<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
gpt-expt-sp-v3-K-300-9-mixed-with-tv
This model is a fine-tuned version of gpt2 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0349
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: 0.0005
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 1000
- num_epochs: 500
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
0.3825 | 4.28 | 5000 | 0.1047 |
0.0883 | 8.56 | 10000 | 0.0668 |
0.0633 | 12.84 | 15000 | 0.0505 |
0.0545 | 17.12 | 20000 | 0.0472 |
0.0503 | 21.4 | 25000 | 0.0441 |
0.0477 | 25.68 | 30000 | 0.0420 |
0.0457 | 29.97 | 35000 | 0.0408 |
0.0444 | 34.25 | 40000 | 0.0401 |
0.0435 | 38.53 | 45000 | 0.0399 |
0.0428 | 42.81 | 50000 | 0.0391 |
0.0423 | 47.09 | 55000 | 0.0388 |
0.0418 | 51.37 | 60000 | 0.0386 |
0.0414 | 55.65 | 65000 | 0.0384 |
0.0411 | 59.93 | 70000 | 0.0381 |
0.0408 | 64.21 | 75000 | 0.0381 |
0.0405 | 68.49 | 80000 | 0.0379 |
0.0403 | 72.77 | 85000 | 0.0376 |
0.0401 | 77.05 | 90000 | 0.0375 |
0.0399 | 81.34 | 95000 | 0.0375 |
0.0398 | 85.62 | 100000 | 0.0374 |
0.0396 | 89.9 | 105000 | 0.0373 |
0.0395 | 94.18 | 110000 | 0.0373 |
0.0393 | 98.46 | 115000 | 0.0372 |
0.0392 | 102.74 | 120000 | 0.0370 |
0.0391 | 107.02 | 125000 | 0.0369 |
0.039 | 111.3 | 130000 | 0.0370 |
0.0389 | 115.58 | 135000 | 0.0369 |
0.0388 | 119.86 | 140000 | 0.0369 |
0.0387 | 124.14 | 145000 | 0.0368 |
0.0386 | 128.42 | 150000 | 0.0367 |
0.0385 | 132.71 | 155000 | 0.0367 |
0.0388 | 136.99 | 160000 | 0.0511 |
0.0384 | 141.27 | 165000 | 0.0367 |
0.0383 | 145.55 | 170000 | 0.0366 |
0.0382 | 149.83 | 175000 | 0.0365 |
0.0381 | 154.11 | 180000 | 0.0365 |
0.0381 | 158.39 | 185000 | 0.0364 |
0.038 | 162.67 | 190000 | 0.0364 |
0.0379 | 166.95 | 195000 | 0.0364 |
0.0378 | 171.23 | 200000 | 0.0365 |
0.0378 | 175.51 | 205000 | 0.0363 |
0.0377 | 179.79 | 210000 | 0.0362 |
0.0377 | 184.08 | 215000 | 0.0362 |
0.0376 | 188.36 | 220000 | 0.0362 |
0.0375 | 192.64 | 225000 | 0.0361 |
0.0375 | 196.92 | 230000 | 0.0361 |
0.0374 | 201.2 | 235000 | 0.0361 |
0.0373 | 205.48 | 240000 | 0.0360 |
0.0373 | 209.76 | 245000 | 0.0360 |
0.0372 | 214.04 | 250000 | 0.0360 |
0.0372 | 218.32 | 255000 | 0.0359 |
0.0371 | 222.6 | 260000 | 0.0359 |
0.0371 | 226.88 | 265000 | 0.0359 |
0.037 | 231.16 | 270000 | 0.0359 |
0.037 | 235.45 | 275000 | 0.0358 |
0.0375 | 239.73 | 280000 | 0.0358 |
0.0369 | 244.01 | 285000 | 0.0357 |
0.0368 | 248.29 | 290000 | 0.0357 |
0.0367 | 252.57 | 295000 | 0.0358 |
0.0367 | 256.85 | 300000 | 0.0357 |
0.0368 | 261.13 | 305000 | 0.0356 |
0.0366 | 265.41 | 310000 | 0.0357 |
0.0366 | 269.69 | 315000 | 0.0356 |
0.0365 | 273.97 | 320000 | 0.0355 |
0.0364 | 278.25 | 325000 | 0.0356 |
0.0364 | 282.53 | 330000 | 0.0356 |
0.0363 | 286.81 | 335000 | 0.0355 |
0.0363 | 291.1 | 340000 | 0.0355 |
0.0362 | 295.38 | 345000 | 0.0354 |
0.0362 | 299.66 | 350000 | 0.0355 |
0.0361 | 303.94 | 355000 | 0.0354 |
0.0361 | 308.22 | 360000 | 0.0354 |
0.036 | 312.5 | 365000 | 0.0354 |
0.036 | 316.78 | 370000 | 0.0354 |
0.036 | 321.06 | 375000 | 0.0353 |
0.0359 | 325.34 | 380000 | 0.0353 |
0.0359 | 329.62 | 385000 | 0.0354 |
0.0358 | 333.9 | 390000 | 0.0353 |
0.0358 | 338.18 | 395000 | 0.0353 |
0.0357 | 342.47 | 400000 | 0.0352 |
0.0358 | 346.75 | 405000 | 0.0351 |
0.0356 | 351.03 | 410000 | 0.0352 |
0.0356 | 355.31 | 415000 | 0.0352 |
0.0356 | 359.59 | 420000 | 0.0352 |
0.0355 | 363.87 | 425000 | 0.0352 |
0.0355 | 368.15 | 430000 | 0.0351 |
0.0355 | 372.43 | 435000 | 0.0351 |
0.0354 | 376.71 | 440000 | 0.0351 |
0.0354 | 380.99 | 445000 | 0.0350 |
0.0354 | 385.27 | 450000 | 0.0351 |
0.0353 | 389.55 | 455000 | 0.0351 |
0.0353 | 393.84 | 460000 | 0.0350 |
0.0353 | 398.12 | 465000 | 0.0350 |
0.0352 | 402.4 | 470000 | 0.0350 |
0.0352 | 406.68 | 475000 | 0.0350 |
0.0352 | 410.96 | 480000 | 0.0350 |
0.0352 | 415.24 | 485000 | 0.0350 |
0.0351 | 419.52 | 490000 | 0.0350 |
0.0351 | 423.8 | 495000 | 0.0349 |
0.0351 | 428.08 | 500000 | 0.0349 |
0.0351 | 432.36 | 505000 | 0.0349 |
0.0351 | 436.64 | 510000 | 0.0349 |
0.035 | 440.92 | 515000 | 0.0349 |
0.035 | 445.21 | 520000 | 0.0349 |
0.035 | 449.49 | 525000 | 0.0349 |
0.035 | 453.77 | 530000 | 0.0349 |
0.035 | 458.05 | 535000 | 0.0349 |
0.035 | 462.33 | 540000 | 0.0349 |
0.0349 | 466.61 | 545000 | 0.0349 |
0.0349 | 470.89 | 550000 | 0.0349 |
0.0349 | 475.17 | 555000 | 0.0349 |
0.0349 | 479.45 | 560000 | 0.0349 |
0.0349 | 483.73 | 565000 | 0.0349 |
0.0349 | 488.01 | 570000 | 0.0349 |
0.0349 | 492.29 | 575000 | 0.0349 |
0.0349 | 496.58 | 580000 | 0.0349 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.1+cu117
- Datasets 2.8.0
- Tokenizers 0.13.2