e

big-bird__loss-window-2__all_results.json

@@ -0,0 +1,21 @@
+{
+    "epoch": 3.0,
+    "eval_e": 0.0,
+    "eval_f1": 0.0019538069857852587,
+    "eval_loss": 7.420947551727295,
+    "eval_runtime": 82.0111,
+    "eval_samples": 3576,
+    "eval_samples_per_second": 43.604,
+    "eval_steps_per_second": 43.604,
+    "test_e": 0.5774608501118568,
+    "test_f1": 0.532088049820999,
+    "test_loss": 3.9567770957946777,
+    "test_runtime": 81.5031,
+    "test_samples_per_second": 43.876,
+    "test_steps_per_second": 43.876,
+    "train_loss": 6.972438398601898,
+    "train_runtime": 3412.8799,
+    "train_samples": 10138,
+    "train_samples_per_second": 8.912,
+    "train_steps_per_second": 8.912
+}

big-bird__loss-window-2__config.json

@@ -0,0 +1,37 @@
+{
+  "_name_or_path": "google/bigbird-roberta-base",
+  "architectures": [
+    "QAModel"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "attention_type": "original_full",
+  "block_size": 64,
+  "bos_token_id": 1,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "freeze_layers": null,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu_new",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "include_nones_as_positives": false,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 4096,
+  "model_type": "big_bird",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "num_random_blocks": 3,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "qa_head": {},
+  "rescale_embeddings": false,
+  "sep_token_id": 66,
+  "torch_dtype": "float32",
+  "transformers_version": "4.12.2",
+  "type_vocab_size": 2,
+  "use_bias": true,
+  "use_cache": true,
+  "vocab_size": 50358
+}

big-bird__qa-model__all_results.json

@@ -0,0 +1,8 @@
+{
+    "eval_e": 0.0,
+    "eval_f1": 0.0026853597694833906,
+    "eval_loss": 7.507201671600342,
+    "eval_runtime": 82.025,
+    "eval_samples_per_second": 43.596,
+    "eval_steps_per_second": 43.596
+}

big-bird__qa-model__roberta-large__all_results.json

@@ -0,0 +1,8 @@
+{
+    "eval_e": 0.0,
+    "eval_f1": 0.0025534260097588124,
+    "eval_loss": 7.366059303283691,
+    "eval_runtime": 216.7168,
+    "eval_samples_per_second": 16.501,
+    "eval_steps_per_second": 16.501
+}

big-bird__salience-model__all_results.json

@@ -0,0 +1,21 @@
+{
+    "epoch": 3.0,
+    "eval_e": 0.0,
+    "eval_f1": 0.0019965838147566663,
+    "eval_loss": 7.391232013702393,
+    "eval_runtime": 157.8075,
+    "eval_samples": 3576,
+    "eval_samples_per_second": 22.661,
+    "eval_steps_per_second": 22.661,
+    "test_e": 0.5914429530201343,
+    "test_f1": 0.5363322750706369,
+    "test_loss": 3.4301834106445312,
+    "test_runtime": 157.9622,
+    "test_samples_per_second": 22.638,
+    "test_steps_per_second": 22.638,
+    "train_loss": 1.4217724323805765,
+    "train_runtime": 5802.1755,
+    "train_samples": 10138,
+    "train_samples_per_second": 5.242,
+    "train_steps_per_second": 5.242
+}

big-bird__salience-model__augmented-data__all_results.json

@@ -0,0 +1,21 @@
+{
+    "epoch": 3.0,
+    "eval_e": 0.0,
+    "eval_f1": 0.0018995468801266367,
+    "eval_loss": 7.458363056182861,
+    "eval_runtime": 156.6623,
+    "eval_samples": 3576,
+    "eval_samples_per_second": 22.826,
+    "eval_steps_per_second": 22.826,
+    "test_e": 0.5584451901565995,
+    "test_f1": 0.5157599466272186,
+    "test_loss": 3.5660927295684814,
+    "test_runtime": 156.6477,
+    "test_samples_per_second": 22.828,
+    "test_steps_per_second": 22.828,
+    "train_loss": 1.9798311326282656,
+    "train_runtime": 7385.0011,
+    "train_samples": 13605,
+    "train_samples_per_second": 5.527,
+    "train_steps_per_second": 5.527
+}

