metadata
license: other
base_model: KnutJaegersberg/Qwen-1_8B-Llamafied
tags:
- alignment-handbook
- trl
- sft
- generated_from_trainer
- trl
- sft
- generated_from_trainer
datasets:
- HuggingFaceH4/ultrachat_200k
model-index:
- name: qwen_1_8B_llamafied
results: []
qwen_1_8B_llamafied
This model is a fine-tuned version of KnutJaegersberg/Qwen-1_8B-Llamafied on the HuggingFaceH4/ultrachat_200k dataset. It achieves the following results on the evaluation set:
- Loss: 1.2486
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: 2e-05
- train_batch_size: 12
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 96
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 120
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.3881 | 0.1 | 100 | 1.3937 |
| 1.3499 | 0.2 | 200 | 1.3372 |
| 1.3138 | 0.3 | 300 | 1.3168 |
| 1.3152 | 0.4 | 400 | 1.3045 |
| 1.2897 | 0.5 | 500 | 1.2954 |
| 1.28 | 0.6 | 600 | 1.2882 |
| 1.2669 | 0.7 | 700 | 1.2820 |
| 1.2591 | 0.8 | 800 | 1.2768 |
| 1.2447 | 0.9 | 900 | 1.2721 |
| 1.2867 | 1.0 | 1000 | 1.2680 |
| 1.1918 | 1.1 | 1100 | 1.2684 |
| 1.2002 | 1.2 | 1200 | 1.2660 |
| 1.1943 | 1.3 | 1300 | 1.2633 |
| 1.199 | 1.4 | 1400 | 1.2607 |
| 1.1887 | 1.5 | 1500 | 1.2581 |
| 1.1987 | 1.6 | 1600 | 1.2556 |
| 1.1954 | 1.7 | 1700 | 1.2534 |
| 1.1869 | 1.8 | 1800 | 1.2511 |
| 1.1744 | 1.9 | 1900 | 1.2492 |
| 1.1718 | 2.0 | 2000 | 1.2486 |
| 1.1456 | 2.1 | 2100 | 1.2532 |
| 1.1204 | 2.2 | 2200 | 1.2529 |
| 1.1347 | 2.3 | 2300 | 1.2519 |
| 1.1312 | 2.4 | 2400 | 1.2513 |
| 1.1229 | 2.5 | 2500 | 1.2508 |
| 1.1287 | 2.6 | 2600 | 1.2500 |
| 1.1252 | 2.7 | 2700 | 1.2500 |
| 1.139 | 2.8 | 2800 | 1.2498 |
| 1.1282 | 2.9 | 2900 | 1.2497 |
Framework versions
- Transformers 4.37.0
- Pytorch 2.1.2+cu118
- Datasets 2.16.1
- Tokenizers 0.15.0