fusion-bench 0.2.8__py3-none-any.whl → 0.2.10__py3-none-any.whl

fusion_bench/utils/rich_utils.py

@@ -1,3 +1,4 @@
+import logging
 from pathlib import Path
 from typing import Sequence
 
@@ -10,6 +11,7 @@ from omegaconf import DictConfig, OmegaConf, open_dict
 from rich import print
 from rich.columns import Columns
 from rich.console import Console
+from rich.logging import RichHandler
 from rich.panel import Panel
 from rich.prompt import Prompt
 from rich.syntax import Syntax
@@ -184,3 +186,16 @@ if __name__ == "__main__":
 
     print("\nAvailable Styles:")
     display_available_styles()
+
+
+def setup_colorlogging(force=False, **config_kwargs):
+    FORMAT = "%(message)s"
+
+    logging.basicConfig(
+        level=logging.INFO,
+        format=FORMAT,
+        datefmt="[%X]",
+        handlers=[RichHandler()],
+        force=force,
+        **config_kwargs,
+    )

fusion_bench/utils/set.py

@@ -0,0 +1,8 @@
+__all__ = ["union"]
+
+
+def union(*iters) -> set:
+    if len(iters) == 0:
+        return set()
+    s = set().union(*iters)
+    return s

fusion_bench/utils/tensorboard.py

@@ -0,0 +1,51 @@
+"""
+functions deal with tensorboard logs.    
+"""
+
+from typing import Dict, Iterable, List
+
+import numpy as np
+import pandas as pd
+from tensorboard.backend.event_processing import event_accumulator
+
+
+def parse_tensorboard_as_dict(path: str, scalars: Iterable[str]):
+    """
+    returns a dictionary of pandas dataframes for each requested scalar.
+
+    Args:
+        path(str): A file path to a directory containing tf events files, or a single
+                   tf events file. The accumulator will load events from this path.
+        scalars:   scalars
+
+    Returns:
+        Dict[str, pandas.DataFrame]: a dictionary of pandas dataframes for each requested scalar
+    """
+    ea = event_accumulator.EventAccumulator(
+        path,
+        size_guidance={event_accumulator.SCALARS: 0},
+    )
+    _absorb_print = ea.Reload()
+    # make sure the scalars are in the event accumulator tags
+    assert all(
+        s in ea.Tags()["scalars"] for s in scalars
+    ), "some scalars were not found in the event accumulator"
+    return {k: pd.DataFrame(ea.Scalars(k)) for k in scalars}
+
+
+def parse_tensorboard_as_list(path: str, scalars: Iterable[str]):
+    """
+    returns a list of pandas dataframes for each requested scalar.
+
+    see also: :py:func:`parse_tensorboard_as_dict`
+
+    Args:
+        path(str): A file path to a directory containing tf events files, or a single
+                   tf events file. The accumulator will load events from this path.
+        scalars:   scalars
+
+    Returns:
+        List[pandas.DataFrame]: a list of pandas dataframes for each requested scalar.
+    """
+    d = parse_tensorboard_as_dict(path, scalars)
+    return [d[s] for s in scalars]

{fusion_bench-0.2.8.dist-info → fusion_bench-0.2.10.dist-info}/METADATA

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.2
 Name: fusion_bench
-Version: 0.2.8
+Version: 0.2.10
 Summary: A Comprehensive Benchmark of Deep Model Fusion
 Author-email: Anke Tang <tang.anke@foxmail.com>
 License: MIT License
@@ -70,22 +70,21 @@ FusionBench is a benchmark suite designed to evaluate the performance of various
 Projects based on FusionBench and news from the community (descending order of date):
 
 <details>
-  <summary>Hongling Zheng, Li Shen, Anke Tang, Yong Luo et al. Learn From Model Beyond Fine-Tuning: A Survey. has been accepted for publication in Nature Machine Intelligence. Nov, 2024. https://arxiv.org/abs/2310.08184</summary>
-
-  > Foundation models (FM) have demonstrated remarkable performance across a wide range of tasks (especially in the fields
-  of natural language processing and computer vision), primarily attributed to their ability to comprehend instructions and access
-  extensive, high-quality data. This not only showcases their current effectiveness but also sets a promising trajectory towards the
-  development of artificial general intelligence. Unfortunately, due to multiple constraints, the raw data of the model used for large model
-  training are often inaccessible, so the use of end-to-end models for downstream tasks has become a new research trend, which we call
-  Learn From Model (LFM) in this article. LFM focuses on the research, modification, and design of FM based on the model interface,
-  so as to better understand the model structure and weights (in a black box environment), and to generalize the model to downstream
-  tasks. The study of LFM techniques can be broadly categorized into five major areas: model tuning, model distillation, model reuse,
-  meta learning and model editing. Each category encompasses a repertoire of methods and strategies that aim to enhance the
-  capabilities and performance of FM. This paper gives a comprehensive review of the current methods based on FM from the
-  perspective of LFM, in order to help readers better understand the current research status and ideas. To conclude, we summarize the
-  survey by highlighting several critical areas for future exploration and addressing open issues that require further attention from the
-  research community. The relevant papers we investigated in this article can be accessed at
-  https://github.com/ruthless-man/Awesome-Learn-from-Model.
+  <summary>Anke Tang, et al. Merging Models on the Fly Without Retraining: A Sequential Approach to Scalable Continual Model Merging. Jan 2025. https://arxiv.org/pdf/2501.09522</summary>
+
+Deep model merging represents an emerging research direction that combines multiple fine-tuned models to harness their specialized capabilities across different tasks and domains. Current model merging techniques focus on merging all available models simultaneously, with weight interpolation-based methods being the predominant approaches. However, these conventional approaches are not well-suited for scenarios where models become available sequentially, and they often suffer from high memory requirements and potential interference between tasks. In this study, we propose a training-free projection-based continual merging method that processes models sequentially through orthogonal projections of weight matrices and adaptive scaling mechanisms. Our method operates by projecting new parameter updates onto subspaces orthogonal to existing merged parameter updates while using an adaptive scaling mechanism to maintain stable parameter distances, enabling efficient sequential integration of task-specific knowledge. Our approach maintains constant memory complexity to the number of models, minimizes interference between tasks through orthogonal projections, and retains the performance of previously merged models through adaptive task vector scaling. Extensive experiments on CLIP-ViT models demonstrate that our method achieves a 5-8% average accuracy improvement while maintaining robust performance in different task orderings.
+</details>
+
+<details>
+  <summary>Yongxian Wei, et al. Modeling Multi-Task Model Merging as Adaptive Projective Gradient Descent. Jan 2025. https://arxiv.org/abs/2501.01230</summary>
+
+Merging multiple expert models offers a promising approach for performing multi-task learning without accessing their original data. Existing methods attempt to alleviate task conflicts by sparsifying task vectors or promoting orthogonality among them. However, they overlook the fundamental requirement of model merging: ensuring the merged model performs comparably to task-specific models on respective tasks. We find these methods inevitably discard task-specific information that, while causing conflicts, is crucial for performance. Based on our findings, we frame model merging as a constrained optimization problem (i.e., minimizing the gap between the merged model and individual models, subject to the constraint of retaining shared knowledge) and solve it via adaptive projective gradient descent. Specifically, we align the merged model with individual models by decomposing and reconstituting the loss function, alleviating conflicts through data-free optimization of task vectors. To retain shared knowledge, we optimize this objective by projecting gradients within a shared subspace spanning all tasks. Moreover, we view merging coefficients as adaptive learning rates and propose a task-aware, training-free strategy. Experiments show that our plug-andplay approach consistently outperforms previous methods, achieving state-of-the-art results across diverse architectures and tasks in both vision and NLP domains. Our code is available here.
+</details>
+
+<details>
+  <summary>Hongling Zheng, Li Shen, Anke Tang, Yong Luo et al. Learn From Model Beyond Fine-Tuning: A Survey. Nature Machine Intelligence. Jan, 2025. https://www.nature.com/articles/s42256-024-00961-0</summary>
+
+  > Foundation models (FM) have demonstrated remarkable performance across a wide range of tasks (especially in the fields of natural language processing and computer vision), primarily attributed to their ability to comprehend instructions and access extensive, high-quality data. This not only showcases their current effectiveness but also sets a promising trajectory towards the development of artificial general intelligence. Unfortunately, due to multiple constraints, the raw data of the model used for large model training are often inaccessible, so the use of end-to-end models for downstream tasks has become a new research trend, which we call Learn From Model (LFM) in this article. LFM focuses on the research, modification, and design of FM based on the model interface, so as to better understand the model structure and weights (in a black box environment), and to generalize the model to downstream tasks. The study of LFM techniques can be broadly categorized into five major areas: model tuning, model distillation, model reuse, meta learning and model editing. Each category encompasses a repertoire of methods and strategies that aim to enhance the capabilities and performance of FM. This paper gives a comprehensive review of the current methods based on FM from the perspective of LFM, in order to help readers better understand the current research status and ideas. To conclude, we summarize the survey by highlighting several critical areas for future exploration and addressing open issues that require further attention from the research community. The relevant papers we investigated in this article can be accessed at https://github.com/ruthless-man/Awesome-Learn-from-Model.
 </details>
 
