Bang Liu

Biography

Bang Liu is an assistant professor in the Department of Computer Science and Operations Research (DIRO), and a core member of the Applied Research in Computational Linguistics Lab (RALI) at Université de Montréal. He is also an associate academic member of Mila – Quebec Artificial Intelligence Institute and a Canada CIFAR AI Chair.

Liu received his BEng from the University of Science and Technology of China in 2013, and his MSc and PhD degrees from the University of Alberta in 2015 and 2020, respectively. His research interests lie primarily in the areas of natural language processing, multimodal and embodied learning, theory and techniques for AGI (e.g., understanding and improving large language models), and AI for science (e.g., health, material science, XR).

Current Students

Farshid Effaty

Postdoctorate - Université de Montréal

f.effaty@gmail.com

Wenhao Huang

PhD - Université de Montréal

huang.wenhao@mila.quebec

Yizhan Li

PhD - Université de Montréal

yizhan.li@mila.quebec

Yinhan Lu

Master's Research - Université de Montréal

Co-supervisor :

David Ifeoluwa Adelani

yinhan.lu@mila.quebec

Kyle Roth

PhD - Université de Montréal

kyle.roth@mila.quebec

Haochen Shi

PhD - Université de Montréal

haochen.shi@mila.quebec

Xiran Song

PhD - Université de Montréal

xiran.song@mila.quebec

yihong.tang@mail.mcgill.ca

Jia'ao Sun

PhD - Université de Montréal

sunjiaao@mila.quebec

Winston Tang

PhD - McGill University

xiaoqiang.wang@mila.quebec

PhD - Université de Montréal

suyuchen.wang@mila.quebec

Suyuchen Wang

PhD - Université de Montréal

mengyang.xiong@mila.quebec

Dekun Wu

PhD - Université de Montréal

dekun.wu@mila.quebec

Mengyang Xiong

Research Intern - McGill University

Yan Zhang

PhD - Université de Montréal

yan.zhang2@mila.quebec

Armin Zolfagharidariani

Master's Research - Université de Montréal

armin.dariani@mila.quebec

Revolutionizing Materials Science with NLP: Introducing MatSci-NLP and HoneyBee

Blog Posts

MatSci-Instruct and HoneyBee training workflow.

October 21, 2024

Bang Liu

Read the article

Publications

HarnessRL: Co-Evolving Policies with a Just-in-Time Training Harness for Open-Ended RL

Ting-En Lin

Yuchuan Wu

Pinyi Zhang

Caixin Kang

Ling Li

Bingxue Qiu

Jieping Ye

Yongbin Li

Rubric-based reinforcement learning (RL) trains policies for open-ended generation with natural-language criteria as rewards, but these rewa… (see more)rds can become ineffective as training progresses. We identify two failure modes. Under non-discriminative rewards, scores collapse to uniformly low or high values due to insufficient exploration or a mismatch between task difficulty and policy capability. Under spurious rewards, higher scores do not correspond to better responses. Both reflect a training configuration whose task sampling, exploration, and evaluation stay fixed as the policy evolves. We introduce HarnessRL, a training-time harness that co-evolves with the policy through three interfaces for task sampling, rollout guidance, and rubric criteria. HarnessRL maintains a training-time memory from which a harness evolver retrieves within-task history and cross-task evidence. Using this evidence, the evolver attributes each reward failure to the interface whose update can address it. It then runs a proposer and just-in-time critic loop in which the critic accepts a candidate update only after measuring its effect on the current frozen policy, returning rejected candidates for revision. We also curate Rubric-Mix-10k, a five-domain training and evaluation suite for open-ended RL across medical, science, writing, role-play, and instruction-following tasks. On Rubric-Mix-10k, HarnessRL outperforms the strongest baselines by 2.8 to 3.1 points across both 4B and 9B model scales and nearly halves the training steps to its best checkpoint. Code and data are available at https://github.com/Tongyi-ConvAI/Qwen-Character/tree/main/HarnessRL.

