Llama text-generation architecture
The captured configuration identifies LlamaForCausalLM with model type llama and Transformers support.
Open Source Model Profile · Saxo
Linkbricks-Horizon-AI-Llama-3.3-Korean-70B-sft-dpo
Linkbricks-Horizon-AI-Llama-3.3-Korean-70B-sft-dpo is a 70.55B-parameter Llama-family text-generation fine-tune from Saxo. According to the model card, it is a Korean SFT-to-DPO model.
- Publisher
- Saxo
- Task
- text-generation
- Model type
- llama
- License
- apache-2.0
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
Linkbricks-Horizon-AI-Llama-3.3-Korean-70B-sft-dpo is published by Saxo as a text-generation model. The captured configuration identifies LlamaForCausalLM with model type llama, and Safetensors metadata reports about 70.55B parameters. According to the model card, it applies Korean SFT then DPO to Llama-3.3-70B-Instruct with Korean, Japanese, Chinese, and English cross-training data.
Recorded capabilities
Llama-3.3-70B-Instruct lineage
According to the model card and hub tags, the model derives from meta-llama/Llama-3.3-70B-Instruct.
Reported 128k context and tool calling
According to the model card, the publisher reports a 128k context window with function-call and tool-calling support.
Korean-focused dataset links
Hub tags record Korean translation, summarization, sentiment, government QA, news, UltraFeedback, function-calling, and GSM8K-Korean datasets.
Use cases in the source record
- Korean-focused conversational and analytical workflows using the publisher-described Korean SFT-to-DPO training.
- Function-calling experiments using the publisher-reported function-call and tool-calling support.
Limitations and unknowns
- No evaluation results were extracted from this record.
- According to the model card, training used eight H100-80G GPUs with DeepSpeed Stage 3, rslora, and BAdam layer mode; Ethen did not independently verify this.
- Provider state is historical snapshot data and should be refreshed before being presented as current.
Source and provenance
Source: Saxo/Linkbricks-Horizon-AI-Llama-3.3-Korean-70B-sft-dpo
Captured: Unknown. Processed: 2026-09-07T19:34:36.597175+00:00.
Model Card for Model ID AI 와 빅데이터 분석 전문 기업인 Linkbricks의 데이터사이언티스트인 지윤성(Saxo) 이사가 meta-llama/Llama-3.3-70B-Instruct 베이스모델을 사용해서 H100-80G 8개를 통해 한국어 SFT->DPO 한 한국어 강화 언어 모델 4천만건의 한글 뉴스 및 위키 코퍼스를 기준으로 다양한 테스크별 한국어-일본어-중국어-영어 교차 학습 데이터와 수학 및 논리판단 데이터를 통하여 한중일영 언어 교차 증강 처리와 복잡한 논리 문제 역시 대응 가능하도록 훈련한 모델이다. -토크나이저는 단어 확장 없이 베이스 모델 그대로 사용 -고객 리뷰나 소셜 포스팅 고차원 분석 및 코딩과 작문, 수학, 논리판단 등이 강화된 모델 -128k-Context Window -Function Call 및 Tool Calling 지원 -128k-Context Window -Deepspeed Stage=3, rslora 및 BAdam Layer Mode 사용 -"transformers_version": "4.46.3" Finetuned by Mr. Yunsung Ji (Saxo), a data scientist at Linkbricks, a company specializing in AI a…
F001F002F003F004F005F006F007F008F009F010F011F012F013F014