Hugging Face's logo

junthbasnet
/
TEST-1_ADAPTER_IMDB

TensorBoard
Safetensors
t5
Model card Files
xet
Metrics Community
junthbasnet commited on
Commit
33ea2c0
·
verified ·
1 Parent(s): 6bdafb1

Training in progress, epoch 1

Browse files
Files changed (13) hide show
  1. config.json +129 -0
  2. model.safetensors +3 -0
  3. runs/Apr18_08-58-22_448e4276cffe/events.out.tfevents.1744966708.448e4276cffe.40881.0 +3 -0
  4. runs/Apr18_09-12-08_448e4276cffe/events.out.tfevents.1744967529.448e4276cffe.40881.1 +3 -0
  5. runs/Apr18_09-38-24_448e4276cffe/events.out.tfevents.1744969105.448e4276cffe.40881.2 +3 -0
  6. runs/Apr18_09-46-26_448e4276cffe/events.out.tfevents.1744969587.448e4276cffe.40881.3 +3 -0
  7. runs/Apr18_09-54-31_448e4276cffe/events.out.tfevents.1744970073.448e4276cffe.40881.4 +3 -0
  8. runs/Apr18_09-59-09_448e4276cffe/events.out.tfevents.1744970351.448e4276cffe.40881.5 +3 -0
  9. runs/Apr18_10-08-24_448e4276cffe/events.out.tfevents.1744970905.448e4276cffe.40881.6 +3 -0
  10. runs/Apr18_10-17-55_448e4276cffe/events.out.tfevents.1744971476.448e4276cffe.40881.7 +3 -0
  11. runs/Apr18_10-24-14_448e4276cffe/events.out.tfevents.1744971855.448e4276cffe.40881.8 +3 -0
  12. runs/Apr18_10-28-36_448e4276cffe/events.out.tfevents.1744972117.448e4276cffe.40881.9 +3 -0
  13. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,129 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "t5-small",
3
+ "adapters": {
4
+ "adapters": {
5
+ "imdb_pfeiffer": "9076f36a74755ac4"
6
+ },
7
+ "config_map": {
8
+ "9076f36a74755ac4": {
9
+ "adapter_residual_before_ln": false,
10
+ "cross_adapter": false,
11
+ "dropout": 0.0,
12
+ "factorized_phm_W": true,
13
+ "factorized_phm_rule": false,
14
+ "hypercomplex_nonlinearity": "glorot-uniform",
15
+ "init_weights": "bert",
16
+ "inv_adapter": null,
17
+ "inv_adapter_reduction_factor": null,
18
+ "is_parallel": false,
19
+ "learn_phm": true,
20
+ "leave_out": [],
21
+ "ln_after": false,
22
+ "ln_before": false,
23
+ "mh_adapter": false,
24
+ "non_linearity": "relu",
25
+ "original_ln_after": true,
26
+ "original_ln_before": true,
27
+ "output_adapter": true,
28
+ "phm_bias": true,
29
+ "phm_c_init": "normal",
30
+ "phm_dim": 4,
31
+ "phm_init_range": 0.0001,
32
+ "phm_layer": false,
33
+ "phm_rank": 1,
34
+ "reduction_factor": 16,
35
+ "residual_before_ln": true,
36
+ "scaling": 1.0,
37
+ "shared_W_phm": false,
38
+ "shared_phm_rule": true,
39
+ "stochastic_depth": 0.0,
40
+ "use_gating": false
41
+ }
42
+ },
43
+ "fusion_config_map": {},
44
+ "fusion_name_map": {},
45
+ "fusions": {}
46
+ },
47
+ "architectures": [
48
+ "T5AdapterModel"
49
+ ],
50
+ "classifier_dropout": 0.0,
51
+ "d_ff": 2048,
52
+ "d_kv": 64,
53
+ "d_model": 512,
54
+ "decoder_start_token_id": 0,
55
+ "dense_act_fn": "relu",
56
+ "dropout_rate": 0.1,
57
+ "eos_token_id": 1,
58
+ "feed_forward_proj": "relu",
59
+ "id2label": null,
60
+ "initializer_factor": 1.0,
61
+ "is_encoder_decoder": true,
62
+ "is_gated_act": false,
63
+ "label2id": null,
64
+ "layer_norm_epsilon": 1e-06,
65
+ "model_type": "t5",
66
+ "n_positions": 512,
67
+ "num_decoder_layers": 6,
68
+ "num_heads": 8,
69
+ "num_layers": 6,
70
+ "output_past": true,
71
+ "pad_token_id": 0,
72
+ "prediction_heads": {
73
+ "default": {
74
+ "activation_function": null,
75
+ "bias": false,
76
+ "head_type": "seq2seq_lm",
77
+ "label2id": null,
78
+ "layer_norm": false,
79
+ "layers": 1,
80
+ "shift_labels": false,
81
+ "vocab_size": 32128
82
+ },
83
+ "imdb_pfeiffer": {
84
+ "activation_function": null,
85
+ "bias": false,
86
+ "head_type": "seq2seq_lm",
