Hugging Face's logo

GOVINDFROM
/
parallel-consistent-masked-diffusion

masked-diffusion
discrete-diffusion
constraint-satisfaction
sudoku
recurrent-transformer
from-scratch
Model card Files
xet
 Community
GOVINDFROM commited on
Commit
3366706
·
verified ·
1 Parent(s): 5135496

Automated push from Zaratan

Browse files
Files changed (22) hide show
  1. results/g4-agree-gam0.5-dw1-s0.json +49 -0
  2. results/g4-agree-gam0.5-dw1-s1.json +49 -0
  3. results/g4-agree-gam2.0-dw0-s0.json +49 -0
  4. results/g4-agree-gam2.0-dw0-s1.json +49 -0
  5. results/g4-agree-gam2.0-dw1-s0.json +49 -0
  6. results/g4-agree-gam2.0-dw1-s1.json +49 -0
  7. results/g4-agree-gam8.0-dw0-s0.json +49 -0
  8. results/g4-agree-gam8.0-dw0-s1.json +49 -0
  9. results/g4-agree-gam8.0-dw1-s0.json +49 -0
  10. results/g4-agree-gam8.0-dw1-s1.json +49 -0
  11. results/g4-parity-gam0.5-dw0-s0.json +49 -0
  12. results/g4-parity-gam0.5-dw0-s1.json +49 -0
  13. results/g4-parity-gam0.5-dw1-s0.json +49 -0
  14. results/g4-parity-gam0.5-dw1-s1.json +49 -0
  15. results/g4-parity-gam2.0-dw0-s0.json +49 -0
  16. results/g4-parity-gam2.0-dw0-s1.json +49 -0
  17. results/g4-parity-gam2.0-dw1-s0.json +49 -0
  18. results/g4-parity-gam2.0-dw1-s1.json +49 -0
  19. results/g4-parity-gam8.0-dw0-s0.json +49 -0
  20. results/g4-parity-gam8.0-dw0-s1.json +49 -0
  21. results/g4-parity-gam8.0-dw1-s0.json +49 -0
  22. results/g4-parity-gam8.0-dw1-s1.json +49 -0
results/g4-agree-gam0.5-dw1-s0.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 1.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 0.5,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "agree",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 0,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-agree-gam0.5-dw1-s0"
41
+ },
42
+ "minutes": 8.66148560444514,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-agree-gam0.5-dw1-s1.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 1.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 0.5,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "agree",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 1,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-agree-gam0.5-dw1-s1"
41
+ },
42
+ "minutes": 8.824847221374512,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-agree-gam2.0-dw0-s0.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 2.0,
17
+ "dep_weighted": 0,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "agree",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 0,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-agree-gam2.0-dw0-s0"
41
+ },
42
+ "minutes": 6.714557294050852,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-agree-gam2.0-dw0-s1.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 2.0,
17
+ "dep_weighted": 0,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "agree",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 1,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-agree-gam2.0-dw0-s1"
41
+ },
42
+ "minutes": 6.607120327154795,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-agree-gam2.0-dw1-s0.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 2.0,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "agree",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 0,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-agree-gam2.0-dw1-s0"
41
+ },
42
+ "minutes": 8.701955270767211,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-agree-gam2.0-dw1-s1.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 2.0,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "agree",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 1,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-agree-gam2.0-dw1-s1"
41
+ },
42
+ "minutes": 8.732966057459514,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-agree-gam8.0-dw0-s0.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 8.0,
17
+ "dep_weighted": 0,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "agree",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 0,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-agree-gam8.0-dw0-s0"
41
+ },
42
+ "minutes": 6.586321103572845,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-agree-gam8.0-dw0-s1.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 8.0,
17
+ "dep_weighted": 0,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "agree",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 1,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-agree-gam8.0-dw0-s1"
41
+ },
42
+ "minutes": 6.608458415667216,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-agree-gam8.0-dw1-s0.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 8.0,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "agree",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 0,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-agree-gam8.0-dw1-s0"
41
+ },
42
