<!-- 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. -->
qwen_30w12attnproj_oasst1_r8
This model is a fine-tuned version of Qwen/Qwen-14B on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.6472
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: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 0.02
- num_epochs: 1
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
1.4695 | 0.01 | 25 | 1.6788 |
1.5695 | 0.02 | 50 | 1.6711 |
1.6381 | 0.03 | 75 | 1.6688 |
1.5259 | 0.04 | 100 | 1.6682 |
1.5199 | 0.05 | 125 | 1.6691 |
1.5327 | 0.06 | 150 | 1.6657 |
1.5062 | 0.07 | 175 | 1.6659 |
1.5092 | 0.08 | 200 | 1.6679 |
1.4798 | 0.09 | 225 | 1.6633 |
1.5766 | 0.1 | 250 | 1.6632 |
1.5689 | 0.11 | 275 | 1.6616 |
1.5645 | 0.12 | 300 | 1.6630 |
1.4897 | 0.13 | 325 | 1.6620 |
1.5019 | 0.15 | 350 | 1.6609 |
1.5323 | 0.16 | 375 | 1.6600 |
1.5042 | 0.17 | 400 | 1.6589 |
1.5411 | 0.18 | 425 | 1.6608 |
1.4553 | 0.19 | 450 | 1.6580 |
1.4988 | 0.2 | 475 | 1.6597 |
1.6388 | 0.21 | 500 | 1.6576 |
1.5004 | 0.22 | 525 | 1.6566 |
1.5077 | 0.23 | 550 | 1.6555 |
1.5663 | 0.24 | 575 | 1.6554 |
1.4776 | 0.25 | 600 | 1.6561 |
1.6239 | 0.26 | 625 | 1.6578 |
1.5283 | 0.27 | 650 | 1.6550 |
1.551 | 0.28 | 675 | 1.6539 |
1.5924 | 0.29 | 700 | 1.6563 |
1.5488 | 0.3 | 725 | 1.6553 |
1.5221 | 0.31 | 750 | 1.6534 |
1.5351 | 0.32 | 775 | 1.6556 |
1.5076 | 0.33 | 800 | 1.6562 |
1.4865 | 0.34 | 825 | 1.6531 |
1.4816 | 0.35 | 850 | 1.6511 |
1.5093 | 0.36 | 875 | 1.6522 |
1.4883 | 0.37 | 900 | 1.6516 |
1.5056 | 0.38 | 925 | 1.6513 |
1.4656 | 0.39 | 950 | 1.6502 |
1.509 | 0.4 | 975 | 1.6506 |
1.4546 | 0.41 | 1000 | 1.6523 |
1.5579 | 0.43 | 1025 | 1.6507 |
1.5544 | 0.44 | 1050 | 1.6520 |
1.4805 | 0.45 | 1075 | 1.6524 |
1.4934 | 0.46 | 1100 | 1.6525 |
1.5964 | 0.47 | 1125 | 1.6502 |
1.5364 | 0.48 | 1150 | 1.6503 |
1.5445 | 0.49 | 1175 | 1.6481 |
1.555 | 0.5 | 1200 | 1.6485 |
1.5675 | 0.51 | 1225 | 1.6493 |
1.5562 | 0.52 | 1250 | 1.6514 |
1.503 | 0.53 | 1275 | 1.6500 |
1.4849 | 0.54 | 1300 | 1.6505 |
1.5352 | 0.55 | 1325 | 1.6519 |
1.4582 | 0.56 | 1350 | 1.6506 |
1.5267 | 0.57 | 1375 | 1.6480 |
1.5664 | 0.58 | 1400 | 1.6493 |
1.5127 | 0.59 | 1425 | 1.6484 |
1.4997 | 0.6 | 1450 | 1.6489 |
1.5298 | 0.61 | 1475 | 1.6485 |
1.5134 | 0.62 | 1500 | 1.6490 |
1.5701 | 0.63 | 1525 | 1.6498 |
1.4571 | 0.64 | 1550 | 1.6488 |
1.4327 | 0.65 | 1575 | 1.6485 |
1.5029 | 0.66 | 1600 | 1.6482 |
1.5496 | 0.67 | 1625 | 1.6483 |
1.4887 | 0.68 | 1650 | 1.6497 |
1.4946 | 0.69 | 1675 | 1.6490 |
1.5836 | 0.71 | 1700 | 1.6487 |
1.4919 | 0.72 | 1725 | 1.6487 |
1.4796 | 0.73 | 1750 | 1.6488 |
1.4999 | 0.74 | 1775 | 1.6488 |
1.5163 | 0.75 | 1800 | 1.6491 |
1.4711 | 0.76 | 1825 | 1.6492 |
1.4933 | 0.77 | 1850 | 1.6489 |
1.4555 | 0.78 | 1875 | 1.6485 |
1.4461 | 0.79 | 1900 | 1.6484 |
1.5561 | 0.8 | 1925 | 1.6476 |
1.5021 | 0.81 | 1950 | 1.6477 |
1.452 | 0.82 | 1975 | 1.6477 |
1.4898 | 0.83 | 2000 | 1.6474 |
1.4848 | 0.84 | 2025 | 1.6474 |
1.5348 | 0.85 | 2050 | 1.6472 |
1.4752 | 0.86 | 2075 | 1.6477 |
1.504 | 0.87 | 2100 | 1.6476 |
1.5331 | 0.88 | 2125 | 1.6474 |
1.4958 | 0.89 | 2150 | 1.6471 |
1.4691 | 0.9 | 2175 | 1.6472 |
1.5563 | 0.91 | 2200 | 1.6471 |
1.5297 | 0.92 | 2225 | 1.6471 |
1.4896 | 0.93 | 2250 | 1.6471 |
1.5028 | 0.94 | 2275 | 1.6472 |
1.528 | 0.95 | 2300 | 1.6473 |
1.4338 | 0.96 | 2325 | 1.6472 |
1.4975 | 0.98 | 2350 | 1.6473 |
1.4971 | 0.99 | 2375 | 1.6472 |
1.5189 | 1.0 | 2400 | 1.6472 |
Framework versions
- Transformers 4.35.0.dev0
- Pytorch 2.0.1+cu117
- Datasets 2.5.2
- Tokenizers 0.14.0