Papers › Dataless Knowledge Fusion by Merging Weights of Language Models

Dataless Knowledge Fusion by Merging Weights of Language Models

19 Dec 2022arXiv:2212.09849archive 2025-07-28

Xisen Jin, Xiang Ren, Daniel Preotiuc-Pietro, Pengxiang Cheng

Fine-tuning pre-trained language models has become the prevalent paradigm for building downstream NLP models. Oftentimes fine-tuned models are readily available but their training data is not, due to data privacy or intellectual property concerns. This creates a barrier to fusing knowledge across individual models to yield a better single model. In this paper, we study the problem of merging individual models built on different training data sets to obtain a single model that performs well both across all data set domains and can generalize on out-of-domain data. We propose a dataless knowledge fusion method that merges models in their parameter space, guided by weights that minimize prediction differences between the merged model and the individual models. Over a battery of evaluation settings, we show that the proposed method significantly outperforms baselines such as Fisher-weighted averaging or model ensembling. Further, we find that our method is a promising alternative to multi-task learning that can preserve or sometimes improve over the individual models without access to the training data. Finally, model merging is more efficient than training a multi-task model, thus making it applicable to a wider set of scenarios.

PaperPDFCodeCode Syntology ranMethod as code

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Syntology holds a reference implementation of dataless knowledge fusion, generated from the paper text (not the authors' code), at level 3 on Syntology's ladder: V2 “two implementations GENERATED from the paper's signature and spec agree numerically (1e-6) on fixture inputs” “neither of the two ever read the authors' own code, so V2 measures how unambiguously the paper specifies the method, not fidelity to the authors' implementation (measured 2026-09-11, commit 75eaa77: cross_checked_by is a second LLM on 2,628 of 2,831 served samples)”; V3 “paper-derived property tests pass on both of those same two generations.” (quoted from Syntology's server card). No rung compares this code against the authors' implementation. For agents: get_reference_implementation(name='dataless knowledge fusion') (how to connect); the free check have('2212.09849') needs no token.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 19 code samples harvested from 1 repository linked to this paper: 10 ran (of which 4 constructed an object rather than computing a result; 4 with no instrument failure: 0 honoured, 0 violated, 4 with no contract checked; 6 where Syntology's instrument failed) · 9 unverified. Where Syntology's instrument failed: 6 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 19 samples from 1 repository, 10 ran (4 with no instrument failure; 6 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written on 2026-09-22 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

bloomberg/dataless-model-merging officialmentioned in paperpytorchSyntology: 19 samples, 10 ran (6 instrument-failed) · 9 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

19 samples harvested: 10 ran (of which 4 constructed an object rather than computing a result; 4 with no instrument failure: 0 honoured, 0 violated, 4 with no contract checked; 6 where Syntology's instrument failed) · 9 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

6ran · our draft was wrong
4ran
9unverified

Licence: 0 of the 19 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from bloomberg/dataless-model-merging. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.

bloomberg/dataless-model-merging official repository 19 samples, 10 ran (6 instrument-failed) · 9 unverified
FedAvgMerger src/model_merge/ot_merger.py ran Apache-2.0 (permissive) · 10b9df811ddc6d24 · report fixture: returned FedAvgMerger · level recorded 2026-09-22
GroundMetric src/model_merge/ot_merger.py ran Apache-2.0 (permissive) · f3635898362e71cb · report fixture: returned GroundMetric · level recorded 2026-09-22
ModelMergerBase src/model_merge/ot_merger.py ran Apache-2.0 (permissive) · c314ef759f51e47d · report fixture: returned ModelMergerBase · level recorded 2026-09-22
TmpLocalModel src/model_merge/ot_merger.py ran Apache-2.0 (permissive) · e823833341bb66b2 · report fixture: returned TmpLocalModel · level recorded 2026-09-22
compute_bce_loss src/model_merge/ot_merger.py ran · our draft was wrong Apache-2.0 (permissive) · 9512ea87c384d970 · report fixture: signature config:Config/0/?, logits:Tensor/2/float, labels:Tensor/2/int · returned Tensor · level recorded 2026-09-22
filter_modules_by_regex src/model_merge/ot_merger.py ran · our draft was wrong Apache-2.0 (permissive) · af8fc9a6b079dd80 · report fixture: signature base_module:callable/0/?, include_patterns:list/1/str, include_type:list/1/callable · returned dict · level recorded 2026-09-22
filter_params_to_merge src/model_merge/ot_merger.py ran · our draft was wrong Apache-2.0 (permissive) · 542bdbc6b44e26fe · report fixture: signature param_names:list/1/str, exclude_param_regex:list/1/str · returned list · level recorded 2026-09-22
get_component_configs src/model_merge/ot_merger.py ran · our draft was wrong Apache-2.0 (permissive) · b7a34bad497441a4 · report fixture: signature config:MockConfig/0/?, model_names:list/1/str · returned list · level recorded 2026-09-22
get_reweight_factor src/model_merge/ot_merger.py ran · our draft was wrong Apache-2.0 (permissive) · d150cc5adcf9981b · report fixture: signature freq_dist:dict/0/?, labels:Tensor/2/int, schema:str/0/str · returned Tensor · level recorded 2026-09-22
get_submodule in 6 papers src/model_merge/ot_merger.py ran · our draft was wrong Apache-2.0 (permissive) · 253168af406c0497 · report fixture: signature m:callable/0/?, name:str/0/str · returned Linear · level recorded 2026-09-22
DebertaV2ForMultiLabelClassification src/model_merge/ot_merger.py unverified Apache-2.0 (permissive) · 9f88183c29de2263 · report
DistilBERTForMultiLabelClassification src/model_merge/ot_merger.py unverified Apache-2.0 (permissive) · cb4e87b0878653c2 · report
GPT2ForMultiLabelClassification src/model_merge/ot_merger.py unverified Apache-2.0 (permissive) · dec050be98e95d6a · report
MTLModel src/model_merge/ot_merger.py unverified Apache-2.0 (permissive) · 9410ed9ff867ec7d · report
MultiDomainModel src/model_merge/ot_merger.py unverified Apache-2.0 (permissive) · 80ffc739f078f603 · report
OptimalTransportMerger src/model_merge/ot_merger.py unverified Apache-2.0 (permissive) · cc27cfd051c590d3 · report
RoBERTaForMultiLabelClassification src/model_merge/ot_merger.py unverified Apache-2.0 (permissive) · 7ff12c9ca05e8578 · report
create_model in 163 papers src/model_merge/ot_merger.py unverified Apache-2.0 (permissive) · 3b2e0a4e703d6201 · report
create_stl_model src/model_merge/ot_merger.py unverified Apache-2.0 (permissive) · 4b49e7003c26de74 · report