 <details>

{fusion_bench-0.2.8.dist-info → fusion_bench-0.2.10.dist-info}/RECORD

@@ -1,4 +1,5 @@
 fusion_bench/__init__.py,sha256=68dF-zPvb8E2MgYnmgIJsxIHJBy1MApKeOrRZvQEVlg,421
+fusion_bench/__main__.py,sha256=weUjxpP3ULnDgUxCehdbmoCM9cqfkhDhGB85tAF5qoE,81
 fusion_bench/compat/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 fusion_bench/compat/method/__init__.py,sha256=KUKHpX7AfvB7fmOAlruWp0r1z17xpkI9l29PMvLWR9A,4956
 fusion_bench/compat/method/base_algorithm.py,sha256=63_AQDj1eJOO6RyTSGXVC6G2DsG8yg9E4pT3RJXgP3A,1952
@@ -14,7 +15,7 @@ fusion_bench/constants/__init__.py,sha256=Pyc4dLbl6oNduOCdnpeXQ9LDyVoIrkdl9eZ_l2
 fusion_bench/constants/paths.py,sha256=DVZyQ9FLhkyUdw6ARpXUCAMf_B8hFyJ6UNI-oYly3pE,591
 fusion_bench/dataset/__init__.py,sha256=OJiYmcqz0Vm5O7mE4PB5QFJeL_KjrsseQTRsQATGTm4,1050
 fusion_bench/dataset/clip_dataset.py,sha256=XLpCOiXlLEP3DffAlBn4P2PpUenbEFl-Yk9MNy6nbbI,2790
-fusion_bench/dataset/fer2013.py,sha256=bAdujQSj1PcUVFlKJgqcHAuE9AWz7JE1fzZ6scFVvmc,403
+fusion_bench/dataset/fer2013.py,sha256=Lub_xVhHfqaiPprvOsDVspJNioh1FjSrkhn3gL_UXDA,404
 fusion_bench/dataset/gpt2_glue.py,sha256=Qq1ZkEIQsTjj8tImvkZDNlduocSYwlEfVrDReZqDWdw,8761
 fusion_bench/dataset/gsm8k.py,sha256=CmANZ0A89PfPwVu_myKhXk1D9IwypOpjH3iqDo1KxcQ,2233
 fusion_bench/dataset/image_dataset.py,sha256=MSZE_UESyRRQDwnkm2KpyIARUg9SWcwqnH4fDNstzS4,1870
@@ -40,7 +41,7 @@ fusion_bench/dataset/llama/stanford_shp.py,sha256=6ueXKnFXIBBobacU1h5WxGLZrSOtBk
 fusion_bench/dataset/llama/ultrachat.py,sha256=Go7WvrDAYnm184fdazHGRYLbSY6Xd7jrESyQeUJtOww,1736
 fusion_bench/dataset/llama/wikitext.py,sha256=9ZHR-nMfXRumd3o-PIj3n7B83YlVeqpGkZ2zJs2B-9Y,2883
 fusion_bench/dataset/llama/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-fusion_bench/method/__init__.py,sha256=vhjXm9EbYLz9YmO3U6reo45wcbHXKeLh3g1v3pO-JD8,5904
+fusion_bench/method/__init__.py,sha256=py1hn-gRqwzESwS5KvDapclyUvIW42r0nDsuINgFBB4,6601
 fusion_bench/method/base_algorithm.py,sha256=5dutGZfPqNhO8F8FOlo3UFR91TZu2Xj7O0pTB40JvWo,1135
 fusion_bench/method/dummy.py,sha256=hb1y6LR_geRZ5eRgGwt5zJUcHYorCeIbs5i76CvurUc,1031
 fusion_bench/method/ensemble.py,sha256=rGxvJTeorfcBuE_e0XO-0-MAc9un7ZCC46ikKGuAcN4,3077
@@ -62,14 +63,16 @@ fusion_bench/method/adamerging/utils.py,sha256=Yq8ovlpLJY-5MkSmpoB-_EMYG8cr6eyO-
 fusion_bench/method/analysis/__init__.py,sha256=EQzOCShS0hF958drq1yg2oSVsS0hvBznPxtTAWB9SGY,122
 fusion_bench/method/analysis/task_vector_cos_similarity.py,sha256=pL-XsWTo258yZTEsER_6KXS7JePneVNEHN_nv8Db0qo,5468
 fusion_bench/method/analysis/task_vector_violin_plot.py,sha256=ie8hPl6QsVz9MQ6C2OEpzIBxQnmVKNf1FPc5bThmQGM,7606
-fusion_bench/method/classification/__init__.py,sha256=XAEvv4GZSssIa_BBReXVpj-cymTj5vF7ymketuS6FqA,84
-fusion_bench/method/classification/clip_finetune.py,sha256=G0ZFGvV-Nbhd8ddFKLlhaTxDMGpfVCf3CwrSdPfino8,15778
+fusion_bench/method/classification/__init__.py,sha256=emB06UOMDHK5pfQ1WuvLG9Fm0aEEtZxSjpVw8fVE0fM,167
+fusion_bench/method/classification/clip_finetune.py,sha256=DlV1isp8vz6jwXNYQ6zbblAoUfnssL-WBpDeaXI5BVw,15727
+fusion_bench/method/classification/continual_clip_finetune.py,sha256=OLhZKS-6aCnafevZkZYcNMKTWDDj3DATB27eZl_i8EY,11530
 fusion_bench/method/concrete_subspace/__init__.py,sha256=yjadcpquHZbeZYsbfYhe2JlX46kObfiWJRsIoVcOEg4,223
 fusion_bench/method/concrete_subspace/clip_concrete_adamerging.py,sha256=90_0HkOIl0XQG89xMa0UiBhrwfV2YqfLxlS04AouR3o,24755
 fusion_bench/method/concrete_subspace/clip_concrete_task_arithmetic.py,sha256=Nx-3AiAeIt5zmcC21Ta2_-4cAQg9hOWvThurXNZzA-w,10580
-fusion_bench/method/dare/__init__.py,sha256=o0XNxMy-Mn66HZmD-YhQ_Mb1VX4XrJO9A6UR8mI30fE,114
+fusion_bench/method/dare/__init__.py,sha256=63Xwkawyl_Ooy4xFxoDlP6wf-rgEWNqPuWTT9-6Ku5o,156
 fusion_bench/method/dare/simple_average.py,sha256=jR08PokPIr5PWSZbGVOp3IApgKvxAIovg3vnB2KiTwk,906
-fusion_bench/method/dare/task_arithmetic.py,sha256=2dR2OgKHAG2jTGcthFYUYiuDlO-jEdibrmvlvSTJ_Ek,2619
+fusion_bench/method/dare/task_arithmetic.py,sha256=Seno_2BhuogdRxXOni8alnHG-fdW15_OWoAvMoBoJj0,2780
+fusion_bench/method/dare/ties_merging.py,sha256=aAIMdIpsBs0vnSKGhqDTFKEChBTmcvczt9JmK_Dr4D4,3424
 fusion_bench/method/dare/utils.py,sha256=TSZMZidnwqVHG36A0UI9Wz_rXNvojXnww7_E7-YfeRI,2888
 fusion_bench/method/dawe/__init__.py,sha256=JrhtX-qAHymU8z44QtFMxtM5Qx5iH1Kxo5cptH0KNgo,83
 fusion_bench/method/dawe/dawe_for_clip.py,sha256=bF4U0_skxyPR-5RCdGQCgudqhC1Hj2x62w_xUibFg1c,9828
