Text Classification
Transformers
Safetensors
Portuguese
deberta-v2
patchdedup
experimental
nli
text-embeddings-inference
Instructions to use enzoferracini/patchdedup-ptbr-mdeberta with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use enzoferracini/patchdedup-ptbr-mdeberta with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="enzoferracini/patchdedup-ptbr-mdeberta")# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForSequenceClassification tokenizer = AutoTokenizer.from_pretrained("enzoferracini/patchdedup-ptbr-mdeberta") model = AutoModelForSequenceClassification.from_pretrained("enzoferracini/patchdedup-ptbr-mdeberta", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download evaluation/multitask_minilm-training.json from enzoferracini/patchdedup-ptbr-mdeberta: direct link, hf CLI and curl.
- Browser
- Download file 5.42 kB
-
https://huggingface.co/enzoferracini/patchdedup-ptbr-mdeberta/resolve/main/evaluation/multitask_minilm-training.json
- Command line
-
hf download hf://enzoferracini/patchdedup-ptbr-mdeberta/evaluation/multitask_minilm-training.json
-
curl -L -o multitask_minilm-training.json https://huggingface.co/enzoferracini/patchdedup-ptbr-mdeberta/resolve/main/evaluation/multitask_minilm-training.json
5.42 kB
| { | |
| "config": { | |
| "epochs": 3, | |
| "batch_size": 8, | |
| "gradient_accumulation": 2, | |
| "learning_rate": 1e-05, | |
| "freeze_embeddings": true, | |
| "warmup_ratio": 0.1, | |
| "weight_decay": 0.01, | |
| "label_smoothing": 0.05, | |
| "patience": 2, | |
| "bf16": true, | |
| "layer_lr_decay": 0.95, | |
| "seed": 42, | |
| "freeze_lower_layers": 3, | |
| "rdrop": 0.05, | |
| "ema_decay": 0.98 | |
| }, | |
| "history": [ | |
| { | |
| "epoch": 1, | |
| "train_loss": 1.032079094199724, | |
| "validation": { | |
| "support": 132, | |
| "positive_support": 57, | |
| "precision": 0.726027397260274, | |
| "recall": 0.9298245614035088, | |
| "f1": 0.8153846153846154, | |
| "macro_f1": 0.8181400688863376, | |
| "average_precision": 0.8211117616527146, | |
| "confusion_matrix": [ | |
| [ | |
| 55, | |
| 20 | |
| ], | |
| [ | |
| 4, | |
| 53 | |
| ] | |
| ], | |
| "coverage": 0.5909090909090909, | |
| "review_rate": 0.4090909090909091, | |
| "review_count": 54, | |
| "redundant_count": 17, | |
| "not_redundant_count": 61, | |
| "selective_precision": 0.9411764705882353, | |
| "negative_predictive_value": 0.9016393442622951 | |
| } | |
| }, | |
| { | |
| "epoch": 2, | |
| "train_loss": 0.8466487510870861, | |
| "validation": { | |
| "support": 132, | |
| "positive_support": 57, | |
| "precision": 0.7397260273972602, | |
| "recall": 0.9473684210526315, | |
| "f1": 0.8307692307692308, | |
| "macro_f1": 0.8332950631458094, | |
| "average_precision": 0.844672436806174, | |
| "confusion_matrix": [ | |
| [ | |
| 56, | |
| 19 | |
| ], | |
| [ | |
| 3, | |
| 54 | |
| ] | |
| ], | |
| "coverage": 0.6212121212121212, | |
| "review_rate": 0.3787878787878788, | |
| "review_count": 50, | |
| "redundant_count": 20, | |
| "not_redundant_count": 62, | |
| "selective_precision": 0.9, | |
| "negative_predictive_value": 0.9032258064516129 | |
| } | |
| }, | |
| { | |
| "epoch": 3, | |
| "train_loss": 0.8288910722219816, | |
| "validation": { | |
| "support": 132, | |
| "positive_support": 57, | |
| "precision": 0.7397260273972602, | |
