89 lines
2.8 KiB
Markdown
89 lines
2.8 KiB
Markdown
---
|
|
library_name: transformers
|
|
license: other
|
|
base_model: Qwen/Qwen3-4B-Instruct-2507
|
|
tags:
|
|
- llama-factory
|
|
- full
|
|
- generated_from_trainer
|
|
model-index:
|
|
- name: qwen3-4B-instruct-refiner-sft
|
|
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. -->
|
|
|
|
# qwen3-4B-instruct-refiner-sft
|
|
|
|
This model is a fine-tuned version of [Qwen/Qwen3-4B-Instruct-2507](https://huggingface.co/Qwen/Qwen3-4B-Instruct-2507) on the refiner_sft_hard_filtered_train dataset.
|
|
It achieves the following results on the evaluation set:
|
|
- Loss: 1.1232
|
|
|
|
## 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: 2
|
|
- eval_batch_size: 2
|
|
- seed: 42
|
|
- gradient_accumulation_steps: 16
|
|
- total_train_batch_size: 32
|
|
- 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: 0.05
|
|
- num_epochs: 5
|
|
|
|
### Training results
|
|
|
|
| Training Loss | Epoch | Step | Validation Loss |
|
|
|:-------------:|:------:|:----:|:---------------:|
|
|
| 0.4937 | 0.1874 | 100 | 0.6320 |
|
|
| 0.511 | 0.3749 | 200 | 0.6321 |
|
|
| 0.4657 | 0.5623 | 300 | 0.6459 |
|
|
| 0.4577 | 0.7498 | 400 | 0.6420 |
|
|
| 0.4634 | 0.9372 | 500 | 0.6470 |
|
|
| 0.2661 | 1.1256 | 600 | 0.6921 |
|
|
| 0.2427 | 1.3130 | 700 | 0.6904 |
|
|
| 0.2608 | 1.5005 | 800 | 0.6896 |
|
|
| 0.2811 | 1.6879 | 900 | 0.6763 |
|
|
| 0.2506 | 1.8754 | 1000 | 0.6782 |
|
|
| 0.1031 | 2.0619 | 1100 | 0.7820 |
|
|
| 0.1053 | 2.2493 | 1200 | 0.7939 |
|
|
| 0.1009 | 2.4367 | 1300 | 0.7773 |
|
|
| 0.1022 | 2.6242 | 1400 | 0.7983 |
|
|
| 0.1087 | 2.8116 | 1500 | 0.8067 |
|
|
| 0.1046 | 2.9991 | 1600 | 0.8037 |
|
|
| 0.0311 | 3.1856 | 1700 | 0.9448 |
|
|
| 0.0343 | 3.3730 | 1800 | 0.9443 |
|
|
| 0.0322 | 3.5604 | 1900 | 0.9526 |
|
|
| 0.0299 | 3.7479 | 2000 | 0.9680 |
|
|
| 0.0335 | 3.9353 | 2100 | 0.9606 |
|
|
| 0.0073 | 4.1218 | 2200 | 1.0976 |
|
|
| 0.0069 | 4.3093 | 2300 | 1.1145 |
|
|
| 0.0064 | 4.4967 | 2400 | 1.1218 |
|
|
| 0.0086 | 4.6842 | 2500 | 1.1228 |
|
|
| 0.0072 | 4.8716 | 2600 | 1.1233 |
|
|
|
|
|
|
### Framework versions
|
|
|
|
- Transformers 4.52.4
|
|
- Pytorch 2.10.0+cu128
|
|
- Datasets 4.8.4
|
|
- Tokenizers 0.21.1
|