File size: 4,359 Bytes
a81aace |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 |
---
library_name: peft
license: other
base_model: Qwen/Qwen3-32B
tags:
- llama-factory
- lora
- generated_from_trainer
model-index:
- name: Qwen3-32B-alpaca-th-52k-dolly-th-15k-wangchan-instruct
results: []
---
<!-- 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. -->
# Qwen3-32B-alpaca-th-52k-dolly-th-15k-wangchan-instruct
This model is a fine-tuned version of [Qwen/Qwen3-32B](https://huggingface.co/Qwen/Qwen3-32B) on the alpaca-th-52k, the dolly-th-15k and the wangchan-instruct datasets.
It achieves the following results on the evaluation set:
- Loss: 0.6417
## 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: 0.0002
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- distributed_type: multi-GPU
- num_devices: 32
- gradient_accumulation_steps: 8
- total_train_batch_size: 512
- total_eval_batch_size: 64
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:------:|:----:|:---------------:|
| 0.9564 | 0.0575 | 10 | 1.0507 |
| 0.806 | 0.1149 | 20 | 0.8268 |
| 0.7551 | 0.1724 | 30 | 0.7598 |
| 0.7158 | 0.2299 | 40 | 0.7396 |
| 0.7217 | 0.2874 | 50 | 0.7252 |
| 0.7078 | 0.3448 | 60 | 0.7130 |
| 0.6719 | 0.4023 | 70 | 0.7029 |
| 0.6855 | 0.4598 | 80 | 0.6964 |
| 0.7328 | 0.5172 | 90 | 0.6907 |
| 0.6663 | 0.5747 | 100 | 0.6848 |
| 0.7049 | 0.6322 | 110 | 0.6792 |
| 0.6772 | 0.6897 | 120 | 0.6751 |
| 0.687 | 0.7471 | 130 | 0.6721 |
| 0.6786 | 0.8046 | 140 | 0.6700 |
| 0.6389 | 0.8621 | 150 | 0.6672 |
| 0.6673 | 0.9195 | 160 | 0.6649 |
| 0.6711 | 0.9770 | 170 | 0.6633 |
| 0.6614 | 1.0345 | 180 | 0.6615 |
| 0.6219 | 1.0920 | 190 | 0.6602 |
| 0.6542 | 1.1494 | 200 | 0.6587 |
| 0.6596 | 1.2069 | 210 | 0.6572 |
| 0.6526 | 1.2644 | 220 | 0.6567 |
| 0.657 | 1.3218 | 230 | 0.6551 |
| 0.6124 | 1.3793 | 240 | 0.6537 |
| 0.6489 | 1.4368 | 250 | 0.6526 |
| 0.614 | 1.4943 | 260 | 0.6515 |
| 0.656 | 1.5517 | 270 | 0.6504 |
| 0.6255 | 1.6092 | 280 | 0.6492 |
| 0.6419 | 1.6667 | 290 | 0.6486 |
| 0.6275 | 1.7241 | 300 | 0.6473 |
| 0.6324 | 1.7816 | 310 | 0.6466 |
| 0.6334 | 1.8391 | 320 | 0.6461 |
| 0.6213 | 1.8966 | 330 | 0.6452 |
| 0.6269 | 1.9540 | 340 | 0.6443 |
| 0.6408 | 2.0115 | 350 | 0.6437 |
| 0.6213 | 2.0690 | 360 | 0.6441 |
| 0.6146 | 2.1264 | 370 | 0.6440 |
| 0.6572 | 2.1839 | 380 | 0.6438 |
| 0.6264 | 2.2414 | 390 | 0.6435 |
| 0.6051 | 2.2989 | 400 | 0.6434 |
| 0.5983 | 2.3563 | 410 | 0.6429 |
| 0.6388 | 2.4138 | 420 | 0.6425 |
| 0.6227 | 2.4713 | 430 | 0.6425 |
| 0.6335 | 2.5287 | 440 | 0.6421 |
| 0.6247 | 2.5862 | 450 | 0.6420 |
| 0.6404 | 2.6437 | 460 | 0.6418 |
| 0.6218 | 2.7011 | 470 | 0.6418 |
| 0.6368 | 2.7586 | 480 | 0.6417 |
| 0.6191 | 2.8161 | 490 | 0.6417 |
| 0.6234 | 2.8736 | 500 | 0.6417 |
| 0.6079 | 2.9310 | 510 | 0.6417 |
| 0.6243 | 2.9885 | 520 | 0.6417 |
### Framework versions
- PEFT 0.15.2
- Transformers 4.52.3
- Pytorch 2.7.0+cu126
- Datasets 3.6.0
- Tokenizers 0.21.1 |