Upload folder using huggingface_hub
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/gliner_config.json +105 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/optimizer.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/pytorch_model.bin +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/rng_state.pth +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/scheduler.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/tokenizer.json +0 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/tokenizer_config.json +24 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/trainer_state.json +48 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/gliner_config.json +105 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/optimizer.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/pytorch_model.bin +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/rng_state.pth +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/scheduler.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/tokenizer.json +0 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/tokenizer_config.json +24 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/trainer_state.json +174 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/gliner_config.json +105 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/optimizer.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/pytorch_model.bin +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/rng_state.pth +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/scheduler.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/tokenizer.json +0 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/tokenizer_config.json +24 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/trainer_state.json +181 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/gliner_config.json +105 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/optimizer.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/pytorch_model.bin +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/rng_state.pth +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/scheduler.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/tokenizer.json +0 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/tokenizer_config.json +24 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/trainer_state.json +188 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/gliner_config.json +105 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/optimizer.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/pytorch_model.bin +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/rng_state.pth +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/scheduler.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/tokenizer.json +0 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/tokenizer_config.json +24 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/trainer_state.json +195 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/gliner_config.json +105 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/optimizer.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/pytorch_model.bin +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/rng_state.pth +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/scheduler.pt +3 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/tokenizer.json +0 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/tokenizer_config.json +24 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/trainer_state.json +202 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12500/gliner_config.json +105 -0
- open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12500/optimizer.pt +3 -0
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/gliner_config.json
ADDED
|
@@ -0,0 +1,105 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"class_token_index": 128002,
|
| 4 |
+
"dropout": 0.4,
|
| 5 |
+
"embed_ent_token": true,
|
| 6 |
+
"encoder_config": {
|
| 7 |
+
"_name_or_path": "microsoft/deberta-v3-large",
|
| 8 |
+
"add_cross_attention": false,
|
| 9 |
+
"architectures": null,
|
| 10 |
+
"attention_probs_dropout_prob": 0.1,
|
| 11 |
+
"bos_token_id": null,
|
| 12 |
+
"chunk_size_feed_forward": 0,
|
| 13 |
+
"cross_attention_hidden_size": null,
|
| 14 |
+
"decoder_start_token_id": null,
|
| 15 |
+
"dtype": null,
|
| 16 |
+
"eos_token_id": null,
|
| 17 |
+
"finetuning_task": null,
|
| 18 |
+
"hidden_act": "gelu",
|
| 19 |
+
"hidden_dropout_prob": 0.1,
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"id2label": {
|
| 22 |
+
"0": "LABEL_0",
|
| 23 |
+
"1": "LABEL_1"
|
| 24 |
+
},
|
| 25 |
+
"initializer_range": 0.02,
|
| 26 |
+
"intermediate_size": 4096,
|
| 27 |
+
"is_decoder": false,
|
| 28 |
+
"is_encoder_decoder": false,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"LABEL_0": 0,
|
| 31 |
+
"LABEL_1": 1
|
| 32 |
+
},
|
| 33 |
+
"layer_norm_eps": 1e-07,
|
| 34 |
+
"legacy": true,
|
| 35 |
+
"max_position_embeddings": 512,
|
| 36 |
+
"max_relative_positions": -1,
|
| 37 |
+
"model_type": "deberta-v2",
|
| 38 |
+
"norm_rel_ebd": "layer_norm",
|
| 39 |
+
"num_attention_heads": 16,
|
| 40 |
+
"num_hidden_layers": 24,
|
| 41 |
+
"output_attentions": false,
|
| 42 |
+
"output_hidden_states": false,
|
| 43 |
+
"pad_token_id": 0,
|
| 44 |
+
"pooler_dropout": 0,
|
| 45 |
+
"pooler_hidden_act": "gelu",
|
| 46 |
+
"pooler_hidden_size": 1024,
|
| 47 |
+
"pos_att_type": [
|
| 48 |
+
"p2c",
|
| 49 |
+
"c2p"
|
| 50 |
+
],
|
| 51 |
+
"position_biased_input": false,
|
| 52 |
+
"position_buckets": 256,
|
| 53 |
+
"prefix": null,
|
| 54 |
+
"problem_type": null,
|
| 55 |
+
"pruned_heads": {},
|
| 56 |
+
"relative_attention": true,
|
| 57 |
+
"return_dict": true,
|
| 58 |
+
"sep_token_id": null,
|
| 59 |
+
"share_att_key": true,
|
| 60 |
+
"task_specific_params": null,
|
| 61 |
+
"tf_legacy_loss": false,
|
| 62 |
+
"tie_encoder_decoder": false,
|
| 63 |
+
"tie_word_embeddings": true,
|
| 64 |
+
"tokenizer_class": null,
|
| 65 |
+
"torchscript": false,
|
| 66 |
+
"type_vocab_size": 0,
|
| 67 |
+
"use_bfloat16": false,
|
| 68 |
+
"vocab_size": 128004
|
| 69 |
+
},
|
| 70 |
+
"ent_token": "<<ENT>>",
|
| 71 |
+
"eos_token_id": 2,
|
| 72 |
+
"eval_every": 5000,
|
| 73 |
+
"fine_tune": true,
|
| 74 |
+
"fuse_layers": false,
|
| 75 |
+
"has_rnn": true,
|
| 76 |
+
"hidden_size": 512,
|
| 77 |
+
"labels_encoder": null,
|
| 78 |
+
"labels_encoder_config": null,
|
| 79 |
+
"lr_encoder": "1e-5",
|
| 80 |
+
"lr_others": "5e-5",
|
| 81 |
+
"max_len": 384,
|
| 82 |
+
"max_neg_type_ratio": 1,
|
| 83 |
+
"max_types": 25,
|
| 84 |
+
"max_width": 12,
|
| 85 |
+
"model_name": "microsoft/deberta-v3-large",
|
| 86 |
+
"model_type": null,
|
| 87 |
+
"name": "correct",
|
| 88 |
+
"num_post_fusion_layers": 1,
|
| 89 |
+
"num_rnn_layers": 1,
|
| 90 |
+
"num_steps": 30000,
|
| 91 |
+
"pad_token_id": 0,
|
| 92 |
+
"post_fusion_schema": "",
|
| 93 |
+
"random_drop": true,
|
| 94 |
+
"sep_token": "<<SEP>>",
|
| 95 |
+
"shuffle_types": true,
|
| 96 |
+
"size_sup": -1,
|
| 97 |
+
"span_mode": "markerV0",
|
| 98 |
+
"subtoken_pooling": "first",
|
| 99 |
+
"train_batch_size": 8,
|
| 100 |
+
"transformers_version": "5.3.0",
|
| 101 |
+
"use_cache": false,
|
| 102 |
+
"vocab_size": 128004,
|
| 103 |
+
"warmup_ratio": 3000,
|
| 104 |
+
"words_splitter_type": "whitespace"
|
| 105 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:81a6fc856dbcb404fbedeb1f3f682fe8e8dd01d77d45cefc9a7867a4386937ff
|
| 3 |
+
size 2873
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0abe4edc77f96f88e966f61f2c72b785f8d6f1d5aca2bd845e7ac15dac597a5d
|
| 3 |
+
size 1782005383
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d80de80a6f8d437b18fe581066fb16e653d9b7059aaa91ae9ebaa0447c8fadd2
|
| 3 |
+
size 14645
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4c753eba7fea9d380b53fdedf2c8c05880c880d1f2e0a2fccafacaa4c259ac33
|
| 3 |
+
size 1529
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/tokenizer_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": true,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "[CLS]",
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"cls_token": "[CLS]",
|
| 7 |
+
"do_lower_case": false,
|
| 8 |
+
"eos_token": "[SEP]",
|
| 9 |
+
"is_local": true,
