case": true,
47
- "mask_token": "[MASK]",
48
- "model_max_length": 512,
49
- "pad_token": "[PAD]",
50
- "sep_token": "[SEP]",
51
- "strip_accents": null,
52
- "tokenize_chinese_chars": true,
53
- "tokenizer_class": "BertTokenizer",
54
- "unk_token": "[UNK]"
55
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f62a57a75856b93282501c92a86f62b169997c81e93cf6f75b7cc15d6285968e
3
+ size 1190
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_27290/adapter_config.json CHANGED
@@ -1,32 +1,3 @@
1
- {
2
- "alpha_pattern": {},
3
- "auto_mapping": null,
4
- "base_model_name_or_path": "bert-base-uncased",
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": 16,
14
- "lora_dropout": 0.1,
15
- "megatron_config": null,
16
- "megatron_core": "megatron.core",
17
- "modules_to_save": [
18
- "classifier",
19
- "score"
20
- ],
21
- "peft_type": "LORA",
22
- "r": 8,
23
- "rank_pattern": {},
24
- "revision": null,
25
- "target_modules": [
26
- "value",
27
- "query"
28
- ],
29
- "task_type": "SEQ_CLS",
30
- "use_dora": false,
31
- "use_rslora": false
32
- }
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81a29a631d083a248a7fe8b5acc0d972f1e62158c7e9d95bdbbeffc2630802a8
3
+ size 668
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_27290/all_results.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.8645312886470443, "eval_f1": 0.8208198383878039}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc643f884b586fc34176513468477242eae688c5736d12a8fb5978b2db4685bd
3
+ size 68
outputs/qqp/bert-base-uncased_loratrain_val_8_16_0.1_0.0001_65/step_27290/all_results_la_kron_all_homo_mc_corr_1000.json CHANGED
@@ -1 +1,3 @@
1
- {"eval_accuracy": 0.8644818204303735, "eval_f1": 0.8207544083488729}
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dc5cc5223d349f01abe3cc3ffb4d04d5bede67d7003918482f511d6342c75c13
3
+ size 68