SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.05910 · ICLR · 2024

SALMON: Self-Alignment with Instructable Reward Models

Chuang Gan, Yiming Yang, Zhiqing Sun, David Cox, Zhenfang Chen, Yikang Shen, Qinhong Zhou, Hongxin Zhang

arXiv · PDF · Open in the Atlas

Code that ran

13 of 13 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 3 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 10 failures of our instrument; 0 no recorded run); 3 of them were also run on the host, not in the jail, by the peer-agreement lane that set their level, measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2310.05910.json · badge: /api/ran/2310.05910.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
ibm/salmon canonical (classifier label)
classifier basis: census
13 ran · 0 no recorded run
FunctionStatusWhere it lives
batch_select Ran
ran; our contract draft was wrong, not the code
ibm/salmon/training/models/reward_model.py
pointer only (licence: GPL-3.0) · get_code("2154e44a6b59e432")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

repeat_kv Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ibm/salmon/training/models/llama_with_flash_attn.py
pointer only (licence: GPL-3.0) · get_code("30d7eec482ebf6b1")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

unpack_dict Ran
ran; our contract draft was wrong, not the code
ibm/salmon/training/models/reward_model.py
pointer only (licence: GPL-3.0) · get_code("a6989412aca721c1")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

all_gather_and_cat Ran
ran, honoured the contract we drafted
ibm/salmon/training/models/distributed_utils.py
pointer only (licence: GPL-3.0) · get_code("5aa59c85970b28a7")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

apply_rotary_pos_emb Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ibm/salmon/training/models/llama_with_flash_attn.py
pointer only (licence: GPL-3.0) · get_code("e34097675d132bc1")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

cast_with_native_amp Ran
ran; our contract draft was wrong, not the code
ibm/salmon/training/models/rl_trainer.py
pointer only (licence: GPL-3.0) · get_code("af2735cffc6ee536")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

clean_after_stop_token Ran
ran; our contract draft was wrong, not the code
ibm/salmon/training/models/ppo_trainer.py
pointer only (licence: GPL-3.0) · get_code("d5a396157801a324")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

create_scheduler Ran
ran; our contract draft was wrong, not the code
ibm/salmon/training/models/trainer_utils.py
pointer only (licence: GPL-3.0) · get_code("42596ac916e883a5")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_last_checkpoint Ran
ran; our contract draft was wrong, not the code
ibm/salmon/training/qlora_utils.py
pointer only (licence: GPL-3.0) · get_code("89ce33341af66037")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

rotate_half Ran
ran; our contract draft was wrong, not the code
ibm/salmon/training/models/llama_with_flash_attn.py
pointer only (licence: GPL-3.0) · get_code("b99eea6376d1e212")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

setup Ran
ran, honoured the contract we drafted
ibm/salmon/training/models/distributed_utils.py
pointer only (licence: GPL-3.0) · get_code("4098e9c2afde4a5d")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

truncate_after_eos Ran
ran, honoured the contract we drafted
ibm/salmon/training/models/rl_trainer.py
pointer only (licence: GPL-3.0) · get_code("1088838ef514bf22")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

