README.md

metadata

library_name: transformers
license: apache-2.0
base_model: google/flan-t5-base
tags:
  - generated_from_trainer
metrics:
  - rouge
model-index:
  - name: flan-t5-rouge-durga-3b
    results: []

flan-t5-rouge-durga-3b

This model is a fine-tuned version of google/flan-t5-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2135
• Rouge1: 0.4326
• Rouge2: 0.3704
• Rougel: 0.4285
• Rougelsum: 0.4288

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: 0.0003
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
2.2165	1.0	2000	1.9218	0.2604	0.1215	0.2356	0.2356
1.8357	2.0	4000	1.6767	0.2496	0.1181	0.2262	0.2264
1.11	3.0	6000	1.4820	0.2709	0.1363	0.2460	0.2461
1.9651	4.0	8000	1.3179	0.2750	0.1410	0.2505	0.2506
1.0764	5.0	10000	1.1559	0.2768	0.1493	0.2586	0.2586
1.0195	6.0	12000	1.0138	0.2885	0.1612	0.2685	0.2689
0.7455	7.0	14000	0.8792	0.2965	0.1739	0.2788	0.2788
0.9384	8.0	16000	0.7466	0.2998	0.1841	0.2849	0.2849
1.3216	9.0	18000	0.6505	0.3168	0.2041	0.3007	0.3010
0.9445	10.0	20000	0.5516	0.3268	0.2238	0.3133	0.3140
1.2079	11.0	22000	0.4781	0.3370	0.2381	0.3236	0.3237
0.7909	12.0	24000	0.4186	0.3394	0.2439	0.3249	0.3250
0.4352	13.0	26000	0.3459	0.3630	0.2749	0.3503	0.3507
0.4925	14.0	28000	0.3134	0.3665	0.2798	0.3548	0.3553
0.0707	15.0	30000	0.2752	0.3828	0.3025	0.3724	0.3728
0.2351	16.0	32000	0.2579	0.3815	0.3009	0.3708	0.3711
0.9694	17.0	34000	0.2439	0.3874	0.3103	0.3788	0.3795
0.6217	18.0	36000	0.2273	0.4037	0.3282	0.3956	0.3959
0.0914	19.0	38000	0.2121	0.4055	0.3343	0.3981	0.3987
0.0534	20.0	40000	0.2006	0.4134	0.3444	0.4066	0.4067
0.1081	21.0	42000	0.2009	0.4181	0.3514	0.4131	0.4132
0.0786	22.0	44000	0.2086	0.4182	0.3488	0.4112	0.4118
0.0364	23.0	46000	0.2047	0.4182	0.3508	0.4110	0.4114
0.0568	24.0	48000	0.2058	0.4227	0.3570	0.4175	0.4177
0.1106	25.0	50000	0.2073	0.4211	0.3566	0.4159	0.4164
0.0733	26.0	52000	0.2075	0.4248	0.3603	0.4195	0.4199
0.1169	27.0	54000	0.2123	0.4286	0.3650	0.4244	0.4247
0.0892	28.0	56000	0.2109	0.4327	0.3704	0.4281	0.4285
0.045	29.0	58000	0.2107	0.4339	0.3715	0.4300	0.4302
0.0908	30.0	60000	0.2135	0.4326	0.3704	0.4285	0.4288

Framework versions

• Transformers 4.46.0
• Pytorch 2.5.0+cu121
• Datasets 3.0.2
• Tokenizers 0.20.1