Hugging Face's logo

noanabeshima
/
tiny_model

Model card Files Community
noanabeshima commited on
Commit
78c7da6
•
1 Parent(s): 22b3030

Upload folder using huggingface_hub

Browse files
Files changed (48) hide show
  1. mlp/M0_S-1_R1_P0.pt +3 -0
  2. mlp/M0_S-1_R1_P0_config.json +38 -0
  3. mlp/M0_S-2_R1_P0.pt +3 -0
  4. mlp/M0_S-2_R1_P0_config.json +38 -0
  5. mlp/M0_S-3_R1_P0.pt +3 -0
  6. mlp/M0_S-3_R1_P0_config.json +38 -0
  7. mlp/M0_S-4_R1_P0.pt +3 -0
  8. mlp/M0_S-4_R1_P0_config.json +38 -0
  9. mlp/M0_S-5_R1_P0.pt +3 -0
  10. mlp/M0_S-5_R1_P0_config.json +38 -0
  11. mlp/M0_S-6_R1_P0.pt +3 -0
  12. mlp/M0_S-6_R1_P0_config.json +38 -0
  13. mlp/M1_S-1_R1_P0.pt +3 -0
  14. mlp/M1_S-1_R1_P0_config.json +38 -0
  15. mlp/M1_S-2_R1_P0.pt +3 -0
  16. mlp/M1_S-2_R1_P0_config.json +38 -0
  17. mlp/M1_S-3_R1_P0.pt +3 -0
  18. mlp/M1_S-3_R1_P0_config.json +38 -0
  19. mlp/M1_S-4_R1_P0.pt +3 -0
  20. mlp/M1_S-4_R1_P0_config.json +38 -0
  21. mlp/M1_S-5_R1_P0.pt +3 -0
  22. mlp/M1_S-5_R1_P0_config.json +38 -0
  23. mlp/M1_S-6_R1_P0.pt +3 -0
  24. mlp/M1_S-6_R1_P0_config.json +38 -0
  25. mlp/M2_S-1_R1_P0.pt +3 -0
  26. mlp/M2_S-1_R1_P0_config.json +38 -0
  27. mlp/M2_S-2_R1_P0.pt +3 -0
  28. mlp/M2_S-2_R1_P0_config.json +38 -0
  29. mlp/M2_S-3_R1_P0.pt +3 -0
  30. mlp/M2_S-3_R1_P0_config.json +38 -0
  31. mlp/M2_S-4_R1_P0.pt +3 -0
  32. mlp/M2_S-4_R1_P0_config.json +38 -0
  33. mlp/M2_S-5_R1_P0.pt +3 -0
  34. mlp/M2_S-5_R1_P0_config.json +38 -0
  35. mlp/M2_S-6_R1_P0.pt +3 -0
  36. mlp/M2_S-6_R1_P0_config.json +38 -0
  37. mlp/M3_S-1_R1_P0.pt +3 -0
  38. mlp/M3_S-1_R1_P0_config.json +38 -0
  39. mlp/M3_S-2_R1_P0.pt +3 -0
  40. mlp/M3_S-2_R1_P0_config.json +38 -0
  41. mlp/M3_S-3_R1_P0.pt +3 -0
  42. mlp/M3_S-3_R1_P0_config.json +38 -0
  43. mlp/M3_S-4_R1_P0.pt +3 -0
  44. mlp/M3_S-4_R1_P0_config.json +38 -0
  45. mlp/M3_S-5_R1_P0.pt +3 -0
  46. mlp/M3_S-5_R1_P0_config.json +38 -0
  47. mlp/M3_S-6_R1_P0.pt +3 -0
  48. mlp/M3_S-6_R1_P0_config.json +38 -0
mlp/M0_S-1_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bcda59b9785874b2b9eeae92c397e7425610983d1f016ecb9e3a44fb758804f0
3
+ size 153705080
mlp/M0_S-1_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 0,
20
+ "l1_exp": -1,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo0_S-1_R1_P0"
38
+ }
mlp/M0_S-2_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aebed9ea9ffe0ecbfeeffdffe5c1be42542af6d034ddf64505a9609533cea8b9
3
+ size 153705080
mlp/M0_S-2_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 0,
20
+ "l1_exp": -2,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo0_S-2_R1_P0"
38
+ }
mlp/M0_S-3_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2106f5872e56bf8ed8038860c28e5edb3fb8769b19459b287f537be85d1504ee
3
+ size 153705080
mlp/M0_S-3_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 0,
20
+ "l1_exp": -3,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo0_S-3_R1_P0"
38
+ }
mlp/M0_S-4_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3be61e61581bc5b130727e693446be29bf02677eb1f03f04a8612121df2af6e
3
+ size 153705080
mlp/M0_S-4_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 0,
20
+ "l1_exp": -4,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo0_S-4_R1_P0"
38
+ }
mlp/M0_S-5_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e982b4d7aceacbfbc992b601c6717e9b0968c50125b44ae27d1e09069ed3fa13
3
+ size 153705080
mlp/M0_S-5_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 0,
20