87
+ "label2id": null,
88
+ "layer_norm": false,
89
+ "layers": 1,
90
+ "shift_labels": false,
91
+ "vocab_size": 32128
92
+ }
93
+ },
94
+ "relative_attention_max_distance": 128,
95
+ "relative_attention_num_buckets": 32,
96
+ "task_specific_params": {
97
+ "summarization": {
98
+ "early_stopping": true,
99
+ "length_penalty": 2.0,
100
+ "max_length": 200,
101
+ "min_length": 30,
102
+ "no_repeat_ngram_size": 3,
103
+ "num_beams": 4,
104
+ "prefix": "summarize: "
105
+ },
106
+ "translation_en_to_de": {
107
+ "early_stopping": true,
108
+ "max_length": 300,
109
+ "num_beams": 4,
110
+ "prefix": "translate English to German: "
111
+ },
112
+ "translation_en_to_fr": {
113
+ "early_stopping": true,
114
+ "max_length": 300,
115
+ "num_beams": 4,
116
+ "prefix": "translate English to French: "
117
+ },
118
+ "translation_en_to_ro": {
119
+ "early_stopping": true,
120
+ "max_length": 300,
121
+ "num_beams": 4,
122
+ "prefix": "translate English to Romanian: "
123
+ }
124
+ },
125
+ "torch_dtype": "float32",
126
+ "transformers_version": "4.48.3",
127
+ "use_cache": true,
128
+ "vocab_size": 32128
129
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b2e8592bb8ebd28cec314cb92e1a35843883717564fd0a94d4da50b105cb821
3
+ size 243649520
runs/Apr18_08-58-22_448e4276cffe/events.out.tfevents.1744966708.448e4276cffe.40881.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ece5fd22611c565cbf67f3478bf095824a051e50ea39ada496a6f4a1c6c03327
3
+ size 6445
runs/Apr18_09-12-08_448e4276cffe/events.out.tfevents.1744967529.448e4276cffe.40881.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f2a593973c1be24b1abb92738c0fc528bf26c06016274230ea635482a34f4a0
3
+ size 6445
runs/Apr18_09-38-24_448e4276cffe/events.out.tfevents.1744969105.448e4276cffe.40881.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59f2cd8c2500abb439337e914a9443cff2975d4c1df9794bbef24d0d38fde4d7
3
+ size 6445
runs/Apr18_09-46-26_448e4276cffe/events.out.tfevents.1744969587.448e4276cffe.40881.3 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1919d2c3a28cec848eda2d6a1e60201bd4335f8e63002176097db9fa2b82b903
3
+ size 6445
runs/Apr18_09-54-31_448e4276cffe/events.out.tfevents.1744970073.448e4276cffe.40881.4 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:91e3e0261a7be0575f043e82a6b97da84c2b667aeabf9b20d3c3bfce871d5b33
3
+ size 6445
runs/Apr18_09-59-09_448e4276cffe/events.out.tfevents.1744970351.448e4276cffe.40881.5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e5b4a7c3d37e73fa4a388db0d79c01fa4b78d96b43a1fe2a762502e040db3b3
3
+ size 6445
runs/Apr18_10-08-24_448e4276cffe/events.out.tfevents.1744970905.448e4276cffe.40881.6 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d5d379503f067e1f9bd5d08b884d2b8a0bdaf58d357676d41cb5eebf87d5f23
3
+ size 6445
runs/Apr18_10-17-55_448e4276cffe/events.out.tfevents.1744971476.448e4276cffe.40881.7 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5c759f8a81d58608f29c616d7ee692f7ec90b9e686d9c995d0ce63fe26bfcffa
3
+ size 6445
runs/Apr18_10-24-14_448e4276cffe/events.out.tfevents.1744971855.448e4276cffe.40881.8 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab321e40d940bec224faf4363847a4f923b5a81b256bbfa4c76099fc8db7ab7e
3
+ size 6445
runs/Apr18_10-28-36_448e4276cffe/events.out.tfevents.1744972117.448e4276cffe.40881.9 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:13236e608c64dd54e1efafb013ef8193ed66581fc4f8a7d268f2e98a9e376062
3
+ size 6762
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7583de184e2beb0dc73eedd63571ec8daafb059f31aff0e63af2f61b42b1cc7f
3
+ size 5368