2026-10-18

arXiv (preprint)

Input-Conditioned Search for Efficient Test-Time Scaling of Recursive Reasoners

Tianwen Fu

Yajie Zhao

Xingdi Yuan

Hengyang Lu

People often tackle difficult problems by thinking more deeply or exploring more possibilities. Similarly, recursive reasoners, which reason… (see more) by repeatedly applying a learned update rule to latent states, can also scale inference through more updates (depth) or additional trajectories generated by perturbing latent states (width). When people hit a dead end in their thinking, they may step back and approach the problem from a fresh angle. However, neither depth scaling nor perturbation-based width scaling of recursive reasoners can reliably achieve this: both may converge to incorrect attractors. To address this, we propose to explore diverse reasoning traces by applying different update maps to the same latent state, without any additional training. We realize this idea through \emph{Input-Conditioned Search (ICS)}, which searches over input conditions to induce alternative update maps from the pretrained reasoner. Across six reasoning tasks, ICS achieves the highest mean accuracy, while requiring up to two orders of magnitude fewer model FLOPs than the evaluated scaling baselines. Controlled experiments confirm that ICS can redirect trajectories from the same state. Our theoretical analysis and rigorous numerical verification also confirm that ICS can locally eliminate incorrect attracting fixed points. Together, these findings establish ICS as an effective axis for efficient test-time scaling and motivate further research on shaping recursive reasoning during training and at test time.

2026-10-12

arXiv (preprint)

LatentHarness: Learning Latent Actions for Memory and Reasoning via Counterfactual Policy Distillation

Suyuchen Wang

Long-context reasoning faces two complementary bottlenecks: retaining evidence across long inputs and sustaining computation across many rea… (see more)soning steps. Existing approaches largely address them separately, with external memory extending access to distant evidence and latent reasoning compressing multi-step computation. We introduce LatentHarness, which unifies memory access and latent reasoning as sequential latent action selection. At each internal step, the model chooses THINK for further computation, RECALL from a fast-weight memory of input evidence and intermediate reasoning states, or EXIT to emit the next token. We train this policy with counterfactual policy distillation, which branches every action for one step and scores its effect on the emitted token. These gains teach the policy when memory is more useful than further reasoning, while gradients through counterfactual recall teach which intermediate states should be retained in memory for future use. Across six general and long-context reasoning benchmarks, LatentHarness at 1.4B improves on the strongest baselines by 2.8% and 10.0% relative, respectively, and runs 5.9x faster than the strongest long-context baseline.

2026-09-29

arXiv (preprint)

MemCodex: Self-Programming Hierarchical Memory for Language Agents

Agent memory faces heterogeneous access needs: a single-hop question may require one piece of evidence, whereas a multi-hop question must co… (see more)mbine evidence from multiple sources. Predefined memory workflows cannot adapt to these varying needs. Recent adaptive methods search or learn over memory components and their compositions, but the design space itself remains predefined. We introduce MemCodex, a self-evolving hierarchical memory system that organizes experience into executable memory programs for summaries, relational knowledge, reusable skills, and latent memory. Open-ended program evolution searches the open design space of layer programs by rewriting how each layer is constructed, indexed, retrieved, and routed, thereby adapting both within-layer implementations and cross-layer composition. At query time, reads traverse the hierarchy from coarse to fine and stop once sufficient evidence is found, descending to the original history when needed. We further develop MemArena, a unified runtime that places heterogeneous data and memory systems behind a common interface. MemCodex improves average task success by 10.1% relative to the strongest adaptive-memory baseline, while using 3.4x fewer context tokens and achieving 2.1x faster inference.

2026-09-29

arXiv (preprint)

Circuit Hypernetworks for Quantum-Augmented Diffusion Language Models

Language models can be adapted by changing the computations applied to individual tokens. Quantum circuits offer one such approach, but eval… (see more)uating wider circuits inside a large model can be computationally demanding. Here we introduce HyperQ, which adds token-conditioned quantum residual branches to a frozen masked-diffusion language model. A quantum residual branch is a module in each transformer block that reads a token's hidden state, emits the coordinates of that token's circuit, executes it, and adds the measured values back through a residual connection. The backbone remains frozen, and only the added branches are trained. Within each branch, a lightweight circuit hypernetwork emits token-specific rotation angles, coupling strengths, and measurement axes in a shared sparse circuit structure. The required expectation values have an exact classical expression whose evaluation cost grows linearly with the qubit count, enabling circuits from 16 to 64 qubits to be trained within a 1.1-billion-parameter backbone. Across downstream benchmarks, increasing circuit width raises the average score from 47.65 to 54.30. At 64 qubits, HyperQ exceeds the backbone and its low-rank-adapted counterpart by 4.71 and 3.67 points, respectively. HyperQ is fine-tuned on 20,000 prompt-response pairs, compared with 200,000 for the classical baselines. These findings support token-conditioned circuit emission as a tractable architectural approach to quantum-augmented language modelling.