+ "l1_exp": -5,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo0_S-5_R1_P0"
38
+ }
mlp/M0_S-6_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:695badca456d4db5a317ce8470e33f7bab393cf7ca5d1207ffcab5b4bcd259db
3
+ size 153705080
mlp/M0_S-6_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 0,
20
+ "l1_exp": -6,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo0_S-6_R1_P0"
38
+ }
mlp/M1_S-1_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b3a48f741f8318af05e9d4141c5bae377b1af8a4fe79e2d459bc692dbf4b8700
3
+ size 153705080
mlp/M1_S-1_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 1,
20
+ "l1_exp": -1,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo1_S-1_R1_P0"
38
+ }
mlp/M1_S-2_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e4630e009c4d2c3cc9ccb563744c679e6726138f27beda74ba1fac2a47768991
3
+ size 153705080
mlp/M1_S-2_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 1,
20
+ "l1_exp": -2,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo1_S-2_R1_P0"
38
+ }
mlp/M1_S-3_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8c430af5b000c58973a893ad8ec61c5bbdc3d6dd9cc1d1b01a949403b29cefcc
3
+ size 153705080
mlp/M1_S-3_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 1,
20
+ "l1_exp": -3,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo1_S-3_R1_P0"
38
+ }
mlp/M1_S-4_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e8787e12c7fe5edf2a56a87b1c3df9acf4495df59a78c95cceebb86dc7ccf87
3
+ size 153705080
mlp/M1_S-4_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 1,
20
+ "l1_exp": -4,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo1_S-4_R1_P0"
38
+ }
mlp/M1_S-5_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f0f5e9f230869bbaf44eac1550a099ea58c1d86ea8df73c3cd9f316dd5f1650
3
+ size 153705080
mlp/M1_S-5_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 1,
20
+ "l1_exp": -5,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo1_S-5_R1_P0"
38
+ }
mlp/M1_S-6_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61d6441e8dbc5b092dc2e11bd525a13eee4d72111e0b0173deb3cb89b6e18075
3
+ size 153705080
mlp/M1_S-6_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 1,
20
+ "l1_exp": -6,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo1_S-6_R1_P0"
38
+ }
mlp/M2_S-1_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0b242e709f9c64bbe0aa82a8cc904b25cb1aeb7388ea89529b2685f8a528bf37
3
+ size 153705080
mlp/M2_S-1_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 2,
20
+ "l1_exp": -1,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo2_S-1_R1_P0"
38
+ }
mlp/M2_S-2_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:25eec217d1c72e4d83180698a252235692a536bbe598c68e97a0fa5140a0ebdd
3
+ size 153705080
mlp/M2_S-2_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 2,
20
+ "l1_exp": -2,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo2_S-2_R1_P0"
38
+ }
mlp/M2_S-3_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:71a43b0400729a9bcde458b0629c32feacb53eaa9622d476334d65842a6b9833
3
+ size 153705080
mlp/M2_S-3_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 2,
20
+ "l1_exp": -3,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo2_S-3_R1_P0"
38
+ }
mlp/M2_S-4_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d98cefdab793343dd1a8be86c606414af42262f651d63fcb13096efff490356e
3
+ size 153705080
mlp/M2_S-4_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 2,
20
+ "l1_exp": -4,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo2_S-4_R1_P0"
