jjourney1125/M-SOLAR-10.7B-v1.0
Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:10.7BQuant:FP8Ctx Length:4kLicense:cc-by-nc-sa-4.0Architecture:Transformer0.0K Open Weights Warm

M-SOLAR-10.7B-v1.0 is a 10.7 billion parameter instruction-tuned language model developed through a joint research effort by Megastudy Education, Prediction, and Myes. It is fine-tuned on a diverse set of Korean datasets, including KOR-OpenOrca-Platypus-v3, KorQuAD 2.1, AIHUB technical science summaries, and proprietary in-house educational data. The model demonstrates capabilities in text generation, summarization, and understanding complex instructional prompts, with a focus on Korean language tasks.

Loading preview...

M-SOLAR-10.7B-v1.0: An Instruction-Tuned Korean LLM

M-SOLAR-10.7B-v1.0 is a 10.7 billion parameter language model developed collaboratively by Megastudy Education, Prediction, and Myes. This model has undergone extensive instruction fine-tuning using a variety of Korean datasets, emphasizing diverse linguistic tasks and domain-specific knowledge.

Key Capabilities

  • Instruction Following: Trained on a range of instruction-based datasets, including kyujinpy/KOR-OpenOrca-Platypus-v3, enabling it to understand and execute complex prompts.
  • Text Summarization and Generation: Utilizes AIHUB technical science summary data to reconstruct original texts from summaries and clues, demonstrating strong summarization and text expansion abilities.
  • Contextual Understanding: Incorporates a unique "Random Split Generation" strategy where the model learns to infer the original order of shuffled sentences, enhancing its understanding of textual coherence and logical flow.
  • Domain Adaptation: Leverages proprietary in-house educational data from Megastudy Education and Prediction, suggesting potential strengths in educational content generation and analysis.

Good For

  • Korean Language Processing: Optimized for tasks requiring deep understanding and generation of Korean text.
  • Instruction-Based Tasks: Excels in scenarios where precise instruction following is critical.
  • Educational Applications: Its training on educational domain data makes it suitable for tasks related to learning content, question answering, and summarization in an academic context.
Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p