├── .github └── workflows │ ├── cla.yml │ └── pre-commit.yml ├── .gitignore ├── .pre-commit-config.yaml ├── CLA.md ├── CONTRIBUTING.md ├── LICENSE ├── README.md ├── docs ├── create_a_merge_method.md ├── evolve.md ├── merge_methods.md ├── moe.md ├── multimerge.md └── tokensurgeon.md ├── examples ├── arcee_fusion.yml ├── bio-merge.yml ├── gradient-slerp.yml ├── linear.yml ├── orcamini-platy-44layer.yml └── ties.yml ├── mergekit ├── __init__.py ├── _data │ ├── __init__.py │ ├── architectures │ │ ├── __init__.py │ │ ├── afmoe_partial.json │ │ ├── arcee.json │ │ ├── baichuan.json │ │ ├── bert-masked-lm.json │ │ ├── bert-sequence-classification.json │ │ ├── bert.json │ │ ├── chatglm.json │ │ ├── cohere.json │ │ ├── cohere2.json │ │ ├── distilbert-masked-lm.json │ │ ├── distilbert-sequence-classification.json │ │ ├── distilbert-token-classification.json │ │ ├── distilbert.json │ │ ├── exaone.json │ │ ├── falcon.json │ │ ├── gemma.json │ │ ├── gemma2.json │ │ ├── gemma3.json │ │ ├── gemma3vl.json │ │ ├── glm4.json │ │ ├── gpt-neox.json │ │ ├── gpt2-sequence-classification.json │ │ ├── gpt2.json │ │ ├── gptbigcode.json │ │ ├── granite4moehybrid.json │ │ ├── internlm2.json │ │ ├── jais.json │ │ ├── lfm2.json │ │ ├── llama-sequence-classification.json │ │ ├── llama.json │ │ ├── llama4.json │ │ ├── mamba.json │ │ ├── mistral.json │ │ ├── mistral3.json │ │ ├── olmo2.json │ │ ├── phi-1.json │ │ ├── phi2-old.json │ │ ├── phi2.json │ │ ├── phi3-small.json │ │ ├── phi3.json │ │ ├── qwen.json │ │ ├── qwen2.json │ │ ├── qwen2_5_vl.json │ │ ├── qwen2_vl.json │ │ ├── qwen3.json │ │ ├── roberta-masked-lm.json │ │ ├── roberta-sequence-classification.json │ │ ├── roberta-token-classification.json │ │ ├── roberta.json │ │ ├── solar.json │ │ ├── stablelm.json │ │ ├── stablelm2.json │ │ ├── starcoder2.json │ │ ├── t5.json │ │ └── whisper.json │ └── chat_templates │ │ ├── __init__.py │ │ ├── alpaca.jinja │ │ ├── chatml.jinja │ │ ├── exaone.jinja │ │ ├── llama3.jinja │ │ └── mistral.jinja ├── architecture │ ├── __init__.py │ ├── auto.py │ ├── base.py │ ├── json_definitions.py │ └── moe_defs.py ├── card.py ├── common.py ├── config.py ├── evo │ ├── __init__.py │ ├── actors.py │ ├── config.py │ ├── genome.py │ ├── helpers.py │ ├── monkeypatch.py │ └── strategy.py ├── graph.py ├── io │ ├── __init__.py │ ├── lazy_tensor_loader.py │ ├── lazy_unpickle.py │ ├── loader.py │ ├── tasks.py │ └── tensor_writer.py ├── merge.py ├── merge_methods │ ├── __init__.py │ ├── arcee_fusion.py │ ├── base.py │ ├── easy_define.py │ ├── generalized_task_arithmetic.py │ ├── karcher.py │ ├── linear.py │ ├── model_stock.py │ ├── multislerp.py │ ├── nearswap.py │ ├── nuslerp.py │ ├── passthrough.py │ ├── rectify_embed.py │ ├── registry.py │ ├── sce.py │ └── slerp.py ├── moe │ ├── __init__.py │ ├── arch.py │ ├── common.py │ ├── config.py │ ├── deepseek.py │ ├── mixtral.py │ ├── qwen.py │ ├── qwen3.py │ └── router.py ├── multigpu_executor.py ├── options.py ├── plan.py ├── scripts │ ├── __init__.py │ ├── bakllama.py │ ├── evolve.py │ ├── extract_lora.py │ ├── fill_missing_params.py │ ├── layershuffle.py │ ├── legacy.py │ ├── merge_raw_pytorch.py │ ├── moe.py │ ├── multimerge.py │ ├── run_yaml.py │ └── tokensurgeon.py ├── sparsify.py ├── tokenizer │ ├── __init__.py │ ├── build.py │ ├── config.py │ ├── embed.py │ └── normalization.py └── tokensurgeon │ ├── __init__.py │ ├── common_interpolation.py │ ├── magikarp.py │ ├── omp.py │ ├── pca.py │ ├── rope_helpers.py │ ├── subword.py │ └── token_basis.py ├── notebook.ipynb ├── pyproject.toml └── tests ├── __init__.py ├── common.py ├── test_basic_merges.py ├── test_chat_template.py ├── test_graph.py ├── test_io.py ├── test_lazy_unpickle.py ├── test_modelref.py ├── test_sparsify.py └── test_tokenizer.py /.github/workflows/cla.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/.github/workflows/cla.yml -------------------------------------------------------------------------------- /.github/workflows/pre-commit.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/.github/workflows/pre-commit.yml -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/.gitignore -------------------------------------------------------------------------------- /.pre-commit-config.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/.pre-commit-config.yaml -------------------------------------------------------------------------------- /CLA.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/CLA.md -------------------------------------------------------------------------------- /CONTRIBUTING.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/CONTRIBUTING.md -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/LICENSE -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/README.md -------------------------------------------------------------------------------- /docs/create_a_merge_method.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/docs/create_a_merge_method.md -------------------------------------------------------------------------------- /docs/evolve.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/docs/evolve.md -------------------------------------------------------------------------------- /docs/merge_methods.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/docs/merge_methods.md -------------------------------------------------------------------------------- /docs/moe.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/docs/moe.md -------------------------------------------------------------------------------- /docs/multimerge.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/docs/multimerge.md -------------------------------------------------------------------------------- /docs/tokensurgeon.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/docs/tokensurgeon.md -------------------------------------------------------------------------------- /examples/arcee_fusion.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/examples/arcee_fusion.yml -------------------------------------------------------------------------------- /examples/bio-merge.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/examples/bio-merge.yml -------------------------------------------------------------------------------- /examples/gradient-slerp.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/examples/gradient-slerp.yml -------------------------------------------------------------------------------- /examples/linear.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/examples/linear.yml -------------------------------------------------------------------------------- /examples/orcamini-platy-44layer.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/examples/orcamini-platy-44layer.yml -------------------------------------------------------------------------------- /examples/ties.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/examples/ties.yml -------------------------------------------------------------------------------- /mergekit/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /mergekit/_data/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /mergekit/_data/architectures/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /mergekit/_data/architectures/afmoe_partial.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/afmoe_partial.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/arcee.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/arcee.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/baichuan.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/baichuan.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/bert-masked-lm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/bert-masked-lm.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/bert-sequence-classification.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/bert-sequence-classification.