Hugging Face's logo

Lab608
/
Paratope-GLoMe

TensorBoard
Safetensors
Model card Files
xet
 Metrics Community
Jokey14 commited on
Commit
6f6a044
·
1 Parent(s): f7a3fe8

first_commit

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/config.json +52 -0
  2. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/model.safetensors +3 -0
  3. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/optimizer.pt +3 -0
  4. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/rng_state.pth +3 -0
  5. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/scaler.pt +3 -0
  6. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/scheduler.pt +3 -0
  7. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/trainer_state.json +0 -0
  8. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/training_args.bin +3 -0
  9. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/config.json +52 -0
  10. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/model.safetensors +3 -0
  11. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/optimizer.pt +3 -0
  12. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/rng_state.pth +3 -0
  13. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/scaler.pt +3 -0
  14. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/scheduler.pt +3 -0
  15. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/trainer_state.json +0 -0
  16. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/training_args.bin +3 -0
  17. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/config.json +52 -0
  18. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/model.safetensors +3 -0
  19. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/optimizer.pt +3 -0
  20. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/rng_state.pth +3 -0
  21. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/scaler.pt +3 -0
  22. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/scheduler.pt +3 -0
  23. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/trainer_state.json +0 -0
  24. finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/training_args.bin +3 -0
  25. finetune_glome_nano_site_dice_0_10_1_9_tiny/logs/events.out.tfevents.1772419140.amax.1813661.0 +3 -0
  26. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/config.json +52 -0
  27. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/model.safetensors +3 -0
  28. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/optimizer.pt +3 -0
  29. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/rng_state.pth +3 -0
  30. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/scaler.pt +3 -0
  31. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/scheduler.pt +3 -0
  32. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/trainer_state.json +0 -0
  33. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/training_args.bin +3 -0
  34. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/config.json +52 -0
  35. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/model.safetensors +3 -0
  36. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/optimizer.pt +3 -0
  37. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/rng_state.pth +3 -0
  38. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/scaler.pt +3 -0
  39. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/scheduler.pt +3 -0
  40. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/trainer_state.json +0 -0
  41. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/training_args.bin +3 -0
  42. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/config.json +52 -0
  43. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/model.safetensors +3 -0
  44. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/optimizer.pt +3 -0
  45. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/rng_state.pth +3 -0
  46. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/scaler.pt +3 -0
  47. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/scheduler.pt +3 -0
  48. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/trainer_state.json +0 -0
  49. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/training_args.bin +3 -0
  50. finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/logs/events.out.tfevents.1772413682.amax.578393.0 +3 -0
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "GloMeModelForTokenClassification"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.1,
7
+ "bos_token_id": 28,
8
+ "cdr_weight": 0.0,
9
+ "class_weights": [
10
+ 0.1,
11
+ 0.9
12
+ ],
13
+ "classifier_activation": "gelu",
14
