Skip to content

EthenEthenEthen

LoginPlatformTry Ethen

Open Source Model Profile · aisingapore

Llama-SEA-LION-v3-8B-IT

Llama-SEA-LION-v3-8B-IT is an 8.03B-parameter Llama-family instruct-tune from AI Singapore. Its model card describes instruction tuning for the Southeast Asia region across thirteen supported languages.

Publisher
aisingapore
Task
text-generation
Model type
llama
License
llama3.1
Library
transformers
Publication status
Accepted · not indexed

Model overview

Llama-SEA-LION-v3-8B-IT is published by aisingapore as a Llama-based text-generation model. The captured configuration identifies LlamaForCausalLM and Safetensors metadata reports 8030261248 parameters. According to the model card, SEA-LION stands for Southeast Asian Languages In One Network, and this instruct-tuned version was created from the continued pre-trained Llama-SEA-LION-v3-8B using the Llama 3.1 architecture.

Recorded capabilities

8.03B Llama scale

Captured config identifies LlamaForCausalLM and Safetensors metadata reports 8030261248 parameters.

Thirteen SEA languages

According to the model card, supported languages span Burmese, Chinese, English, Filipino, Indonesia, Javanese, Khmer, Lao, Malay, Sundanese, Tamil, Thai, and Vietnamese.

SEA instruction tuning

According to the model card, instruction tuning in English plus SEA languages was applied to the continued pre-trained Llama-SEA-LION-v3-8B built on the Llama 3.1 architecture.

Documented multilingual evaluation

According to the model card, evaluation covers QA, sentiment, toxicity, translation, summarisation, reasoning, NLI, and LINDSEA tasks zero-shot, plus SEA-IFEval constraint-following normalised by correct-language rate.

Use cases in the source record

  • Multilingual Southeast Asian text workflows such as question answering, sentiment analysis, translation, and summarisation in the card's listed languages.
  • Instruction-following experiments evaluated zero-shot with native prompts, including SEA-IFEval constraint adherence in the correct response language.

Limitations and unknowns

  • No scored evaluation results were extracted from this record; the card describes tasks and method but no numeric scores.
  • No context-window value was extracted from this record.
  • Provider state is historical snapshot data, not independently refreshed current availability.
  • Training counts and hardware figures come from the publisher model card, which notes they reflect this release run and may differ from figures in accompanying research papers.

Source and provenance

Source: aisingapore/Llama-SEA-LION-v3-8B-IT

Captured: Unknown. Processed: 2026-09-07T19:34:39.438688+00:00.

Current Version: 14.04.2025 Llama-SEA-LION-v3-8B-IT SEA-LION is a collection of Large Language Models (LLMs) which have been pretrained and instruct-tuned for the Southeast Asia (SEA) region. SEA-LION stands for Southeast Asian Languages In One Network . Developed by: Products Pillar, AI Singapore Funded by: Singapore NRF Model type: Decoder Languages supported: Burmese, Chinese, English, Filipino, Indonesia, Javanese, Khmer, Lao, Malay, Sundanese, Tamil, Thai, Vietnamese License: Llama 3.1 Community License Model Details Model Description We performed instruction tuning in English and also in SEA languages such as Indonesian, Javan…

F001F002F003F004F005F006F007F009F010F011F012F013F014F015F016F017F018F019