Modelpublic

Orcawise/eu_ai_act_orcawise_july12

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face

0likes6downloads

Model Card

results

This model is a fine-tuned version of google/flan-t5-base on the None dataset. It achieves the following results on the evaluation set:

—Loss: 1.9615

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: 0.001
—trainbatchsize: 6
—evalbatchsize: 8
—seed: 42
—gradientaccumulationsteps: 2
—totaltrainbatch_size: 12
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—lrschedulerwarmup_steps: 23
—training_steps: 2373

Training results

Training Loss	Epoch	Step	Validation Loss
3.0035	0.42	50	2.5115
2.7023	0.84	100	2.3978
2.6198	1.26	150	2.3258
2.5523	1.68	200	2.2768
2.4817	2.1	250	2.2360
2.4591	2.52	300	2.2041
2.4	2.94	350	2.1844
2.3709	3.36	400	2.1547
2.3591	3.78	450	2.1366
2.3232	4.2	500	2.1210
2.3016	4.62	550	2.1119
2.3041	5.04	600	2.0993
2.2646	5.46	650	2.0908
2.247	5.88	700	2.0794
2.1935	6.3	750	2.0612
2.2334	6.72	800	2.0573
2.2054	7.14	850	2.0498
2.212	7.56	900	2.0460
2.1687	7.98	950	2.0388
2.1454	8.4	1000	2.0347
2.1344	8.82	1050	2.0243
2.1522	9.24	1100	2.0155
2.1051	9.66	1150	2.0144
2.1435	10.08	1200	2.0152
2.1251	10.5	1250	2.0133
2.0664	10.92	1300	2.0000
2.0656	11.34	1350	2.0002
2.1186	11.76	1400	1.9933
2.0719	12.18	1450	1.9906
2.0389	12.61	1500	1.9913
2.0655	13.03	1550	1.9874
2.0371	13.45	1600	1.9824
2.0581	13.87	1650	1.9789
2.0068	14.29	1700	1.9801
2.0536	14.71	1750	1.9750
2.0311	15.13	1800	1.9729
2.0292	15.55	1850	1.9716
1.9955	15.97	1900	1.9714
2.0056	16.39	1950	1.9671
2.0391	16.81	2000	1.9642
2.0059	17.23	2050	1.9687
2.0155	17.65	2100	1.9644
1.9745	18.07	2150	1.9617
1.9929	18.49	2200	1.9621
1.9978	18.91	2250	1.9639
2.023	19.33	2300	1.9617
1.992	19.75	2350	1.9615

Framework versions

—PEFT 0.8.2
—Transformers 4.38.1
—Pytorch 2.3.0+cu121
—Datasets 2.17.0
—Tokenizers 0.15.2