Edit model card

test

This model is a fine-tuned version of ahmedabdelwahed/Mojiz-sft on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0000
  • Rewards/chosen: 20.7508
  • Rewards/rejected: -10.7382
  • Rewards/accuracies: 1.0
  • Rewards/margins: 31.4890
  • Logps/rejected: -92.5158
  • Logps/chosen: -284.8114
  • Logits/rejected: -11.6194
  • Logits/chosen: -12.6924

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: 0.0001
  • train_batch_size: 4
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 150
  • training_steps: 8000

Training results

Training Loss Epoch Step Validation Loss Rewards/chosen Rewards/rejected Rewards/accuracies Rewards/margins Logps/rejected Logps/chosen Logits/rejected Logits/chosen
0.0017 0.41 100 0.0000 9.9359 -3.7597 1.0 13.6956 -78.5589 -306.4413 -11.4127 -12.4541
0.0002 0.82 200 0.0000 14.2382 -5.8180 1.0 20.0562 -82.6754 -297.8366 -11.3000 -12.2684
0.0035 1.22 300 0.0000 14.9451 -6.8831 1.0 21.8282 -84.8057 -296.4229 -11.2631 -12.2221
0.0 1.63 400 0.0000 15.6239 -8.0940 1.0 23.7178 -87.2274 -295.0653 -11.2114 -12.1338
0.0 2.04 500 0.0000 15.9950 -8.3192 1.0 24.3142 -87.6779 -294.3232 -11.2217 -12.1499
0.0 2.45 600 0.0000 16.4967 -8.2808 1.0 24.7775 -87.6010 -293.3195 -11.2633 -12.2118
0.0 2.86 700 0.0000 16.2905 -9.0144 1.0 25.3049 -89.0682 -293.7320 -11.2314 -12.1373
0.0 3.27 800 0.0000 17.3895 -7.9312 1.0 25.3208 -86.9019 -291.5340 -11.3726 -12.3633
0.0 3.67 900 0.0000 17.3977 -7.9560 1.0 25.3537 -86.9514 -291.5177 -11.3723 -12.3628
0.0 4.08 1000 0.0000 17.4673 -8.1543 1.0 25.6216 -87.3481 -291.3784 -11.3750 -12.3654
0.0 4.49 1100 0.0000 17.3363 -8.9657 1.0 26.3020 -88.9709 -291.6405 -11.3670 -12.3470
0.0 4.9 1200 0.0000 17.3540 -9.0028 1.0 26.3568 -89.0451 -291.6051 -11.3671 -12.3466
0.0 5.31 1300 0.0000 17.4850 -9.3043 1.0 26.7893 -89.6480 -291.3430 -11.3838 -12.3759
0.0 5.71 1400 0.0000 17.6089 -9.3554 1.0 26.9643 -89.7502 -291.0953 -11.3893 -12.3826
0.0 6.12 1500 0.0000 17.6418 -9.3848 1.0 27.0266 -89.8090 -291.0294 -11.3872 -12.3788
0.0001 6.53 1600 0.0000 17.7200 -9.3570 1.0 27.0770 -89.7534 -290.8731 -11.3975 -12.3941
0.0 6.94 1700 0.0000 17.7617 -9.3377 1.0 27.0994 -89.7148 -290.7896 -11.4020 -12.4017
0.0 7.35 1800 0.0000 17.8247 -9.3772 1.0 27.2019 -89.7938 -290.6637 -11.4033 -12.4039
0.0 7.76 1900 0.0000 17.8638 -9.3928 1.0 27.2566 -89.8251 -290.5855 -11.4046 -12.4052
0.0 8.16 2000 0.0000 18.1144 -9.2188 1.0 27.3332 -89.4771 -290.0843 -11.4242 -12.4400
0.0 8.57 2100 0.0000 18.1229 -9.2243 1.0 27.3472 -89.4881 -290.0672 -11.4242 -12.4401
0.0 8.98 2200 0.0000 18.1432 -9.2739 1.0 27.4171 -89.5872 -290.0266 -11.4281 -12.4420
