Model Card for Model ID
AI μ λΉ
λ°μ΄ν° λΆμ μ λ¬Έ κΈ°μ
μΈ Linkbricksμ λ°μ΄ν°μ¬μ΄μΈν°μ€νΈμΈ μ§μ€μ±(Saxo) μ΄μ¬κ°
Hermes-3-Llama-3.1-8B λ² μ΄μ€λͺ¨λΈμ μ¬μ©ν΄μ H100-80G 8κ°λ₯Ό ν΅ν΄ μ½ 25%μ λμ νλΌλ―Έν°λ₯Ό νκ΅μ΄ CPT(Continued-Pretraining)->SFT->DPO ν νκΈ μΈμ΄ λͺ¨λΈ
μ²λ§κ±΄μ νκΈ λ΄μ€ μ½νΌμ€λ₯Ό κΈ°μ€μΌλ‘ λ€μν ν
μ€ν¬λ³ νκ΅μ΄-μ€κ΅μ΄-μμ΄-μΌλ³Έμ΄ κ΅μ°¨ νμ΅ λ°μ΄ν°μ μν λ° λ
Όλ¦¬νλ¨ λ°μ΄ν°λ₯Ό ν΅νμ¬ νμ€μΌμ μΈμ΄ κ΅μ°¨ μ¦κ° μ²λ¦¬μ 볡μ‘ν λ
Όλ¦¬ λ¬Έμ μμ λμ κ°λ₯νλλ‘ νλ ¨ν λͺ¨λΈμ΄λ€.
-ν ν¬λμ΄μ λ λ¨μ΄ νμ₯ μμ΄ λ² μ΄μ€ λͺ¨λΈ κ·Έλλ‘ μ¬μ©
-κ³ κ° λ¦¬λ·°λ μμ
ν¬μ€ν
κ³ μ°¨μ λΆμ λ° μ½λ©κ³Ό μλ¬Έ, μν, λ
Όλ¦¬νλ¨ λ±μ΄ κ°νλ λͺ¨λΈ
-128k-Context Window
-νκΈ Function Call λ° Tool Calling μ§μ
-Deepspeed Stage=3, rslora λ° BAdam Layer Mode μ¬μ©
-ollama run benedict/linkbricks-hermes3-llama3.1-8b-korean-advanced-q4
-ollama run benedict/linkbricks-hermes3-llama3.1-8b-korean-advanced-q8
Finetuned by Mr. Yunsung Ji (Saxo), a data scientist at Linkbricks, a company specializing in AI and big data analytics
about 25% of total parameters Korean CPT(Continued-Pretraining)->SFT->DPO training model based on Hermes-3-Llama-3.1-8B through 8 H100-80Gs as a Korean language model
It is a model that has been trained to handle Korean-Chinese-English-Japanese cross-training data and 10M korean news corpus and logic judgment data for various tasks to enable cross-fertilization processing and complex Korean logic & math problems.
-Tokenizer uses the base model without word expansion
-Models enhanced with high-dimensional analysis of customer reviews and social posts, as well as coding, writing, math and decision making
-128k-Context Window
-Support for Korean Functioncall and Tool Calling
-Deepspeed Stage=3, use rslora and BAdam Layer Mode
- Downloads last month
- 700
Model tree for Saxo/Linkbricks-Horizon-AI-Korean-Advanced-8B
Base model
meta-llama/Llama-3.1-8B