+ "classifier_bias": false,
15
+ "classifier_dropout": 0.1,
16
+ "classifier_pooling": "cls",
17
+ "cls_token_id": 28,
18
+ "compress_block_size": 16,
19
+ "compress_block_sliding_stride": 16,
20
+ "decoder_bias": true,
21
+ "dice_weight": 0.1,
22
+ "embedding_dropout": 0.1,
23
+ "eos_token_id": 29,
24
+ "hidden_activation": "gelu",
25
+ "hidden_size": 320,
26
+ "inner_rank": 32,
27
+ "intermediate_size": 1280,
28
+ "kv_heads": 10,
29
+ "mask_token_id": 31,
30
+ "mlp_bias": false,
31
+ "mlp_dropout": 0.1,
32
+ "model_size": "tiny",
33
+ "model_type": "glome",
34
+ "norm_bias": false,
35
+ "norm_eps": 1e-05,
36
+ "num_attention_heads": 20,
37
+ "num_hidden_layers": 6,
38
+ "num_selected_blocks": 8,
39
+ "num_slots": 64,
40
+ "pad_token_id": 30,
41
+ "reference_compile": null,
42
+ "selection_block_size": 16,
43
+ "sep_token_id": 29,
44
+ "sliding_window_size": 0,
45
+ "tie_word_embeddings": false,
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.52.3",
48
+ "unk_token_id": 27,
49
+ "use_glome": true,
50
+ "use_nsa": true,
51
+ "vocab_size": 36
52
+ }
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30c034c905b9a0eea1b7cc3c4c77fbdf290e876b5a00c00f09397921c11013c3
3
+ size 61385376
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:725b67bba8a7c8b0286053847c3081378277810c7a29c183c24fbe0871520125
3
+ size 122881658
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:85c805248eed07d14f2eedb8b706397ce25315fc3b96df0bb8479411b871bd7b
3
+ size 14244
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b30172cf14f5dbe00280d63e36224a9f28dc7a0e8b38a74ceb5eb284e84da363
3
+ size 988
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a0f85b1b4566ce3af61937d9c59a498414651a1e667865d284be6f03114163df
3
+ size 1064
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8856/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:499ba8a39afec206dd7194e2d216bf0be2633330bfcda3d90a12ddcbc04cdaca
3
+ size 5368
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "GloMeModelForTokenClassification"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.1,
7
+ "bos_token_id": 28,
8
+ "cdr_weight": 0.0,
9
+ "class_weights": [
10
+ 0.1,
11
+ 0.9
12
+ ],
13
+ "classifier_activation": "gelu",
14
+ "classifier_bias": false,
15
+ "classifier_dropout": 0.1,
16
+ "classifier_pooling": "cls",
17
+ "cls_token_id": 28,
18
+ "compress_block_size": 16,
19
+ "compress_block_sliding_stride": 16,
20
+ "decoder_bias": true,
21
+ "dice_weight": 0.1,
22
+ "embedding_dropout": 0.1,
23
+ "eos_token_id": 29,
24
+ "hidden_activation": "gelu",
25
+ "hidden_size": 320,
26
+ "inner_rank": 32,
27
+ "intermediate_size": 1280,
28
+ "kv_heads": 10,
29
+ "mask_token_id": 31,
30
+ "mlp_bias": false,
31
+ "mlp_dropout": 0.1,
32
+ "model_size": "tiny",
33
+ "model_type": "glome",
34
+ "norm_bias": false,
35
+ "norm_eps": 1e-05,
36
+ "num_attention_heads": 20,
37
+ "num_hidden_layers": 6,
38
+ "num_selected_blocks": 8,
39
+ "num_slots": 64,
40
+ "pad_token_id": 30,
41
+ "reference_compile": null,
42
+ "selection_block_size": 16,
43
+ "sep_token_id": 29,
44
+ "sliding_window_size": 0,
45
+ "tie_word_embeddings": false,
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.52.3",
48
+ "unk_token_id": 27,
49
+ "use_glome": true,
50
+ "use_nsa": true,
51
+ "vocab_size": 36
52
+ }
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bd6b4e359b5d6cb87fd11e6785513fb93cd33e2621e530c3d5d0372c61965079
3
+ size 61385376
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fd7657ee28cb041646b9be46a93baffab2afaf1966b7a9f81a83df51bd473eff
3
+ size 122881658
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae123764811139e2cb14efaf37527d6fd81efe423a4a92fa24d82e5d54b5501c
3
+ size 14244
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b30172cf14f5dbe00280d63e36224a9f28dc7a0e8b38a74ceb5eb284e84da363
3
+ size 988
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b2cddf9c2f8aa06961889d4d18e579770b9648558e3b3dd47f47b9be5707aaa0
3
+ size 1064
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-8982/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:499ba8a39afec206dd7194e2d216bf0be2633330bfcda3d90a12ddcbc04cdaca
3
+ size 5368
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "GloMeModelForTokenClassification"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.1,
7
+ "bos_token_id": 28,
8
+ "cdr_weight": 0.0,
9
+ "class_weights": [
10
+ 0.1,
11
+ 0.9
12
+ ],
13
+ "classifier_activation": "gelu",
14
+ "classifier_bias": false,
15
+ "classifier_dropout": 0.1,
16
+ "classifier_pooling": "cls",
