d50-50-50-50-50-50

cshim-cmu/d50-50-50-50-50-50

d50-50-50-50-50-50 is a machine learning model from cshim-cmu. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

4% of all-time downloads

All-time downloads

Public

Repo size

19.3 GB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.bin306 MB · 99%

At a glance

License: apache-2.0
Model type: marian
Access: Public
Created: Apr 27, 2025
Updated: Apr 27, 2025
SHA: 4dfb1fa9

Model card

From the Hugging Face model README

d50-50-50-50-50-50

This model is a fine-tuned version of Helsinki-NLP/opus-mt-es-fi on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.5082
Bleu: 1.9022
Chrf: 30.1958
Gen Len: 36.6217

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: 2e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 20
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Chrf	Gen Len
0.7979	0.09	1000	0.7414	0.646	14.8713	46.995
0.6749	0.18	2000	0.6947	0.9869	17.9392	37.5845
0.6243	0.27	3000	0.6673	0.7999	20.0172	42.3008
0.5971	0.36	4000	0.6503	0.8563	21.2795	39.6429
0.5775	0.45	5000	0.6366	0.8301	21.4941	40.8481
0.5474	0.54	6000	0.6242	0.7819	22.2129	41.5724
0.5374	0.63	7000	0.6168	1.0101	23.4331	45.4386
0.5137	0.72	8000	0.6057	0.9301	22.8011	45.0563
0.5079	0.81	9000	0.6008	1.0671	23.4414	41.6972
0.4927	0.9	10000	0.5920	0.9584	24.1556	37.83
0.4787	0.99	11000	0.5849	1.1807	24.8041	40.0533
0.4652	1.08	12000	0.5814	1.0595	24.6162	42.2334
0.4626	1.18	13000	0.5756	1.3123	25.4887	39.2284
0.4547	1.27	14000	0.5700	1.2358	26.0034	39.8209
0.4417	1.36	15000	0.5671	0.9928	25.548	40.3823
0.438	1.45	16000	0.5643	1.2989	26.6306	41.3461
0.4338	1.54	17000	0.5589	1.219	26.5789	39.4074
0.4255	1.63	18000	0.5526	1.2619	26.695	38.8984
0.4248	1.72	19000	0.5509	1.093	27.6438	38.4306
0.4121	1.81	20000	0.5468	1.3449	27.4379	41.9085
0.4122	1.9	21000	0.5452	1.6956	27.7589	35.4899
0.3985	1.99	22000	0.5432	1.4275	27.9258	40.9235
0.4	2.08	23000	0.5399	1.2224	28.3093	37.3139
0.3929	2.17	24000	0.5386	1.3539	28.2056	39.4356
0.3868	2.26	25000	0.5365	1.3508	28.0525	39.7213
0.3864	2.35	26000	0.5330	1.3298	27.7498	41.2787
0.384	2.44	27000	0.5323	1.5147	28.5066	38.5463
0.3765	2.53	28000	0.5305	1.5193	28.9916	39.5252
0.3794	2.62	29000	0.5314	1.397	28.872	37.6509
0.3734	2.71	30000	0.5269	1.546	28.9768	38.1308
0.374	2.8	31000	0.5277	1.4875	29.2508	36.9759
0.3711	2.89	32000	0.5238	1.5487	29.4803	36.7616
0.3672	2.98	33000	0.5214	1.5123	29.6784	38.5201
0.3637	3.07	34000	0.5207	1.6029	29.8429	37.5825
0.3597	3.16	35000	0.5214	1.4863	29.4697	38.4598
0.3534	3.25	36000	0.5211	1.5307	29.6469	36.0101
0.349	3.34	37000	0.5172	1.7734	29.3285	39.1137
0.3464	3.43	38000	0.5180	1.6407	29.8678	37.5926
0.3475	3.53	39000	0.5160	1.9238	30.0747	37.7928
0.3485	3.62	40000	0.5164	1.7015	29.2713	37.9879
0.3498	3.71	41000	0.5138	1.7873	30.0377	37.3109
0.3465	3.8	42000	0.5148	1.7825	29.866	37.5392
0.3423	3.89	43000	0.5129	1.7935	30.2257	38.7666
0.3447	3.98	44000	0.5106	1.6434	29.5052	39.8028
0.342	4.07	45000	0.5116	1.5715	29.8137	42.5191
0.3376	4.16	46000	0.5112	1.7514	29.9028	38.6911
0.3383	4.25	47000	0.5109	1.6486	30.3114	38.7052
0.3345	4.34	48000	0.5087	1.869	30.0484	37.838
0.3337	4.43	49000	0.5059	1.6667	30.2201	37.6831
0.3333	4.52	50000	0.5076	1.8281	29.9147	38.2918
0.3345	4.61	51000	0.5087	1.8993	30.1002	38.0785
0.3296	4.7	52000	0.5082	1.9022	30.1958	36.6217

Framework versions

Transformers 4.28.1
Pytorch 2.0.0+cu117
Datasets 2.11.0
Tokenizers 0.13.3

Type: marian
License: apache-2.0
Created: Apr 27, 2025
Updated: Apr 27, 2025