@@ -82,6 +85,9 @@ fusion_bench/method/fisher_merging/__init__.py,sha256=KWsjrtxKkPYwcUA5rB_6UNIqve
 fusion_bench/method/fisher_merging/clip_fisher_merging.py,sha256=QCutGqjkfW3OWETPZsCChqLRAhvfJp4QKD9TGSpTyV0,7635
 fusion_bench/method/fisher_merging/fisher_merging.py,sha256=CPU-tJiDv9FCIBYl7Pn0zA5cdRB1Md5kWchRDlJgly0,20456
 fusion_bench/method/fisher_merging/gpt2_fisher_merging.py,sha256=LZmz41jZ5dSsAHxfOUpr3u2rlCgUPTDR7xMsIlQM-jc,7576
+fusion_bench/method/isotropic_merging/__init__.py,sha256=Bg12OiltvZLMmZm066quvtG0LOWSVqI5RggYeaMDGFA,585
+fusion_bench/method/isotropic_merging/iso.py,sha256=GILofZQiTcOnJRQ28RmzOjqkso5Xih9WuFuB2JDWA_M,3773
+fusion_bench/method/isotropic_merging/iso_utils.py,sha256=7L8PYUIJROwHJQmhFY-tdEhkLAnzVKXr-ae55FQ1QSo,6928
 fusion_bench/method/linear/__init__.py,sha256=ChfkoOEAb-rUKwpowFPel-a1hRfS8gCrbnWD-jlRbe4,283
 fusion_bench/method/linear/expo.py,sha256=LCHTWlsPm1Mjhrq0mfpWLVC7skkI9ZksGduy3TxULoU,3939
 fusion_bench/method/linear/linear_interpolation.py,sha256=IONw9BPiRJouY8bE9Abfyz7qVI_1B1n8KGZa0f7Pza8,2157
@@ -96,6 +102,12 @@ fusion_bench/method/lm_finetune/peftfinetune_sft.py,sha256=klZ_IDr5-1xoYvyVZwug9
 fusion_bench/method/mixture_of_experts/__init__.py,sha256=r95iu1-3tgIUP7sWuAbLuqV7xexNYMYPZkM4_8egfp8,198
 fusion_bench/method/mixture_of_experts/mixtral_merging.py,sha256=-n1CLP1o08VyMSfaTq42kRutbw-cFDSCWHTu0iNh6ok,4237
 fusion_bench/method/mixture_of_experts/mixtral_upcycling.py,sha256=tQYAeS8MLFEfH3zDFfNZrML7lRnpGLN-HquQvjPtHNw,11208
+fusion_bench/method/opcm/__init__.py,sha256=0QcltOnjIYV1XEPDEagChLixLAhjiBnYwfWK00am29k,202
+fusion_bench/method/opcm/opcm.py,sha256=USPPMFFVQ9UbcGvvK1573tgkO1kgcrhA5jzKdbNTy9g,10693
+fusion_bench/method/opcm/task_arithmetic.py,sha256=SNuuSyzHqvOT_e3i0z0MHNWaMP6xnDdkI9c2t1OcxO4,4328
+fusion_bench/method/opcm/ties_merging.py,sha256=38ogIysnRfePhB9SAfr1BPwtHyM8gEdhU2td_yTiB2g,6080
+fusion_bench/method/opcm/utils.py,sha256=_q7yy3ENNFUh1qUd5J5DThRL4J1tIxEcknCO2AKmeYM,2102
+fusion_bench/method/opcm/weight_average.py,sha256=JfQoIU5J1jvrNKpO9k_t4Zj0y8PtteIfyoSQWx1yg2k,4379
 fusion_bench/method/pruning/__init__.py,sha256=3gtmay2bkdIAEGjpAhbY2ztMZOZLKhiJcKV3mCe2H5w,252
 fusion_bench/method/pruning/llama_magnitude_prune.py,sha256=40Gmy665S9XqIw027En6E5IlomOIcKECIRje7NDkH00,6300
 fusion_bench/method/pruning/llama_random_prune.py,sha256=EW7zfE-1a5VlPPrQ5xO1k1aqFcpPUfs5eSO_a4M1K90,4566
@@ -124,7 +136,7 @@ fusion_bench/method/regmean/clip_regmean.py,sha256=xhT7dYSCg9sPLL5ZUCCtcA-Ypw4PB
 fusion_bench/method/regmean/gpt2_regmean.py,sha256=p2D3E8YAZsltsI6GM474UWNqPZfBqihLZ93ZLUpOJ_c,5565
 fusion_bench/method/regmean/regmean.py,sha256=NTozStNkMV7RLyxj1TvzPkHjiFPynGiWgQITs8J9CiA,16210
 fusion_bench/method/slerp/__init__.py,sha256=Wgl9gg01Xou6jyZeBRD98kRnB_dAADDaPqRTHoONx9o,59
-fusion_bench/method/slerp/slerp.py,sha256=U5XG-Bikaji9zTXhdE9XlS6Tv26sbncrTlORVT5UF4g,3406
+fusion_bench/method/slerp/slerp.py,sha256=2_n10REnRoV5DuwCC0bDX8RM3MLL4Q_5rZiU0hICw2w,3406
 fusion_bench/method/slerp/slerp_utils.py,sha256=vksRo6n7FqY7By9aqbwTL4XV3BjcU_GrUl_r85Kpfjc,3504
 fusion_bench/method/smile_upscaling/__init__.py,sha256=CbJugcuQ7AcoPGFRF7TdxUztYTi3Ii9bnA6Msj8QH0A,149
 fusion_bench/method/smile_upscaling/singular_projection_merging.py,sha256=0neZS9oZnl64wu1xb9ruGB7lbhYXyy4zj8l3E1QYRGQ,6670
@@ -142,13 +154,13 @@ fusion_bench/method/tall_mask/utils.py,sha256=Wlp8WcPwR_lCaBIZ9rgG6ewLfSzz3G7kPk
 fusion_bench/method/task_arithmetic/__init__.py,sha256=pSx_NV5Ra_6UXpyYWCi6ANQoAnEtymZt_X1dDN9wT4Y,96
 fusion_bench/method/task_arithmetic/task_arithmetic.py,sha256=1D0uuNtqyA1VS35jh6AnEVsX72HnT02THyerck_lmso,5441
 fusion_bench/method/task_singular_vector/TSVC.py,sha256=yn4SrZNvtA6PoGYJmbmtNeDyDbGnRCgfZ7ZCg914AZU,410
-fusion_bench/method/task_singular_vector/TSVM.py,sha256=ANBGC1GM8c9oy_xlY-ZEyoWO9mnZh5aiF_rrvVH73l8,1925
+fusion_bench/method/task_singular_vector/TSVM.py,sha256=2MqeJazsZNBTKghrtZDqXE2XoO_BShK60n3SEMjV74k,2787
 fusion_bench/method/task_singular_vector/__init__.py,sha256=WMucyl9pu_Ev2kcdrfT4moqMMbzD7hHQVFME5Su5jMA,298
 fusion_bench/method/task_singular_vector/utils/TSVC_utils.py,sha256=FytKbal48EW6iGIA-2zV7QSVbYTVflXr4Mr56q0W75k,2286
-fusion_bench/method/task_singular_vector/utils/TSVM_utils.py,sha256=gZu0QMYLh9WTOWEaUv9V0geliXX9qkGr2HTZOywEHb0,27223
+fusion_bench/method/task_singular_vector/utils/TSVM_utils.py,sha256=dsTMQ15zFJ1MPqDOt2TJ01O9Bwq_klyG9xL9hRD2aI0,27521
