lhpku20010120/llama3.1-8b-k12kgraph

TEXT GENERATIONPricing:Input $0.2 / Cached $0.028 / Output $0.32Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jul 28, 2026License:llama3.1Architecture:Transformer Featherless Exclusive Cold

lhpku20010120/llama3.1-8b-k12kgraph is an 8 billion parameter LlamaForCausalLM model, fine-tuned from meta-llama/Llama-3.1-8B using full-parameter supervised fine-tuning. It was trained on the K12-Train dataset, specifically optimized for K-12 educational question answering and knowledge-intensive reasoning. This model excels in generating responses for educational content in both Chinese and English, leveraging a 32,768 token context length.

Loading preview...

Model Overview

lhpku20010120/llama3.1-8b-k12kgraph is an 8 billion parameter LlamaForCausalLM, built upon the meta-llama/Llama-3.1-8B base model. It underwent full-parameter supervised fine-tuning (SFT) using the K12-Train dataset, which is associated with the K12-KGraph project. This specialized training focuses on K-12 educational question answering and knowledge-intensive reasoning.

Key Capabilities

  • Specialized Domain: Optimized for K-12 educational content, providing answers and reasoning for school-level questions.
  • Bilingual Support: Capable of processing and generating content in both Chinese and English.
  • Extended Context: Utilizes a context length of 32,768 tokens during preprocessing, allowing for more comprehensive understanding of longer educational texts.
  • Llama 3.1 Compatibility: Uses the Llama 3.1 chat format and is built with the Llama 3.1 Community License.

Intended Use Cases

This model is primarily intended for:

  • Research on K-12 educational question answering.
  • Knowledge-intensive reasoning within educational contexts.
  • Evaluation of models trained with knowledge-graph-derived educational data.

Limitations

Users should be aware that the model may:

  • Reproduce biases or errors from its base model or training data.
  • Generate incorrect, incomplete, or hallucinated answers, and should not be considered an authoritative source for high-stakes decisions.
  • Experience degraded performance outside its strongest domain of K-12 educational content.