|
| 10 |
+
"mask_token": "[MASK]",
|
| 11 |
+
"max_length": null,
|
| 12 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 13 |
+
"pad_to_multiple_of": null,
|
| 14 |
+
"pad_token": "[PAD]",
|
| 15 |
+
"pad_token_type_id": 0,
|
| 16 |
+
"padding_side": "right",
|
| 17 |
+
"sep_token": "[SEP]",
|
| 18 |
+
"sp_model_kwargs": {},
|
| 19 |
+
"split_by_punct": false,
|
| 20 |
+
"tokenizer_class": "DebertaV2Tokenizer",
|
| 21 |
+
"unk_id": 3,
|
| 22 |
+
"unk_token": "[UNK]",
|
| 23 |
+
"vocab_type": "spm"
|
| 24 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-1000/trainer_state.json
ADDED
|
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.03447087211306446,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01723543605653223,
|
| 14 |
+
"grad_norm": 0.0,
|
| 15 |
+
"learning_rate": 9.827990348155809e-06,
|
| 16 |
+
"loss": 0.0,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03447087211306446,
|
| 21 |
+
"grad_norm": 0.0,
|
| 22 |
+
"learning_rate": 9.655635987590488e-06,
|
| 23 |
+
"loss": 0.0,
|
| 24 |
+
"step": 1000
|
| 25 |
+
}
|
| 26 |
+
],
|
| 27 |
+
"logging_steps": 500,
|
| 28 |
+
"max_steps": 29010,
|
| 29 |
+
"num_input_tokens_seen": 0,
|
| 30 |
+
"num_train_epochs": 1,
|
| 31 |
+
"save_steps": 500,
|
| 32 |
+
"stateful_callbacks": {
|
| 33 |
+
"TrainerControl": {
|
| 34 |
+
"args": {
|
| 35 |
+
"should_epoch_stop": false,
|
| 36 |
+
"should_evaluate": false,
|
| 37 |
+
"should_log": false,
|
| 38 |
+
"should_save": true,
|
| 39 |
+
"should_training_stop": false
|
| 40 |
+
},
|
| 41 |
+
"attributes": {}
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"total_flos": 0.0,
|
| 45 |
+
"train_batch_size": 16,
|
| 46 |
+
"trial_name": null,
|
| 47 |
+
"trial_params": null
|
| 48 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/gliner_config.json
ADDED
|
@@ -0,0 +1,105 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"class_token_index": 128002,
|
| 4 |
+
"dropout": 0.4,
|
| 5 |
+
"embed_ent_token": true,
|
| 6 |
+
"encoder_config": {
|
| 7 |
+
"_name_or_path": "microsoft/deberta-v3-large",
|
| 8 |
+
"add_cross_attention": false,
|
| 9 |
+
"architectures": null,
|
| 10 |
+
"attention_probs_dropout_prob": 0.1,
|
| 11 |
+
"bos_token_id": null,
|
| 12 |
+
"chunk_size_feed_forward": 0,
|
| 13 |
+
"cross_attention_hidden_size": null,
|
| 14 |
+
"decoder_start_token_id": null,
|
| 15 |
+
"dtype": null,
|
| 16 |
+
"eos_token_id": null,
|
| 17 |
+
"finetuning_task": null,
|
| 18 |
+
"hidden_act": "gelu",
|
| 19 |
+
"hidden_dropout_prob": 0.1,
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"id2label": {
|
| 22 |
+
"0": "LABEL_0",
|
| 23 |
+
"1": "LABEL_1"
|
| 24 |
+
},
|
| 25 |
+
"initializer_range": 0.02,
|
| 26 |
+
"intermediate_size": 4096,
|
| 27 |
+
"is_decoder": false,
|
| 28 |
+
"is_encoder_decoder": false,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"LABEL_0": 0,
|
| 31 |
+
"LABEL_1": 1
|
| 32 |
+
},
|
| 33 |
+
"layer_norm_eps": 1e-07,
|
| 34 |
+
"legacy": true,
|
| 35 |
+
"max_position_embeddings": 512,
|
| 36 |
+
"max_relative_positions": -1,
|
| 37 |
+
"model_type": "deberta-v2",
|
| 38 |
+
"norm_rel_ebd": "layer_norm",
|
| 39 |
+
"num_attention_heads": 16,
|
| 40 |
+
"num_hidden_layers": 24,
|
| 41 |
+
"output_attentions": false,
|
| 42 |
+
"output_hidden_states": false,
|
| 43 |
+
"pad_token_id": 0,
|
| 44 |
+
"pooler_dropout": 0,
|
| 45 |
+
"pooler_hidden_act": "gelu",
|
| 46 |
+
"pooler_hidden_size": 1024,
|
| 47 |
+
"pos_att_type": [
|
| 48 |
+
"p2c",
|
| 49 |
+
"c2p"
|
| 50 |
+
],
|
| 51 |
+
"position_biased_input": false,
|
| 52 |
+
"position_buckets": 256,
|
| 53 |
+
"prefix": null,
|
| 54 |
+
"problem_type": null,
|
| 55 |
+
"pruned_heads": {},
|
| 56 |
+
"relative_attention": true,
|
| 57 |
+
"return_dict": true,
|
| 58 |
+
"sep_token_id": null,
|
| 59 |
+
"share_att_key": true,
|
| 60 |
+
"task_specific_params": null,
|
| 61 |
+
"tf_legacy_loss": false,
|
| 62 |
+
"tie_encoder_decoder": false,
|
| 63 |
+
"tie_word_embeddings": true,
|
| 64 |
+
"tokenizer_class": null,
|
| 65 |
+
"torchscript": false,
|
| 66 |
+
"type_vocab_size": 0,
|
| 67 |
+
"use_bfloat16": false,
|
| 68 |
+
"vocab_size": 128004
|
| 69 |
+
},
|
| 70 |
+
"ent_token": "<<ENT>>",
|
| 71 |
+
"eos_token_id": 2,
|
| 72 |
+
"eval_every": 5000,
|
| 73 |
+
"fine_tune": true,
|
| 74 |
+
"fuse_layers": false,
|
| 75 |
+
"has_rnn": true,
|
| 76 |
+
"hidden_size": 512,
|
| 77 |
+
"labels_encoder": null,
|
| 78 |
+
"labels_encoder_config": null,
|
| 79 |
+
"lr_encoder": "1e-5",
|
| 80 |
+
"lr_others": "5e-5",
|
| 81 |
+
"max_len": 384,
|
| 82 |
+
"max_neg_type_ratio": 1,
|
| 83 |
+
"max_types": 25,
|
| 84 |
+
"max_width": 12,
|
| 85 |
+
"model_name": "microsoft/deberta-v3-large",
|
| 86 |
+
"model_type": null,
|
| 87 |
+
"name": "correct",
|
| 88 |
+
"num_post_fusion_layers": 1,
|
| 89 |
+
"num_rnn_layers": 1,
|
| 90 |
+
"num_steps": 30000,
|
| 91 |
+
"pad_token_id": 0,
|
| 92 |
+
"post_fusion_schema": "",
|
| 93 |
+
"random_drop": true,
|
| 94 |
+
"sep_token": "<<SEP>>",
|
| 95 |
+
"shuffle_types": true,
|
| 96 |
+
"size_sup": -1,
|
| 97 |
+
"span_mode": "markerV0",
|
| 98 |
+
"subtoken_pooling": "first",
|
| 99 |
+
"train_batch_size": 8,
|
| 100 |
+
"transformers_version": "5.3.0",
|
| 101 |
+
"use_cache": false,
|
| 102 |
+
"vocab_size": 128004,
|
| 103 |
+
"warmup_ratio": 3000,
|
| 104 |
+
"words_splitter_type": "whitespace"
|
| 105 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:275f2a0f126a536b093454415e0f07e6daa8b8a0f2897523161a3188d9d2273a
|
| 3 |
+
size 2873
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0abe4edc77f96f88e966f61f2c72b785f8d6f1d5aca2bd845e7ac15dac597a5d
|
| 3 |
+
size 1782005383
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e158d61e8c1dafca10df4089c74d2dba536f5c3732605b56757d78eac6ccba96
|
| 3 |
+
size 14709
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8ae2677708bb6d3421ef86bcae73770248d82eb908486f44ccbc6fc7c919c4f9
|
| 3 |
+
size 1529
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/tokenizer_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": true,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "[CLS]",
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"cls_token": "[CLS]",
|
| 7 |
+
"do_lower_case": false,
|
| 8 |
+
"eos_token": "[SEP]",
|
| 9 |
+
"is_local": true,
|
| 10 |
+
"mask_token": "[MASK]",
|
| 11 |
+
"max_length": null,
|
| 12 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 13 |
+
"pad_to_multiple_of": null,
|
| 14 |
+
"pad_token": "[PAD]",
|
| 15 |
+
"pad_token_type_id": 0,
|
| 16 |
+
"padding_side": "right",
|
| 17 |
+
"sep_token": "[SEP]",
|
| 18 |
+
"sp_model_kwargs": {},
|
| 19 |
+
"split_by_punct": false,
|
| 20 |
+
"tokenizer_class": "DebertaV2Tokenizer",
|
| 21 |
+
"unk_id": 3,
|
| 22 |
+
"unk_token": "[UNK]",
|
| 23 |
+
"vocab_type": "spm"
|
| 24 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10000/trainer_state.json
ADDED
|
@@ -0,0 +1,174 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.3447087211306446,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 10000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01723543605653223,
|
| 14 |
+
"grad_norm": 0.0,
|
| 15 |
+
"learning_rate": 9.827990348155809e-06,
|
| 16 |
+
"loss": 0.0,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03447087211306446,
|
| 21 |
+
"grad_norm": 0.0,
|
| 22 |
+
"learning_rate": 9.655635987590488e-06,
|
| 23 |
+
"loss": 0.0,
|
| 24 |
+
"step": 1000
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05170630816959669,
|
| 28 |
+
"grad_norm": 0.0,
|
| 29 |
+
"learning_rate": 9.483281627025164e-06,
|
| 30 |
+
"loss": 0.0,
|
| 31 |
+
"step": 1500
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06894174422612892,
|
| 35 |
+
"grad_norm": 0.0,
|
| 36 |
+
"learning_rate": 9.310927266459843e-06,
|
| 37 |
+
"loss": 0.0,
|
| 38 |
+
"step": 2000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08617718028266115,
|
| 42 |
+
"grad_norm": 0.0,
|
| 43 |
+
"learning_rate": 9.13857290589452e-06,
|
| 44 |
+
"loss": 0.0,
|
| 45 |
+
"step": 2500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.10341261633919338,
|
| 49 |
+
"grad_norm": 0.0,
|
| 50 |
+
"learning_rate": 8.966218545329196e-06,
|
| 51 |
+
"loss": 0.0,
|
| 52 |
+
"step": 3000
