Home › Code › dpo_loss

dpo_loss

Syntologyentry name in harvested coderead from the graph 2026-09-24

dpo_loss appears in the code Syntology harvested for 21 papers, as 17 distinct code bodies found in 21 places (a place is one code body under one paper). At least one of them ran in 10 of the papers; 0 of the code bodies carry a behaviour fingerprint.

What this page is not. Routines are grouped here by the exact string of their function or class name. Nothing asserts that two samples named dpo_loss do the same thing, share code, or are comparable; the name is a string, not an identity. Behaviour outputs (what a fingerprinted sample returned on the shared battery) are not in this export and are not shown here; the graph at syntology.ai holds them. "Ran" means executed on a synthesized fixture, not that the code is correct or reproduces a paper.

Samples Syntology

Syntology ran 8 of the 17 distinct code bodies named dpo_loss; 9 are unverified. One tile per status, in the site's fixed vocabulary, each code body counted once:

0ran · honoured contract
0ran · violated contract
1ran · our draft was wrong
1ran · fixture could not drive it
6ran
9unverified
0fingerprinted

Licence is a property of each copy, so it is counted per place: 8 of the 21 places are pointer only (Syntology does not serve that copy's text). This site shows no code text for any sample; every row below links to the file in its repository where the record names one.

“Ran” means the sample executed on a synthesized input; it does not mean the output is correct. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself; “our draft was wrong” and “fixture could not drive it” are failures of Syntology's instrument, not of the code, and those samples did run. The ran count above is every status except unverified, the same rule as each paper page.

Papers

21 papers shown of 21, newest first; 21 places in the table. A paper with no recorded date is placed by the month its arXiv id encodes, shown in the Date column as YYYY-MM (from id). One row per place: a paper whose repository defines the name more than once appears more than once, and the same code body held for several papers appears once under each, with the same status. Titles and dates are the archive's archive 2025-07-28 for papers in the archive, and the graph's for 1 papers added by Syntology; 1 papers have no page here and are shown by arXiv id only. Status and fingerprint are Syntology's record of each code body; licence is recorded for each place. The File cell ends with the code body's code_sha256, Syntology's identity for that exact code: an agent fetches the stored sample with Syntology's MCP tool get_code(code_sha256="…") (how to connect); click an identity to copy that call.

PaperDateFileStatus SyntologyLicence
Self-Improving Tabular Language Models via Iterative Reward-Guided Post-Training added by Syntology 2026-04 (from id) Yunbo-max/Post-Training-Tabular-Language-Models/loss.py 5aa6cff234d655cd unverified MIT (permissive)
arXiv:2506.11712 2025-06 (from id) Liuwq-bit/SymMPO/muffin/train/trainers.py a9ca56191b3904f6 ran · fixture could not drive it no licence file found · pointer only
NLSR: Neuron-Level Safety Realignment of Large Language Models Against Harmful Fine-Tuning 17 Dec 2024 xinykou/nlsr/src/prune_regions/modeling_llama.py b8d15a97350e90e6 unverified no licence file found · pointer only
Beyond Toxic Neurons: A Mechanistic Analysis of DPO for Toxicity Reduction 10 Nov 2024 yushi-y/dpo-toxic-neurons/train_dpo/trainers.py 6e420696373414b1 unverified MIT (permissive)
COMAL: A Convergent Meta-Algorithm for Aligning LLMs with General Preferences 30 Oct 2024 yale-nlp/comal/losses.py a71ddf2f2e1209ca unverified no licence file found · pointer only
$f$-PO: Generalizing Preference Optimization with $f$-divergence Minimization 29 Oct 2024 minkaixu/fpo/src/align_stage/loss.py f87ec21e6cc554fb unverified MIT (permissive)
Compute-Constrained Data Selection 21 Oct 2024 oseyosey/CCDS/ccds/evaluation/open_instruct/dpo_utils.py 48ff6fdc0e5dae2e unverified Apache-2.0 (permissive)
Learning Fine-Grained Grounded Citations for Attributed Large Language Models 8 Aug 2024 luckyyysta/fine-grained-attribution/training/stage2_consistency_aware_alignment/dpo_utils.py b2b6de249cebe189 ran no licence file found · pointer only
Understanding Reference Policies in Direct Preference Optimization 18 Jul 2024 yale-nlp/refdpo/dpo_utils.py 8c5617ecaa98baa1 ran no licence file found · pointer only
Lottery Ticket Adaptation: Mitigating Destructive Interference in LLMs 24 Jun 2024 kiddyboots216/lottery-ticket-adaptation/rlaif/single_gpu_trainers.py 16a2450aff4a3577 ran Apache-2.0 (permissive)
Unpacking DPO and PPO: Disentangling Best Practices for Learning from Preference Feedback 13 Jun 2024 hamishivi/EasyLM/EasyLM/models/llama/llama_train_dpo.py a6d706f6e0699b5c unverified Apache-2.0 (permissive)
SimPO: Simple Preference Optimization with a Reference-Free Reward 23 May 2024 openpsi-project/realhf/realhf/impl/model/utils/dpo_functional.py 6b05e6f1ad3cb9b6 ran · our draft was wrong Apache-2.0 (permissive)
ADELIE: Aligning Large Language Models on Information Extraction 8 May 2024 THU-KEG/ADELIE/train4llama/open_instruct/dpo_utils.py b2b6de249cebe189 ran no licence file found · pointer only
A Critical Evaluation of AI Feedback for Aligning Large Language Models 19 Feb 2024 architsharma97/dpo-rlaif/trainers.py 16a2450aff4a3577 ran Apache-2.0 (permissive)
Towards Efficient Exact Optimization of Language Model Alignment 1 Feb 2024 haozheji/exact-optimization/src/align_stage/loss.py f87ec21e6cc554fb unverified MIT (permissive)
A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity 3 Jan 2024 ajyl/dpo_toxic/toxicity/train_dpo/trainers.py 6e420696373414b1 unverified MIT (permissive)
Sample Efficient Preference Alignment in LLMs via Active Exploration 1 Dec 2023 belakaria/active-llm-alignment/trainers.py cb2b1a4a7f1ecabf ran MIT (permissive)
Embodied Multi-Modal Agent trained by an LLM from a Parallel TextWorld 28 Nov 2023 stevenyangyj/emma-alfworld/LAVIS/lavis/models/blip2_models/blip2_emma.py 112c839f24908701 ran no licence file found · pointer only
Preference Optimization for Molecular Language Models 18 Oct 2023 harmonic-discovery/pref-opt-for-mols/pref_opt_for_mols/models/dpo.py 3d3a8ce9515f501f ran no licence file found · pointer only
Beyond Reverse KL: Generalizing Direct Preference Optimization with Diverse Divergence Constraints 28 Sep 2023 alecwangcq/f-divergence-dpo/trainers.py fc5f3e7e6f97a222 unverified Apache-2.0 (permissive)
Leftover Lunch: Advantage-based Offline Reinforcement Learning for Language Models 24 May 2023 abaheti95/LoL-RL/dpo_qlora_llama_hh.py 1ed7c9ceaa90eb2c unverified MIT (permissive)

This site shows no code text; each File cell links to the file on GitHub at the repository's current default branch, which may have changed since the harvest. "Pointer only" means Syntology does not serve that copy's text, for one of four 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; or the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar). 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 cell for the reason. Per-sample records for a paper are on its paper page under "Code Syntology ran".

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