2026-09-20

arXiv (preprint)

SMC1A is required for fate determinations of human spermatogonial stem cells and male fertility by interacting with YBX1 and stabilizing HMGA2 mRNA via an m5C modification

Wei Liu

Li Du

Yanna Zhu

Yuxuan Xu

Dai Zhou

Jun Gao

Yinghong Cui

Wei Chen

Chunyun Li

Yongzhe Chen

Fen Jiang

Zuping He

Spermatogonial stem cells (SSCs) are required for initiating and maintaining normal spermatogenesis, and notably, they have significant appl… (see more)ications in both reproductive and regenerative medicine owing to their great plasticity with de-differentiation and trans-differentiation potentials. Nevertheless, molecular mechanisms regulating human SSC fate determinations and male fertility remain elusive. Here we report for the first time that structural maintenance of chromosomes 1A (SMC1A) interacts with RNA-binding protein YBX1 to control fate decisions of human SSCs and an association exists between SMC1A dysfunction with male infertility. SMC1A/YBX1 complex stabilizes high mobility group AT-hook 2 (HMGA2) mRNA in an m5C modification-dependent manner to activate the G1/S phase transition, which ultimately mediates human SSC self-renewal and apoptosis. Transplantation of SSCs into the seminiferous tubules of recipient mice without male germ cells reveals that Smc1a silencing impairs SSC colonization and causes differentiation defect. Notably, Smc1a-conditional knockout (Smc1a-cKO) mice assume spermatogenesis disorder and obviously compromise male fertility. Single-cell transcriptomics demonstrates that Smc1a-cKO mice exhibit the downregulation of genes associated with mitosis, meiosis, and m5C regulation. Significantly, whole-exome sequencing (WES) of 2028 non-obstructive azoospermia (NOA) patients identify two deleterious SMC1A variants and one deleterious HMGA2 variant. SMC1A and HMGA2 expression levels were remarkably decreased in testicular tissues from NOA patients with spermatogenic failure compared to obstructive azoospermia (OA) patients with normal spermatogenesis. Collectively, our findings highlight the critical roles and a novel mechanism by the SMC1A/YBX1/HMGA2 axis in regulating human SSC fate decisions and male fertility via an m5C modification. This study is thus of the utmost significance because it would provide novel molecular mechanisms underlying fate determinations of adult stem cells and human spermatogenesis and it could offer network targets for gene targeting male infertility and other diseases.

2026-09-14

Cell Death and Differentiation (published)

ReactHuman: A Physics-Grounded Benchmark for Human-Like Reactive Decision-Making in Embodied Multimodal LLMs

Yizhan Li

Jianxin You

Mengyang Xiong

Yinhuan Chen

Zicheng Zhao

Dekun Wu

Dongqing Zhang

Reacting to sudden physical hazards (catching a slipping plate, dodging a falling knife) is both a meaningful test of embodied intelligence … (see more)and a hard requirement for deploying multimodal large language models (MLLMs) as the decision coreof household robots. Existing evaluations, however, probe intuitive physics passively through question answering over videos, or target deliberate, long-horizon tasks such as navigation and rearrangement; none measure whether a model can turn physical understanding into immediate, safety-critical action. We introduce ReactHuman, the first physics-grounded benchmark for human-like reactive decision-making, in which the evaluated MLLM acts as the brain of a simulated humanoid facing sudden household hazards; it spans 17 event families and over 1,000 bit-for-bit reproducible scenes with exact, annotation-free ground truth derived from 240 Hz rigid-body simulation, including adversarial objects whose appearance contradicts their physics (a foam anvil, a steel apple). We further design a five-metric suite that scores each reaction along three axes: reasonable, safe, and physically grounded. We physically execute every committed plan so that decisions have observable consequences. With this harness we evaluate seven representative MLLMs. Results show that reactive safety is far from solved: models mishandle roughly one hazard in three, act from fixed dispositions rather than the observed scene, trust appearance over motion, and miss interception points at meter scale even when the chosen action is correct; none of these failures shrink with model scale. ReactHuman thus offers both a fine-grained diagnosis and a scalable training signal toward physically grounded, safety-aware embodied agents. The benchmark can be found here: https://huggingface.co/datasets/Alan123/reacthuman-benchmark-scaled

2026-09-08

arXiv (preprint)

One Policy Is Enough: Single-Agent Reinforcement Learning Outperforms Tree Search for Chemistry Tool Learning