+ "minutes": 8.838382252057393,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-agree-gam8.0-dw1-s1.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 8.0,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "agree",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 1,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-agree-gam8.0-dw1-s1"
41
+ },
42
+ "minutes": 8.63929298321406,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam0.5-dw0-s0.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 1.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 0.5,
17
+ "dep_weighted": 0,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 0,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam0.5-dw0-s0"
41
+ },
42
+ "minutes": 6.648975769678752,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam0.5-dw0-s1.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 0.5,
17
+ "dep_weighted": 0,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 1,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam0.5-dw0-s1"
41
+ },
42
+ "minutes": 6.6801059484481815,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam0.5-dw1-s0.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 1.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 0.5,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 0,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam0.5-dw1-s0"
41
+ },
42
+ "minutes": 8.79375748236974,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam0.5-dw1-s1.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 0.5,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 1,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam0.5-dw1-s1"
41
+ },
42
+ "minutes": 8.721678797403971,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam2.0-dw0-s0.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 2.0,
17
+ "dep_weighted": 0,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 0,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam2.0-dw0-s0"
41
+ },
42
+ "minutes": 6.624940903981527,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam2.0-dw0-s1.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 2.0,
17
+ "dep_weighted": 0,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 1,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam2.0-dw0-s1"
41
+ },
42
+ "minutes": 6.710831908384959,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam2.0-dw1-s0.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 1.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 2.0,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 0,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam2.0-dw1-s0"
41
+ },
42
+ "minutes": 8.75299590031306,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam2.0-dw1-s1.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 1.0,
10
+ "128": 1.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 2.0,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 1,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam2.0-dw1-s1"
41
+ },
42
+ "minutes": 8.76179831425349,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam8.0-dw0-s0.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 8.0,
17
+ "dep_weighted": 0,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 0,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam8.0-dw0-s0"
41
+ },
42
+ "minutes": 6.6732826113700865,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam8.0-dw0-s1.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 8.0,
17
+ "dep_weighted": 0,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 1,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam8.0-dw0-s1"
41
+ },
42
+ "minutes": 6.690542948246002,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam8.0-dw1-s0.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 8.0,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 0,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam8.0-dw1-s0"
41
+ },
42
+ "minutes": 8.767544798056285,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }
results/g4-parity-gam8.0-dw1-s1.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "accuracy_by_passes": {
3
+ "1": 0.0,
4
+ "2": 0.0,
5
+ "4": 0.0,
6
+ "8": 0.0,
7
+ "16": 0.0,
8
+ "32": 0.0,
9
+ "64": 0.0,
10
+ "128": 0.0
11
+ },
12
+ "args": {
13
+ "objective": "coord",
14
+ "alpha": 0.5,
15
+ "beta": 0.0,
16
+ "gamma": 8.0,
17
+ "dep_weighted": 1,
18
+ "elbo_w": 1.0,
19
+ "group": 4,
20
+ "coord_every": 4,
21
+ "k_frac": 0.25,
22
+ "correct_weight": 1.0,
23
+ "ramp_frac": 0.2,
24
+ "task": "parity",
25
+ "coupling": 4,
26
+ "seq_len": 128,
27
+ "n_train": 200000,
28
+ "n_eval": 500,
29
+ "d": 384,
30
+ "layers": 6,
31
+ "heads": 6,
32
+ "pe": "rope",
33
+ "bs": 64,
34
+ "steps": 20000,
35
+ "lr": 0.0003,
36
+ "warmup": 500,
37
+ "seed": 1,
38
+ "data_root": "data",
39
+ "eval_bs": 100,
40
+ "out": "/scratch/zt1/project/msml612/user/govind02/pcmd/runs/g4-parity-gam8.0-dw1-s1"
41
+ },
42
+ "minutes": 8.782358932495118,
43
+ "compute": {
44
+ "params": 10660608,
45
+ "train_tokens": 163840000,
46
+ "train_flops_approx": 10479804088320000,
47
+ "fwd_per_step": 1
48
+ }
49
+ }