8.03B Llama scale
Captured config identifies LlamaForCausalLM and Safetensors metadata reports 8030261248 parameters.
Open Source Model Profile · werty1248
Llama-3-Ko-8B-Instruct-AOG
Llama-3-Ko-8B-Instruct-AOG is an 8.03B-parameter Korean Llama instruct-tune from werty1248. Its model card documents Axolotl LoRA training for 3 epochs on Korean Alpaca-format datasets.
- Publisher
- werty1248
- Task
- text-generation
- Model type
- llama
- License
- llama3
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
Llama-3-Ko-8B-Instruct-AOG is published by werty1248 as a Llama-based text-generation model. The captured configuration identifies LlamaForCausalLM and Safetensors metadata reports 8030261248 parameters. According to the model card, it is an instruction-tuning version of beomi/Llama-3-Open-Ko-8B, with hub tags marking Korean language and conversational use.
Recorded capabilities
Korean instruction tuning
According to the model card, it is an instruction-tuning version of beomi/Llama-3-Open-Ko-8B.
Axolotl LoRA, 3 epochs
According to the model card, training used Axolotl with LoRA for 3 epochs.
Alpaca-format Korean datasets
According to the model card, training data includes KoAlpaca-v1.1a, OpenOrca-KO, and openassistant-guanaco-ko in Alpaca instruction format, described as commercially usable.
Use cases in the source record
- Korean-language instruction-following text generation using the documented Alpaca instruction format.
- Comparative Korean fine-tune experiments on non-math, non-coding instruction data, given the card's stated dataset exclusions.
Limitations and unknowns
- No evaluation results were extracted from this record.
- No context-window value was extracted from this record.
- Provider state is historical snapshot data, not independently refreshed current availability.
- Lineage and training detail come from a brief publisher model card and have not been independently verified by Ethen.
Source and provenance
Source: werty1248/Llama-3-Ko-8B-Instruct-AOG
Captured: Unknown. Processed: 2026-09-07T19:35:00.765150+00:00.
Model Card 아오지 Model Details Model Description beomi/Llama-3-Open-Ko-8B (최신 버전)의 Instruction tuning 버전 모델입니다. Dataset: beomi/Ko A lpaca-v1.1a kyujinpy/Open O rca-KO nlpai-lab/openassistant- g uanaco-ko Instruction format: alpaca Dataset 상업적으로 이용 가능한 데이터 셋을 사용하였습니다. 향후 비교를 위하여, 수학/코딩 관련 질문이 많은 데이터 셋(ex: kyujinpy/KOpen-platypus)은 제외하였습니다. 멀티 턴 대화 데이터(nlpai-lab/openassistant-guanaco-ko)를 추가해보았습니다. beomi/KoAlpaca-v1.1a 80% 확률로 instruction 추가("당신은 인공지능 비서입니다. 사용자가 답을 이해하기 위해 외부에서 검색할 필요가 없도록 상세한 답변을 제공하세요.") kyujinpy/OpenOrca-KO 변경사항 없음 nlpai-lab/openassistant-guanaco-ko 멀티 턴 대화의 경우, 마지막 Assistant의 답변을 목표로 함 Alpaca format에 맞춰 이전 대화 내용에 대…
F001F002F003F004F005F006F007F009F010F012F013F014F015