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/bert.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/bert.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/chatglm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/chatglm.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/cohere.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/cohere.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/cohere2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/cohere2.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/distilbert-masked-lm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/distilbert-masked-lm.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/distilbert-sequence-classification.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/distilbert-sequence-classification.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/distilbert-token-classification.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/distilbert-token-classification.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/distilbert.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/distilbert.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/exaone.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/exaone.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/falcon.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/falcon.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/gemma.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/gemma.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/gemma2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/gemma2.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/gemma3.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/gemma3.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/gemma3vl.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/gemma3vl.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/glm4.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/glm4.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/gpt-neox.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/gpt-neox.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/gpt2-sequence-classification.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/gpt2-sequence-classification.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/gpt2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/gpt2.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/gptbigcode.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/gptbigcode.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/granite4moehybrid.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/granite4moehybrid.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/internlm2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/internlm2.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/jais.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/jais.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/lfm2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/lfm2.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/llama-sequence-classification.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/llama-sequence-classification.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/llama.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/llama.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/llama4.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/llama4.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/mamba.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/mamba.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/mistral.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/mistral.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/mistral3.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/mistral3.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/olmo2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/olmo2.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/phi-1.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/phi-1.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/phi2-old.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/phi2-old.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/phi2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/phi2.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/phi3-small.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/phi3-small.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/phi3.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/phi3.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/qwen.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/qwen.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/qwen2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/qwen2.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/qwen2_5_vl.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/qwen2_5_vl.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/qwen2_vl.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/qwen2_vl.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/qwen3.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/qwen3.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/roberta-masked-lm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/roberta-masked-lm.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/roberta-sequence-classification.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/roberta-sequence-classification.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/roberta-token-classification.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/roberta-token-classification.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/roberta.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/roberta.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/solar.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/solar.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/stablelm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/stablelm.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/stablelm2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/stablelm2.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/starcoder2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/starcoder2.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/t5.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/t5.json -------------------------------------------------------------------------------- /mergekit/_data/architectures/whisper.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/architectures/whisper.json -------------------------------------------------------------------------------- /mergekit/_data/chat_templates/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /mergekit/_data/chat_templates/alpaca.jinja: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/chat_templates/alpaca.jinja -------------------------------------------------------------------------------- /mergekit/_data/chat_templates/chatml.jinja: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/chat_templates/chatml.jinja -------------------------------------------------------------------------------- /mergekit/_data/chat_templates/exaone.jinja: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/chat_templates/exaone.jinja -------------------------------------------------------------------------------- /mergekit/_data/chat_templates/llama3.jinja: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/chat_templates/llama3.jinja -------------------------------------------------------------------------------- /mergekit/_data/chat_templates/mistral.jinja: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/_data/chat_templates/mistral.jinja -------------------------------------------------------------------------------- /mergekit/architecture/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/architecture/__init__.py -------------------------------------------------------------------------------- /mergekit/architecture/auto.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/architecture/auto.py -------------------------------------------------------------------------------- /mergekit/architecture/base.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/architecture/base.py -------------------------------------------------------------------------------- /mergekit/architecture/json_definitions.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/architecture/json_definitions.py -------------------------------------------------------------------------------- /mergekit/architecture/moe_defs.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/architecture/moe_defs.py -------------------------------------------------------------------------------- /mergekit/card.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/card.py -------------------------------------------------------------------------------- /mergekit/common.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/common.py -------------------------------------------------------------------------------- /mergekit/config.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/config.py -------------------------------------------------------------------------------- /mergekit/evo/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /mergekit/evo/actors.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/evo/actors.py -------------------------------------------------------------------------------- /mergekit/evo/config.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/evo/config.py -------------------------------------------------------------------------------- /mergekit/evo/genome.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/evo/genome.py -------------------------------------------------------------------------------- /mergekit/evo/helpers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/evo/helpers.py -------------------------------------------------------------------------------- /mergekit/evo/monkeypatch.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/evo/monkeypatch.py -------------------------------------------------------------------------------- /mergekit/evo/strategy.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/evo/strategy.py -------------------------------------------------------------------------------- /mergekit/graph.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/graph.py -------------------------------------------------------------------------------- /mergekit/io/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/io/__init__.py -------------------------------------------------------------------------------- /mergekit/io/lazy_tensor_loader.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/io/lazy_tensor_loader.py -------------------------------------------------------------------------------- /mergekit/io/lazy_unpickle.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/io/lazy_unpickle.py -------------------------------------------------------------------------------- /mergekit/io/loader.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/io/loader.py -------------------------------------------------------------------------------- /mergekit/io/tasks.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/io/tasks.py -------------------------------------------------------------------------------- /mergekit/io/tensor_writer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/io/tensor_writer.py -------------------------------------------------------------------------------- /mergekit/merge.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge.py -------------------------------------------------------------------------------- /mergekit/merge_methods/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/__init__.py -------------------------------------------------------------------------------- /mergekit/merge_methods/arcee_fusion.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/arcee_fusion.py -------------------------------------------------------------------------------- /mergekit/merge_methods/base.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/base.py -------------------------------------------------------------------------------- /mergekit/merge_methods/easy_define.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/easy_define.py -------------------------------------------------------------------------------- /mergekit/merge_methods/generalized_task_arithmetic.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/generalized_task_arithmetic.py -------------------------------------------------------------------------------- /mergekit/merge_methods/karcher.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/karcher.py -------------------------------------------------------------------------------- /mergekit/merge_methods/linear.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/linear.py -------------------------------------------------------------------------------- /mergekit/merge_methods/model_stock.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/model_stock.py -------------------------------------------------------------------------------- /mergekit/merge_methods/multislerp.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/multislerp.py -------------------------------------------------------------------------------- /mergekit/merge_methods/nearswap.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/nearswap.py -------------------------------------------------------------------------------- /mergekit/merge_methods/nuslerp.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/nuslerp.py -------------------------------------------------------------------------------- /mergekit/merge_methods/passthrough.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/passthrough.py -------------------------------------------------------------------------------- /mergekit/merge_methods/rectify_embed.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/rectify_embed.py -------------------------------------------------------------------------------- /mergekit/merge_methods/registry.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/registry.py -------------------------------------------------------------------------------- /mergekit/merge_methods/sce.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/sce.py -------------------------------------------------------------------------------- /mergekit/merge_methods/slerp.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/merge_methods/slerp.py -------------------------------------------------------------------------------- /mergekit/moe/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/moe/__init__.py -------------------------------------------------------------------------------- /mergekit/moe/arch.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/moe/arch.py -------------------------------------------------------------------------------- /mergekit/moe/common.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/moe/common.py -------------------------------------------------------------------------------- /mergekit/moe/config.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/moe/config.py -------------------------------------------------------------------------------- /mergekit/moe/deepseek.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/moe/deepseek.py -------------------------------------------------------------------------------- /mergekit/moe/mixtral.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/moe/mixtral.py -------------------------------------------------------------------------------- /mergekit/moe/qwen.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/moe/qwen.py -------------------------------------------------------------------------------- /mergekit/moe/qwen3.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/moe/qwen3.py -------------------------------------------------------------------------------- /mergekit/moe/router.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/moe/router.py -------------------------------------------------------------------------------- /mergekit/multigpu_executor.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/multigpu_executor.py -------------------------------------------------------------------------------- /mergekit/options.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/options.py -------------------------------------------------------------------------------- /mergekit/plan.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/plan.py -------------------------------------------------------------------------------- /mergekit/scripts/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /mergekit/scripts/bakllama.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/scripts/bakllama.py -------------------------------------------------------------------------------- /mergekit/scripts/evolve.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/scripts/evolve.py -------------------------------------------------------------------------------- /mergekit/scripts/extract_lora.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/scripts/extract_lora.py -------------------------------------------------------------------------------- /mergekit/scripts/fill_missing_params.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/scripts/fill_missing_params.py -------------------------------------------------------------------------------- /mergekit/scripts/layershuffle.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/scripts/layershuffle.py -------------------------------------------------------------------------------- /mergekit/scripts/legacy.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/scripts/legacy.py -------------------------------------------------------------------------------- /mergekit/scripts/merge_raw_pytorch.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/scripts/merge_raw_pytorch.py -------------------------------------------------------------------------------- /mergekit/scripts/moe.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/scripts/moe.py -------------------------------------------------------------------------------- /mergekit/scripts/multimerge.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/scripts/multimerge.py -------------------------------------------------------------------------------- /mergekit/scripts/run_yaml.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/scripts/run_yaml.py -------------------------------------------------------------------------------- /mergekit/scripts/tokensurgeon.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/scripts/tokensurgeon.py -------------------------------------------------------------------------------- /mergekit/sparsify.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/sparsify.py -------------------------------------------------------------------------------- /mergekit/tokenizer/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokenizer/__init__.py -------------------------------------------------------------------------------- /mergekit/tokenizer/build.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokenizer/build.py -------------------------------------------------------------------------------- /mergekit/tokenizer/config.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokenizer/config.py -------------------------------------------------------------------------------- /mergekit/tokenizer/embed.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokenizer/embed.py -------------------------------------------------------------------------------- /mergekit/tokenizer/normalization.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokenizer/normalization.py -------------------------------------------------------------------------------- /mergekit/tokensurgeon/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokensurgeon/__init__.py -------------------------------------------------------------------------------- /mergekit/tokensurgeon/common_interpolation.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokensurgeon/common_interpolation.py -------------------------------------------------------------------------------- /mergekit/tokensurgeon/magikarp.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokensurgeon/magikarp.py -------------------------------------------------------------------------------- /mergekit/tokensurgeon/omp.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokensurgeon/omp.py -------------------------------------------------------------------------------- /mergekit/tokensurgeon/pca.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokensurgeon/pca.py -------------------------------------------------------------------------------- /mergekit/tokensurgeon/rope_helpers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokensurgeon/rope_helpers.py -------------------------------------------------------------------------------- /mergekit/tokensurgeon/subword.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokensurgeon/subword.py -------------------------------------------------------------------------------- /mergekit/tokensurgeon/token_basis.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/mergekit/tokensurgeon/token_basis.py -------------------------------------------------------------------------------- /notebook.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/notebook.ipynb -------------------------------------------------------------------------------- /pyproject.toml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/pyproject.toml -------------------------------------------------------------------------------- /tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tests/common.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/tests/common.py -------------------------------------------------------------------------------- /tests/test_basic_merges.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/tests/test_basic_merges.py -------------------------------------------------------------------------------- /tests/test_chat_template.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/tests/test_chat_template.py -------------------------------------------------------------------------------- /tests/test_graph.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/tests/test_graph.py -------------------------------------------------------------------------------- /tests/test_io.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/tests/test_io.py -------------------------------------------------------------------------------- /tests/test_lazy_unpickle.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/tests/test_lazy_unpickle.py -------------------------------------------------------------------------------- /tests/test_modelref.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/tests/test_modelref.py -------------------------------------------------------------------------------- /tests/test_sparsify.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/tests/test_sparsify.py -------------------------------------------------------------------------------- /tests/test_tokenizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/arcee-ai/mergekit/HEAD/tests/test_tokenizer.py --------------------------------------------------------------------------------