Cil Ordinal Coral Seed3 | Sweet Tea Studio
Resources / Cil Ordinal Coral Seed3 Cil Ordinal Coral Seed3 This model is a fine-tuned version of xlm-roberta-base on the None dataset. It achieves the following results on the evaluation set: Loss: 0.3482 Mae: 0.4476 Coral Mae: 0.5951 Rounded Mae: 0.3903 Accuracy: 0.5135 Model description
Verified source
Kind Other Base model xlm-roberta-base Version v1902554814cf480658f8bf050b2164082765171a License mit Publisher @MichaHenh C grade Model source
Kind Other
Base model xlm-roberta-base
Version v1902554814cf480658f8bf050b2164082765171a
License mit
Source Hugging Face --- library_name: peft license: mit base_model: xlm-roberta-base tags: - base_model:adapter:xlm-roberta-base - lora - transformers metrics: - accuracy model-index: - name: cil-ordinal-coral-seed3 results: [] --- # cil-ordinal-coral-seed3 This model is a fine-tuned version of [xlm-roberta-base](https://huggingface.co/xlm-roberta-base) on the None dataset. It achieves the following results on the evaluation set: - Loss: 0.3482 - Mae: 0.4476 - Coral Mae: 0.5951 - Rounded Mae: 0.3903 - Accuracy: 0.5135 ## Model description More information needed ## Intended uses & limitations More information needed ## Training and evaluation data More information needed ## Training procedure ### Training hyperparameters The following hyperparameters were used during training: - learning_rate: 0.00015 - train_batch_size: 64 - eval_batch_size: 1024 - seed: 3 - optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments - lr_scheduler_type: cosine - lr_scheduler_warmup_steps: 100 - num_epochs: 3 - mixed_precision_training: Native AMP ### Training results | Training Loss | Epoch | Step | Validation Loss | Mae | Coral Mae | Rounded Mae | Accuracy | |:-------------:|:------:|:-----:|:---------------:|:------:|:---------:|:-----------:|:--------:| | 0.5183 | 0.1411 | 500 | 0.4711 | 0.5556 | 0.8826 | 0.4996 | 0.4046 | | 0.4581 | 0.2822 | 1000 | 0.4483 | 0.4983 | 0.8392 | 0.4429 | 0.4150 | | 0.4421 | 0.4233 | 1500 | 0.4299 | 0.4863 | 0.7942 | 0.4291 | 0.4306 | | 0.4253 | 0.5643 | 2000 | 0.4181 | 0.4994 | 0.7490 | 0.4317 | 0.4476 | | 0.4142 | 0.7054 | 2500 | 0.4074 | 0.4846 | 0.7446 | 0.4192 | 0.4496 | | 0.4043 | 0.8465 | 3000 | 0.3975 | 0.4702 | 0.6925 | 0.4086 | 0.4694 | | 0.3951 | 0.9876 | 3500 | 0.3877 | 0.4715 | 0.6921 | 0.4031 | 0.4685 | | 0.3837 | 1.1287 | 4000 | 0.3822 | 0.4667 | 0.6532 | 0.4088 | 0.4870 | | 0.3756 | 1.2698 | 4500 | 0.3735 | 0.4749 | 0.6486 | 0.4076 | 0.4881 | | 0.3702 | 1.4108 | 5000 | 0.3682 | 0.4612 | 0.6284 | 0.4005 | 0.4973 | | 0.3640 | 1.5519 | 5500 | 0.3631 | 0.4610 | 0.6194 | 0.3979 | 0.5025 | | 0.3640 | 1.6930 | 6000 | 0.3588 | 0.4607 | 0.6224 | 0.3963 | 0.5001 | | 0.3548 | 1.8341 | 6500 | 0.3569 | 0.4626 | 0.6369 | 0.3956 | 0.4929 | | 0.3552 | 1.9752 | 7000 | 0.3542 | 0.4521 | 0.5985 | 0.3944 | 0.5122 | | 0.3493 | 2.1163 | 7500 | 0.3535 | 0.4504 | 0.6046 | 0.3931 | 0.5090 | | 0.3463 | 2.2573 | 8000 | 0.3547 | 0.4443 | 0.6105 | 0.3943 | 0.5063 | | 0.3442 | 2.3984 | 8500 | 0.3508 | 0.4459 | 0.5954 | 0.3903 | 0.5135 | | 0.3433 | 2.5395 | 9000 | 0.3487 | 0.4483 | 0.5944 | 0.3910 | 0.5140 | | 0.3415 | 2.6806 | 9500 | 0.3486 | 0.4468 | 0.5959 | 0.3903 | 0.5130 | | 0.3430 | 2.8217 | 10000 | 0.3482 | 0.4480 | 0.5949 | 0.3907 | 0.5131 | | 0.3435 | 2.9628 | 10500 | 0.3482 | 0.4476 | 0.5952 | 0.3905 | 0.5135 | | 0.3435 | 3.0 | 10632 | 0.3482 | 0.4476 | 0.5951 | 0.3903 | 0.5135 | ### Framework versions - PEFT 0.19.1 - Transformers 5.8.1 - Pytorch 2.11.0+cu128 - Datasets 4.8.5 - Tokenizers 0.22.2
Sources & provenance
1 active source Source evidence
3 excerpts This model is a fine-tuned version of xlm-roberta-base on the None dataset. It achieves the following results on the evaluation set: Loss: 0.3482 Mae: 0.4476 Coral Mae: 0.5951 Rounded Mae: 0.3903 Accuracy: 0.5135 Model description
Jul 11
libraryname: peft license: mit basemodel: xlm-roberta-base tags: basemodel:adapter:xlm-roberta-base lora transformers metrics: accuracy model-index: name: cil-ordinal-coral-seed3 results: []
MichaHenh/cil-ordinal-coral-seed3
