---
library_name: transformers
license: apache-2.0
basemodel: beomi/llama-2-koen-13b
datasets:
- Saxo/total_ko_train_set_small_basic
- beomi/KoAlpaca-v1.1a
- kyujinpy/KOR-OpenOrca-Platypus-v2
- nlpai-lab/databricks-dolly-15k-ko
language:
- ko
- en
pipeline_tag: text-generation
---

# Model Card for Model ID

<div align="center">
<img src="https://www.linkbricks.com/wp-content/uploads/2022/03/%E1%84%85%E1%85%B5%E1%86%BC%E1%84%8F%E1%85%B3%E1%84%87%E1%85%B3%E1%84%85%E1%85%B5%E1%86%A8%E1%84%89%E1%85%B3%E1%84%85%E1%85%A9%E1%84%80%E1%85%A9-2-1024x804.png" />
</div>


AI 와 빅데이터 분석 전문 기업인 Linkbricks의 데이터사이언티스트인 지윤성 박사(Saxo)가 beomi/llama-2-koen-13b 베이스모델을 GCP상의 A100-40G 4개를 통해 SFT 훈련을 한(2048 Tokens) 인스트럭션 모델.
 Accelerate, Deepspeed Zero-3 라이브러리를 사용했으며 Flash Attention 은 Disable  로 설정

Dr. Yunsung Ji (Saxo), a data scientist at Linkbricks, a company specializing in AI and big data analytics, trained the beomi/llama-2-koen-13b base model on 4 A100-40Gs on GCP for 4 hours of instructional training (2048 Tokens).
Accelerate, Deepspeed Zero-3 libraries were used. 

www.linkbricks.com, www.linkbricks.vc

## Configuration including BitsandBytes
---
learning_rate = 2e-4
num_epochs = 5 
batch_size = 4 
block_size = 2048 
trainer = "sft" 
warmup_ratio = 0.1 
weight_decay = 0.01 
gradient_accumulation = 4 
mixed_precision = "fp16" 
peft = True 
quantization = "int4" 
lora_r = 64 
lora_alpha = 16 
lora_dropout = 0.1 
model_max_length = 2048 
---
## Dataset Format
Alpaca Format Prompt Text