Hugging Face's logo

Datasets:
yimingzhang
/
allie-models

Modalities:
Text
Formats:
text
Size:
100K - 1M
Libraries:
Datasets
Dataset card Data Studio Files
xet
Community
1
yimingzhang commited on
Commit
29ae29c
·
verified ·
1 Parent(s): f05d31d

Upload folder using huggingface_hub

Browse files
Files changed (25) hide show
  1. ablations/half/best.pt +3 -0
  2. ablations/half/config.yaml +24 -0
  3. ablations/half/debug.txt +0 -0
  4. ablations/half/last.pt +3 -0
  5. ablations/half/log.txt +0 -0
  6. ablations/large/best.pt +3 -0
  7. ablations/large/config.yaml +23 -0
  8. ablations/large/debug.txt +0 -0
  9. ablations/large/last.pt +3 -0
  10. ablations/large/log.txt +0 -0
  11. ablations/short/best.pt +3 -0
  12. ablations/short/config.yaml +23 -0
  13. ablations/short/debug.txt +0 -0
  14. ablations/short/last.pt +3 -0
  15. ablations/short/log.txt +0 -0
  16. ablations/small/best.pt +3 -0
  17. ablations/small/config.yaml +23 -0
  18. ablations/small/debug.txt +0 -0
  19. ablations/small/last.pt +3 -0
  20. ablations/small/log.txt +0 -0
  21. medium/best.pt +3 -0
  22. medium/config.yaml +23 -0
  23. medium/debug.txt +0 -0
  24. medium/last.pt +3 -0
  25. medium/log.txt +0 -0
ablations/half/best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2adde01507aa606af47691ef052fd3f226247c3cdccf2316fe69de975e99bbaa
3
+ size 3665220334
ablations/half/config.yaml ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ trainer_config:
2
+ output_dir: models/ablations/half
3
+ wandb_run_name: half-medium
4
+ wandb_run_id: half-medium
5
+ dtype: bfloat16
6
+ torch_compile: true
7
+ batch_size: 32
8
+ gradient_accumulation_steps: 1
9
+ max_iters: 2000000
10
+ decay_iters: 2000000
11
+ eval_interval: 10000
12
+ log_interval: 50
13
+ save_interval: 10000
14
+ min_lr: 1.0e-05
15
+ weight_decay: 0.01
16
+ data_config:
17
+ tokenizer_config:
18
+ max_length: 512
19
+ train_file: data/lichess-2022-blitz-train/2022-half.bin
20
+ model_config:
21
+ base_model: gpt2-medium
22
+ use_control_token: true
23
+ use_regression_head: true
24
+ use_pretrained: true
ablations/half/debug.txt ADDED
The diff for this file is too large to render. See raw diff
 
ablations/half/last.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5b2214db047864da4c3e50edca126079e85545b21962207ab0165b50a176ac4
3
+ size 3665220334
ablations/half/log.txt ADDED
The diff for this file is too large to render. See raw diff
 
ablations/large/best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4805761d8652ca91080232d7ee0b662d4940122aafe6b6f4d2757a9e15111a20
3
+ size 8547639802
ablations/large/config.yaml ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ trainer_config:
2
+ output_dir: models/ablations/large
3
+ wandb_run_name: full-large
4
+ wandb_run_id: full-large
5
+ dtype: bfloat16
6
+ torch_compile: true
7
+ batch_size: 32
8
+ gradient_accumulation_steps: 1
9
+ max_iters: 2000000
10
+ decay_iters: 2000000
11
+ eval_interval: 10000
12
+ log_interval: 50
13
+ save_interval: 10000
14
+ min_lr: 1.0e-05
15
+ weight_decay: 0.01
16
+ data_config:
17
+ tokenizer_config:
18
+ max_length: 512
19
+ model_config:
20
+ base_model: gpt2-large
21
+ use_control_token: true
22
+ use_regression_head: true
23
+ use_pretrained: true
ablations/large/debug.txt ADDED
The diff for this file is too large to render. See raw diff
 