 fusion_bench/method/task_singular_vector/utils/__init__.py,sha256=Pgthb9Ld1x0Qis1wKWottwgzlBcyuzByFZCMIoI6Fys,240
 fusion_bench/method/ties_merging/__init__.py,sha256=9u9teBbdILbupr9jbwk-qCXSzssCssC5FUV2BfpyZM4,67
-fusion_bench/method/ties_merging/ties_merging.py,sha256=Pa37P1tpbaPSnnorbcKiDvfN1XsZIaWGSQMtDn4Kqyc,4000
+fusion_bench/method/ties_merging/ties_merging.py,sha256=rQGYr1HoZLbgwgfsfpIqln0tcoXj9s2J985FtRWO2dA,5502
 fusion_bench/method/ties_merging/ties_merging_utils.py,sha256=EZyltS9hUM8NmcvXjAqhBpj-ucMlMtR95082kPDsJPU,10296
 fusion_bench/method/trust_region/__init__.py,sha256=4ao0E-jTlmTQPArbFWD_dFn_4yve3urNIuSMT8JtRIM,91
 fusion_bench/method/trust_region/clip_task_arithmetic.py,sha256=SWP7sRMiXzkDZ3KdNht3zqjaTcAtB4wpnnd8KYbcKZI,7441
@@ -160,6 +172,7 @@ fusion_bench/method/weighted_average/__init__.py,sha256=bLxIuuB72hH05J_Spz4MZbiL
 fusion_bench/method/weighted_average/llama.py,sha256=BgJKZ15aqjFuNidOqAoGc74rpw3YRDnJ6gYOQJhe6tc,4161
 fusion_bench/method/weighted_average/weighted_average.py,sha256=E4byEA2VfXozu7S_gnYVvwI3qg8AFWaSeNRHGbs2Tno,3340
 fusion_bench/metrics/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+fusion_bench/metrics/continual_learning/backward_transfer.py,sha256=3e_eMN9eXZaGorankaPIQ-WfOHpW-bA1EywH6LDInWk,555
 fusion_bench/metrics/nyuv2/__init__.py,sha256=OjIB2vNcEoFrT-b8hqdcCgEgc2uCx3TyKUYS3aoRNNU,282
 fusion_bench/metrics/nyuv2/depth.py,sha256=oKc8NUZd0yAsk7-f78CbOzyuqmFpBhVFA8GOOdaNP7c,1548
 fusion_bench/metrics/nyuv2/loss.py,sha256=oBrkU8cDTczjIKixHjUGLJ_rghx0Vw59OL9DB-kZT3M,914
@@ -171,7 +184,7 @@ fusion_bench/metrics/text_to_image_generation/aesthetic_scorer.py,sha256=-ZaD84E
 fusion_bench/metrics/text_to_image_generation/compressibility.py,sha256=x4dNTFnAN4naChBDZBO-jUghnHAyobRVOupctKYRg1w,1656
 fusion_bench/metrics/text_to_image_generation/pickscore_scorer.py,sha256=aSWzl8k7z80Cirg5qdfkPsp3sMFEv_PjA1NJv3PPWXY,3115
 fusion_bench/mixins/__init__.py,sha256=AsUNvrHdNd6xht7-dfuVipmJuRfMNFlkgG-fn3ojt1U,892
-fusion_bench/mixins/clip_classification.py,sha256=Bk0M5AdHFDc6HYO45guQY70BS_O116S-bCiOns3GCJ0,10049
+fusion_bench/mixins/clip_classification.py,sha256=2Q20bEfRcRx9cg79ubCVpsey3TtpWa8jxk-N_JZVueY,10162
 fusion_bench/mixins/fabric_training.py,sha256=ZmycEhCaNCgVi5oM9m0q6msxgk3quowmFvDAcvskFrg,13017
 fusion_bench/mixins/lightning_fabric.py,sha256=6S1-rV6ItNQDSu7GM4qB99s8rnNXdO4PZDiQI4w0-DU,6593
 fusion_bench/mixins/rich_live.py,sha256=j7wNgrgwfdpl6nCXZGF_2DLtNq2aqCb_52Qhe9QSltc,495
@@ -251,23 +264,23 @@ fusion_bench/optim/lr_scheduler/utils/__init__.py,sha256=GfZk9VYL3cFE1Qy2xQpGc1G
 fusion_bench/optim/lr_scheduler/utils/visualization.py,sha256=Ea1n9ElNizAe0iUnjynyfteuZunv2-UBMN_NfEU2imA,3490
 fusion_bench/programs/__init__.py,sha256=oGoRp2TMI6ELxyfkeTg2h27hZJEDz9x31AsmvwvNvJw,508
 fusion_bench/programs/base_program.py,sha256=0dX_KcMWASo53pr-ldzfUBWIjEXy6oeDWZBrfc7FIk8,195
-fusion_bench/programs/fabric_fusion_program.py,sha256=gkt9BUj1KELotxtxjnXzhJeM9pEA5zyLVmMfZowbY0c,12406
+fusion_bench/programs/fabric_fusion_program.py,sha256=WOA9a2hxAKq0aykT4FLwHAyaFTo1XkYLU8fpiyOSX0o,12885
 fusion_bench/scripts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-fusion_bench/scripts/cli.py,sha256=497nhqnJAwxkqU2WCMUqcAIvqTmGRdQaByWGNTX_onY,1131
+fusion_bench/scripts/cli.py,sha256=hw32XtmixFxYXwgAY7iRBMzma_XQjdf_FxPiXKL6dIc,1154
 fusion_bench/scripts/imgui.py,sha256=r9Glbfbwu3JCsX9TKQFwcHarvwA_G7ff0jWBUPW1S1U,7613
 fusion_bench/scripts/nyuv2_mtl_train.py,sha256=hB_P_4DIT83CGOXoyyaBnh9fYnxTJtvAPbKa3_arF6E,3603
 fusion_bench/scripts/webui.py,sha256=ryA-2leSnHcYA88tTAYzJGDhiljbi0vl1Fibejzndlw,14398
 fusion_bench/scripts/clip/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 fusion_bench/scripts/clip/convert_checkpoint.py,sha256=zncgRAhInFpJDSHIm3GO4F6BzgsdAQVj3LLmV7g-JiQ,1221
 fusion_bench/taskpool/__init__.py,sha256=_qaYgzYnvrJDrZ2DjKXMvOFbelsLrujCKa_gP3UQBBg,1094
-fusion_bench/taskpool/base_pool.py,sha256=FaP0nndeSsrwbdd9mKa_CedbX9T5AHJmxk7Lc0NEVNY,835
+fusion_bench/taskpool/base_pool.py,sha256=Cbe3ZgJ34DWSDZeZEjlgqR0b84aM1i68D9-vomaooo8,852
 fusion_bench/taskpool/dummy.py,sha256=Di9JZO3XyDYn6wAGukrJMTnkS_NaxGTeQYo_3j1JD3Y,1675
 fusion_bench/taskpool/gpt2_text_classification.py,sha256=S4YyrcJhD4JOgvHF-AVG-gENgVGl-wpQZr1SbiThM04,4886
 fusion_bench/taskpool/nyuv2_taskpool.py,sha256=Y-TI-rzh9udCjX3FJ11ZbIG7CGrjDccGc-Ch1Ug6cRY,2059
 fusion_bench/taskpool/clip_vision/__init__.py,sha256=4xGO7rRbRpXF-I34A3WEMU4vydgfdtvXQ57ThaFcpmE,214
