README.md

metadata

license: mit
base_model: unicamp-dl/ptt5-base-portuguese-vocab
tags:
  - generated_from_trainer
datasets:
  - xlsum
metrics:
  - rouge
model-index:
  - name: ptt5-xlsumm-30epochs
    results:
      - task:
          name: Sequence-to-sequence Language Modeling
          type: text2text-generation
        dataset:
          name: xlsum
          type: xlsum
          config: portuguese
          split: test
          args: portuguese
        metrics:
          - name: Rouge1
            type: rouge
            value: 0.3246

ptt5-xlsumm-30epochs

This model is a fine-tuned version of unicamp-dl/ptt5-base-portuguese-vocab on the xlsum dataset. It achieves the following results on the evaluation set:

• Loss: 2.1787
• Rouge1: 0.3246
• Rouge2: 0.1471
• Rougel: 0.2617
• Rougelsum: 0.2641
• Gen Len: 18.7065

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: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.5549	1.0	28701	2.2631	0.3049	0.1232	0.2416	0.2436	18.8318
2.4878	2.0	57402	2.2000	0.3134	0.1322	0.2497	0.2519	18.7965
2.3021	3.0	86103	2.1654	0.3161	0.1354	0.2528	0.255	18.732
2.2808	4.0	114804	2.1423	0.3179	0.1376	0.2544	0.2565	18.7217
2.2731	5.0	143505	2.1291	0.3202	0.1405	0.2567	0.259	18.7037
2.1654	6.0	172206	2.1209	0.3209	0.1417	0.2577	0.2598	18.6956
2.1716	7.0	200907	2.1173	0.3213	0.1423	0.2584	0.2606	18.7256
2.0696	8.0	229608	2.1136	0.3234	0.1441	0.2603	0.2627	18.7352
2.0492	9.0	258309	2.1123	0.3214	0.1425	0.2589	0.261	18.6357
2.0953	10.0	287010	2.1136	0.3244	0.146	0.2611	0.2634	18.7001
2.0358	11.0	315711	2.1180	0.3248	0.1466	0.2617	0.2639	18.6868
1.9475	12.0	344412	2.1191	0.3243	0.1463	0.2614	0.2637	18.6707
2.0194	13.0	373113	2.1181	0.3253	0.1466	0.2616	0.264	18.6939
1.925	14.0	401814	2.1236	0.3232	0.1454	0.2604	0.2629	18.6843
1.9194	15.0	430515	2.1294	0.3239	0.1464	0.2612	0.2636	18.6792
1.9163	16.0	459216	2.1301	0.3248	0.1464	0.261	0.2635	18.701
1.8482	17.0	487917	2.1366	0.325	0.1473	0.2619	0.2644	18.6786
1.8637	18.0	516618	2.1387	0.3263	0.1483	0.2624	0.2648	18.6811
1.8496	19.0	545319	2.1425	0.3244	0.1461	0.2613	0.2637	18.6934
1.8565	20.0	574020	2.1513	0.3257	0.1479	0.2626	0.2649	18.702
1.7683	21.0	602721	2.1559	0.3261	0.1482	0.2622	0.2646	18.718
1.7483	22.0	631422	2.1577	0.3254	0.1482	0.2625	0.2649	18.6939
1.7832	23.0	660123	2.1614	0.3234	0.147	0.2616	0.264	18.7033
1.8002	24.0	688824	2.1625	0.3246	0.1477	0.2626	0.2649	18.682
1.7381	25.0	717525	2.1689	0.3253	0.1473	0.2617	0.2641	18.7289
1.7367	26.0	746226	2.1677	0.3255	0.1475	0.2626	0.2649	18.7015
1.752	27.0	774927	2.1760	0.3255	0.1482	0.2631	0.2654	18.7146
1.7595	28.0	803628	2.1753	0.3241	0.1468	0.2616	0.264	18.7036
1.777	29.0	832329	2.1785	0.3246	0.1474	0.2618	0.2643	18.7089
1.7142	30.0	861030	2.1787	0.3246	0.1471	0.2617	0.2641	18.7065

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3