SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2404.03608 · EMNLP · 2024

Sailor: Open Language Models for South-East Asia

Qian Liu, Min Lin, Jia Guo, Guangtao Zeng, Longxu Dou, Jiahui Zhou, Wei Lu, Ziqi Jin, Xin Mao

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 12 functions out of this paper's own repositories and ran 8 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
epfllm/megatron-llm canonical 7 of 11
sail-sg/sailor-llm canonical 1 of 1
FunctionStatusWhere it lives
all_gather_item Ran epfllm/megatron-llm/megatron/dist_signal_handler.py
pointer only (licence: NOASSERTION) · get_code("684c66f5cec0794e")
allocate_mem_buff Ran epfllm/megatron-llm/megatron/memory.py
pointer only (licence: NONE) · get_code("db8deebff0d5256c")
assert_viewless_tensor Ran epfllm/megatron-llm/megatron/core/utils.py
pointer only (licence: NOASSERTION) · get_code("cc64b6ee963afc52")
build_num_microbatches_calculator Ran epfllm/megatron-llm/megatron/microbatches.py
pointer only (licence: NOASSERTION) · get_code("99baeb4271f72548")
divide Ran epfllm/megatron-llm/megatron/core/utils.py
pointer only (licence: NONE) · get_code("33de97c702d46371")
first_option_parse Ran sail-sg/sailor-llm/eval/sailor_text_postprocessors.py
code served (permissive licence) · get_code("84fc895d549d7b93")
get_device Ran epfllm/megatron-llm/megatron/dist_signal_handler.py
pointer only (licence: NOASSERTION) · get_code("11ee5870494a1570")
make_viewless_tensor Ran epfllm/megatron-llm/megatron/core/utils.py
pointer only (licence: NOASSERTION) · get_code("9114f74bf4588d72")
get_mem_buff Not yet run epfllm/megatron-llm/megatron/memory.py
pointer only (licence: NONE) · get_code("8a1aaa5bb993da97")
is_pipeline_first_stage Not yet run epfllm/megatron-llm/megatron/core/parallel_state.py
pointer only (licence: NOASSERTION) · get_code("6ac246b8e8d0329c")
is_pipeline_last_stage Not yet run epfllm/megatron-llm/megatron/core/parallel_state.py
pointer only (licence: NOASSERTION) · get_code("b64cb10233a3ce60")
is_rank_in_embedding_group Not yet run epfllm/megatron-llm/megatron/core/parallel_state.py
pointer only (licence: NOASSERTION) · get_code("d6baa2c0dd7f70d7")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We present Sailor, a family of open language models ranging from 0.5B to 14B parameters, tailored for South-East Asian (SEA) languages. From Qwen1.5, Sailor models accept 200B to 400B tokens during continual pre-training, primarily covering the languages of English, Chinese, Vietnamese, Thai, Indonesian, Malay, and Lao. The training leverages several techniques, including BPE dropout for improving the model robustness, aggressive data cleaning and deduplication, and small proxy models to optimize the data mixture. Experimental results on four typical tasks indicate that Sailor models demonstrate strong performance across different benchmarks, including commonsense reasoning, question answering, reading comprehension and examination. We share our insights to spark a wider interest in developing large language models for multilingual use cases. Our demo can be found at https: //hf.co/spaces/sail/Sailor-14B-Chat.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2404.03608")
get_code_for_paper("2404.03608")
have("2404.03608")

Connect an agent — have() is free.