 fusion_bench/taskpool/clip_vision/clip_rankone_moe_taskpool.py,sha256=JKbRrGaRYztgZ-P0U767HISe40UpDVQ7fn6Tf2rrug0,4891
 fusion_bench/taskpool/clip_vision/clip_sparse_wemoe_taskpool.py,sha256=hVDTtg-oXqRFmAE2wZPFpk_kvtdk_wS-2-ev2ujEJBs,5390
-fusion_bench/taskpool/clip_vision/taskpool.py,sha256=xbJHQXUYd2ZDs-oIyE-3knCsPdiUbZCKN7O86kPwpsQ,14907
+fusion_bench/taskpool/clip_vision/taskpool.py,sha256=f6PlUlkeVOZr_S1jYP5g1t4dit1CSzL6m6llPSF9aro,14990
 fusion_bench/taskpool/llama/__init__.py,sha256=iB4ESMgnsl0m-z0YtRdPZiwGGv96-86R8pbSnkdet8Q,57
 fusion_bench/taskpool/llama/reward_model.py,sha256=ZpRSX4esBAuE0MdTjPHjqS2TnvGb6P8arOGxBeXnq6Y,5028
 fusion_bench/taskpool/llama/test_generation.py,sha256=kJ_5GruG12FsuJHDh_S7pbQgwEojTqhGpA_wVNH5KPc,6675
@@ -307,27 +320,33 @@ fusion_bench/tasks/flan_t5_text_generation/glue_evaluation.py,sha256=-B1wqVGp3wZ
 fusion_bench/tasks/flan_t5_text_generation/glue_load_dataset.py,sha256=sVihXHbqwi8IlDpiIxzvmDv-Ob7WKvi23GIRYbBUKOc,1833
 fusion_bench/tasks/flan_t5_text_generation/glue_preprocessors.py,sha256=GhRmGmcJGF4oVgZQarsBtx8GNKrNEZUkrillNz3iBuY,13183
 fusion_bench/tasks/flan_t5_text_generation/glue_prompt_templates.py,sha256=mKMTXIr5o-BqS_Hvv1bbMvvjQLLeKNVw7BKS9qgQ8Dw,1890
-fusion_bench/utils/__init__.py,sha256=yFhiBlrdcsJqZe-C5wdlZZ3wpmSN8Tipfpa2-R7CFbc,337
+fusion_bench/utils/__init__.py,sha256=fogKvcbYQx2nhj4-NWeeUg0xWUKA30lVJBxJpoPdFA0,398
 fusion_bench/utils/auto.py,sha256=uACQLE62_kNyhl4BGduvcbyeTE61qXpIJx3Ccl8kh68,920
 fusion_bench/utils/cache_utils.py,sha256=rU8x4-RFUtaCZWKd4Kft_7xgPTr1bpXnqUDMkrIdpj8,1653
 fusion_bench/utils/data.py,sha256=51nbgOnayyerLBUGHrlm9iilGjhJsBkXKKGXOKgLRW8,6104
 fusion_bench/utils/devices.py,sha256=72HeUVVlVGTt97JA7KFG3D8BM8VHqR-y1nkr9Bm-PRE,7578
+fusion_bench/utils/dict.py,sha256=ZCK0CRRT_B1Z18WY_GOYcmth7k5x9Jn1k7XhAVWRu98,1379
 fusion_bench/utils/dtype.py,sha256=kYoEGqsXitnwOU3W7ivqhQ0OjdI7MGu1VsyMJS4cSyQ,4299
+fusion_bench/utils/expr.py,sha256=zwHNrtIbOMnIChU-0ZI5qLbDva8zvHbizL-4F2TwM14,2386
+fusion_bench/utils/fabric.py,sha256=X2B_QPT2kqDPceQo3tp4XYAKbBpIs07w94Je_h2_81w,355
 fusion_bench/utils/functools.py,sha256=7_tYJ2WD88_2DDuOOj5aZz3cYuslYH5tsVyIgCeLtmk,1318
 fusion_bench/utils/hydra_utils.py,sha256=TklUDKDEZlg4keI-TEZiqh4gFjr9-61Rt1RMlqkoSGk,1174
-fusion_bench/utils/instantiate.py,sha256=v8L9JDfh2YoEOFpIQIHomvBoqdboZdYIaHEATnD2gdQ,16972
-fusion_bench/utils/json.py,sha256=iNeZHFvpzbb4oX-52dX15De_dMcux7vQtAUFZqW12GA,1907
+fusion_bench/utils/instantiate.py,sha256=8DFPFECJRWsJ8vKMWItR_vm19q5ClMP7bDL9y44TbWk,17134
+fusion_bench/utils/json.py,sha256=sVCqbm9mmyHybiui-O57KFt_ULrjLtN2wipSo6VDvqE,2533
 fusion_bench/utils/lazy_imports.py,sha256=v5l9cpHXPMaz1IVBmB5oOqefYr9vA3XvP340xT7Wy18,2796
 fusion_bench/utils/misc.py,sha256=liZPUnORPus45tZ9bOuyMJj3TshP6lOPxTVroshjmjc,385
 fusion_bench/utils/packages.py,sha256=DBmq7t02V0gCE8W05gaZwJnaz0NUrH7K75tRh4kwgyU,1833
-fusion_bench/utils/parameters.py,sha256=pR-Emteo2-SI3hFAvTQsCfkzqEh5ssPSNZ1Gf3J5o6Q,10263
-fusion_bench/utils/path.py,sha256=fG-3_bkjiYsZHx7uXM5UKNz20AkoXeINbprD5DuR098,157
+fusion_bench/utils/parameters.py,sha256=ZQG3AGo6B0DmbX13LQFJEESLUipaIqnp9hoWhr0t8-Y,10803
+fusion_bench/utils/path.py,sha256=hRA1CPHNnTYBUmzbftH77sHvn4aTuybEK5Tth1skP-k,531
 fusion_bench/utils/pylogger.py,sha256=a5tHfpEFbsdzw0vhQxt4BJ6CfTXaxyuwzoDFhyNy4KI,2468
-fusion_bench/utils/rich_utils.py,sha256=V_BjY3o8bXMp-kWfxle4cK48GGHDnKbVonZX65qbXAA,5464
+fusion_bench/utils/rich_utils.py,sha256=B8DhAYuVp23pG6ZnnYrUhcL-ikHZoQeTNqlM7u4pwwU,5786
+fusion_bench/utils/set.py,sha256=_43ZvGKJ_BK9sUslsSNhi7xEfuAQuyj3vViImnGpnCY,134
 fusion_bench/utils/state_dict_arithmetic.py,sha256=dVPBkO8Te9_VANPbetV59ORAQTw7D3css_-d0lYgK4k,9062
+fusion_bench/utils/tensorboard.py,sha256=VTEN-uoYBmfhCEJOOXCCZpJe2ErZg0VKcTGyfWZJkSM,1673
 fusion_bench/utils/timer.py,sha256=RC2hP8JqaibdL0FnRyUCBRf4m7CXyfn5tE16zBWZ7hg,1338
 fusion_bench/utils/type.py,sha256=Jz--BmTAzQkxcXXZfGiQLzLR0IPktrFGdjiWhkE93Qg,567
 fusion_bench/utils/plot/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+fusion_bench/utils/plot/color_data.py,sha256=5QO2tlf-9bCKywsIZJXxl6klWb8EntXFilTas_8je5c,48260
 fusion_bench/utils/plot/token.py,sha256=QGmL_qX8drmWnN_VNLD_0YjKc1o_qahJE-svXVor8dU,1634
 fusion_bench/utils/plot/token_notebook.py,sha256=bsntXf46Zz_RavTxNiB9c3-KvHw7LFwyLeG0MIwOwF4,3739
 fusion_bench/utils/strenum/__init__.py,sha256=id9ORi1uXrDxhbmVxitJ1KDwLS4H3AAwFpaK5h1cQzw,8531
