EnumaInc/ko-TinyLlama-1.1B-vocab-extend-mix-corpus-basic-tokenizer-eeve-stage1

---
license: apache-2.0
base_model: EnumaInc/ko-TinyLlama-1.1B-intermediate-step-1431k-3Tb-vocab-extend-45000-untrained-v1
tags:
- generated_from_trainer
model-index:
- name: out
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. -->

[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
<details><summary>See axolotl config</summary>

axolotl version: `0.4.0`
```yaml
base_model: EnumaInc/ko-TinyLlama-1.1B-intermediate-step-1431k-3Tb-vocab-extend-45000-untrained-v1
model_type: LlamaForCausalLM
tokenizer_type: LlamaTokenizer

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
- path: /root/axolotl/datasets/mix_corpus_extended_validated.json
type: completion
field: text
dataset_prepared_path:

val_set_size: 0.01
output_dir: ./out

sequence_len: 2048
sample_packing: true
pad_to_sequence_len: true
eval_sample_packing: false

wandb_project: language-transfer-eeve-v2
wandb_entity:
wandb_watch:
wandb_name: eeve-v2-stage1
wandb_log_model:

gradient_accumulation_steps: 8
micro_batch_size: 32
num_epochs: 1
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.00015

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 500
evals_per_epoch: 1
eval_table_size:
eval_max_new_tokens: 128

save_strategy: steps
save_steps: 100
save_total_limit: 5

#saves_per_epoch: 1

debug:
deepspeed:
weight_decay: 0.1
fsdp:
fsdp_config:
special_tokens:

# for curriculum learning
shuffle_merged_datasets: false

unfrozen_parameters:
- ^model.embed_tokens.weight$[32000:]
# - model.layers.2[0-9]+.block_sparse_moe.gate
# - model.layers.2[0-9]+.block_sparse_moe.experts
# - model.layers.3[0-9]+.block_sparse_moe.gate
# - model.layers.3[0-9]+.block_sparse_moe.experts

```

</details><br>

# out

This model is a fine-tuned version of [EnumaInc/ko-TinyLlama-1.1B-intermediate-step-1431k-3Tb-vocab-extend-45000-untrained-v1](https://huggingface.co/EnumaInc/ko-TinyLlama-1.1B-intermediate-step-1431k-3Tb-vocab-extend-45000-untrained-v1) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.6306

## 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: 0.00015
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 8
- total_train_batch_size: 1024
- total_eval_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 500
- num_epochs: 1

### Training results

| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 1.651 | 1.0 | 1918 | 1.6306 |

### Framework versions

- Transformers 4.40.0.dev0
- Pytorch 2.1.2+cu118
- Datasets 2.18.0
- Tokenizers 0.15.0

原始信息汇总

数据集概述

数据集基本信息

• 许可证: Apache-2.0
• 基础模型: EnumaInc/ko-TinyLlama-1.1B-intermediate-step-1431k-3Tb-vocab-extend-45000-untrained-v1
• 模型类型: LlamaForCausalLM
• 分词器类型: LlamaTokenizer

训练配置

• 数据集路径: /root/axolotl/datasets/mix_corpus_extended_validated.json
• 数据集类型: completion
• 字段: text
• 验证集大小: 0.01
• 输出目录: ./out
• 序列长度: 2048
• 微批量大小: 32
• 优化器: adamw_bnb_8bit
• 学习率: 0.00015
• 梯度累积步数: 8
• 训练周期数: 1
• 权重衰减: 0.1

评估配置

• 评估样本打包: false
• 评估最大新令牌数: 128

保存策略

• 保存策略: steps
• 保存步数: 100
• 保存总数限制: 5

训练结果

• 训练损失: 1.651
• 验证损失: 1.6306
• 训练步骤: 1918
• 评估批次大小: 32

框架版本

• Transformers: 4.40.0.dev0
• Pytorch: 2.1.2+cu118
• Datasets: 2.18.0
• Tokenizers: 0.15.0