step_1_1300w_bge_0.8

<!-- 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. -->

# train_translate_step1_1321w_bge

This model is a fine-tuned version of [/mnt/workspace/cv_multimodal/jinmu/mangguo_tv/pretrained_weight/Sailor2-L-20B-Chat](https://huggingface.co//mnt/workspace/cv_multimodal/jinmu/mangguo_tv/pretrained_weight/Sailor2-L-20B-Chat) on the train_translate_step1_1321w_bge dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6573

## 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: 16
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 8
- total_train_batch_size: 1024
- total_eval_batch_size: 64
- optimizer: Use 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: 1.0

### Training results

| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:-----:|:---------------:|
| 0.6646 | 1.0 | 12776 | 0.6573 |

### Framework versions

- PEFT 0.15.2
- Transformers 4.51.3
- Pytorch 2.4.0+cu124
- Datasets 3.6.0
- Tokenizers 0.21.1

<!-- 本模型卡片由训练脚本根据训练过程中可获取的信息自动生成。建议您校对并完善该卡片内容后，移除此注释。 -->
# train_translate_step1_1321w_bge

本模型是基于train_translate_step1_1321w_bge数据集，对[/mnt/workspace/cv_multimodal/jinmu/mangguo_tv/pretrained_weight/Sailor2-L-20B-Chat](https://huggingface.co//mnt/workspace/cv_multimodal/jinmu/mangguo_tv/pretrained_weight/Sailor2-L-20B-Chat)进行微调得到的版本。

该模型在验证集上取得如下指标：
- 损失值：0.6573

## 模型描述

详细信息待补充

## 预期用途与局限性

详细信息待补充

## 训练与评估数据

详细信息待补充

## 训练流程

### 训练超参数

本次训练采用的超参数如下：
- 学习率：2×10^-5
- 训练批次大小：16
- 评估批次大小：8
- 随机种子：42
- 分布式训练类型：多GPU
- 训练设备数量：8
- 梯度累积步数：8
- 总训练批次大小：1024
- 总评估批次大小：64
- 优化器：采用adamw_torch优化器，betas参数设置为(0.9, 0.999)，epsilon参数设置为1×10^-8，无额外优化器参数
- 学习率调度器类型：余弦调度器
- 学习率调度器预热比例：0.1
- 训练轮次：1.0

### 训练结果

| 训练损失 | 训练轮次 | 步数 | 验证损失 |
|:-------:|:-------:|:---:|:--------:|
| 0.6646 | 1.0 | 12776 | 0.6573 |

## 框架版本

- 参数高效微调（Parameter-Efficient Fine-Tuning，PEFT） 0.15.2
- Transformers库 4.51.3
- PyTorch 2.4.0+cu124
- Datasets库 3.6.0
- Tokenizers库 0.21.1