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.12064805239572561,
|
| 56 |
+
"grad_norm": 0.0,
|
| 57 |
+
"learning_rate": 8.793864184763875e-06,
|
| 58 |
+
"loss": 0.0,
|
| 59 |
+
"step": 3500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.13788348845225784,
|
| 63 |
+
"grad_norm": 0.0,
|
| 64 |
+
"learning_rate": 8.621509824198553e-06,
|
| 65 |
+
"loss": 0.0,
|
| 66 |
+
"step": 4000
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.15511892450879008,
|
| 70 |
+
"grad_norm": 0.0,
|
| 71 |
+
"learning_rate": 8.44915546363323e-06,
|
| 72 |
+
"loss": 0.0,
|
| 73 |
+
"step": 4500
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.1723543605653223,
|
| 77 |
+
"grad_norm": 0.0,
|
| 78 |
+
"learning_rate": 8.276801103067909e-06,
|
| 79 |
+
"loss": 0.0,
|
| 80 |
+
"step": 5000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.18958979662185454,
|
| 84 |
+
"grad_norm": 0.0,
|
| 85 |
+
"learning_rate": 8.104446742502586e-06,
|
| 86 |
+
"loss": 0.0,
|
| 87 |
+
"step": 5500
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.20682523267838676,
|
| 91 |
+
"grad_norm": 0.0,
|
| 92 |
+
"learning_rate": 7.932092381937264e-06,
|
| 93 |
+
"loss": 0.0,
|
| 94 |
+
"step": 6000
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.224060668734919,
|
| 98 |
+
"grad_norm": 0.0,
|
| 99 |
+
"learning_rate": 7.75973802137194e-06,
|
| 100 |
+
"loss": 0.0,
|
| 101 |
+
"step": 6500
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.24129610479145122,
|
| 105 |
+
"grad_norm": 0.0,
|
| 106 |
+
"learning_rate": 7.5873836608066185e-06,
|
| 107 |
+
"loss": 0.0,
|
| 108 |
+
"step": 7000
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.25853154084798347,
|
| 112 |
+
"grad_norm": 0.0,
|
| 113 |
+
"learning_rate": 7.415029300241296e-06,
|
| 114 |
+
"loss": 0.0,
|
| 115 |
+
"step": 7500
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.2757669769045157,
|
| 119 |
+
"grad_norm": 0.0,
|
| 120 |
+
"learning_rate": 7.242674939675975e-06,
|
| 121 |
+
"loss": 0.0,
|
| 122 |
+
"step": 8000
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2930024129610479,
|
| 126 |
+
"grad_norm": 0.0,
|
| 127 |
+
"learning_rate": 7.070320579110652e-06,
|
| 128 |
+
"loss": 0.0,
|
| 129 |
+
"step": 8500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.31023784901758017,
|
| 133 |
+
"grad_norm": 0.0,
|
| 134 |
+
"learning_rate": 6.89796621854533e-06,
|
| 135 |
+
"loss": 0.0,
|
| 136 |
+
"step": 9000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.3274732850741124,
|
| 140 |
+
"grad_norm": 0.0,
|
| 141 |
+
"learning_rate": 6.7256118579800075e-06,
|
| 142 |
+
"loss": 0.0,
|
| 143 |
+
"step": 9500
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3447087211306446,
|
| 147 |
+
"grad_norm": 0.0,
|
| 148 |
+
"learning_rate": 6.553257497414686e-06,
|
| 149 |
+
"loss": 0.0,
|
| 150 |
+
"step": 10000
|
| 151 |
+
}
|
| 152 |
+
],
|
| 153 |
+
"logging_steps": 500,
|
| 154 |
+
"max_steps": 29010,
|
| 155 |
+
"num_input_tokens_seen": 0,
|
| 156 |
+
"num_train_epochs": 1,
|
| 157 |
+
"save_steps": 500,
|
| 158 |
+
"stateful_callbacks": {
|
| 159 |
+
"TrainerControl": {
|
| 160 |
+
"args": {
|
| 161 |
+
"should_epoch_stop": false,
|
| 162 |
+
"should_evaluate": false,
|
| 163 |
+
"should_log": false,
|
| 164 |
+
"should_save": true,
|
| 165 |
+
"should_training_stop": false
|
| 166 |
+
},
|
| 167 |
+
"attributes": {}
|
| 168 |
+
}
|
| 169 |
+
},
|
| 170 |
+
"total_flos": 0.0,
|
| 171 |
+
"train_batch_size": 16,
|
| 172 |
+
"trial_name": null,
|
| 173 |
+
"trial_params": null
|
| 174 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/gliner_config.json
ADDED
|
@@ -0,0 +1,105 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"class_token_index": 128002,
|
| 4 |
+
"dropout": 0.4,
|
| 5 |
+
"embed_ent_token": true,
|
| 6 |
+
"encoder_config": {
|
| 7 |
+
"_name_or_path": "microsoft/deberta-v3-large",
|
| 8 |
+
"add_cross_attention": false,
|
| 9 |
+
"architectures": null,
|
| 10 |
+
"attention_probs_dropout_prob": 0.1,
|
| 11 |
+
"bos_token_id": null,
|
| 12 |
+
"chunk_size_feed_forward": 0,
|
| 13 |
+
"cross_attention_hidden_size": null,
|
| 14 |
+
"decoder_start_token_id": null,
|
| 15 |
+
"dtype": null,
|
| 16 |
+
"eos_token_id": null,
|
| 17 |
+
"finetuning_task": null,
|
| 18 |
+
"hidden_act": "gelu",
|
| 19 |
+
"hidden_dropout_prob": 0.1,
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"id2label": {
|
| 22 |
+
"0": "LABEL_0",
|
| 23 |
+
"1": "LABEL_1"
|
| 24 |
+
},
|
| 25 |
+
"initializer_range": 0.02,
|
| 26 |
+
"intermediate_size": 4096,
|
| 27 |
+
"is_decoder": false,
|
| 28 |
+
"is_encoder_decoder": false,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"LABEL_0": 0,
|
| 31 |
+
"LABEL_1": 1
|
| 32 |
+
},
|
| 33 |
+
"layer_norm_eps": 1e-07,
|
| 34 |
+
"legacy": true,
|
| 35 |
+
"max_position_embeddings": 512,
|
| 36 |
+
"max_relative_positions": -1,
|
| 37 |
+
"model_type": "deberta-v2",
|
| 38 |
+
"norm_rel_ebd": "layer_norm",
|
| 39 |
+
"num_attention_heads": 16,
|
| 40 |
+
"num_hidden_layers": 24,
|
| 41 |
+
"output_attentions": false,
|
| 42 |
+
"output_hidden_states": false,
|
| 43 |
+
"pad_token_id": 0,
|
| 44 |
+
"pooler_dropout": 0,
|
| 45 |
+
"pooler_hidden_act": "gelu",
|
| 46 |
+
"pooler_hidden_size": 1024,
|
| 47 |
+
"pos_att_type": [
|
| 48 |
+
"p2c",
|
| 49 |
+
"c2p"
|
| 50 |
+
],
|
| 51 |
+
"position_biased_input": false,
|
| 52 |
+
"position_buckets": 256,
|
| 53 |
+
"prefix": null,
|
| 54 |
+
"problem_type": null,
|
| 55 |
+
"pruned_heads": {},
|
| 56 |
+
"relative_attention": true,
|
| 57 |
+
"return_dict": true,
|
| 58 |
+
"sep_token_id": null,
|
| 59 |
+
"share_att_key": true,
|
| 60 |
+
"task_specific_params": null,
|
| 61 |
+
"tf_legacy_loss": false,
|
| 62 |
+
"tie_encoder_decoder": false,
|
| 63 |
+
"tie_word_embeddings": true,
|
| 64 |
+
"tokenizer_class": null,
|
| 65 |
+
"torchscript": false,
|
| 66 |
+
"type_vocab_size": 0,
|
| 67 |
+
"use_bfloat16": false,
|
| 68 |
+
"vocab_size": 128004
|
| 69 |
+
},
|
| 70 |
+
"ent_token": "<<ENT>>",
|
| 71 |
+
"eos_token_id": 2,
|
| 72 |
+
"eval_every": 5000,
|
| 73 |
+
"fine_tune": true,
|
| 74 |
+
"fuse_layers": false,
|
| 75 |
+
"has_rnn": true,
|
| 76 |
+
"hidden_size": 512,
|
| 77 |
+
"labels_encoder": null,
|
| 78 |
+
"labels_encoder_config": null,
|
| 79 |
+
"lr_encoder": "1e-5",
|
| 80 |
+
"lr_others": "5e-5",
|
| 81 |
+
"max_len": 384,
|
| 82 |
+
"max_neg_type_ratio": 1,
|
| 83 |
+
"max_types": 25,
|
| 84 |
+
"max_width": 12,
|
| 85 |
+
"model_name": "microsoft/deberta-v3-large",
|
| 86 |
+
"model_type": null,
|
| 87 |
+
"name": "correct",
|
| 88 |
+
"num_post_fusion_layers": 1,
|
| 89 |
+
"num_rnn_layers": 1,
|
| 90 |
+
"num_steps": 30000,
|
| 91 |
+
"pad_token_id": 0,
|
| 92 |
+
"post_fusion_schema": "",
|
| 93 |
+
"random_drop": true,
|
| 94 |
+
"sep_token": "<<SEP>>",
|
| 95 |
+
"shuffle_types": true,
|
| 96 |
+
"size_sup": -1,
|
| 97 |
+
"span_mode": "markerV0",
|
| 98 |
+
"subtoken_pooling": "first",
|
| 99 |
+
"train_batch_size": 8,
|
| 100 |
+
"transformers_version": "5.3.0",
|
| 101 |
+
"use_cache": false,
|
| 102 |
+
"vocab_size": 128004,
|
| 103 |
+
"warmup_ratio": 3000,
|
| 104 |
+
"words_splitter_type": "whitespace"
|
| 105 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c5aa5ffa4068f19e453a42537e5cbf12e9c7713477a9a8e069e7ff7e4bcf761d
|
| 3 |
+
size 2873
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0abe4edc77f96f88e966f61f2c72b785f8d6f1d5aca2bd845e7ac15dac597a5d
|
| 3 |
+
size 1782005383
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aeac33503c434601c8a2c9e42c325c7c42a5ab1528f19137df0b725b722aa010
|
| 3 |
+
size 14709
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a68f55c0274ce4ff97ff1a99fda1fe9c49096da529314b65121a5ec0c380bab
|
| 3 |
+
size 1529
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/tokenizer_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": true,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "[CLS]",
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"cls_token": "[CLS]",
|
| 7 |
+
"do_lower_case": false,
|
| 8 |
+
"eos_token": "[SEP]",
|
| 9 |
+
"is_local": true,
|
| 10 |
+
"mask_token": "[MASK]",
|
| 11 |
+
"max_length": null,
|
| 12 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 13 |
+
"pad_to_multiple_of": null,
|
| 14 |
+