whiten Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ibm/salmon/training/models/ppo_trainer.py
pointer only (licence: GPL-3.0) · get_code("9a0b92f09a2d6c76")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, ibm/salmon, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 11 are also present in other papers whose code we hold (5 with at least one paper dated earlier, 2 the same month, 10 later) and 2 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
rotate_half 360 other papers
28 dated earlier, 10 the same month, 319 later, 3 undated · 30 only as an assembly brick
1901.02860 (2019-01)
Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context
2203.08913 (2022-03)
Memorizing Transformers
2302.13971 (2023-02)
LLaMA: Open and Efficient Foundation Language Models
and 357 more not named here
repeat_kv 243 other papers
12 dated earlier, 2 the same month, 227 later, 2 undated · 22 only as an assembly brick
1901.02860 (2019-01)
Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context
2203.08913 (2022-03)
Memorizing Transformers
2305.17888 (2023-05)
LLM-QAT: Data-Free Quantization Aware Training for Large Language Models
and 240 more not named here
get_last_checkpoint 5 other papers
1 dated earlier, 4 later
2305.14314 (2023-05)
QLORA: Efficient Finetuning of Quantized LLMs
2405.11165 (2024-05)
Automated Multi-level Preference for MLLMs
2406.11280 (2024-06)
ISR-DPO: Aligning Large Multimodal Models for Videos by Iterative Self-Retrospective DPO
and 2 more not named here
create_scheduler 4 other papers
1 dated earlier, 3 later
2309.09055 (2023-09)
Exploring the impact of low-rank adaptation on the performance, efficiency, and regularization of RLHF
2406.11280 (2024-06)
ISR-DPO: Aligning Large Multimodal Models for Videos by Iterative Self-Retrospective DPO
2406.12030 (2024-06)
SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Model
and 1 more not named here
all_gather_and_cat 3 other papers
3 later
2405.11165 (2024-05)
Automated Multi-level Preference for MLLMs
2406.11280 (2024-06)
ISR-DPO: Aligning Large Multimodal Models for Videos by Iterative Self-Retrospective DPO
2406.12030 (2024-06)
SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Model
batch_select 3 other papers
3 later
2403.09472 (2024-03)
Easy-to-Hard Generalization: Scalable Alignment Beyond Human Supervision
2406.11280 (2024-06)
ISR-DPO: Aligning Large Multimodal Models for Videos by Iterative Self-Retrospective DPO
2406.12030 (2024-06)
SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Model
setup 3 other papers
3 later
2405.11165 (2024-05)
Automated Multi-level Preference for MLLMs
2406.11280 (2024-06)
ISR-DPO: Aligning Large Multimodal Models for Videos by Iterative Self-Retrospective DPO
2406.12030 (2024-06)
SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Model
unpack_dict 3 other papers
3 later
2403.09472 (2024-03)
Easy-to-Hard Generalization: Scalable Alignment Beyond Human Supervision
2406.11280 (2024-06)
ISR-DPO: Aligning Large Multimodal Models for Videos by Iterative Self-Retrospective DPO
2406.12030 (2024-06)
SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Model
apply_rotary_pos_emb 1 other paper
1 dated earlier
2307.09288 (2023-07)
Llama 2: Open Foundation and Fine-Tuned Chat Models
cast_with_native_amp 1 other paper
1 later
2406.11280 (2024-06)
ISR-DPO: Aligning Large Multimodal Models for Videos by Iterative Self-Retrospective DPO
whiten 1 other paper
1 later
2406.11280 (2024-06)
ISR-DPO: Aligning Large Multimodal Models for Videos by Iterative Self-Retrospective DPO

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Supervised Fine-Tuning (SFT) on response demonstrations combined with Reinforcement Learning from Human Feedback (RLHF) constitutes a powerful paradigm for aligning LLM-based AI agents. However, a significant limitation of such an approach is its dependency on high-quality human annotations, making its application to intricate tasks challenging due to difficulties in obtaining consistent response demonstrations and in-distribution response preferences. This paper presents a novel approach, namely SALMON, to align base language models with minimal human supervision, using only a small set of human-defined principles, yet achieving superior performance. Central to our approach is an instructable reward model. Trained on synthetic preference data, this model can generate reward scores based on arbitrary human-defined principles. By merely adjusting these principles during the RL training phase, we gain full control over the preferences with the instructable reward model, subsequently influencing the behavior of the RL-trained policy models, and reducing the reliance on the collection of online human preferences. Applying our method to the LLaMA-2-70b base language model, we developed an AI assistant named Dromedary-2. With only 6 exemplars for in-context learning and 31 human-defined principles, Dromedary-2 significantly surpasses the performance of several state-of-the-art AI systems, including LLaMA-2-Chat-70b, on various benchmark datasets. We have opensourced the code and model weights to encourage further research into aligning LLM-based AI agents with enhanced supervision efficiency, improved controllability, and scalable oversight.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2310.05910")
get_code_for_paper("2310.05910")
have("2310.05910")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2310.05910.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2310.05910.svg)](https://syntology.ai/paper/2310.05910)

Connect an agent — have() is free.