0.0 9.39 2300 0.0000 18.2729 -9.3131 1.0 27.5860 -89.6657 -289.7673 -11.4278 -12.4441
0.0 9.8 2400 0.0000 18.2914 -9.3532 1.0 27.6446 -89.7459 -289.7303 -11.4279 -12.4436
0.0 10.2 2500 0.0000 18.3550 -9.3675 1.0 27.7225 -89.7745 -289.6031 -11.4324 -12.4488
0.0 10.61 2600 0.0000 18.5092 -9.4395 1.0 27.9487 -89.9185 -289.2947 -11.4477 -12.4716
0.0 11.02 2700 0.0000 18.5278 -9.4387 1.0 27.9666 -89.9169 -289.2574 -11.4484 -12.4728
0.0 11.43 2800 0.0000 18.9266 -9.3672 1.0 28.2938 -89.7738 -288.4599 -11.4894 -12.5273
0.0 11.84 2900 0.0000 18.9978 -9.4237 1.0 28.4215 -89.8868 -288.3174 -11.5000 -12.5400
0.0 12.24 3000 0.0000 19.0186 -9.4479 1.0 28.4665 -89.9352 -288.2759 -11.4983 -12.5375
0.0 12.65 3100 0.0000 19.0213 -9.4485 1.0 28.4698 -89.9365 -288.2705 -11.4994 -12.5392
0.0 13.06 3200 0.0000 19.0656 -9.5104 1.0 28.5759 -90.0602 -288.1819 -11.4988 -12.5380
0.0 13.47 3300 0.0000 19.0811 -9.5638 1.0 28.6449 -90.1670 -288.1508 -11.4994 -12.5412
0.0 13.88 3400 0.0000 19.0755 -9.6303 1.0 28.7058 -90.3000 -288.1620 -11.4984 -12.5391
0.0 14.29 3500 0.0000 19.0764 -9.6361 1.0 28.7124 -90.3116 -288.1603 -11.4984 -12.5390
0.0 14.69 3600 0.0000 19.7645 -9.6207 1.0 29.3852 -90.2808 -286.7841 -11.5674 -12.6283
0.0 15.1 3700 0.0000 19.7594 -9.7019 1.0 29.4613 -90.4432 -286.7942 -11.5659 -12.6252
0.0 15.51 3800 0.0000 19.8213 -9.7241 1.0 29.5454 -90.4877 -286.6704 -11.5693 -12.6319
0.0 15.92 3900 0.0000 19.8591 -9.7267 1.0 29.5857 -90.4928 -286.5949 -11.5754 -12.6423
0.0 16.33 4000 0.0000 20.1637 -10.0565 1.0 30.2202 -91.1524 -285.9856 -11.6035 -12.6809
0.0 16.73 4100 0.0000 20.1671 -10.0572 1.0 30.2244 -91.1539 -285.9789 -11.6039 -12.6816
0.0 17.14 4200 0.0000 20.1791 -10.1186 1.0 30.2977 -91.2767 -285.9549 -11.6032 -12.6803
0.0 17.55 4300 0.0000 20.1786 -10.1726 1.0 30.3512 -91.3847 -285.9559 -11.6026 -12.6788
0.0 17.96 4400 0.0000 20.1663 -10.2017 1.0 30.3680 -91.4428 -285.9804 -11.6022 -12.6778
0.0 18.37 4500 0.0000 20.1651 -10.2076 1.0 30.3727 -91.4546 -285.9829 -11.6021 -12.6777
0.0 18.78 4600 0.0000 20.1509 -10.2578 1.0 30.4087 -91.5550 -286.0112 -11.6017 -12.6762
0.0 19.18 4700 0.0000 20.1784 -10.2457 1.0 30.4241 -91.5308 -285.9563 -11.6037 -12.6793
0.0 19.59 4800 0.0000 20.1812 -10.2503 1.0 30.4315 -91.5400 -285.9507 -11.6040 -12.6798
0.0 20.0 4900 0.0000 20.1823 -10.2604 1.0 30.4428 -91.5603 -285.9484 -11.6041 -12.6798
0.0 20.41 5000 0.0000 20.1883 -10.2616 1.0 30.4499 -91.5626 -285.9364 -11.6051 -12.6818
0.0 20.82 5100 0.0000 20.1896 -10.2675 1.0 30.4571 -91.5745 -285.9339 -11.6051 -12.6819
0.0 21.22 5200 0.0000 20.1736 -10.3226 1.0 30.4962 -91.6847 -285.9659 -11.6057 -12.6823