"pad_token": "[PAD]",
|
| 15 |
+
"pad_token_type_id": 0,
|
| 16 |
+
"padding_side": "right",
|
| 17 |
+
"sep_token": "[SEP]",
|
| 18 |
+
"sp_model_kwargs": {},
|
| 19 |
+
"split_by_punct": false,
|
| 20 |
+
"tokenizer_class": "DebertaV2Tokenizer",
|
| 21 |
+
"unk_id": 3,
|
| 22 |
+
"unk_token": "[UNK]",
|
| 23 |
+
"vocab_type": "spm"
|
| 24 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-10500/trainer_state.json
ADDED
|
@@ -0,0 +1,181 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.3619441571871768,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 10500,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01723543605653223,
|
| 14 |
+
"grad_norm": 0.0,
|
| 15 |
+
"learning_rate": 9.827990348155809e-06,
|
| 16 |
+
"loss": 0.0,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03447087211306446,
|
| 21 |
+
"grad_norm": 0.0,
|
| 22 |
+
"learning_rate": 9.655635987590488e-06,
|
| 23 |
+
"loss": 0.0,
|
| 24 |
+
"step": 1000
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05170630816959669,
|
| 28 |
+
"grad_norm": 0.0,
|
| 29 |
+
"learning_rate": 9.483281627025164e-06,
|
| 30 |
+
"loss": 0.0,
|
| 31 |
+
"step": 1500
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06894174422612892,
|
| 35 |
+
"grad_norm": 0.0,
|
| 36 |
+
"learning_rate": 9.310927266459843e-06,
|
| 37 |
+
"loss": 0.0,
|
| 38 |
+
"step": 2000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08617718028266115,
|
| 42 |
+
"grad_norm": 0.0,
|
| 43 |
+
"learning_rate": 9.13857290589452e-06,
|
| 44 |
+
"loss": 0.0,
|
| 45 |
+
"step": 2500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.10341261633919338,
|
| 49 |
+
"grad_norm": 0.0,
|
| 50 |
+
"learning_rate": 8.966218545329196e-06,
|
| 51 |
+
"loss": 0.0,
|
| 52 |
+
"step": 3000
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.12064805239572561,
|
| 56 |
+
"grad_norm": 0.0,
|
| 57 |
+
"learning_rate": 8.793864184763875e-06,
|
| 58 |
+
"loss": 0.0,
|
| 59 |
+
"step": 3500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.13788348845225784,
|
| 63 |
+
"grad_norm": 0.0,
|
| 64 |
+
"learning_rate": 8.621509824198553e-06,
|
| 65 |
+
"loss": 0.0,
|
| 66 |
+
"step": 4000
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.15511892450879008,
|
| 70 |
+
"grad_norm": 0.0,
|
| 71 |
+
"learning_rate": 8.44915546363323e-06,
|
| 72 |
+
"loss": 0.0,
|
| 73 |
+
"step": 4500
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.1723543605653223,
|
| 77 |
+
"grad_norm": 0.0,
|
| 78 |
+
"learning_rate": 8.276801103067909e-06,
|
| 79 |
+
"loss": 0.0,
|
| 80 |
+
"step": 5000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.18958979662185454,
|
| 84 |
+
"grad_norm": 0.0,
|
| 85 |
+
"learning_rate": 8.104446742502586e-06,
|
| 86 |
+
"loss": 0.0,
|
| 87 |
+
"step": 5500
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.20682523267838676,
|
| 91 |
+
"grad_norm": 0.0,
|
| 92 |
+
"learning_rate": 7.932092381937264e-06,
|
| 93 |
+
"loss": 0.0,
|
| 94 |
+
"step": 6000
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.224060668734919,
|
| 98 |
+
"grad_norm": 0.0,
|
| 99 |
+
"learning_rate": 7.75973802137194e-06,
|
| 100 |
+
"loss": 0.0,
|
| 101 |
+
"step": 6500
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.24129610479145122,
|
| 105 |
+
"grad_norm": 0.0,
|
| 106 |
+
"learning_rate": 7.5873836608066185e-06,
|
| 107 |
+
"loss": 0.0,
|
| 108 |
+
"step": 7000
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.25853154084798347,
|
| 112 |
+
"grad_norm": 0.0,
|
| 113 |
+
"learning_rate": 7.415029300241296e-06,
|
| 114 |
+
"loss": 0.0,
|
| 115 |
+
"step": 7500
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.2757669769045157,
|
| 119 |
+
"grad_norm": 0.0,
|
| 120 |
+
"learning_rate": 7.242674939675975e-06,
|
| 121 |
+
"loss": 0.0,
|
| 122 |
+
"step": 8000
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2930024129610479,
|
| 126 |
+
"grad_norm": 0.0,
|
| 127 |
+
"learning_rate": 7.070320579110652e-06,
|
| 128 |
+
"loss": 0.0,
|
| 129 |
+
"step": 8500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.31023784901758017,
|
| 133 |
+
"grad_norm": 0.0,
|
| 134 |
+
"learning_rate": 6.89796621854533e-06,
|
| 135 |
+
"loss": 0.0,
|
| 136 |
+
"step": 9000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.3274732850741124,
|
| 140 |
+
"grad_norm": 0.0,
|
| 141 |
+
"learning_rate": 6.7256118579800075e-06,
|
| 142 |
+
"loss": 0.0,
|
| 143 |
+
"step": 9500
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3447087211306446,
|
| 147 |
+
"grad_norm": 0.0,
|
| 148 |
+
"learning_rate": 6.553257497414686e-06,
|
| 149 |
+
"loss": 0.0,
|
| 150 |
+
"step": 10000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3619441571871768,
|
| 154 |
+
"grad_norm": 0.0,
|
| 155 |
+
"learning_rate": 6.380903136849362e-06,
|
| 156 |
+
"loss": 0.0,
|
| 157 |
+
"step": 10500
|
| 158 |
+
}
|
| 159 |
+
],
|
| 160 |
+
"logging_steps": 500,
|
| 161 |
+
"max_steps": 29010,
|
| 162 |
+
"num_input_tokens_seen": 0,
|
| 163 |
+
"num_train_epochs": 1,
|
| 164 |
+
"save_steps": 500,
|
| 165 |
+
"stateful_callbacks": {
|
| 166 |
+
"TrainerControl": {
|
| 167 |
+
"args": {
|
| 168 |
+
"should_epoch_stop": false,
|
| 169 |
+
"should_evaluate": false,
|
| 170 |
+
"should_log": false,
|
| 171 |
+
"should_save": true,
|
| 172 |
+
"should_training_stop": false
|
| 173 |
+
},
|
| 174 |
+
"attributes": {}
|
| 175 |
+
}
|
| 176 |
+
},
|
| 177 |
+
"total_flos": 0.0,
|
| 178 |
+
"train_batch_size": 16,
|
| 179 |
+
"trial_name": null,
|
| 180 |
+
"trial_params": null
|
| 181 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/gliner_config.json
ADDED
|
@@ -0,0 +1,105 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"class_token_index": 128002,
|
| 4 |
+
"dropout": 0.4,
|
| 5 |
+
"embed_ent_token": true,
|
| 6 |
+
"encoder_config": {
|
| 7 |
+
"_name_or_path": "microsoft/deberta-v3-large",
|
| 8 |
+
"add_cross_attention": false,
|
| 9 |
+
"architectures": null,
|
| 10 |
+
"attention_probs_dropout_prob": 0.1,
|
| 11 |
+
"bos_token_id": null,
|
| 12 |
+
"chunk_size_feed_forward": 0,
|
| 13 |
+
"cross_attention_hidden_size": null,
|
| 14 |
+
"decoder_start_token_id": null,
|
| 15 |
+
"dtype": null,
|
| 16 |
+
"eos_token_id": null,
|
| 17 |
+
"finetuning_task": null,
|
| 18 |
+
"hidden_act": "gelu",
|
| 19 |
+
"hidden_dropout_prob": 0.1,
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"id2label": {
|
| 22 |
+
"0": "LABEL_0",
|
| 23 |
+
"1": "LABEL_1"
|
| 24 |
+
},
|
| 25 |
+
"initializer_range": 0.02,
|
| 26 |
+
"intermediate_size": 4096,
|
| 27 |
+
"is_decoder": false,
|
| 28 |
+
"is_encoder_decoder": false,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"LABEL_0": 0,
|
| 31 |
+
"LABEL_1": 1
|
| 32 |
+
},
|
| 33 |
+
"layer_norm_eps": 1e-07,
|
| 34 |
+
"legacy": true,
|
| 35 |
+
"max_position_embeddings": 512,
|
| 36 |
+
"max_relative_positions": -1,
|
| 37 |
+
"model_type": "deberta-v2",
|
| 38 |
+
"norm_rel_ebd": "layer_norm",
|
| 39 |
+
"num_attention_heads": 16,
|
| 40 |
+
"num_hidden_layers": 24,
|
| 41 |
+
"output_attentions": false,
|
| 42 |
+
"output_hidden_states": false,
|
| 43 |
+
"pad_token_id": 0,
|
| 44 |
+
"pooler_dropout": 0,
|
| 45 |
+
"pooler_hidden_act": "gelu",
|
| 46 |
+
"pooler_hidden_size": 1024,
|
| 47 |
+
"pos_att_type": [
|
| 48 |
+
"p2c",
|
| 49 |
+
"c2p"
|
| 50 |
+
],
|
| 51 |
+
"position_biased_input": false,
|
| 52 |
+
"position_buckets": 256,
|
| 53 |
+
"prefix": null,
|
| 54 |
+
"problem_type": null,
|
| 55 |
+
"pruned_heads": {},
|
| 56 |
+
"relative_attention": true,
|
| 57 |
+
"return_dict": true,
|
| 58 |
+
"sep_token_id": null,
|
| 59 |
+
"share_att_key": true,
|
| 60 |
+
"task_specific_params": null,
|
| 61 |
+
"tf_legacy_loss": false,
|
| 62 |
+
"tie_encoder_decoder": false,
|
| 63 |
+
"tie_word_embeddings": true,
|
| 64 |
+
"tokenizer_class": null,
|
| 65 |
+
"torchscript": false,
|
| 66 |
+
"type_vocab_size": 0,
|
| 67 |
+
"use_bfloat16": false,
|
| 68 |
+
"vocab_size": 128004
|
| 69 |
+
},
|
| 70 |
+
"ent_token": "<<ENT>>",
|
| 71 |
+
"eos_token_id": 2,
|
| 72 |
+
"eval_every": 5000,
|
| 73 |
+
"fine_tune": true,
|
| 74 |
+
"fuse_layers": false,
|
| 75 |
+
"has_rnn": true,
|
| 76 |
+
"hidden_size": 512,
|
| 77 |
+
"labels_encoder": null,
|
| 78 |
+
"labels_encoder_config": null,
|
| 79 |
+
"lr_encoder": "1e-5",