17
+ "cls_token_id": 28,
18
+ "compress_block_size": 16,
19
+ "compress_block_sliding_stride": 16,
20
+ "decoder_bias": true,
21
+ "dice_weight": 0.1,
22
+ "embedding_dropout": 0.1,
23
+ "eos_token_id": 29,
24
+ "hidden_activation": "gelu",
25
+ "hidden_size": 320,
26
+ "inner_rank": 32,
27
+ "intermediate_size": 1280,
28
+ "kv_heads": 10,
29
+ "mask_token_id": 31,
30
+ "mlp_bias": false,
31
+ "mlp_dropout": 0.1,
32
+ "model_size": "tiny",
33
+ "model_type": "glome",
34
+ "norm_bias": false,
35
+ "norm_eps": 1e-05,
36
+ "num_attention_heads": 20,
37
+ "num_hidden_layers": 6,
38
+ "num_selected_blocks": 8,
39
+ "num_slots": 64,
40
+ "pad_token_id": 30,
41
+ "reference_compile": null,
42
+ "selection_block_size": 16,
43
+ "sep_token_id": 29,
44
+ "sliding_window_size": 0,
45
+ "tie_word_embeddings": false,
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.52.3",
48
+ "unk_token_id": 27,
49
+ "use_glome": true,
50
+ "use_nsa": true,
51
+ "vocab_size": 36
52
+ }
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:423cf42c264d2567fd289ade1e6c1f0e506d54000000d68db8e861743d892f63
3
+ size 61385376
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:84e048a6921928cf224d4195130617e539f1adeb7a4f1611cbdb14ccefd71846
3
+ size 122881658
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:09c0fb83ad832f6cdeeca05a528dfc9d2b3afad5151c7ac2d59232a6247360ef
3
+ size 14244
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b30172cf14f5dbe00280d63e36224a9f28dc7a0e8b38a74ceb5eb284e84da363
3
+ size 988
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d23023944b37031b4b7a09c8ee9683bdb1128842a447d5968d6407368b043a9
3
+ size 1064
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
finetune_glome_nano_site_dice_0_10_1_9_tiny/checkpoint-9000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:499ba8a39afec206dd7194e2d216bf0be2633330bfcda3d90a12ddcbc04cdaca
3
+ size 5368
finetune_glome_nano_site_dice_0_10_1_9_tiny/logs/events.out.tfevents.1772419140.amax.1813661.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31b4f0499f3c83af64510e723eab0e1b174cbd078260871728cbc3136654212c
3
+ size 582065
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "GloMeModelForTokenClassification"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.1,
7
+ "bos_token_id": 28,
8
+ "cdr_weight": 0.0,
9
+ "class_weights": [
10
+ 0.1,
11
+ 0.9
12
+ ],
13
+ "classifier_activation": "gelu",
14
+ "classifier_bias": false,
15
+ "classifier_dropout": 0.1,
16
+ "classifier_pooling": "cls",
17
+ "cls_token_id": 28,
18
+ "compress_block_size": 16,
19
+ "compress_block_sliding_stride": 16,
20
+ "decoder_bias": true,
21
+ "dice_weight": 0.1,
22
+ "embedding_dropout": 0.1,
23
+ "eos_token_id": 29,
24
+ "hidden_activation": "gelu",
25
+ "hidden_size": 320,
26
+ "inner_rank": 32,
27
+ "intermediate_size": 1280,
28
+ "kv_heads": 10,
29
+ "mask_token_id": 31,
30
+ "mlp_bias": false,
31
+ "mlp_dropout": 0.1,
32
+ "model_size": "tiny",
33
+ "model_type": "glome",
34
+ "norm_bias": false,
35
+ "norm_eps": 1e-05,
36
+ "num_attention_heads": 20,
37
+ "num_hidden_layers": 6,
38
+ "num_selected_blocks": 8,
39
+ "num_slots": 64,
40
+ "pad_token_id": 30,
41
+ "reference_compile": null,
42
+ "selection_block_size": 16,
43
+ "sep_token_id": 29,
44
+ "sliding_window_size": 0,
45
+ "tie_word_embeddings": false,
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.52.3",
48
+ "unk_token_id": 27,
49
+ "use_glome": true,
50
+ "use_nsa": true,
51
+ "vocab_size": 36
52
+ }
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c7a6c9e7bebd86d828e33c602f601585df15ab364132986ff74a2c37ee1a20b5
3
+ size 61385376
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cedfd25f8179cb06ee2cdef74c23b5465b2feb98a5d90c0c18310779a53873fa
3
+ size 122881658
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4b5e0b21ceddfc06d6dfc155153f6b1cb2682aebf418b0515e022b082bfcbcf5
3
+ size 14244
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b30172cf14f5dbe00280d63e36224a9f28dc7a0e8b38a74ceb5eb284e84da363
3
+ size 988
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:231ac688ea3b75d272ed7ef26f577919491d334410210be20bdcc895a1659ccb
3
+ size 1064
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12818/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2fce922531bcc60b40ec3cfe0214120623a297c18ab37c3a2e94007f715374c7
3
+ size 5368
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "GloMeModelForTokenClassification"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.1,
7
+ "bos_token_id": 28,
8