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. For a generated assembly, an independent peer is a second implementation generated separately from the same paper text; neither implementation read the authors' code. At level 2 the two agreed numerically on the fixture's inputs; level 3 is recorded when property tests derived from the paper also passed on both, and each level 3 card prints how many its report lists. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it. A generated assembly's label was set when its code was generated for the method, and no model judged it; its card prints it as "by construction" or "not judged by a model". Where an assembly's phrase has no constructed figure, a brick that ran is not among the harvested samples this build reads, and whether a brick's run only constructed an object is read from that brick's harvested sample, so the figure was not counted; it does not say there were none.

2 assemblies on this paper: 1 harvested, 1 generated.

generated Assembly for dataless knowledge fusion (graph method: dataless knowledge fusion), generated from the paper text, not the repository's code. 1 brick; individually on a fixture: 1 ran (0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified. Ran end-to-end against an independent peer (level 3; 5 property tests passed on both).

label matches by construction: the code was generated for this method (not judged by a model)

  1. compute_merged_linear_weight entry · level 3 · ran; the contract check could not run · 92d043c8ae674263c62d3159a3f04ad4d6b0ae51b20decc0b5958fb230cdf91e

harvested Assembly for dataless knowledge fusion (graph method: dataless knowledge fusion). 19 bricks; individually on a fixture: 10 ran (of which 4 constructed an object rather than computing a result; 4 with no instrument failure: 0 honoured, 0 violated, 4 with no contract checked; 6 where Syntology's instrument failed) · 9 unverified. The assembly has not been run end-to-end.

  1. OptimalTransportMerger entry · level 0 · unverified · cc27cfd051c590d3
  2. FedAvgMerger helper · level 1 · ran · 10b9df811ddc6d24
  3. GroundMetric helper · level 1 · ran · f3635898362e71cb
  4. TmpLocalModel helper · level 1 · ran · e823833341bb66b2
  5. create_model helper · level 0 · unverified · 3b2e0a4e703d6201
  6. filter_modules_by_regex helper · level 1 · ran · our draft was wrong · af8fc9a6b079dd80
  7. get_submodule helper · level 1 · ran · our draft was wrong · 253168af406c0497
  8. ModelMergerBase helper · level 1 · ran · c314ef759f51e47d
  9. filter_params_to_merge helper · level 1 · ran · our draft was wrong · 542bdbc6b44e26fe
  10. MTLModel helper · level 0 · unverified · 9410ed9ff867ec7d
  11. MultiDomainModel helper · level 0 · unverified · 80ffc739f078f603
  12. create_stl_model helper · level 0 · unverified · 4b49e7003c26de74
  13. get_component_configs helper · level 1 · ran · our draft was wrong · b7a34bad497441a4
  14. DebertaV2ForMultiLabelClassification helper · level 0 · unverified · 9f88183c29de2263
  15. DistilBERTForMultiLabelClassification helper · level 0 · unverified · cb4e87b0878653c2
  16. GPT2ForMultiLabelClassification helper · level 0 · unverified · dec050be98e95d6a
  17. RoBERTaForMultiLabelClassification helper · level 0 · unverified · 7ff12c9ca05e8578
  18. compute_bce_loss helper · level 1 · ran · our draft was wrong · 9512ea87c384d970
  19. get_reweight_factor helper · level 1 · ran · our draft was wrong · d150cc5adcf9981b

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2212.09849") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Multi-Task Learning

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections