--- tags: - merge - mergekit - lazymergekit - btherien/JOB-3150994_410M_it-132366_tr-pile-train_scratch License: apache-2.0 --- # 405M_TIES-merge_pile_300B_into_slimp_300B_from_pile_replay5_density-0.95 405M_TIES-merge_pile_300B_into_slimp_300B_from_pile_replay5_density-0.95 is a merge of the following models using [mergekit](https://github.com/cg123/mergekit): * [btherien/JOB-3150994_410M_it-132366_tr-pile-train_scratch](https://huggingface.co/btherien/JOB-3150994_410M_it-132366_tr-pile-train_scratch) ## 🧩 Configuration \```yamlmodels: - model: btherien/Model_-410M_It_-132366_Tr_-slim-pajama-300B-replay5_finetune # no parameters necessary for base model - model: btherien/JOB-3150994_410M_it-132366_tr-pile-train_scratch parameters: density: 0.95 weight: 1.0 merge_method: ties base_model: btherien/Model_-410M_It_-132366_Tr_-slim-pajama-300B-replay5_finetune parameters: normalize: true dtype: float16\```