38
+ }
mlp/M2_S-5_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06caf3e157a9254766c09eeaf0dab9c3c69d7947be202b62a30db79dd7a74d7b
3
+ size 153705080
mlp/M2_S-5_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 2,
20
+ "l1_exp": -5,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo2_S-5_R1_P0"
38
+ }
mlp/M2_S-6_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e9a47f28e4720773cf93b81565d38fc85c22e4574e3dd1f44095b1d55648ae16
3
+ size 153705080
mlp/M2_S-6_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 2,
20
+ "l1_exp": -6,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo2_S-6_R1_P0"
38
+ }
mlp/M3_S-1_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b90f2559cfeb894ec51fa727a7c87fabf0d1262f2b3e3814cc4acbe09a947bba
3
+ size 153705080
mlp/M3_S-1_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 3,
20
+ "l1_exp": -1,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo3_S-1_R1_P0"
38
+ }
mlp/M3_S-2_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fa88e4135dea096926082aab18328e5ec40330c45c1b12c59cc365a28a473000
3
+ size 153705080
mlp/M3_S-2_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 3,
20
+ "l1_exp": -2,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo3_S-2_R1_P0"
38
+ }
mlp/M3_S-3_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7b8f9ce9ac97dfc06c90b27082e7756766234835cabcc79cf7330714fa3972db
3
+ size 153705080
mlp/M3_S-3_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 3,
20
+ "l1_exp": -3,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo3_S-3_R1_P0"
38
+ }
mlp/M3_S-4_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6ddc41001f648e468699a0010a9647f9eaece946e1ae265ed1c8af1772249732
3
+ size 153705080
mlp/M3_S-4_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 3,
20
+ "l1_exp": -4,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo3_S-4_R1_P0"
38
+ }
mlp/M3_S-5_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:83cb301d90dac4a9d06d5c99ca5b62fe40b7492d8744ce5f49caadd316551956
3
+ size 153705080
mlp/M3_S-5_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 3,
20
+ "l1_exp": -5,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo3_S-5_R1_P0"
38
+ }
mlp/M3_S-6_R1_P0.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c7fb31406999c1274c69e37022648b502b704c7fcb9d6fd805c84f4ce779ca71
3
+ size 153705080
mlp/M3_S-6_R1_P0_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "n_features": 25000,
3
+ "d_model": 768,
4
+ "lr_exp": -10,
5
+ "disable_comet": false,
6
+ "per_neuron_reinit_interval": 0,
7
+ "reservoir_time_discount": 0.995,
8
+ "reinit_interval": 800,
9
+ "max_reinit_neurons": 5000,
10
+ "reservoir_size": 5000,
11
+ "n_piles": 292,
12
+ "log_interval": 200,
13
+ "reinit_input_norm": "target_scaled",
14
+ "reinit_input": "x",
15
+ "reinit_norm_alpha": 0.3,
16
+ "data_loc": "mlp_data",
17
+ "reinit_threshold": -6,
18
+ "scheduler": "wsd",
19
+ "layer_idx": 3,
20
+ "l1_exp": -6,
21
+ "neuron_reinit_percent": 0.85,
22
+ "beta1": 1,
23
+ "beta2": 4,
24
+ "reinit_target": "error",
25
+ "sparse_adam": false,
26
+ "run_template": "Mo{layer_idx}_S{l1_exp}_R{l1_ratio}_P{l1_p}",
27
+ "project_name": "mlp_out_test",
28
+ "decoder_bias": true,
29
+ "l1_beta": 0.99,
30
+ "alt_sparsity_loss": "log",
31
+ "l1_ratio": 1,
32
+ "l1_p": 0,
33
+ "optimizer": "sparse_adam",
34
+ "model_type": "mlp_out",
35
+ "adam_beta1": 0.5,
36
+ "adam_beta2": 0.9375,
37
+ "run_name": "Mo3_S-6_R1_P0"
38
+ }