Edit model card

Visualize in Weights & Biases

FlanT5Summarization-samsum

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

  • Loss: 0.3001
  • Rouge1: 0.2788
  • Rouge2: 0.1310
  • Rougel: 0.2363
  • Rougelsum: 0.2369

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: 3e-05
  • train_batch_size: 128
  • eval_batch_size: 64
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 512
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 3

Training results

Training Loss Epoch Step Validation Loss Rouge1 Rouge2 Rougel Rougelsum
1.1072 0.0866 5 0.9165 0.2705 0.1135 0.2226 0.2229
1.1039 0.1732 10 0.9080 0.2709 0.1138 0.2230 0.2234
1.0848 0.2597 15 0.8917 0.2706 0.1137 0.2228 0.2231
1.0706 0.3463 20 0.8654 0.2709 0.1142 0.2232 0.2234
1.0461 0.4329 25 0.8336 0.2706 0.1140 0.2228 0.2232
1.0187 0.5195 30 0.7960 0.2718 0.1145 0.2240 0.2243
0.9774 0.6061 35 0.7532 0.2723 0.1152 0.2250 0.2253
0.9326 0.6926 40 0.7064 0.2726 0.1153 0.2253 0.2257
0.8834 0.7792 45 0.6570 0.2728 0.1160 0.2259 0.2261
0.833 0.8658 50 0.6080 0.2734 0.1161 0.2262 0.2263
0.7871 0.9524 55 0.5614 0.2726 0.1156 0.2260 0.2260
0.735 1.0390 60 0.5180 0.2731 0.1169 0.2262 0.2264
0.6978 1.1255 65 0.4802 0.2736 0.1179 0.2275 0.2276
0.6464 1.2121 70 0.4482 0.2741 0.1188 0.2283 0.2286
0.6175 1.2987 75 0.4222 0.2742 0.1193 0.2291 0.2292
0.5722 1.3853 80 0.4007 0.2740 0.1187 0.2287 0.2287
0.5443 1.4719 85 0.3834 0.2730 0.1180 0.2282 0.2282
0.5203 1.5584 90 0.3692 0.2740 0.1192 0.2293 0.2293
0.4851 1.6450 95 0.3568 0.2744 0.1201 0.2300 0.2302
0.4619 1.7316 100 0.3466 0.2746 0.1201 0.2304 0.2305
0.4484 1.8182 105 0.3379 0.2754 0.1218 0.2314 0.2319
0.4357 1.9048 110 0.3305 0.2766 0.1241 0.2325 0.2330
0.4246 1.9913 115 0.3243 0.2772 0.1254 0.2338 0.2341
0.4074 2.0779 120 0.3190 0.2776 0.1263 0.2343 0.2347
0.3965 2.1645 125 0.3144 0.2775 0.1264 0.2342 0.2345
0.3922 2.2511 130 0.3105 0.2776 0.1266 0.2344 0.2347
0.3861 2.3377 135 0.3073 0.2786 0.1289 0.2357 0.2362
0.382 2.4242 140 0.3048 0.2782 0.1289 0.2354 0.2358
0.3807 2.5108 145 0.3029 0.2787 0.1297 0.2359 0.2364
0.3717 2.5974 150 0.3016 0.2787 0.1303 0.2363 0.2367
0.3708 2.6840 155 0.3008 0.2788 0.1305 0.2363 0.2368
0.372 2.7706 160 0.3003 0.2789 0.1310 0.2365 0.2370
0.3696 2.8571 165 0.3002 0.2788 0.1310 0.2363 0.2369
0.3646 2.9437 170 0.3001 0.2788 0.1310 0.2363 0.2369

Framework versions

  • PEFT 0.12.0
  • Transformers 4.43.2
  • Pytorch 2.1.2
  • Datasets 2.20.0
  • Tokenizers 0.19.1
Downloads last month
12
Inference API
Inference API (serverless) does not yet support peft models for this pipeline type.

Model tree for zeyadusf/FlanT5Summarization-samsum

Adapter
this model

Dataset used to train zeyadusf/FlanT5Summarization-samsum

Space using zeyadusf/FlanT5Summarization-samsum 1