| "recall": 0.9473684210526315, | |
| "f1": 0.8307692307692308, | |
| "macro_f1": 0.8332950631458094, | |
| "average_precision": 0.8595292857104215, | |
| "confusion_matrix": [ | |
| [ | |
| 56, | |
| 19 | |
| ], | |
| [ | |
| 3, | |
| 54 | |
| ] | |
| ], | |
| "coverage": 0.6666666666666666, | |
| "review_rate": 0.3333333333333333, | |
| "review_count": 44, | |
| "redundant_count": 23, | |
| "not_redundant_count": 65, | |
| "selective_precision": 0.9130434782608695, | |
| "negative_predictive_value": 0.9076923076923077 | |
| } | |
| } | |
| ], | |
| "selected_epoch": 3, | |
| "selected_parameter_delta_l2": 0.04123555123806, | |
| "optimizer_steps": 141, | |
| "pruned": false, | |
| "policy": { | |
| "binary_threshold": 0.21747919917106628, | |
| "thresholds": { | |
| "low": 0.2793824076652527, | |
| "high": 0.8280277252197266, | |
| "positive_enabled": true, | |
| "negative_enabled": true | |
| }, | |
| "selection_split": "validation", | |
| "precision_target": 0.9, | |
| "negative_predictive_value_target": 0.9, | |
| "minimum_support": 5, | |
| "decision_policy_version": "0.1.0" | |
| }, | |
| "training_pairs": 744, | |
| "validation_pairs": 132, | |
| "trainable_parameters": 5472387, | |
| "total_parameters": 106995075, | |
| "duration_seconds": 58.99627535400214, | |
| "environment": { | |
| "python": "3.11.14", | |
| "platform": "Linux-7.1.8+deb13-amd64-x86_64-with-glibc2.41", | |
| "code_commit": "a7df187600bfb0029864aaf3afb04e83cb101b63", | |
| "source_sha256": { | |
| "cli.py": "8462dd1c90e06165826a5053c740c2e4379bc4db1a3fafacb38eccc866f007ab", | |
| "decision.py": "ce69d18701fb687f0fd94217a121f7ebdd29e9397c0c7a18c899dcede96f5e2a", | |
| "calibration.py": "37f362af38c87eaad5fea83ecef11e935591143b2e463a53a7b08570c4ef2376", | |
| "encoder.py": "332e8df2f51b864971fdd78290fa7d69ca99ea82d4abcdd470da44c4fdb075d4", | |
| "nli.py": "f74b06c83c66bb6d6f500b01ee9831f56422e8aad44e2355bff48ab1c1b27558", | |
| "data.py": "5b8143e617f9166f13159f96081b9048b24fe6d80289b181fc165f1c307b7dde", | |
| "experiment.py": "d6cac58f617129f20a3518724d53678d6ef204fee35a9e998d1dc1aafd8af891", | |
| "training.py": "2a375b311b3a9e3e5d773e0b0f974ff859dd4f232d639c8176302128142aa8c0", | |
| "__main__.py": "b466d3224c1cfa96b82f891d496e36e7bd4ad1eb70fcc530b45c29ce62d3f9c4", | |
| "cross_encoder.py": "d22c187c938cec7caec0fbf56a31fc09eb512eb17898091748fcf8e77f3f7048", | |
| "schemas.py": "1de7bfaedf181f8724387886d88f1b8971689fe47b37570dcfdb52ba3f4c5194", | |
| "evaluation.py": "4880d0ee9fcf6fc15163204b2edcf4856dc06dfe5aaa3930a9dca236b16d0fd0", | |
| "__init__.py": "8f3ef65a08808435e85992bcb80379e0a060dab973ffbe2b5e193bc924766681", | |
| "v03_data.py": "187402e98c0c9b7a5703aaeeff244b60f76410a24a889dddbe8d73e5fdb14767", | |
| "dataset.py": "40c6153527279d43d66d60de10dc0cb1bfbcba07df11a64d18ad810d56bfb149" | |
| }, | |
| "deterministic_warn_only": true, | |
| "cublas_workspace_config": ":4096:8", | |
| "gpu": "NVIDIA RTX A1000 6GB Laptop GPU", | |
| "packages": { | |
| "torch": "2.6.0", | |
| "sentence-transformers": "5.1.2", | |
| "transformers": "4.57.1", | |
| "numpy": "2.3.4", | |
| "scikit-learn": "1.7.2" | |
| } | |
| }, | |
| "objective": "multitask directional NLI CE and symmetric redundancy BCE", | |
| "score": "minimum bidirectional entailment softmax; uncalibrated" | |
| } | |