Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,62 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
datasets:
|
| 4 |
+
- Magpie-Align/Magpie-Qwen2.5-Pro-300K-Filtered
|
| 5 |
+
base_model:
|
| 6 |
+
- Qwen/Qwen2.5-7B-Instruct
|
| 7 |
+
library_name: peft
|
| 8 |
+
tags:
|
| 9 |
+
- generated_from_trainer
|
| 10 |
+
---
|
| 11 |
+
# cybertron-v4-qw7B-MGS
|
| 12 |
+
|
| 13 |
+
Introducing: **cybertron-v4** based on Qwen2.5 7B
|
| 14 |
+
SFT over Magpie-Align/Magpie-Qwen2.5-Pro-1M-v0.1
|
| 15 |
+
|
| 16 |
+
## Training procedure
|
| 17 |
+
1 Epoch as usual.
|
| 18 |
+
[<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
|
| 19 |
+
### Training hyperparameters
|
| 20 |
+
|
| 21 |
+
The following hyperparameters were used during training:
|
| 22 |
+
- seed: 42
|
| 23 |
+
- distributed_type: multi-GPU
|
| 24 |
+
- num_devices: 8
|
| 25 |
+
- total_train_batch_size: 128
|
| 26 |
+
- total_eval_batch_size: 16
|
| 27 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 28 |
+
- num_epochs: 1
|
| 29 |
+
|
| 30 |
+
### Training results
|
| 31 |
+
|
| 32 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 33 |
+
|:-------------:|:------:|:----:|:---------------:|
|
| 34 |
+
| 0.7405 | 0.0007 | 1 | 0.5760 |
|
| 35 |
+
| 0.6146 | 0.0502 | 71 | 0.5045 |
|
| 36 |
+
| 0.5908 | 0.1003 | 142 | 0.4930 |
|
| 37 |
+
| 0.5669 | 0.1505 | 213 | 0.4854 |
|
| 38 |
+
| 0.5575 | 0.2007 | 284 | 0.4811 |
|
| 39 |
+
| 0.535 | 0.2508 | 355 | 0.4765 |
|
| 40 |
+
| 0.5161 | 0.3010 | 426 | 0.4736 |
|
| 41 |
+
| 0.5268 | 0.3511 | 497 | 0.4726 |
|
| 42 |
+
| 0.5119 | 0.4013 | 568 | 0.4701 |
|
| 43 |
+
| 0.5329 | 0.4515 | 639 | 0.4687 |
|
| 44 |
+
| 0.5167 | 0.5016 | 710 | 0.4673 |
|
| 45 |
+
| 0.5105 | 0.5518 | 781 | 0.4660 |
|
| 46 |
+
| 0.5203 | 0.6020 | 852 | 0.4653 |
|
| 47 |
+
| 0.5035 | 0.6521 | 923 | 0.4646 |
|
| 48 |
+
| 0.4903 | 0.7023 | 994 | 0.4641 |
|
| 49 |
+
| 0.5031 | 0.7525 | 1065 | 0.4628 |
|
| 50 |
+
| 0.5147 | 0.8026 | 1136 | 0.4629 |
|
| 51 |
+
| 0.5037 | 0.8528 | 1207 | 0.4620 |
|
| 52 |
+
| 0.5029 | 0.9029 | 1278 | 0.4620 |
|
| 53 |
+
| 0.492 | 0.9531 | 1349 | 0.4621 |
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
### Framework versions
|
| 57 |
+
|
| 58 |
+
- PEFT 0.13.2
|
| 59 |
+
- Transformers 4.45.2
|
| 60 |
+
- Pytorch 2.3.0+cu121
|
| 61 |
+
- Datasets 3.0.1
|
| 62 |
+
- Tokenizers 0.20.1
|