rombodawg/LosslessMegaCoder-llama2-13b-mini
TEXT GENERATIONConcurrency Cost:1Model Size:13BQuant:FP8Ctx Length:4kPublished:Aug 15, 2023License:llama2Architecture:Transformer0.0K Open Weights Cold

LosslessMegaCoder-llama2-13b-mini is a 13 billion parameter Llama 2-based language model developed in collaboration between rombodawg and andreaskoepf, an Open Assistant affiliate. It is one of the first models trained on the LosslessMegaCodeTrainingV2_1m_Evol_Uncensored dataset, specifically a filtered version (megacode2-min100) focused on code. This model is optimized for code-related tasks, demonstrating a HumanEval+ score of 0.29, comparable to LLaMA-2 70B Chat.

Loading preview...