+ "cdr_weight": 0.0,
9
+ "class_weights": [
10
+ 0.1,
11
+ 0.9
12
+ ],
13
+ "classifier_activation": "gelu",
14
+ "classifier_bias": false,
15
+ "classifier_dropout": 0.1,
16
+ "classifier_pooling": "cls",
17
+ "cls_token_id": 28,
18
+ "compress_block_size": 16,
19
+ "compress_block_sliding_stride": 16,
20
+ "decoder_bias": true,
21
+ "dice_weight": 0.1,
22
+ "embedding_dropout": 0.1,
23
+ "eos_token_id": 29,
24
+ "hidden_activation": "gelu",
25
+ "hidden_size": 320,
26
+ "inner_rank": 32,
27
+ "intermediate_size": 1280,
28
+ "kv_heads": 10,
29
+ "mask_token_id": 31,
30
+ "mlp_bias": false,
31
+ "mlp_dropout": 0.1,
32
+ "model_size": "tiny",
33
+ "model_type": "glome",
34
+ "norm_bias": false,
35
+ "norm_eps": 1e-05,
36
+ "num_attention_heads": 20,
37
+ "num_hidden_layers": 6,
38
+ "num_selected_blocks": 8,
39
+ "num_slots": 64,
40
+ "pad_token_id": 30,
41
+ "reference_compile": null,
42
+ "selection_block_size": 16,
43
+ "sep_token_id": 29,
44
+ "sliding_window_size": 0,
45
+ "tie_word_embeddings": false,
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.52.3",
48
+ "unk_token_id": 27,
49
+ "use_glome": true,
50
+ "use_nsa": true,
51
+ "vocab_size": 36
52
+ }
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca38aef4bec71d6d96a0e9bdfd2d7bbd821f572d5ce42d4d2f34673aec533b44
3
+ size 61385376
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d33f9cb88038c34fa64d87c5ddba2ba49323b162a432983750bfc5ffba921e20
3
+ size 122881658
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e4867af6cbff014e37687b6a83467d119b6a2f6f101834ca5a80069323c97a0b
3
+ size 14244
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b30172cf14f5dbe00280d63e36224a9f28dc7a0e8b38a74ceb5eb284e84da363
3
+ size 988
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10c756fdb9db8bf7a472800df8b97780cb11d7cca932ec4aba9d03b9e6436112
3
+ size 1064
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-12974/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2fce922531bcc60b40ec3cfe0214120623a297c18ab37c3a2e94007f715374c7
3
+ size 5368
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "GloMeModelForTokenClassification"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.1,
7
+ "bos_token_id": 28,
8
+ "cdr_weight": 0.0,
9
+ "class_weights": [
10
+ 0.1,
11
+ 0.9
12
+ ],
13
+ "classifier_activation": "gelu",
14
+ "classifier_bias": false,
15
+ "classifier_dropout": 0.1,
16
+ "classifier_pooling": "cls",
17
+ "cls_token_id": 28,
18
+ "compress_block_size": 16,
19
+ "compress_block_sliding_stride": 16,
20
+ "decoder_bias": true,
21
+ "dice_weight": 0.1,
22
+ "embedding_dropout": 0.1,
23
+ "eos_token_id": 29,
24
+ "hidden_activation": "gelu",
25
+ "hidden_size": 320,
26
+ "inner_rank": 32,
27
+ "intermediate_size": 1280,
28
+ "kv_heads": 10,
29
+ "mask_token_id": 31,
30
+ "mlp_bias": false,
31
+ "mlp_dropout": 0.1,
32
+ "model_size": "tiny",
33
+ "model_type": "glome",
34
+ "norm_bias": false,
35
+ "norm_eps": 1e-05,
36
+ "num_attention_heads": 20,
37
+ "num_hidden_layers": 6,
38
+ "num_selected_blocks": 8,
39
+ "num_slots": 64,
40
+ "pad_token_id": 30,
41
+ "reference_compile": null,
42
+ "selection_block_size": 16,
43
+ "sep_token_id": 29,
44
+ "sliding_window_size": 0,
45
+ "tie_word_embeddings": false,
46
+ "torch_dtype": "float32",
47
+ "transformers_version": "4.52.3",
48
+ "unk_token_id": 27,
49
+ "use_glome": true,
50
+ "use_nsa": true,
51
+ "vocab_size": 36
52
+ }
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0af5cd7ea1c961f575b49c1824e246627fe1431012f8c7aefb4c304043a5c505
3
+ size 61385376
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:787d59fa175cff3add21d5089a62d37a22bcfa9310a078ec8ca0ef6cce84aad0
3
+ size 122881658
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ea30593948da7f35739a54c03f3cb3e2f22c1412ec5c93e4c8da21c76e0b76a
3
+ size 14244
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b30172cf14f5dbe00280d63e36224a9f28dc7a0e8b38a74ceb5eb284e84da363
3
+ size 988
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d5e9d89428155a2a0813f903fdb3c0bafa943e8b525b97a6c7f8e2bc07ececa9
3
+ size 1064
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/checkpoint-13000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2fce922531bcc60b40ec3cfe0214120623a297c18ab37c3a2e94007f715374c7
3
+ size 5368
finetune_glome_nano_site_extradata_dice_0_10_1_9_tiny/logs/events.out.tfevents.1772413682.amax.578393.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7aab41102b2a53546cd1c6d1346cd03639b21aeb356f6085991a15eecf0dfac4
3
+ size 583846