|
| 80 |
+
"lr_others": "5e-5",
|
| 81 |
+
"max_len": 384,
|
| 82 |
+
"max_neg_type_ratio": 1,
|
| 83 |
+
"max_types": 25,
|
| 84 |
+
"max_width": 12,
|
| 85 |
+
"model_name": "microsoft/deberta-v3-large",
|
| 86 |
+
"model_type": null,
|
| 87 |
+
"name": "correct",
|
| 88 |
+
"num_post_fusion_layers": 1,
|
| 89 |
+
"num_rnn_layers": 1,
|
| 90 |
+
"num_steps": 30000,
|
| 91 |
+
"pad_token_id": 0,
|
| 92 |
+
"post_fusion_schema": "",
|
| 93 |
+
"random_drop": true,
|
| 94 |
+
"sep_token": "<<SEP>>",
|
| 95 |
+
"shuffle_types": true,
|
| 96 |
+
"size_sup": -1,
|
| 97 |
+
"span_mode": "markerV0",
|
| 98 |
+
"subtoken_pooling": "first",
|
| 99 |
+
"train_batch_size": 8,
|
| 100 |
+
"transformers_version": "5.3.0",
|
| 101 |
+
"use_cache": false,
|
| 102 |
+
"vocab_size": 128004,
|
| 103 |
+
"warmup_ratio": 3000,
|
| 104 |
+
"words_splitter_type": "whitespace"
|
| 105 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:05936a3cf009ce65ee9f0b552390c0ada0ed4353d3cb7497e95e00c697656d32
|
| 3 |
+
size 2873
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0abe4edc77f96f88e966f61f2c72b785f8d6f1d5aca2bd845e7ac15dac597a5d
|
| 3 |
+
size 1782005383
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:01f53d0513b18ad546695e323598dfb666fe33d0e72fa7a77d705c77f3d55395
|
| 3 |
+
size 14645
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a2bdec227e6091ea549261e9c6c1ca5400ea1643f1aaff5a8366dc03a83d7824
|
| 3 |
+
size 1529
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/tokenizer_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": true,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "[CLS]",
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"cls_token": "[CLS]",
|
| 7 |
+
"do_lower_case": false,
|
| 8 |
+
"eos_token": "[SEP]",
|
| 9 |
+
"is_local": true,
|
| 10 |
+
"mask_token": "[MASK]",
|
| 11 |
+
"max_length": null,
|
| 12 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 13 |
+
"pad_to_multiple_of": null,
|
| 14 |
+
"pad_token": "[PAD]",
|
| 15 |
+
"pad_token_type_id": 0,
|
| 16 |
+
"padding_side": "right",
|
| 17 |
+
"sep_token": "[SEP]",
|
| 18 |
+
"sp_model_kwargs": {},
|
| 19 |
+
"split_by_punct": false,
|
| 20 |
+
"tokenizer_class": "DebertaV2Tokenizer",
|
| 21 |
+
"unk_id": 3,
|
| 22 |
+
"unk_token": "[UNK]",
|
| 23 |
+
"vocab_type": "spm"
|
| 24 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11000/trainer_state.json
ADDED
|
@@ -0,0 +1,188 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.3791795932437091,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 11000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01723543605653223,
|
| 14 |
+
"grad_norm": 0.0,
|
| 15 |
+
"learning_rate": 9.827990348155809e-06,
|
| 16 |
+
"loss": 0.0,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03447087211306446,
|
| 21 |
+
"grad_norm": 0.0,
|
| 22 |
+
"learning_rate": 9.655635987590488e-06,
|
| 23 |
+
"loss": 0.0,
|
| 24 |
+
"step": 1000
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05170630816959669,
|
| 28 |
+
"grad_norm": 0.0,
|
| 29 |
+
"learning_rate": 9.483281627025164e-06,
|
| 30 |
+
"loss": 0.0,
|
| 31 |
+
"step": 1500
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06894174422612892,
|
| 35 |
+
"grad_norm": 0.0,
|
| 36 |
+
"learning_rate": 9.310927266459843e-06,
|
| 37 |
+
"loss": 0.0,
|
| 38 |
+
"step": 2000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08617718028266115,
|
| 42 |
+
"grad_norm": 0.0,
|
| 43 |
+
"learning_rate": 9.13857290589452e-06,
|
| 44 |
+
"loss": 0.0,
|
| 45 |
+
"step": 2500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.10341261633919338,
|
| 49 |
+
"grad_norm": 0.0,
|
| 50 |
+
"learning_rate": 8.966218545329196e-06,
|
| 51 |
+
"loss": 0.0,
|
| 52 |
+
"step": 3000
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.12064805239572561,
|
| 56 |
+
"grad_norm": 0.0,
|
| 57 |
+
"learning_rate": 8.793864184763875e-06,
|
| 58 |
+
"loss": 0.0,
|
| 59 |
+
"step": 3500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.13788348845225784,
|
| 63 |
+
"grad_norm": 0.0,
|
| 64 |
+
"learning_rate": 8.621509824198553e-06,
|
| 65 |
+
"loss": 0.0,
|
| 66 |
+
"step": 4000
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.15511892450879008,
|
| 70 |
+
"grad_norm": 0.0,
|
| 71 |
+
"learning_rate": 8.44915546363323e-06,
|
| 72 |
+
"loss": 0.0,
|
| 73 |
+
"step": 4500
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.1723543605653223,
|
| 77 |
+
"grad_norm": 0.0,
|
| 78 |
+
"learning_rate": 8.276801103067909e-06,
|
| 79 |
+
"loss": 0.0,
|
| 80 |
+
"step": 5000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.18958979662185454,
|
| 84 |
+
"grad_norm": 0.0,
|
| 85 |
+
"learning_rate": 8.104446742502586e-06,
|
| 86 |
+
"loss": 0.0,
|
| 87 |
+
"step": 5500
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.20682523267838676,
|
| 91 |
+
"grad_norm": 0.0,
|
| 92 |
+
"learning_rate": 7.932092381937264e-06,
|
| 93 |
+
"loss": 0.0,
|
| 94 |
+
"step": 6000
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.224060668734919,
|
| 98 |
+
"grad_norm": 0.0,
|
| 99 |
+
"learning_rate": 7.75973802137194e-06,
|
| 100 |
+
"loss": 0.0,
|
| 101 |
+
"step": 6500
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.24129610479145122,
|
| 105 |
+
"grad_norm": 0.0,
|
| 106 |
+
"learning_rate": 7.5873836608066185e-06,
|
| 107 |
+
"loss": 0.0,
|
| 108 |
+
"step": 7000
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.25853154084798347,
|
| 112 |
+
"grad_norm": 0.0,
|
| 113 |
+
"learning_rate": 7.415029300241296e-06,
|
| 114 |
+
"loss": 0.0,
|
| 115 |
+
"step": 7500
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.2757669769045157,
|
| 119 |
+
"grad_norm": 0.0,
|
| 120 |
+
"learning_rate": 7.242674939675975e-06,
|
| 121 |
+
"loss": 0.0,
|
| 122 |
+
"step": 8000
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2930024129610479,
|
| 126 |
+
"grad_norm": 0.0,
|
| 127 |
+
"learning_rate": 7.070320579110652e-06,
|
| 128 |
+
"loss": 0.0,
|
| 129 |
+
"step": 8500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.31023784901758017,
|
| 133 |
+
"grad_norm": 0.0,
|
| 134 |
+
"learning_rate": 6.89796621854533e-06,
|
| 135 |
+
"loss": 0.0,
|
| 136 |
+
"step": 9000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.3274732850741124,
|
| 140 |
+
"grad_norm": 0.0,
|
| 141 |
+
"learning_rate": 6.7256118579800075e-06,
|
| 142 |
+
"loss": 0.0,
|
| 143 |
+
"step": 9500
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3447087211306446,
|
| 147 |
+
"grad_norm": 0.0,
|
| 148 |
+
"learning_rate": 6.553257497414686e-06,
|
| 149 |
+
"loss": 0.0,
|
| 150 |
+
"step": 10000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3619441571871768,
|
| 154 |
+
"grad_norm": 0.0,
|
| 155 |
+
"learning_rate": 6.380903136849362e-06,
|
| 156 |
+
"loss": 0.0,
|
| 157 |
+
"step": 10500
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.3791795932437091,
|
| 161 |
+
"grad_norm": 0.0,
|
| 162 |
+
"learning_rate": 6.2085487762840405e-06,
|
| 163 |
+
"loss": 0.0,
|
| 164 |
+
"step": 11000
|
| 165 |
+
}
|
| 166 |
+
],
|
| 167 |
+
"logging_steps": 500,
|
| 168 |
+
"max_steps": 29010,
|
| 169 |
+
"num_input_tokens_seen": 0,
|
| 170 |
+
"num_train_epochs": 1,
|
| 171 |
+
"save_steps": 500,
|
| 172 |
+
"stateful_callbacks": {
|
| 173 |
+
"TrainerControl": {
|
| 174 |
+
"args": {
|
| 175 |
+
"should_epoch_stop": false,
|
| 176 |
+
"should_evaluate": false,
|
| 177 |
+
"should_log": false,
|
| 178 |
+
"should_save": true,
|
| 179 |
+
"should_training_stop": false
|
| 180 |
+
},
|
| 181 |
+
"attributes": {}
|
| 182 |
+
}
|
| 183 |
+
},
|
| 184 |
+
"total_flos": 0.0,
|
| 185 |
+
"train_batch_size": 16,
|
| 186 |
+
"trial_name": null,
|
| 187 |
+
"trial_params": null
|
| 188 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/gliner_config.json
ADDED
|
@@ -0,0 +1,105 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"class_token_index": 128002,
|
| 4 |
+
"dropout": 0.4,
|
| 5 |
+
"embed_ent_token": true,
|
| 6 |
+
"encoder_config": {
|
| 7 |
+
"_name_or_path": "microsoft/deberta-v3-large",
|
| 8 |
+
"add_cross_attention": false,
|
| 9 |
+
"architectures": null,
|
| 10 |
+
"attention_probs_dropout_prob": 0.1,
|
| 11 |
+
"bos_token_id": null,
|
| 12 |
+
"chunk_size_feed_forward": 0,
|
| 13 |
+
"cross_attention_hidden_size": null,
