LLM EXPLORER
 LLM Explorer: A Curated Large Language Model Directory and Analytics
Dark Theme
  LLM  List    LLM Hosting    LLM Leaderboards    Blog    Newsfeed    Advertise

Phi 2 Ko V0.1 by daekeun-ml

 »  All LLMs  »  daekeun-ml  »  Phi 2 Ko V0.1   URL Share it on

  290   
  Autotrain compatible   Custom code Dataset:maywell/korean textboo...   Dataset:nampdn-ai/tiny-codes   Dataset:open-orca/openorca   Dataset:wikimedia/wikipedia   En   Ko   Phi   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/daekeun-ml/phi-2-ko-v0.1 

Phi 2 Ko V0.1 Benchmarks

ARC: 30.72 vs 96.7 (so35)-68.2%
HellaSwag: 37.26 vs 95.3 (gpt4)-60.9%
MMLU: 27.34 vs 88.3 (so35)-69%
TruthfulQA: 43.64 vs 59 (gpt4)-26%
WinoGrande: 52.72 vs 87.5 (gpt4)-39.7%
GSM8K: 1.29 vs 96.4 (so35)-98.7%
LLME Score: 0.21868
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Phi 2 Ko V0.1 (daekeun-ml/phi-2-ko-v0.1)

Phi 2 Ko V0.1 Parameters and Internals

Model Type 
text generation
Use Cases 
Areas:
research, personal experimentation, proof of concept
Limitations:
Model may not operate correctly as separate verification was not performed
Additional Notes 
This model includes a vocab expansion with Korean vocab. It was developed mainly for personal experimentation, not for commercial use.
Supported Languages 
ko (Korean), en (English)
Training Details 
Data Sources:
wikimedia/wikipedia, maywell/korean_textbooks, nampdn-ai/tiny-codes, Open-Orca/OpenOrca
Data Volume:
5GB
Methodology:
Continued pre-training with vocabulary expansion
Hardware Used:
AWS GPU resources
Model Architecture:
SBERT architecture (BBPE)
Input Output 
Input Format:
Tokenized Korean and English text
Accepted Modalities:
text
Output Format:
Text generation
Performance Tips:
Consider fine-tuning such as instruction tuning or alignment tuning according to your use case.
LLM NamePhi 2 Ko V0.1
Repository 🤗https://huggingface.co/daekeun-ml/phi-2-ko-v0.1 
Model Size2.9b
Required VRAM5.8 GB
Updated2025-02-22
Maintainerdaekeun-ml
Model Typephi
Model Files  5.0 GB: 1-of-2   0.8 GB: 2-of-2
Supported Languagesko en
Model ArchitecturePhiForCausalLM
Licensecc-by-sa-3.0
Context Length2048
Model Max Length2048
Transformers Version4.37.2
Tokenizer ClassCodeGenTokenizer
Padding Token!
Vocabulary Size66676
Torch Data Typebfloat16

Best Alternatives to Phi 2 Ko V0.1

Best Alternatives
Context / RAM
Downloads
Likes
...eechtokenizer Init Data 2404032K / 5.8 GB300
Phi2 Ko Instruction Tune2K / 11.4 GB390

Rank the Phi 2 Ko V0.1 Capabilities

🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟

Instruction Following and Task Automation  
Factuality and Completeness of Knowledge  
Censorship and Alignment  
Data Analysis and Insight Generation  
Text Generation  
Text Summarization and Feature Extraction  
Code Generation  
Multi-Language Support and Translation  

What open-source LLMs or SLMs are you in search of? 43470 in total.

Email us: info@extractum.io. Our Privacy Policy  | Terms and Conditions  | Suggest an improvement.
Our Social Media →  
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241227
  Support LLM Explorer