@@ -436,7 +455,7 @@ fusion_bench_config/hydra/default.yaml,sha256=TT0RaUwYgfB7pKpbHgEbmuCVTB2fx2eXxv
 fusion_bench_config/hydra/help/fusion_bench_help.yaml,sha256=v8s891Cr5wyxBXGDn_VBBwwRmb0JXOL874Sl-zNoCWA,1880
 fusion_bench_config/hydra/job_logging/rich_logging.yaml,sha256=_dYGeFTCqaPrRowLXBNMXwzYhw8ns1TkQFfALwK1aCw,441
 fusion_bench_config/method/adamerging.yaml,sha256=fBG7jBBepygKpCbM3fmUeVAr2zzx0g8C21rGGfnEPkA,730
-fusion_bench_config/method/clip_finetune.yaml,sha256=LGrTAdY6we2tq-QHCbnmtH5HXBDnWepaOx0EDJoISFU,675
+fusion_bench_config/method/clip_finetune.yaml,sha256=yWjcdKYaKvy53sGaygg2ElAjb9-YFCyCGE1s9aB_dPM,677
 fusion_bench_config/method/depth_upscaling.yaml,sha256=tLU1DdSoV4StGcG7EFpf_3RGPiSL2Dfqr0txASAAwDY,623
 fusion_bench_config/method/dummy.yaml,sha256=5qs6OuIfriKOH7FgqvcMXMUoRLmXDZmjA4irpAsc5xo,45
 fusion_bench_config/method/mixtral_moe_merging.yaml,sha256=AdVhXD6Crw-B3QyNpP4ToHRSg-EeSCIGtazA7lQvPOU,148
@@ -453,11 +472,14 @@ fusion_bench_config/method/adamerging/layer_wise_gpt2.yaml,sha256=MsyTLP6-ad4tS-
 fusion_bench_config/method/adamerging/llama_sft.yaml,sha256=GNTNElki6Lsu5QvQxzu_sQ8fmxR-Zb60c-pD-gq2umA,1024
 fusion_bench_config/method/analysis/task_vector_cos_similarity.yaml,sha256=5eeJmJpbHdOOz20RyFcUJDCib_VG3CznvNm77_F2iNQ,140
 fusion_bench_config/method/analysis/task_vector_violin_plot.yaml,sha256=A8I51WRBTpI8-1LjP-pZVqJQ0K-OpVSMraT0AmT88_A,135
+fusion_bench_config/method/classification/clip_continual_finetune.yaml,sha256=NNSgKqtAUifbGNNQCFigyGkiMmYS3r99e37gmCd-JsY,792
+fusion_bench_config/method/classification/clip_finetune.yaml,sha256=yWjcdKYaKvy53sGaygg2ElAjb9-YFCyCGE1s9aB_dPM,677
 fusion_bench_config/method/concrete_subspace/clip_concrete_layer_wise_adamerging.yaml,sha256=XsHzr_5NoUZs0Us3eVwP3lUYXYvyJwGEEG9aDI_Z0rU,740
 fusion_bench_config/method/concrete_subspace/clip_concrete_task_arithmetic.yaml,sha256=eNoqcY1iMbs0Y5kKi_ya3rmQQMHqU7ht3EU7G_xmwN0,746
 fusion_bench_config/method/concrete_subspace/clip_concrete_task_wise_adamerging.yaml,sha256=WgTJj28FlVjR0_mCGJC5B8aJa9yezI3QusoXXHOrFoU,739
 fusion_bench_config/method/dare/simple_average.yaml,sha256=oTFSCHul86NTjTtJYK5pNr3tuxW7XxNI-y6fL9Yo4VI,113
 fusion_bench_config/method/dare/task_arithmetic.yaml,sha256=Cvsam89yquamn_GkITT6q8qFKN_Yb5nv8p-XgvnVrgU,134
+fusion_bench_config/method/dare/ties_merging.yaml,sha256=50mPiRkzLN7gxaIs56sPWkAUSvqvdxjQJ8eVl1yUGOg,418
 fusion_bench_config/method/dawe/dawe_for_clip.yaml,sha256=8-Z_kwwGCy1AO4brW-R_pe8oJ0yqoD4WCLI9ZtJ4KOo,1026
 fusion_bench_config/method/ensemble/max_model_predictor.yaml,sha256=fsWuNJwr1ohVB2aJ5L2fsiDLztm5GieE9JS99w--two,56
 fusion_bench_config/method/ensemble/simple_ensemble.yaml,sha256=bw9FabjhQYNbttsiMgTVd-Z4KIowf050Uy97vKtm2ys,55
@@ -465,6 +487,8 @@ fusion_bench_config/method/ensemble/weighted_ensemble.yaml,sha256=U_wQXtogtgiqOT
 fusion_bench_config/method/fisher_merging/clip_fisher_merging.yaml,sha256=rl7kfVvdo2pG-DnglQUbjzkyBqnq1FpfoSDSjFtdLwk,633
 fusion_bench_config/method/fisher_merging/fisher_merging.yaml,sha256=B1wrv9mhaOID4KcAUEMZNxlvY3tR3Q3UGualFslvx-Y,475
 fusion_bench_config/method/fisher_merging/gpt2_fisher_merging.yaml,sha256=AE7XZqRDj4__J_ipEcjPs7qTB2J3xLQyFRlq1W4iHFE,563
+fusion_bench_config/method/isotropic_merging/iso_c.yaml,sha256=Lh_OtTaUJ08--h85fUr2asF85xLe1NMCK8fVAhHOzdQ,82
+fusion_bench_config/method/isotropic_merging/iso_cts.yaml,sha256=x5vZo__kO8njl4_gFdXnOt15X_qFLv6-diSWHOR4clw,111
 fusion_bench_config/method/linear/expo.yaml,sha256=St3NW6cKVRV3vCn8y0gxQ8k66VTdtsLTEWQTbO9wQ0Y,420
 fusion_bench_config/method/linear/linear_interpolation.yaml,sha256=IQgltk5REITSx8xLuLP11ByPbuMgy7dHz_BrxIgwOas,67
 fusion_bench_config/method/linear/llama_expo.yaml,sha256=SEsC-l5gugY0vlsQkTJqzVgWJnMjFzWuTz814UKbFeM,624
@@ -476,6 +500,10 @@ fusion_bench_config/method/linear/weighted_average_for_llama.yaml,sha256=r8BlNqz
 fusion_bench_config/method/lm_finetune/bradley_terry_rm.yaml,sha256=em0Lnodl9bg8dos9MODMXjKtxWCXwQArjLT2z4TC3Q0,1352
 fusion_bench_config/method/lm_finetune/fullfinetune_sft.yaml,sha256=edj3juaYos2I9oQ8J6NKQNcNwqwcQGD74ZMosDsB5SY,1341
 fusion_bench_config/method/lm_finetune/peftfinetune_sft.yaml,sha256=9S-qsWUIALRwWd_gzNF1bwIuPPGP1MmqTpdQ53cwZmc,1628
+fusion_bench_config/method/opcm/opcm.yaml,sha256=FnkudkvzLT9nYSwPkmT36TGwpM0_oFUlyroN-Oesm2c,332
+fusion_bench_config/method/opcm/task_arithmetic.yaml,sha256=q2DnV3DmIqr9YyuShDYaIuI92tkcNEtCrfcRpECFris,328