|
| 14 |
+
"decoder_start_token_id": null,
|
| 15 |
+
"dtype": null,
|
| 16 |
+
"eos_token_id": null,
|
| 17 |
+
"finetuning_task": null,
|
| 18 |
+
"hidden_act": "gelu",
|
| 19 |
+
"hidden_dropout_prob": 0.1,
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"id2label": {
|
| 22 |
+
"0": "LABEL_0",
|
| 23 |
+
"1": "LABEL_1"
|
| 24 |
+
},
|
| 25 |
+
"initializer_range": 0.02,
|
| 26 |
+
"intermediate_size": 4096,
|
| 27 |
+
"is_decoder": false,
|
| 28 |
+
"is_encoder_decoder": false,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"LABEL_0": 0,
|
| 31 |
+
"LABEL_1": 1
|
| 32 |
+
},
|
| 33 |
+
"layer_norm_eps": 1e-07,
|
| 34 |
+
"legacy": true,
|
| 35 |
+
"max_position_embeddings": 512,
|
| 36 |
+
"max_relative_positions": -1,
|
| 37 |
+
"model_type": "deberta-v2",
|
| 38 |
+
"norm_rel_ebd": "layer_norm",
|
| 39 |
+
"num_attention_heads": 16,
|
| 40 |
+
"num_hidden_layers": 24,
|
| 41 |
+
"output_attentions": false,
|
| 42 |
+
"output_hidden_states": false,
|
| 43 |
+
"pad_token_id": 0,
|
| 44 |
+
"pooler_dropout": 0,
|
| 45 |
+
"pooler_hidden_act": "gelu",
|
| 46 |
+
"pooler_hidden_size": 1024,
|
| 47 |
+
"pos_att_type": [
|
| 48 |
+
"p2c",
|
| 49 |
+
"c2p"
|
| 50 |
+
],
|
| 51 |
+
"position_biased_input": false,
|
| 52 |
+
"position_buckets": 256,
|
| 53 |
+
"prefix": null,
|
| 54 |
+
"problem_type": null,
|
| 55 |
+
"pruned_heads": {},
|
| 56 |
+
"relative_attention": true,
|
| 57 |
+
"return_dict": true,
|
| 58 |
+
"sep_token_id": null,
|
| 59 |
+
"share_att_key": true,
|
| 60 |
+
"task_specific_params": null,
|
| 61 |
+
"tf_legacy_loss": false,
|
| 62 |
+
"tie_encoder_decoder": false,
|
| 63 |
+
"tie_word_embeddings": true,
|
| 64 |
+
"tokenizer_class": null,
|
| 65 |
+
"torchscript": false,
|
| 66 |
+
"type_vocab_size": 0,
|
| 67 |
+
"use_bfloat16": false,
|
| 68 |
+
"vocab_size": 128004
|
| 69 |
+
},
|
| 70 |
+
"ent_token": "<<ENT>>",
|
| 71 |
+
"eos_token_id": 2,
|
| 72 |
+
"eval_every": 5000,
|
| 73 |
+
"fine_tune": true,
|
| 74 |
+
"fuse_layers": false,
|
| 75 |
+
"has_rnn": true,
|
| 76 |
+
"hidden_size": 512,
|
| 77 |
+
"labels_encoder": null,
|
| 78 |
+
"labels_encoder_config": null,
|
| 79 |
+
"lr_encoder": "1e-5",
|
| 80 |
+
"lr_others": "5e-5",
|
| 81 |
+
"max_len": 384,
|
| 82 |
+
"max_neg_type_ratio": 1,
|
| 83 |
+
"max_types": 25,
|
| 84 |
+
"max_width": 12,
|
| 85 |
+
"model_name": "microsoft/deberta-v3-large",
|
| 86 |
+
"model_type": null,
|
| 87 |
+
"name": "correct",
|
| 88 |
+
"num_post_fusion_layers": 1,
|
| 89 |
+
"num_rnn_layers": 1,
|
| 90 |
+
"num_steps": 30000,
|
| 91 |
+
"pad_token_id": 0,
|
| 92 |
+
"post_fusion_schema": "",
|
| 93 |
+
"random_drop": true,
|
| 94 |
+
"sep_token": "<<SEP>>",
|
| 95 |
+
"shuffle_types": true,
|
| 96 |
+
"size_sup": -1,
|
| 97 |
+
"span_mode": "markerV0",
|
| 98 |
+
"subtoken_pooling": "first",
|
| 99 |
+
"train_batch_size": 8,
|
| 100 |
+
"transformers_version": "5.3.0",
|
| 101 |
+
"use_cache": false,
|
| 102 |
+
"vocab_size": 128004,
|
| 103 |
+
"warmup_ratio": 3000,
|
| 104 |
+
"words_splitter_type": "whitespace"
|
| 105 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:22b435e723a6e33ca3924b1297f4c1beea4b32170920a24c1cb9e21d402661c9
|
| 3 |
+
size 2873
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0abe4edc77f96f88e966f61f2c72b785f8d6f1d5aca2bd845e7ac15dac597a5d
|
| 3 |
+
size 1782005383
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:39afaeb6606a6d68550ee32dd5c85c902f4cc638cfd03ad426f7f5bafa65c19a
|
| 3 |
+
size 14645
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0c7de4879cf724c007050f83d6c273ed5a17436bbafe24ad67f86eaf720d5996
|
| 3 |
+
size 1529
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/tokenizer_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": true,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "[CLS]",
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"cls_token": "[CLS]",
|
| 7 |
+
"do_lower_case": false,
|
| 8 |
+
"eos_token": "[SEP]",
|
| 9 |
+
"is_local": true,
|
| 10 |
+
"mask_token": "[MASK]",
|
| 11 |
+
"max_length": null,
|
| 12 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 13 |
+
"pad_to_multiple_of": null,
|
| 14 |
+
"pad_token": "[PAD]",
|
| 15 |
+
"pad_token_type_id": 0,
|
| 16 |
+
"padding_side": "right",
|
| 17 |
+
"sep_token": "[SEP]",
|
| 18 |
+
"sp_model_kwargs": {},
|
| 19 |
+
"split_by_punct": false,
|
| 20 |
+
"tokenizer_class": "DebertaV2Tokenizer",
|
| 21 |
+
"unk_id": 3,
|
| 22 |
+
"unk_token": "[UNK]",
|
| 23 |
+
"vocab_type": "spm"
|
| 24 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-11500/trainer_state.json
ADDED
|
@@ -0,0 +1,195 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.3964150293002413,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 11500,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01723543605653223,
|
| 14 |
+
"grad_norm": 0.0,
|
| 15 |
+
"learning_rate": 9.827990348155809e-06,
|
| 16 |
+
"loss": 0.0,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03447087211306446,
|
| 21 |
+
"grad_norm": 0.0,
|
| 22 |
+
"learning_rate": 9.655635987590488e-06,
|
| 23 |
+
"loss": 0.0,
|
| 24 |
+
"step": 1000
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05170630816959669,
|
| 28 |
+
"grad_norm": 0.0,
|
| 29 |
+
"learning_rate": 9.483281627025164e-06,
|
| 30 |
+
"loss": 0.0,
|
| 31 |
+
"step": 1500
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06894174422612892,
|
| 35 |
+
"grad_norm": 0.0,
|
| 36 |
+
"learning_rate": 9.310927266459843e-06,
|
| 37 |
+
"loss": 0.0,
|
| 38 |
+
"step": 2000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08617718028266115,
|
| 42 |
+
"grad_norm": 0.0,
|
| 43 |
+
"learning_rate": 9.13857290589452e-06,
|
| 44 |
+
"loss": 0.0,
|
| 45 |
+
"step": 2500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.10341261633919338,
|
| 49 |
+
"grad_norm": 0.0,
|
| 50 |
+
"learning_rate": 8.966218545329196e-06,
|
| 51 |
+
"loss": 0.0,
|
| 52 |
+
"step": 3000
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.12064805239572561,
|
| 56 |
+
"grad_norm": 0.0,
|
| 57 |
+
"learning_rate": 8.793864184763875e-06,
|
| 58 |
+
"loss": 0.0,
|
| 59 |
+
"step": 3500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.13788348845225784,
|
| 63 |
+
"grad_norm": 0.0,
|
| 64 |
+
"learning_rate": 8.621509824198553e-06,
|
| 65 |
+
"loss": 0.0,
|
| 66 |
+
"step": 4000
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.15511892450879008,
|
| 70 |
+
"grad_norm": 0.0,
|
| 71 |
+
"learning_rate": 8.44915546363323e-06,
|
| 72 |
+
"loss": 0.0,
|
| 73 |
+
"step": 4500
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.1723543605653223,
|
| 77 |
+
"grad_norm": 0.0,
|
| 78 |
+
"learning_rate": 8.276801103067909e-06,
|
| 79 |
+
"loss": 0.0,
|
| 80 |
+
"step": 5000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.18958979662185454,
|
| 84 |
+
"grad_norm": 0.0,
|
| 85 |
+
"learning_rate": 8.104446742502586e-06,
|
| 86 |
+
"loss": 0.0,
|
| 87 |
+
"step": 5500
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.20682523267838676,
|
| 91 |
+
"grad_norm": 0.0,
|
| 92 |
+
"learning_rate": 7.932092381937264e-06,
|
| 93 |
+
"loss": 0.0,
|
| 94 |
+
"step": 6000
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.224060668734919,
|
| 98 |
+
"grad_norm": 0.0,
|
| 99 |
+
"learning_rate": 7.75973802137194e-06,
|
| 100 |
+
"loss": 0.0,
|
| 101 |
+
"step": 6500
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.24129610479145122,
|
| 105 |
+
"grad_norm": 0.0,
|
| 106 |
+
"learning_rate": 7.5873836608066185e-06,
|
| 107 |
+
"loss": 0.0,
|
| 108 |
+
"step": 7000
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.25853154084798347,
|
| 112 |
+
"grad_norm": 0.0,
|
| 113 |
+
"learning_rate": 7.415029300241296e-06,
|
| 114 |
+
"loss": 0.0,
|
| 115 |
+
"step": 7500
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.2757669769045157,
|
| 119 |
+
"grad_norm": 0.0,
|
| 120 |
+
"learning_rate": 7.242674939675975e-06,
|
| 121 |
+
"loss": 0.0,
|
| 122 |
+
"step": 8000
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2930024129610479,
|
| 126 |
+
"grad_norm": 0.0,
|
| 127 |
+
"learning_rate": 7.070320579110652e-06,
|
| 128 |
+
"loss": 0.0,
|
| 129 |
+
"step": 8500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.31023784901758017,
|
| 133 |
+
"grad_norm": 0.0,