ablations/large/last.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0450a902d271a57dfef21d062c3aba7f624fef87f04c58b9e77acc17ee049d95
3
+ size 8547639802
ablations/large/log.txt ADDED
The diff for this file is too large to render. See raw diff
 
ablations/short/best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c015f5bb36e5ddc70e23cb0688a74a5e0dd27c6646d137b2fd3493561202a806
3
+ size 3665220334
ablations/short/config.yaml ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ trainer_config:
2
+ output_dir: models/ablations/short
3
+ wandb_run_name: short-medium
4
+ wandb_run_id: short-medium
5
+ dtype: bfloat16
6
+ torch_compile: true
7
+ batch_size: 32
8
+ gradient_accumulation_steps: 1
9
+ max_iters: 1000000
10
+ decay_iters: 1000000
11
+ eval_interval: 10000
12
+ log_interval: 50
13
+ save_interval: 10000
14
+ min_lr: 1.0e-05
15
+ weight_decay: 0.01
16
+ data_config:
17
+ tokenizer_config:
18
+ max_length: 512
19
+ model_config:
20
+ base_model: gpt2-medium
21
+ use_control_token: true
22
+ use_regression_head: true
23
+ use_pretrained: true
ablations/short/debug.txt ADDED
The diff for this file is too large to render. See raw diff
 
ablations/short/last.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e1514d264dc1e4e1daad95f890567ab84c49bf7f395c789ff5202f520c72c32f
3
+ size 3665220334
ablations/short/log.txt ADDED
The diff for this file is too large to render. See raw diff
 
ablations/small/best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ed031288282b6a83cd7290ff4afcf1ad0b8c3c3d90e009208d14732a03a08d4
3
+ size 1048711282
ablations/small/config.yaml ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ trainer_config:
2
+ output_dir: models/ablations/small
3
+ wandb_run_name: full-small
4
+ wandb_run_id: full-small
5
+ dtype: bfloat16
6
+ torch_compile: true
7
+ batch_size: 32
8
+ gradient_accumulation_steps: 1
9
+ max_iters: 2000000
10
+ decay_iters: 2000000
11
+ eval_interval: 10000
12
+ log_interval: 50
13
+ save_interval: 10000
14
+ min_lr: 1.0e-05
15
+ weight_decay: 0.01
16
+ data_config:
17
+ tokenizer_config:
18
+ max_length: 512
19
+ model_config:
20
+ base_model: gpt2
21
+ use_control_token: true
22
+ use_regression_head: true
23
+ use_pretrained: true
ablations/small/debug.txt ADDED
The diff for this file is too large to render. See raw diff
 
ablations/small/last.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b991d3b85830365a1bd7dff01a70be0e54b6854d40dadf19b97194888dde1b72
3
+ size 1048711282
ablations/small/log.txt ADDED
The diff for this file is too large to render. See raw diff
 
medium/best.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e64e8862ca630dd1b8cc0b3ff9d6dfa78014ea2c93247192b8f6e835ecf4f523
3
+ size 3665220334
medium/config.yaml ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ trainer_config:
2
+ output_dir: models/medium
3
+ wandb_run_name: full-medium
4
+ wandb_run_id: full-medium
5
+ dtype: bfloat16
6
+ torch_compile: true
7
+ batch_size: 32
8
+ gradient_accumulation_steps: 1
9
+ max_iters: 2000000
10
+ decay_iters: 2000000
11
+ eval_interval: 10000
12
+ log_interval: 50
13
+ save_interval: 10000
14
+ min_lr: 1.0e-05
15
+ weight_decay: 0.01
16
+ data_config:
17
+ tokenizer_config:
18
+ max_length: 512
19
+ model_config:
20
+ base_model: gpt2-medium
21
+ use_control_token: true
22
+ use_regression_head: true
23
+ use_pretrained: true
medium/debug.txt ADDED
The diff for this file is too large to render. See raw diff
 
medium/last.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bc29750d3392fb46f0f11af8d66b4be2dbe86592cae2a0712b23ca84d0342203
3
+ size 3665220334
medium/log.txt ADDED
The diff for this file is too large to render. See raw diff