0.0 21.63 5300 0.0000 20.1824 -10.3241 1.0 30.5065 -91.6877 -285.9483 -11.6061 -12.6830
0.0 22.04 5400 0.0000 20.1732 -10.3699 1.0 30.5431 -91.7793 -285.9666 -11.6051 -12.6797
0.0 22.45 5500 0.0000 20.5647 -10.3381 1.0 30.9027 -91.7156 -285.1837 -11.6065 -12.6773
0.0 22.86 5600 0.0000 20.5540 -10.3886 1.0 30.9426 -91.8166 -285.2050 -11.6059 -12.6761
0.0 23.27 5700 0.0000 20.5442 -10.3824 1.0 30.9267 -91.8043 -285.2246 -11.6076 -12.6788
0.0 23.67 5800 0.0000 20.5517 -10.4140 1.0 30.9657 -91.8675 -285.2097 -11.6099 -12.6809
0.0 24.08 5900 0.0000 20.5647 -10.4280 1.0 30.9927 -91.8955 -285.1837 -11.6096 -12.6804
0.0 24.49 6000 0.0000 20.6521 -10.4626 1.0 31.1147 -91.9646 -285.0089 -11.6107 -12.6823
0.0 24.9 6100 0.0000 20.6569 -10.4643 1.0 31.1212 -91.9680 -284.9993 -11.6109 -12.6826
0.0 25.31 6200 0.0000 20.6600 -10.4637 1.0 31.1238 -91.9669 -284.9930 -11.6118 -12.6838
0.0 25.71 6300 0.0000 20.6544 -10.4876 1.0 31.1420 -92.0146 -285.0042 -11.6117 -12.6833
0.0 26.12 6400 0.0000 20.6428 -10.5264 1.0 31.1692 -92.0923 -285.0274 -11.6141 -12.6869
0.0 26.53 6500 0.0000 20.6443 -10.5316 1.0 31.1758 -92.1026 -285.0245 -11.6142 -12.6869
0.0 26.94 6600 0.0000 20.6314 -10.5251 1.0 31.1566 -92.0897 -285.0502 -11.6162 -12.6900
0.0 27.35 6700 0.0000 20.6378 -10.5259 1.0 31.1637 -92.0912 -285.0375 -11.6175 -12.6919
0.0 27.76 6800 0.0000 20.6497 -10.5256 1.0 31.1754 -92.0907 -285.0136 -11.6195 -12.6951
0.0 28.16 6900 0.0000 20.6415 -10.5752 1.0 31.2167 -92.1899 -285.0301 -11.6187 -12.6923
0.0 28.57 7000 0.0000 20.7394 -10.6843 1.0 31.4237 -92.4081 -284.8342 -11.6178 -12.6906
0.0 28.98 7100 0.0000 20.7446 -10.6882 1.0 31.4328 -92.4159 -284.8239 -11.6186 -12.6916
0.0 29.39 7200 0.0000 20.7502 -10.6915 1.0 31.4417 -92.4224 -284.8127 -11.6190 -12.6923
0.0 29.8 7300 0.0000 20.7515 -10.6967 1.0 31.4482 -92.4328 -284.8100 -11.6190 -12.6923
0.0 30.2 7400 0.0000 20.7524 -10.7011 1.0 31.4535 -92.4416 -284.8083 -11.6192 -12.6925
0.0 30.61 7500 0.0000 20.7499 -10.7111 1.0 31.4610 -92.4616 -284.8133 -11.6191 -12.6922
0.0 31.02 7600 0.0000 20.7487 -10.7160 1.0 31.4647 -92.4715 -284.8157 -11.6192 -12.6922
0.0 31.43 7700 0.0000 20.7477 -10.7229 1.0 31.4705 -92.4852 -284.8177 -11.6191 -12.6919
0.0 31.84 7800 0.0000 20.7512 -10.7255 1.0 31.4766 -92.4904 -284.8107 -11.6191 -12.6921
0.0 32.24 7900 0.0000 20.7510 -10.7372 1.0 31.4881 -92.5138 -284.8111 -11.6195 -12.6924
0.0 32.65 8000 0.0000 20.7508 -10.7382 1.0 31.4890 -92.5158 -284.8114 -11.6194 -12.6924

Framework versions

  • PEFT 0.7.1
  • Transformers 4.36.0
  • Pytorch 2.0.0
  • Datasets 2.1.0
  • Tokenizers 0.15.0
Downloads last month
1
Unable to determine this model’s pipeline type. Check the docs .

Adapter for