README.md

metadata

library_name: transformers
license: apache-2.0
base_model: answerdotai/ModernBERT-base
tags:
  - generated_from_trainer
model-index:
  - name: eternis_router_encoder_sft_5Sep
    results: []

eternis_router_encoder_sft_5Sep

This model is a fine-tuned version of answerdotai/ModernBERT-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1954
• Mse: 0.1954
• Mae: 0.1976
• Vector Accuracy: 0.2235
• Complexity Accuracy: 0.8013
• Accuracy Accuracy: 0.9885
• Completeness Accuracy: 0.9928
• Clarity Accuracy: 0.997
• Relevance Accuracy: 0.9978
• Model Accuracy: 0.2898

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: 0.002
• train_batch_size: 16
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use 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_ratio: 0.06
• num_epochs: 6

Training results

Training Loss	Epoch	Step	Validation Loss	Mse	Mae	Vector Accuracy	Complexity Accuracy	Accuracy Accuracy	Completeness Accuracy	Clarity Accuracy	Relevance Accuracy	Model Accuracy
0.425	0.2857	250	0.2167	0.2167	0.2256	0.164	0.7642	0.9885	0.9928	0.997	0.9978	0.2157
0.4162	0.5714	500	0.2129	0.2129	0.2096	0.2405	0.7745	0.9885	0.9928	0.997	0.9978	0.3235
0.3955	0.8571	750	0.2135	0.2135	0.2140	0.1708	0.782	0.9885	0.9928	0.997	0.9978	0.246
0.3864	1.1429	1000	0.2014	0.2014	0.2046	0.195	0.8035	0.9885	0.9928	0.997	0.9978	0.254
0.4043	1.4286	1250	0.2029	0.2029	0.2086	0.1893	0.806	0.9885	0.9928	0.997	0.9978	0.2507
0.3942	1.7143	1500	0.2046	0.2046	0.2022	0.233	0.804	0.9885	0.9928	0.997	0.9978	0.2935
0.3952	2.0	1750	0.2103	0.2103	0.2196	0.1762	0.721	0.9885	0.9928	0.997	0.9978	0.2622
0.3929	2.2857	2000	0.2011	0.2011	0.2014	0.2305	0.788	0.9885	0.9928	0.997	0.9978	0.3023
0.3921	2.5714	2250	0.1986	0.1986	0.2019	0.2258	0.7778	0.9885	0.9928	0.997	0.9978	0.3045
0.3924	2.8571	2500	0.1981	0.1981	0.1980	0.235	0.8043	0.9885	0.9928	0.997	0.9978	0.2988
0.3819	3.1429	2750	0.2035	0.2035	0.2084	0.218	0.7638	0.9885	0.9928	0.997	0.9978	0.294
0.3874	3.4286	3000	0.1970	0.1970	0.1963	0.2233	0.8073	0.9885	0.9928	0.997	0.9978	0.286
0.3934	3.7143	3250	0.1994	0.1994	0.2079	0.184	0.786	0.9885	0.9928	0.997	0.9978	0.2487
0.3813	4.0	3500	0.1985	0.1985	0.1942	0.245	0.8005	0.9885	0.9928	0.997	0.9978	0.314
0.3939	4.2857	3750	0.1986	0.1986	0.2017	0.1905	0.8033	0.9885	0.9928	0.997	0.9978	0.2507
0.3985	4.5714	4000	0.1956	0.1956	0.1993	0.2062	0.797	0.9885	0.9928	0.997	0.9978	0.273
0.378	4.8571	4250	0.1960	0.1960	0.1991	0.227	0.7887	0.9885	0.9928	0.997	0.9978	0.2983
0.3853	5.1429	4500	0.1957	0.1957	0.1982	0.2122	0.803	0.9885	0.9928	0.997	0.9978	0.2747
0.3727	5.4286	4750	0.1955	0.1955	0.1989	0.2122	0.8025	0.9885	0.9928	0.997	0.9978	0.2745
0.3826	5.7143	5000	0.1956	0.1956	0.1975	0.2278	0.8007	0.9885	0.9928	0.997	0.9978	0.2945
0.3746	6.0	5250	0.1954	0.1954	0.1976	0.2235	0.8013	0.9885	0.9928	0.997	0.9978	0.2898

Framework versions

• Transformers 4.56.1
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.0