big-bird__salience-model__augmented-data__config.json

@@ -0,0 +1,37 @@
+{
+  "_name_or_path": "google/bigbird-roberta-base",
+  "architectures": [
+    "QAModelWithSalience"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "attention_type": "original_full",
+  "block_size": 64,
+  "bos_token_id": 1,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "freeze_layers": null,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu_new",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "include_nones_as_positives": false,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 4096,
+  "model_type": "big_bird",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "num_random_blocks": 3,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "qa_head": {},
+  "rescale_embeddings": false,
+  "sep_token_id": 66,
+  "torch_dtype": "float32",
+  "transformers_version": "4.12.2",
+  "type_vocab_size": 2,
+  "use_bias": true,
+  "use_cache": true,
+  "vocab_size": 50358
+}

big-bird__salience-model__config.json

@@ -0,0 +1,37 @@
+{
+  "_name_or_path": "google/bigbird-roberta-base",
+  "architectures": [
+    "QAModelWithSalience"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "attention_type": "original_full",
+  "block_size": 64,
+  "bos_token_id": 1,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "freeze_layers": null,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu_new",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "include_nones_as_positives": false,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 4096,
+  "model_type": "big_bird",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "num_random_blocks": 3,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "qa_head": {},
+  "rescale_embeddings": false,
+  "sep_token_id": 66,
+  "torch_dtype": "float32",
+  "transformers_version": "4.12.2",
+  "type_vocab_size": 2,
+  "use_bias": true,
+  "use_cache": true,
+  "vocab_size": 50358
+}

quote-attribution-qa__big-bird-base__all_results.json

@@ -0,0 +1,21 @@
+{
+    "epoch": 3.0,
+    "eval_e": 0.0011185682326621924,
+    "eval_f1": 0.0030897725029911207,
+    "eval_loss": 7.352117538452148,
+    "eval_runtime": 82.3342,
+    "eval_samples": 3576,
+    "eval_samples_per_second": 43.433,
+    "eval_steps_per_second": 43.433,
+    "test_e": 0.5970357941834452,
+    "test_f1": 0.5385527992826207,
+    "test_loss": 3.199585199356079,
+    "test_runtime": 81.8465,
+    "test_samples_per_second": 43.692,
+    "test_steps_per_second": 43.692,
+    "train_loss": 1.5040293447896245,
+    "train_runtime": 3366.7923,
+    "train_samples": 10138,
+    "train_samples_per_second": 9.034,
+    "train_steps_per_second": 9.034
+}

quote-attribution-qa__big-bird-base__config.json

@@ -0,0 +1,37 @@
+{
+  "_name_or_path": "google/bigbird-roberta-base",
+  "architectures": [
+    "QAModel"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "attention_type": "original_full",
+  "block_size": 64,
+  "bos_token_id": 1,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "frozen_layers": null,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu_new",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "include_nones_as_positives": false,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 4096,
+  "model_type": "big_bird",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "num_random_blocks": 3,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "qa_head": {},
+  "rescale_embeddings": false,
+  "sep_token_id": 66,
+  "torch_dtype": "float32",
+  "transformers_version": "4.12.2",
+  "type_vocab_size": 2,
+  "use_bias": true,
+  "use_cache": true,
+  "vocab_size": 50358
+}

quote-detection__roberta-base-sentence__all_results.json

@@ -0,0 +1,19 @@
+{
+    "epoch": 3.0,
+    "eval_f1": 0.6645335958651243,
+    "eval_loss": 0.6792270531125815,
+    "eval_runtime": 1.8612,
+    "eval_samples": 72,
+    "eval_samples_per_second": 38.686,
+    "eval_steps_per_second": 38.686,
+    "test_f1": 0.8574610244988863,
+    "test_loss": 0.41005876731686347,
+    "test_runtime": 1.8863,
+    "test_samples_per_second": 38.171,
+    "test_steps_per_second": 38.171,
+    "train_loss": 0.34470706577448884,
+    "train_runtime": 143.1766,
+    "train_samples": 516,
+    "train_samples_per_second": 10.812,
+    "train_steps_per_second": 10.812
+}

quote-detection__roberta-base-sentence__config.json

@@ -0,0 +1,41 @@
+{
+  "architectures": [
+    "SentenceClassificationModel"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classification_head": {
+    "num_labels": 1,
+    "pooling_method": "average"
+  },
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "frozen_layers": [
+    0,
+    1,
+    2,
+    3,
+    4,
+    5,
+    6,
+    7,
+    8
+  ],
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 514,
+  "model_type": "roberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "torch_dtype": "float32",
+  "transformers_version": "4.12.2",
+  "type_vocab_size": 1,
+  "use_cache": true,
+  "vocab_size": 50265
+}