SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2510.25441 · ICML · 2026

Grounded in Reality: Learning and Deploying Proactive LLM from Offline Logs

Yaliang Li, Daoyuan Chen, Ce Wang, Yushuo Chen, Xuchen Pan, Yilun Huang, ‡ Bolin, Offline Logs, Fei Wei

arXiv · PDF · Open in the Atlas

Code that ran

3 of 3 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; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 3 failures of our instrument; 0 no recorded run), measured 2026-11-22; 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/2510.25441.json · badge: /api/ran/2510.25441.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
modelscope/Trinity-RFT canonical (classifier label)
classifier basis not recorded on the link
3 ran · 0 no recorded run
FunctionStatusWhere it lives
aggregate_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
modelscope/Trinity-RFT/trinity/algorithm/utils.py
code served (permissive licence) · get_code("29290736aa877212")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create sample tensors
batch_size = 2
seq_length = 4

# values: arbitrary shape tensor of values to aggregate
values = torch.randn(batch_size, seq_length)

# mask: same shape as values, 1/True = include, 0 = ignore
mask = torch.tensor([
    [1.0, 1.0, 0.0, 1.0],
    [1.0, 0.0, 1.0, 1.0]
])

# loss_agg_mode: one of the aggregation modes
loss_agg_mode = 'token-mean'

# normalizer: optional normalizer for 'seq-mean-token-sum-norm' mode
normalizer = None

ARGS = {
    'values': values,
    'mask': mask,
    'loss_agg_mode': loss_agg_mode,
    'normalizer': normalizer
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
masked_mean Ran
ran; our contract draft was wrong, not the code
modelscope/Trinity-RFT/trinity/algorithm/utils.py
code served (permissive licence) · get_code("c73c46c929b295a5")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create small tensors for testing masked_mean
# values: tensor of shape (2, 4) with float values
values = torch.randn(2, 4)

# mask: tensor of shape (2, 4) with boolean values
mask = torch.tensor([[True, True, False, True],
                     [True, False, True, False]], dtype=torch.bool)

# axis: can be None or an integer
axis = None

ARGS = {
    "values": values,
    "mask": mask,
    "axis": axis
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
masked_sum Ran
ran; our contract draft was wrong, not the code
modelscope/Trinity-RFT/trinity/algorithm/utils.py
code served (permissive licence) · get_code("21d59eaacff05781")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 1, dtype family float, shape (2,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create small tensors with distinct dimensions
values = torch.randn(2, 4)  # batch size 2, dimension 4
mask = torch.ones(2, 4)  # same shape as values

# Test with axis parameter
axis = 1

ARGS = {
    "values": values,
    "mask": mask,
    "axis": axis,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family float, shape (2,) — shape (2,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 3 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 3 later) and 0 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
aggregate_loss 2 other papers
2 later
2606.20002 (2026-06)
Connect the Dots: Training LLMs for Long-Lifecycle Agents with Cross-Domain Generalization Via Reinforcement Learning
2606.26917 (2026-06)
GEOALIGN: Geometric Rollout Curation for Robust LLM Reinforcement Learning
masked_mean 2 other papers
2 later
2606.20002 (2026-06)
Connect the Dots: Training LLMs for Long-Lifecycle Agents with Cross-Domain Generalization Via Reinforcement Learning
2606.26917 (2026-06)
GEOALIGN: Geometric Rollout Curation for Robust LLM Reinforcement Learning
masked_sum 2 other papers
2 later
2606.20002 (2026-06)
Connect the Dots: Training LLMs for Long-Lifecycle Agents with Cross-Domain Generalization Via Reinforcement Learning
2606.26917 (2026-06)
GEOALIGN: Geometric Rollout Curation for Robust LLM Reinforcement Learning

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

Large Language Models (LLMs) excel as passive responders, but teaching them to be proactive, goal-oriented partners-a critical capability in high-stakes domains-remains a major challenge. Current paradigms either myopically optimize single-turn attributes or rely on brittle, high-cost user simulators, creating a persistent "reality gap". To bridge this gap, we introduce Learn-to-Ask, a general, simulator-free framework for learning and deploying proactive dialogue agents directly from offline expert data, bypassing the need to model complex user dynamics. Our key insight is to reframe the offline policy learning problem by leveraging the observed future of each expert trajectory. This allows us to infer a dense, turn-by-turn reward signal grounded in the expert's revealed strategy, decomposing the intractable long-horizon problem into a series of supervised learning tasks, and training a policy to output a structured (action, state assessment) tuple, governing both what to ask and, crucially, when to stop. To ensure reward fidelity, our Automated Grader Calibration pipeline systematically purges noise from the LLM-based reward model with minimal human supervision. Empirically, we demonstrate the efficacy of Learn-to-Ask in a real-world medical dataset, using LLMs of varying sizes up to 32B. Our approach culminates in the successful deployment of LLMs into a live, large-scale online AI service. In rigorous in-house evaluations, our model was launched and achieved performance even superior to human experts, proving our framework's ability to translate offline data into tangible, real-world impact. We hope this work provides a practical and economically viable blueprint for transforming passive LLMs into proactive, goal-oriented LLM applications.

For agents

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

get_harvested_code_for_paper("2510.25441")
get_code_for_paper("2510.25441")
have("2510.25441")

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

curl https://syntology.ai/api/ran/2510.25441.json

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

[![Syntology run record](https://syntology.ai/api/ran/2510.25441.svg)](https://syntology.ai/paper/2510.25441)

Connect an agent — have() is free.