|
| 134 |
+
"learning_rate": 6.89796621854533e-06,
|
| 135 |
+
"loss": 0.0,
|
| 136 |
+
"step": 9000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.3274732850741124,
|
| 140 |
+
"grad_norm": 0.0,
|
| 141 |
+
"learning_rate": 6.7256118579800075e-06,
|
| 142 |
+
"loss": 0.0,
|
| 143 |
+
"step": 9500
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3447087211306446,
|
| 147 |
+
"grad_norm": 0.0,
|
| 148 |
+
"learning_rate": 6.553257497414686e-06,
|
| 149 |
+
"loss": 0.0,
|
| 150 |
+
"step": 10000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3619441571871768,
|
| 154 |
+
"grad_norm": 0.0,
|
| 155 |
+
"learning_rate": 6.380903136849362e-06,
|
| 156 |
+
"loss": 0.0,
|
| 157 |
+
"step": 10500
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.3791795932437091,
|
| 161 |
+
"grad_norm": 0.0,
|
| 162 |
+
"learning_rate": 6.2085487762840405e-06,
|
| 163 |
+
"loss": 0.0,
|
| 164 |
+
"step": 11000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.3964150293002413,
|
| 168 |
+
"grad_norm": 0.0,
|
| 169 |
+
"learning_rate": 6.036194415718718e-06,
|
| 170 |
+
"loss": 0.0,
|
| 171 |
+
"step": 11500
|
| 172 |
+
}
|
| 173 |
+
],
|
| 174 |
+
"logging_steps": 500,
|
| 175 |
+
"max_steps": 29010,
|
| 176 |
+
"num_input_tokens_seen": 0,
|
| 177 |
+
"num_train_epochs": 1,
|
| 178 |
+
"save_steps": 500,
|
| 179 |
+
"stateful_callbacks": {
|
| 180 |
+
"TrainerControl": {
|
| 181 |
+
"args": {
|
| 182 |
+
"should_epoch_stop": false,
|
| 183 |
+
"should_evaluate": false,
|
| 184 |
+
"should_log": false,
|
| 185 |
+
"should_save": true,
|
| 186 |
+
"should_training_stop": false
|
| 187 |
+
},
|
| 188 |
+
"attributes": {}
|
| 189 |
+
}
|
| 190 |
+
},
|
| 191 |
+
"total_flos": 0.0,
|
| 192 |
+
"train_batch_size": 16,
|
| 193 |
+
"trial_name": null,
|
| 194 |
+
"trial_params": null
|
| 195 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/gliner_config.json
ADDED
|
@@ -0,0 +1,105 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"class_token_index": 128002,
|
| 4 |
+
"dropout": 0.4,
|
| 5 |
+
"embed_ent_token": true,
|
| 6 |
+
"encoder_config": {
|
| 7 |
+
"_name_or_path": "microsoft/deberta-v3-large",
|
| 8 |
+
"add_cross_attention": false,
|
| 9 |
+
"architectures": null,
|
| 10 |
+
"attention_probs_dropout_prob": 0.1,
|
| 11 |
+
"bos_token_id": null,
|
| 12 |
+
"chunk_size_feed_forward": 0,
|
| 13 |
+
"cross_attention_hidden_size": null,
|
| 14 |
+
"decoder_start_token_id": null,
|
| 15 |
+
"dtype": null,
|
| 16 |
+
"eos_token_id": null,
|
| 17 |
+
"finetuning_task": null,
|
| 18 |
+
"hidden_act": "gelu",
|
| 19 |
+
"hidden_dropout_prob": 0.1,
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"id2label": {
|
| 22 |
+
"0": "LABEL_0",
|
| 23 |
+
"1": "LABEL_1"
|
| 24 |
+
},
|
| 25 |
+
"initializer_range": 0.02,
|
| 26 |
+
"intermediate_size": 4096,
|
| 27 |
+
"is_decoder": false,
|
| 28 |
+
"is_encoder_decoder": false,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"LABEL_0": 0,
|
| 31 |
+
"LABEL_1": 1
|
| 32 |
+
},
|
| 33 |
+
"layer_norm_eps": 1e-07,
|
| 34 |
+
"legacy": true,
|
| 35 |
+
"max_position_embeddings": 512,
|
| 36 |
+
"max_relative_positions": -1,
|
| 37 |
+
"model_type": "deberta-v2",
|
| 38 |
+
"norm_rel_ebd": "layer_norm",
|
| 39 |
+
"num_attention_heads": 16,
|
| 40 |
+
"num_hidden_layers": 24,
|
| 41 |
+
"output_attentions": false,
|
| 42 |
+
"output_hidden_states": false,
|
| 43 |
+
"pad_token_id": 0,
|
| 44 |
+
"pooler_dropout": 0,
|
| 45 |
+
"pooler_hidden_act": "gelu",
|
| 46 |
+
"pooler_hidden_size": 1024,
|
| 47 |
+
"pos_att_type": [
|
| 48 |
+
"p2c",
|
| 49 |
+
"c2p"
|
| 50 |
+
],
|
| 51 |
+
"position_biased_input": false,
|
| 52 |
+
"position_buckets": 256,
|
| 53 |
+
"prefix": null,
|
| 54 |
+
"problem_type": null,
|
| 55 |
+
"pruned_heads": {},
|
| 56 |
+
"relative_attention": true,
|
| 57 |
+
"return_dict": true,
|
| 58 |
+
"sep_token_id": null,
|
| 59 |
+
"share_att_key": true,
|
| 60 |
+
"task_specific_params": null,
|
| 61 |
+
"tf_legacy_loss": false,
|
| 62 |
+
"tie_encoder_decoder": false,
|
| 63 |
+
"tie_word_embeddings": true,
|
| 64 |
+
"tokenizer_class": null,
|
| 65 |
+
"torchscript": false,
|
| 66 |
+
"type_vocab_size": 0,
|
| 67 |
+
"use_bfloat16": false,
|
| 68 |
+
"vocab_size": 128004
|
| 69 |
+
},
|
| 70 |
+
"ent_token": "<<ENT>>",
|
| 71 |
+
"eos_token_id": 2,
|
| 72 |
+
"eval_every": 5000,
|
| 73 |
+
"fine_tune": true,
|
| 74 |
+
"fuse_layers": false,
|
| 75 |
+
"has_rnn": true,
|
| 76 |
+
"hidden_size": 512,
|
| 77 |
+
"labels_encoder": null,
|
| 78 |
+
"labels_encoder_config": null,
|
| 79 |
+
"lr_encoder": "1e-5",
|
| 80 |
+
"lr_others": "5e-5",
|
| 81 |
+
"max_len": 384,
|
| 82 |
+
"max_neg_type_ratio": 1,
|
| 83 |
+
"max_types": 25,
|
| 84 |
+
"max_width": 12,
|
| 85 |
+
"model_name": "microsoft/deberta-v3-large",
|
| 86 |
+
"model_type": null,
|
| 87 |
+
"name": "correct",
|
| 88 |
+
"num_post_fusion_layers": 1,
|
| 89 |
+
"num_rnn_layers": 1,
|
| 90 |
+
"num_steps": 30000,
|
| 91 |
+
"pad_token_id": 0,
|
| 92 |
+
"post_fusion_schema": "",
|
| 93 |
+
"random_drop": true,
|
| 94 |
+
"sep_token": "<<SEP>>",
|
| 95 |
+
"shuffle_types": true,
|
| 96 |
+
"size_sup": -1,
|
| 97 |
+
"span_mode": "markerV0",
|
| 98 |
+
"subtoken_pooling": "first",
|
| 99 |
+
"train_batch_size": 8,
|
| 100 |
+
"transformers_version": "5.3.0",
|
| 101 |
+
"use_cache": false,
|
| 102 |
+
"vocab_size": 128004,
|
| 103 |
+
"warmup_ratio": 3000,
|
| 104 |
+
"words_splitter_type": "whitespace"
|
| 105 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f32282fec0719f2b4ca10dc94ba3a63e3e976ce61b0ed65c518b8b3376345b37
|
| 3 |
+
size 2873
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0abe4edc77f96f88e966f61f2c72b785f8d6f1d5aca2bd845e7ac15dac597a5d
|
| 3 |
+
size 1782005383
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3310fe8f9c35c6ea6bf4c86801e82bb06dbd5193b947f22786eda4df0efc55cd
|
| 3 |
+
size 14645
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7176593041e95e3d48bc50dc130d78bb418b255795577ecb7279ac2b4c1cc7b7
|
| 3 |
+
size 1529
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/tokenizer_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": true,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "[CLS]",
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"cls_token": "[CLS]",
|
| 7 |
+
"do_lower_case": false,
|
| 8 |
+
"eos_token": "[SEP]",
|
| 9 |
+
"is_local": true,
|
| 10 |
+
"mask_token": "[MASK]",
|
| 11 |
+
"max_length": null,
|
| 12 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 13 |
+
"pad_to_multiple_of": null,
|
| 14 |
+
"pad_token": "[PAD]",
|
| 15 |
+
"pad_token_type_id": 0,
|
| 16 |
+
"padding_side": "right",
|
| 17 |
+
"sep_token": "[SEP]",
|
| 18 |
+
"sp_model_kwargs": {},
|
| 19 |
+
"split_by_punct": false,
|
| 20 |
+
"tokenizer_class": "DebertaV2Tokenizer",
|
| 21 |
+
"unk_id": 3,
|
| 22 |
+
"unk_token": "[UNK]",
|
| 23 |
+
"vocab_type": "spm"
|
| 24 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12000/trainer_state.json
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.4136504653567735,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 12000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01723543605653223,
|
| 14 |
+
"grad_norm": 0.0,
|
| 15 |
+
"learning_rate": 9.827990348155809e-06,
|
| 16 |
+
"loss": 0.0,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03447087211306446,
|
| 21 |
+
"grad_norm": 0.0,
|
| 22 |
+
"learning_rate": 9.655635987590488e-06,
|
| 23 |
+
"loss": 0.0,
|
| 24 |
+
"step": 1000
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05170630816959669,
|
| 28 |
+
"grad_norm": 0.0,
|
| 29 |
+
"learning_rate": 9.483281627025164e-06,
|
| 30 |
+
"loss": 0.0,
|
| 31 |
+
"step": 1500
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06894174422612892,
|
| 35 |
+
"grad_norm": 0.0,
|
| 36 |
+
"learning_rate": 9.310927266459843e-06,
|
| 37 |
+
"loss": 0.0,
|
| 38 |
+
"step": 2000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08617718028266115,
|
| 42 |
+
"grad_norm": 0.0,
|
| 43 |
+
"learning_rate": 9.13857290589452e-06,
|
| 44 |
+
"loss": 0.0,
|
| 45 |
+
"step": 2500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.10341261633919338,
|
| 49 |
+
"grad_norm": 0.0,
|
| 50 |
+
"learning_rate": 8.966218545329196e-06,
|
| 51 |
+
