Edit model card

lib_service_4chan

This model is a fine-tuned version of uer/gpt2-chinese-cluecorpussmall on the lip_service_4chan dataset.

Lip Service 满嘴芬芳,吵架陪练员。

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 0.0002
  • train_batch_size: 4
  • eval_batch_size: 8
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 2
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 32
  • total_eval_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 1.0

Training results

Training Loss Epoch Step Validation Loss
2.716 0.01 100 1.9495
1.8985 0.02 200 1.6915
1.7151 0.02 300 1.5763
1.6217 0.03 400 1.5115
1.564 0.04 500 1.4694
1.5461 0.05 600 1.4379
1.4943 0.06 700 1.4127
1.4737 0.07 800 1.3890
1.4399 0.07 900 1.3813
1.4356 0.08 1000 1.3540
1.3999 0.09 1100 1.3329
1.3668 0.1 1200 1.3153
1.3604 0.11 1300 1.3029
1.3352 0.12 1400 1.2834
1.3278 0.12 1500 1.2619
1.315 0.13 1600 1.2539
1.2854 0.14 1700 1.2432
1.292 0.15 1800 1.2288
1.2795 0.16 1900 1.2188
1.2677 0.16 2000 1.2059
1.2599 0.17 2100 1.2019
1.2479 0.18 2200 1.1915
1.2245 0.19 2300 1.1827
1.2326 0.2 2400 1.1734
1.2124 0.21 2500 1.1660
1.2171 0.21 2600 1.1576
1.1917 0.22 2700 1.1518
1.1867 0.23 2800 1.1444
1.1821 0.24 2900 1.1386
1.1741 0.25 3000 1.1347
1.1753 0.25 3100 1.1293
1.1629 0.26 3200 1.1264
1.1694 0.27 3300 1.1201
1.1482 0.28 3400 1.1146
1.156 0.29 3500 1.1052
1.1512 0.3 3600 1.0982
1.142 0.3 3700 1.0971
1.1544 0.31 3800 1.0920
1.1312 0.32 3900 1.0869
1.1394 0.33 4000 1.0808
1.123 0.34 4100 1.0747
1.1154 0.35 4200 1.0715
1.1064 0.35 4300 1.0674
1.1245 0.36 4400 1.0620
1.1036 0.37 4500 1.0575
1.0963 0.38 4600 1.0568
1.0987 0.39 4700 1.0491
1.0859 0.39 4800 1.0443
1.0845 0.4 4900 1.0432
1.0938 0.41 5000 1.0410
1.087 0.42 5100 1.0334
1.077 0.43 5200 1.0324
1.0787 0.44 5300 1.0276
1.068 0.44 5400 1.0220
1.0748 0.45 5500 1.0199
1.0622 0.46 5600 1.0169
1.0555 0.47 5700 1.0153
1.0498 0.48 5800 1.0100
1.055 0.49 5900 1.0074
1.0424 0.49 6000 1.0020
1.0465 0.5 6100 0.9976
1.0414 0.51 6200 0.9942
1.0355 0.52 6300 0.9919
1.0234 0.53 6400 0.9883
1.0205 0.53 6500 0.9857
1.0316 0.54 6600 0.9805
1.0137 0.55 6700 0.9788
1.0222 0.56 6800 0.9773
1.0219 0.57 6900 0.9722
1.0032 0.58 7000 0.9706
1.0039 0.58 7100 0.9669
1.0166 0.59 7200 0.9635
1.0065 0.6 7300 0.9614
1.0087 0.61 7400 0.9574
0.9968 0.62 7500 0.9525
1.0031 0.62 7600 0.9503
0.99 0.63 7700 0.9491
0.9946 0.64 7800 0.9457
0.9944 0.65 7900 0.9424
0.9854 0.66 8000 0.9399
0.9797 0.67 8100 0.9364
0.9804 0.67 8200 0.9341
0.9835 0.68 8300 0.9318
0.9849 0.69 8400 0.9299
0.9753 0.7 8500 0.9274
0.975 0.71 8600 0.9238
0.9649 0.72 8700 0.9225
0.9654 0.72 8800 0.9202
0.958 0.73 8900 0.9167
0.9679 0.74 9000 0.9143
0.9631 0.75 9100 0.9110
0.9633 0.76 9200 0.9086
0.9495 0.76 9300 0.9071
0.9625 0.77 9400 0.9036
0.9519 0.78 9500 0.9023
0.9399 0.79 9600 0.8993
0.9624 0.8 9700 0.8973
0.9418 0.81 9800 0.8963
0.9394 0.81 9900 0.8933
0.947 0.82 10000 0.8919
0.9326 0.83 10100 0.8900
0.9326 0.84 10200 0.8886
0.9343 0.85 10300 0.8860
0.9263 0.85 10400 0.8841
0.9256 0.86 10500 0.8818
0.9373 0.87 10600 0.8807
0.9314 0.88 10700 0.8789
0.9203 0.89 10800 0.8770
0.927 0.9 10900 0.8754
0.934 0.9 11000 0.8744
0.9193 0.91 11100 0.8727
0.9185 0.92 11200 0.8714
0.9188 0.93 11300 0.8702
0.9165 0.94 11400 0.8693
0.9209 0.95 11500 0.8682
0.9241 0.95 11600 0.8670
0.9182 0.96 11700 0.8662
0.9076 0.97 11800 0.8653
0.9225 0.98 11900 0.8643
0.9094 0.99 12000 0.8640
0.913 0.99 12100 0.8635

Framework versions

  • Transformers 4.33.0
  • Pytorch 2.0.0
  • Datasets 2.1.0
  • Tokenizers 0.13.3
Downloads last month
33
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for qgyd2021/lip_service_4chan

Finetuned
(4)
this model

Spaces using qgyd2021/lip_service_4chan 2