Hugging Face's logo

NourFakih
/
Vit-GPT2-COCO2017Sample-Flickr8k

Transformers
TensorBoard
Safetensors
vision-encoder-decoder
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
Vit-GPT2-COCO2017Sample-Flickr8k / README.md
NourFakih's picture
NourFakih
Model save
832248b verified
preview code
|
raw
history blame
No virus
5.4 kB
metadata
license: apache-2.0
base_model: NourFakih/image-captioning-Vit-GPT2-Flickr8k
tags:
  - generated_from_trainer
metrics:
  - rouge
model-index:
  - name: Vit-GPT2-COCO2017Sample-Flickr8k
    results: []

Vit-GPT2-COCO2017Sample-Flickr8k

This model is a fine-tuned version of NourFakih/image-captioning-Vit-GPT2-Flickr8k on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2344
  • Rouge1: 41.2779
  • Rouge2: 15.8081
  • Rougel: 37.3177
  • Rougelsum: 37.2772
  • Gen Len: 11.568

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: 5e-05
  • train_batch_size: 4
  • eval_batch_size: 4
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 3.0

Training results

Training Loss Epoch Step Gen Len Validation Loss Rouge1 Rouge2 Rougel Rougelsum
0.2485 0.08 500 11.382 0.2394 40.2445 13.9525 36.1329 36.1261
0.2336 0.16 1000 11.346 0.2376 39.9631 14.4981 36.1164 36.1265
0.2282 0.24 1500 11.07 0.2367 40.3107 14.533 36.489 36.5082
0.2249 0.32 2000 10.87 0.2338 41.0525 15.3076 37.0189 37.0365
0.2302 0.4 2500 11.31 0.2329 40.7052 14.8288 36.9272 36.9197
0.2255 0.48 3000 11.05 0.2321 40.6896 15.2723 36.9654 36.9515
0.2225 0.56 3500 10.946 0.2305 40.705 15.4878 37.0456 37.0235
0.2233 0.64 4000 11.25 0.2303 41.0229 15.179 37.081 37.0924
0.2177 0.72 4500 11.08 0.2307 40.0156 14.2972 36.0288 36.043
0.2159 0.8 5000 11.336 0.2298 40.4042 15.2531 36.6967 36.7003
0.2189 0.88 5500 11.39 0.2282 40.167 14.4847 36.3855 36.3742
0.2171 0.96 6000 11.002 0.2269 40.8528 15.1811 37.0586 37.0403
0.1962 1.04 6500 11.598 0.2296 40.6676 14.9888 36.7796 36.7703
0.1835 1.12 7000 11.022 0.2311 40.6188 15.2743 36.8519 36.8263
0.1835 1.2 7500 11.248 0.2289 40.6466 15.1727 36.6626 36.6427
0.1864 1.28 8000 11.408 0.2298 40.2408 15.0179 36.5594 36.5756
0.1838 1.36 8500 11.238 0.2295 41.0772 15.2152 37.0647 37.0648
0.1827 1.44 9000 11.28 0.2299 40.3263 14.9976 36.6444 36.6292
0.1828 1.52 9500 11.132 0.2299 40.9308 15.181 36.9028 36.8909
0.179 1.61 10000 11.164 0.2287 40.7406 15.2746 36.85 36.8748
0.1849 1.69 10500 10.988 0.2281 40.931 15.6479 37.0222 37.0071
0.1794 1.77 11000 11.218 0.2281 41.5198 15.9659 37.3709 37.386
0.1787 1.85 11500 11.274 0.2278 40.4006 14.9496 36.4608 36.4675
0.1798 1.93 12000 11.154 0.2279 41.3118 15.4673 37.4917 37.5101
0.1803 2.01 12500 11.23 0.2282 40.5652 15.1467 36.7946 36.7809
0.1519 2.09 13000 11.498 0.2361 40.8978 15.0865 36.7157 36.728
0.1515 2.17 13500 11.37 0.2360 40.9809 15.5877 37.0104 36.9942
0.1519 2.25 14000 11.504 0.2359 40.7947 15.3254 36.9574 36.9431
0.1543 2.33 14500 0.2346 40.7724 15.1837 36.9003 36.848 11.586
0.1548 2.41 15000 0.2355 40.7237 15.2394 37.0767 37.0405 11.294
0.1507 2.49 15500 0.2353 41.2661 15.7703 37.3669 37.32 11.308
0.1512 2.57 16000 0.2351 40.8777 15.2821 36.9591 36.9201 11.43
0.1525 2.65 16500 0.2350 40.6184 15.1824 36.655 36.6117 11.402
0.1522 2.73 17000 0.2343 41.2818 15.7174 37.3059 37.2695 11.502
0.1544 2.81 17500 0.2349 41.0821 15.5164 37.2206 37.1663 11.542
0.1498 2.89 18000 0.2346 41.2128 15.6698 37.2279 37.1874 11.582
0.1497 2.97 18500 0.2344 41.2779 15.8081 37.3177 37.2772 11.568

Framework versions

  • Transformers 4.39.3
  • Pytorch 2.1.2
  • Datasets 2.18.0
  • Tokenizers 0.15.2