"loss": 0.0,
|
| 52 |
+
"step": 3000
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.12064805239572561,
|
| 56 |
+
"grad_norm": 0.0,
|
| 57 |
+
"learning_rate": 8.793864184763875e-06,
|
| 58 |
+
"loss": 0.0,
|
| 59 |
+
"step": 3500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.13788348845225784,
|
| 63 |
+
"grad_norm": 0.0,
|
| 64 |
+
"learning_rate": 8.621509824198553e-06,
|
| 65 |
+
"loss": 0.0,
|
| 66 |
+
"step": 4000
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.15511892450879008,
|
| 70 |
+
"grad_norm": 0.0,
|
| 71 |
+
"learning_rate": 8.44915546363323e-06,
|
| 72 |
+
"loss": 0.0,
|
| 73 |
+
"step": 4500
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.1723543605653223,
|
| 77 |
+
"grad_norm": 0.0,
|
| 78 |
+
"learning_rate": 8.276801103067909e-06,
|
| 79 |
+
"loss": 0.0,
|
| 80 |
+
"step": 5000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.18958979662185454,
|
| 84 |
+
"grad_norm": 0.0,
|
| 85 |
+
"learning_rate": 8.104446742502586e-06,
|
| 86 |
+
"loss": 0.0,
|
| 87 |
+
"step": 5500
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.20682523267838676,
|
| 91 |
+
"grad_norm": 0.0,
|
| 92 |
+
"learning_rate": 7.932092381937264e-06,
|
| 93 |
+
"loss": 0.0,
|
| 94 |
+
"step": 6000
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.224060668734919,
|
| 98 |
+
"grad_norm": 0.0,
|
| 99 |
+
"learning_rate": 7.75973802137194e-06,
|
| 100 |
+
"loss": 0.0,
|
| 101 |
+
"step": 6500
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.24129610479145122,
|
| 105 |
+
"grad_norm": 0.0,
|
| 106 |
+
"learning_rate": 7.5873836608066185e-06,
|
| 107 |
+
"loss": 0.0,
|
| 108 |
+
"step": 7000
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.25853154084798347,
|
| 112 |
+
"grad_norm": 0.0,
|
| 113 |
+
"learning_rate": 7.415029300241296e-06,
|
| 114 |
+
"loss": 0.0,
|
| 115 |
+
"step": 7500
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.2757669769045157,
|
| 119 |
+
"grad_norm": 0.0,
|
| 120 |
+
"learning_rate": 7.242674939675975e-06,
|
| 121 |
+
"loss": 0.0,
|
| 122 |
+
"step": 8000
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2930024129610479,
|
| 126 |
+
"grad_norm": 0.0,
|
| 127 |
+
"learning_rate": 7.070320579110652e-06,
|
| 128 |
+
"loss": 0.0,
|
| 129 |
+
"step": 8500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.31023784901758017,
|
| 133 |
+
"grad_norm": 0.0,
|
| 134 |
+
"learning_rate": 6.89796621854533e-06,
|
| 135 |
+
"loss": 0.0,
|
| 136 |
+
"step": 9000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.3274732850741124,
|
| 140 |
+
"grad_norm": 0.0,
|
| 141 |
+
"learning_rate": 6.7256118579800075e-06,
|
| 142 |
+
"loss": 0.0,
|
| 143 |
+
"step": 9500
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3447087211306446,
|
| 147 |
+
"grad_norm": 0.0,
|
| 148 |
+
"learning_rate": 6.553257497414686e-06,
|
| 149 |
+
"loss": 0.0,
|
| 150 |
+
"step": 10000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3619441571871768,
|
| 154 |
+
"grad_norm": 0.0,
|
| 155 |
+
"learning_rate": 6.380903136849362e-06,
|
| 156 |
+
"loss": 0.0,
|
| 157 |
+
"step": 10500
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.3791795932437091,
|
| 161 |
+
"grad_norm": 0.0,
|
| 162 |
+
"learning_rate": 6.2085487762840405e-06,
|
| 163 |
+
"loss": 0.0,
|
| 164 |
+
"step": 11000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.3964150293002413,
|
| 168 |
+
"grad_norm": 0.0,
|
| 169 |
+
"learning_rate": 6.036194415718718e-06,
|
| 170 |
+
"loss": 0.0,
|
| 171 |
+
"step": 11500
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4136504653567735,
|
| 175 |
+
"grad_norm": 0.0,
|
| 176 |
+
"learning_rate": 5.863840055153396e-06,
|
| 177 |
+
"loss": 0.0,
|
| 178 |
+
"step": 12000
|
| 179 |
+
}
|
| 180 |
+
],
|
| 181 |
+
"logging_steps": 500,
|
| 182 |
+
"max_steps": 29010,
|
| 183 |
+
"num_input_tokens_seen": 0,
|
| 184 |
+
"num_train_epochs": 1,
|
| 185 |
+
"save_steps": 500,
|
| 186 |
+
"stateful_callbacks": {
|
| 187 |
+
"TrainerControl": {
|
| 188 |
+
"args": {
|
| 189 |
+
"should_epoch_stop": false,
|
| 190 |
+
"should_evaluate": false,
|
| 191 |
+
"should_log": false,
|
| 192 |
+
"should_save": true,
|
| 193 |
+
"should_training_stop": false
|
| 194 |
+
},
|
| 195 |
+
"attributes": {}
|
| 196 |
+
}
|
| 197 |
+
},
|
| 198 |
+
"total_flos": 0.0,
|
| 199 |
+
"train_batch_size": 16,
|
| 200 |
+
"trial_name": null,
|
| 201 |
+
"trial_params": null
|
| 202 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12500/gliner_config.json
ADDED
|
@@ -0,0 +1,105 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"class_token_index": 128002,
|
| 4 |
+
"dropout": 0.4,
|
| 5 |
+
"embed_ent_token": true,
|
| 6 |
+
"encoder_config": {
|
| 7 |
+
"_name_or_path": "microsoft/deberta-v3-large",
|
| 8 |
+
"add_cross_attention": false,
|
| 9 |
+
"architectures": null,
|
| 10 |
+
"attention_probs_dropout_prob": 0.1,
|
| 11 |
+
"bos_token_id": null,
|
| 12 |
+
"chunk_size_feed_forward": 0,
|
| 13 |
+
"cross_attention_hidden_size": null,
|
| 14 |
+
"decoder_start_token_id": null,
|
| 15 |
+
"dtype": null,
|
| 16 |
+
"eos_token_id": null,
|
| 17 |
+
"finetuning_task": null,
|
| 18 |
+
"hidden_act": "gelu",
|
| 19 |
+
"hidden_dropout_prob": 0.1,
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"id2label": {
|
| 22 |
+
"0": "LABEL_0",
|
| 23 |
+
"1": "LABEL_1"
|
| 24 |
+
},
|
| 25 |
+
"initializer_range": 0.02,
|
| 26 |
+
"intermediate_size": 4096,
|
| 27 |
+
"is_decoder": false,
|
| 28 |
+
"is_encoder_decoder": false,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"LABEL_0": 0,
|
| 31 |
+
"LABEL_1": 1
|
| 32 |
+
},
|
| 33 |
+
"layer_norm_eps": 1e-07,
|
| 34 |
+
"legacy": true,
|
| 35 |
+
"max_position_embeddings": 512,
|
| 36 |
+
"max_relative_positions": -1,
|
| 37 |
+
"model_type": "deberta-v2",
|
| 38 |
+
"norm_rel_ebd": "layer_norm",
|
| 39 |
+
"num_attention_heads": 16,
|
| 40 |
+
"num_hidden_layers": 24,
|
| 41 |
+
"output_attentions": false,
|
| 42 |
+
"output_hidden_states": false,
|
| 43 |
+
"pad_token_id": 0,
|
| 44 |
+
"pooler_dropout": 0,
|
| 45 |
+
"pooler_hidden_act": "gelu",
|
| 46 |
+
"pooler_hidden_size": 1024,
|
| 47 |
+
"pos_att_type": [
|
| 48 |
+
"p2c",
|
| 49 |
+
"c2p"
|
| 50 |
+
],
|
| 51 |
+
"position_biased_input": false,
|
| 52 |
+
"position_buckets": 256,
|
| 53 |
+
"prefix": null,
|
| 54 |
+
"problem_type": null,
|
| 55 |
+
"pruned_heads": {},
|
| 56 |
+
"relative_attention": true,
|
| 57 |
+
"return_dict": true,
|
| 58 |
+
"sep_token_id": null,
|
| 59 |
+
"share_att_key": true,
|
| 60 |
+
"task_specific_params": null,
|
| 61 |
+
"tf_legacy_loss": false,
|
| 62 |
+
"tie_encoder_decoder": false,
|
| 63 |
+
"tie_word_embeddings": true,
|
| 64 |
+
"tokenizer_class": null,
|
| 65 |
+
"torchscript": false,
|
| 66 |
+
"type_vocab_size": 0,
|
| 67 |
+
"use_bfloat16": false,
|
| 68 |
+
"vocab_size": 128004
|
| 69 |
+
},
|
| 70 |
+
"ent_token": "<<ENT>>",
|
| 71 |
+
"eos_token_id": 2,
|
| 72 |
+
"eval_every": 5000,
|
| 73 |
+
"fine_tune": true,
|
| 74 |
+
"fuse_layers": false,
|
| 75 |
+
"has_rnn": true,
|
| 76 |
+
"hidden_size": 512,
|
| 77 |
+
"labels_encoder": null,
|
| 78 |
+
"labels_encoder_config": null,
|
| 79 |
+
"lr_encoder": "1e-5",
|
| 80 |
+
"lr_others": "5e-5",
|
| 81 |
+
"max_len": 384,
|
| 82 |
+
"max_neg_type_ratio": 1,
|
| 83 |
+
"max_types": 25,
|
| 84 |
+
"max_width": 12,
|
| 85 |
+
"model_name": "microsoft/deberta-v3-large",
|
| 86 |
+
"model_type": null,
|
| 87 |
+
"name": "correct",
|
| 88 |
+
"num_post_fusion_layers": 1,
|
| 89 |
+
"num_rnn_layers": 1,
|
| 90 |
+
"num_steps": 30000,
|
| 91 |
+
"pad_token_id": 0,
|
| 92 |
+
"post_fusion_schema": "",
|
| 93 |
+
"random_drop": true,
|
| 94 |
+
"sep_token": "<<SEP>>",
|
| 95 |
+
"shuffle_types": true,
|
| 96 |
+
"size_sup": -1,
|
| 97 |
+
"span_mode": "markerV0",
|
| 98 |
+
"subtoken_pooling": "first",
|
| 99 |
+
"train_batch_size": 8,
|
| 100 |
+
"transformers_version": "5.3.0",
|
| 101 |
+
"use_cache": false,
|
| 102 |
+
"vocab_size": 128004,
|
| 103 |
+
"warmup_ratio": 3000,
|
| 104 |
+
"words_splitter_type": "whitespace"
|
| 105 |
+
}
|
open-pii-masking-500k-ai4privacy/GLiNER-NVIDIA/checkpoint-12500/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5883114c93c17fb2c976862edbbc8ae8e89a5c4d3cb172153d852877c671d51a
|
| 3 |
+
size 2873
|