+fusion_bench_config/method/opcm/ties_merging.yaml,sha256=llXE1Kp9fx2veik_XUGETmiD3LcVUha0n7SDJ6AOm_A,543
+fusion_bench_config/method/opcm/weight_average.yaml,sha256=vQla5o6wwwLZN_3AOA5o82zN7oiLRRmimpeD6aenwVI,305
 fusion_bench_config/method/pruning/llama_magnitude_pruning.yaml,sha256=Px8LU_UtDz-YHDFfqQ7scEPOproiFOaudKVshrhCTgc,483
 fusion_bench_config/method/pruning/llama_random_pruning.yaml,sha256=0RiZS8d42PXZzwncPG8zcbnyYJ9vtfr2sOSqS8oDyT4,325
 fusion_bench_config/method/pruning/llama_wanda_pruning.yaml,sha256=qKe5yIRsmK2KUyYENENWlw1qlGet9TpDhR-E_uO7vAw,501
@@ -492,7 +520,7 @@ fusion_bench_config/method/sparselo_pruning/llama_iterative_sparselo.yaml,sha256
 fusion_bench_config/method/sparselo_pruning/llama_pcp_sparselo.yaml,sha256=w1OWb38nW08K_hvrRMsCwmRxHWLGQfSSXg5nTiYaP8E,635
 fusion_bench_config/method/sparselo_pruning/llama_sparselo.yaml,sha256=J6vYIwqzh95-B3ekDias3FnCrVr4sig4zxpWyvz8hZ0,613
 fusion_bench_config/method/surgery/adamerging_surgery.yaml,sha256=Ne9JlJFgsRYcygBNCOBSN1ygBcLkE6I-8yusfTxyg-Y,826
-fusion_bench_config/method/task_singular_vector/TaskSingularVectorMerging.yaml,sha256=Se2v7AwwGqulXEVktRRzznpba4nNrWegY2bOwvjrHG8,74
+fusion_bench_config/method/task_singular_vector/TaskSingularVectorMerging.yaml,sha256=CLONjN9TXQ0OQwZHaje0q3WJWxR3LD1b5q5KrWJfZIA,169
 fusion_bench_config/method/trust_region/clip_task_arithmetic.yaml,sha256=mK09Ohsvj0Q6suj5qJM4DyCzRy192QBt4wjHS6W29IY,197
 fusion_bench_config/method/wemoe/sparse_weight_ensembling_moe.yaml,sha256=jiAco7M1XO0aekHFZKLKlXL_jRoCA8bgGD44Z7iB208,1001
 fusion_bench_config/method/wemoe/weight_ensembling_moe.yaml,sha256=OEv5yhyUCe5lXeT2PyXC49yrHXEM7i8SZDw6IQRDtAE,620
@@ -633,6 +661,7 @@ fusion_bench_config/modelpool/CLIPVisionModelPool/clip-vit-base-patch32_robustne
 fusion_bench_config/modelpool/CLIPVisionModelPool/clip-vit-base-patch32_single_finetuned.yaml,sha256=SBTyUX3wJwzdCTvZsW14FqaQP5r_nHPvusggGzP9P4o,148
 fusion_bench_config/modelpool/CLIPVisionModelPool/clip-vit-base-patch32_single_task_projection.yaml,sha256=3dymOGjBsEGzqCvzfDGkfImDNcCS0twJ07XpCrrTGcs,393
 fusion_bench_config/modelpool/CLIPVisionModelPool/clip-vit-base-patch32_svhn_and_mnist.yaml,sha256=fYthV8iwRvF-b4-OCIFW1Rud-BVoLx4Oo3DzVszfqek,175
+fusion_bench_config/modelpool/CLIPVisionModelPool/clip-vit-base-patch32_two_tasks_control_task.yaml,sha256=n60FIiq_VeoDDB-SwkpR3tYiS5r6XJVE8aLA2UubMn0,483
 fusion_bench_config/modelpool/CLIPVisionModelPool/clip-vit-large-patch14_TA8.yaml,sha256=-Tt_YggxkuIGT4_q5FR16zPvW2wWhGJ5LL8omxvHjvw,380
 fusion_bench_config/modelpool/CLIPVisionModelPool/clip-vit-large-patch14_TA8_model_only.yaml,sha256=PrDQr04UnhAciDrdtUutx-prMxF1Cs4jrEar7uJ-1Es,238
 fusion_bench_config/modelpool/CLIPVisionModelPool/clip-vit-large-patch14_TALL14.yaml,sha256=TyF7CKXpBOiDoLtDVvZuBzPI9gEJo_c99j3Gwkl3FWw,510
@@ -695,9 +724,9 @@ fusion_bench_config/taskpool/CLIPVisionModelTaskPool/clip-vit-single-task_sun397
 fusion_bench_config/taskpool/CLIPVisionModelTaskPool/clip-vit-single-task_svhn.yaml,sha256=2AqMiNCRRunLIrssHvFzu1lUzOaQn8uOHM9yjrQq-_A,109
 fusion_bench_config/taskpool/CLIPVisionModelTaskPool/clip_rankone_wemoe_clip-vit-classification_TA8.yaml,sha256=iQMj2VpDTe_D8OfCo94w5Ud2MON-EGa0DzVr6UmphrA,436
 fusion_bench_config/taskpool/CLIPVisionModelTaskPool/clip_sparse_wemoe_clip-vit-classification_TA8.yaml,sha256=i5Bn8bLl2cgqvrgtIGmoovUfSMehk_m-6C2wwcx5JMU,435
-fusion_bench-0.2.8.dist-info/LICENSE,sha256=nhnOJlw4CPuPVE0qvkGmxfFgHmKi-6nzXvTu8t0NUdg,1066
-fusion_bench-0.2.8.dist-info/METADATA,sha256=hMQ_4EyhPeYV0GFDmKHo9xqM2n9EwkKr3HDgeJ32SgU,13528
-fusion_bench-0.2.8.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-fusion_bench-0.2.8.dist-info/entry_points.txt,sha256=iUQ8MCJvda7HP4vYh2n1Teoapb4G9PBVYZkAfcc5SHU,116
-fusion_bench-0.2.8.dist-info/top_level.txt,sha256=BuO4TL6iHL_2yPBUX9-LlIrHRczA_BNMIFwweK0PQEI,13
-fusion_bench-0.2.8.dist-info/RECORD,,
+fusion_bench-0.2.10.dist-info/LICENSE,sha256=nhnOJlw4CPuPVE0qvkGmxfFgHmKi-6nzXvTu8t0NUdg,1066
+fusion_bench-0.2.10.dist-info/METADATA,sha256=kBYozBf6hgA-7ebsn7znqJdhCz4H0dJSv2jVIEkBvyA,16780
+fusion_bench-0.2.10.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+fusion_bench-0.2.10.dist-info/entry_points.txt,sha256=iUQ8MCJvda7HP4vYh2n1Teoapb4G9PBVYZkAfcc5SHU,116
+fusion_bench-0.2.10.dist-info/top_level.txt,sha256=BuO4TL6iHL_2yPBUX9-LlIrHRczA_BNMIFwweK0PQEI,13
+fusion_bench-0.2.10.dist-info/RECORD,,

