larryvrh/Yi-34B-200K-Llamafied
The larryvrh/Yi-34B-200K-Llamafied model is a 34 billion parameter Llamafied version of 01-ai's Yi-34B-200K, designed for ease of use. It features an extended 32,768 token context length, making it suitable for processing longer inputs. This model demonstrates strong performance across various benchmarks, particularly excelling in MMLU, CMMLU, C-Eval, and GAOKAO, indicating robust general reasoning and language understanding capabilities.
Loading preview...
Model Overview
This model, larryvrh/Yi-34B-200K-Llamafied, is a 34 billion parameter variant of 01-ai's Yi-34B-200K, adapted into a Llamafied format for enhanced usability. It boasts an impressive 32,768 token context window, allowing for extensive input and output sequences.
Key Performance Highlights
The Yi-34B-200K model demonstrates competitive performance across a range of benchmarks, often outperforming or matching models in its class. Notable scores include:
- MMLU (5-shot): 76.1
- CMMLU (5-shot): 83.6
- C-Eval (5-shot): 81.9
- GAOKAO (0-shot): 83.4
- Common-sense Reasoning: 79.7
- Reading Comprehension: 76.6
These results indicate strong capabilities in general knowledge, multi-turn conversation, and complex reasoning tasks. The model's evaluation methodology is consistent with the original benchmark, employing greedy decoding without post-processing for generated content.
Usage and Licensing
For detailed usage instructions, users are directed to the 01-ai GitHub repository. The Yi series models are available for academic research and free commercial use, subject to permission and adherence to the Model License Agreement 2.0.