{fusion_bench-0.2.8.dist-info → fusion_bench-0.2.10.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (75.6.0)
+Generator: setuptools (75.8.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

fusion_bench_config/method/classification/clip_continual_finetune.yaml

@@ -0,0 +1,28 @@
+_target_: fusion_bench.method.ContinualImageClassificationFineTuningForCLIP
+
+seed: 42
+# shuffle the order of the tasks
+shuffle_order: true
+learning_rate: 1e-5
+weight_decay: 0
+# number of training steps on each task
+num_steps: 4000
+batch_size: 128
+num_workers: 16
+save_interval: 500
+# if `state_dict_load_path` is not null, the training will be resumed from the state_dict_path
+state_dict_load_path: null
+# if `state_dict_save_path` is not null, the state_dict will be saved to the path after training
+state_dict_save_path: null
+# if `skip_training` is true, use with `state_dict_load_path` to skip training and only evaluate
+skip_training: false
+# === LoRA ===
+use_lora: false
+lora_config:
+  r: 16
+  lora_alpha: 32
+  target_modules:
+    - q_proj
+    - v_proj
+  lora_dropout: 0.1
+  bias: none

fusion_bench_config/method/classification/clip_finetune.yaml

@@ -0,0 +1,26 @@
+name: clip_finetune
+seed: 42
+learning_rate: 1e-5
+weight_decay: 0
+num_steps: 4000
+batch_size: 128
+num_workers: 16
+save_interval: 500
+# if `state_dict_load_path` is not null, the training will be resumed from the state_dict_path
+state_dict_load_path: null
+# if `state_dict_save_path` is not null, the state_dict will be saved to the path after training
+state_dict_save_path: null
+# if `skip_training` is true, use with `state_dict_load_path` to skip training and only evaluate
+skip_training: false
+# === LoRA ===
+use_lora: false
+lora_config:
+  r: 16
+  lora_alpha: 32
+  target_modules:
+    - q_proj
+    - v_proj
+  lora_dropout: 0.1
+  bias: none
+# === L-LoRA ===
+use_l_lora: false

fusion_bench_config/method/clip_finetune.yaml

@@ -3,8 +3,8 @@ seed: 42
 learning_rate: 1e-5
 weight_decay: 0
 num_steps: 4000
-batch_size: 64
-num_workers: 8
+batch_size: 128
+num_workers: 16
 save_interval: 500
 # if `state_dict_load_path` is not null, the training will be resumed from the state_dict_path
 state_dict_load_path: null

fusion_bench_config/method/dare/ties_merging.yaml

@@ -0,0 +1,15 @@
+_target_: fusion_bench.method.dare.DareTiesMerging
+
+# === DARE parameters ===
+sparsity_ratio: 0.5
+only_on_linear_weights: false
+rescale: true
+
+# === Ties merging parameters ===
+# Scaling factor $\lambda$
+scaling_factor: 0.5
+threshold: 20
+# List of keys to remove from the state dict, default is empty
+remove_keys: []
+# Function to merge the models, default is sum. Options are 'sum', 'mean', and 'max'
+merge_func: sum

fusion_bench_config/method/isotropic_merging/iso_c.yaml

@@ -0,0 +1,4 @@
+_target_: fusion_bench.method.ISO_C_Merge
+
+scaling_factor: 1.0
+exclude_keys: null

fusion_bench_config/method/isotropic_merging/iso_cts.yaml

@@ -0,0 +1,5 @@
+_target_: fusion_bench.method.ISO_CTS_Merge
+
+scaling_factor: 1.0
+common_space_fraction: 0.8
+exclude_keys: null

fusion_bench_config/method/opcm/opcm.yaml

@@ -0,0 +1,12 @@
+_target_: fusion_bench.method.opcm.opcm.OPCMForCLIP
+
+# shuffle the order of the models
+shuffle_order: true
+# the scaling factor for the SVD projection
+alpha: 0.5
+# the random seed to use
+seed: null
+# save the merged model on every step
+save_on_every_step: true
+# evaluate the merged model on every step
+evaluate_on_every_step: true

fusion_bench_config/method/opcm/task_arithmetic.yaml

@@ -0,0 +1,12 @@
+_target_: fusion_bench.method.opcm.task_arithmetic.ContinualTaskArithmeticForCLIP
+
+scaling_factor: 0.3
+
+# shuffle the order of the models
+shuffle_order: true
+# the random seed to use
+seed: null
+# save the merged model on every step
+save_on_every_step: true
+# evaluate the merged model on every step
+evaluate_on_every_step: true

fusion_bench_config/method/opcm/ties_merging.yaml

@@ -0,0 +1,18 @@
+_target_: fusion_bench.method.opcm.ties_merging.ContinualTiesMergingForCLIP
+
+# Scaling factor $\lambda$
+scaling_factor: 0.5
+threshold: 20
+# List of keys to remove from the state dict, default is empty
+remove_keys: []
+# Function to merge the models, default is sum. Options are 'sum', 'mean', and 'max'
+merge_func: sum
+
+# shuffle the order of the models
+shuffle_order: true
+# the random seed to use
+seed: null
+# save the merged model on every step
+save_on_every_step: true
+# evaluate the merged model on every step
+evaluate_on_every_step: true

fusion_bench_config/method/opcm/weight_average.yaml

@@ -0,0 +1,10 @@
+_target_: fusion_bench.method.opcm.weight_average.ContinualWeightAverageForCLIP
+
+# shuffle the order of the models
+shuffle_order: true
+# the random seed to use
+seed: null
+# save the merged model on every step
+save_on_every_step: true
+# evaluate the merged model on every step
+evaluate_on_every_step: true

fusion_bench_config/method/task_singular_vector/TaskSingularVectorMerging.yaml

@@ -1,2 +1,8 @@
 _target_: fusion_bench.method.TaskSingularVectorMerging
 remove_keys: null
+
+# alpha is a float or a list of floats
+# example:
+# alpha: 1
+# alpha: [1, 0.5, 0.25]
+alpha: 1

fusion_bench_config/modelpool/CLIPVisionModelPool/clip-vit-base-patch32_two_tasks_control_task.yaml

@@ -0,0 +1,18 @@
+defaults:
+  - _self_
+  - /dataset/image_classification/train@train_datasets:
+      - tiny-imagenet
+
+_target_: fusion_bench.modelpool.CLIPVisionModelPool
+_recursive_: false
+
+models:
+  _pretrained_: openai/clip-vit-base-patch32
+  model_1: tanganke/clip-vit-base-patch32_sun397
+  model_2: tanganke/clip-vit-base-patch32_stanford-cars
+  
+train_datasets: ???
+
+processor:
+  _target_: transformers.CLIPProcessor.from_pretrained
+  pretrained_model_name_or_path: openai/clip-vit-base-patch32