Armin Dariani

Sifan Wu

Entao Yang

Chemistry questions often demand exact computation and database lookups that a language model cannot supply from its parameters, so it must … (see more)reach for external tools. Tool use here is a three-part problem: select the right tool from a large pool, fill it with correctly typed arguments, and chain calls so that each consumes the outputs of the last. CheMatAgent, a previously published system, addresses this with hierarchical evolutionary MCTS: separate policy and execution models searching tool-call trees under two learned critics, one regressed partly onto GPT-assigned scores. We show that a single policy suffices. Our model interleaves reasoning, tool calls, and returns in one left-to-right generation, trained by a supervised warm-up and then outcome-level reinforcement learning against a programmatic reward read directly off the gold call chain, which leaves no learned critic and no judge in the training loop. On ChemToolBench multiple-tool comprehensive chemistry, on both backbones CheMatAgent use, we improve Tool F1 by 5.5% and Return F1 by 9.6% on Qwen-2.5-7B, and by 3.7% and 3.9% on Llama-3.1-8B, compared with their strongest search configuration, at one model invocation per question, against a search whose cost grows with the tree; we also lead answer Pass Rate on Qwen-2.5-7B.

2026-08-30

arXiv (preprint)

ComBodied Agents: a New Paradigm of Human-Centric Agentic AI

Qianggang Ding

Xingyao Wang

Rui Feng

Zhibin Wang

Feixiang Yao

Kelong Mao

Hao Sun

Zhiyao Luo

Jiankai Tang

Lei Li

Jiadong Guo

Minheng Ni

Weicong Lin

Chenxi Yang

Hongxiang Gao

Zhenghua Chen

Yang Bai

Min Wu

Jun Cheng

Huazhu Fu … (see 2 more)

Dacheng Tao

After an older adult misses a medication dose, a software agent can send another reminder and an embodied agent can bring the medication. Ye… (see more)t neither explains whether the person forgot, is confused, has side effects, or deliberately refused, nor what support is appropriate. This reveals a structural gap in Agentic AI: Digital Agents primarily transform software states, while Embodied Agents transform physical states; neither makes a person's evolving state and agency the primary object of modeling, intervention, and evaluation. We introduce Combodied Agents, a human-centered paradigm that perceives, models, predicts, and supports individual human-state trajectories over time, using software tools, sensors, wearables, robots, and human services as action channels rather than end goals. We unify fragmented capabilities across personal assistants, health agents, AI companions, and adaptive human--AI systems into a closed loop: event-based multimodal perception reconstructs meaningful personal events; longitudinal, correctable memory provides temporal context; Personal World Models estimate future personal states and outcomes under alternative decisions and interventions; and an admissible intervention policy selects proportionate support under consent, uncertainty, safety, reversibility, and user control. Feedback from the person and environment updates the loop. Rather than requiring an exhaustive Human Digital Twin, the framework uses purpose-bounded, uncertainty-aware, user-correctable representations. We organize the design space by human-state targets, relational contexts, and agent roles, and propose scenario-centered evaluation, agency-preservation metrics, benchmark requirements, edge-native personal models, and governance directions. Combodied Agents shift Agentic AI from external task completion toward sustained human benefit.

2026-08-10

arXiv (preprint)

Search-on-Graph-R1: Training Large Language Models to Search Knowledge Graphs with Reinforcement Learning

Jia Ao Sun

Yu Hao

Fengran Mo

Zhan Su

Yuchen Hui

Jian-Yun Nie

Knowledge graph question answering (KGQA) requires navigating from topic entities to an answer several relations away. Recent methods prompt… (see more) a frontier LLM to explore the graph through a retrieval tool, but their reliance on frontier-scale inference makes them costly to deploy. We present Search-on-Graph-R1 (\sogrone{}), which internalizes this navigation into a compact 8B model through supervised fine-tuning (SFT) followed by reinforcement learning (RL). Our central idea is to scaffold a frontier teacher with each question's gold SPARQL query, so the teacher traverses a known answer-bearing path with a live \texttt{Search} tool rather than having to discover the path itself. Since every call executes against a live Freebase server, the resulting trajectories are grounded in the knowledge graph by construction. On WebQSP, CWQ, and GrailQA, \sogrone{} at 8B surpasses every frozen frontier-LLM system in our comparison and posts the strongest results on CWQ of any system we compare against. It does so using no auxiliary module at inference and no LLM judge during training. Isolating each training stage shows that SFT and RL contribute complementary gains, our approach transfers across model families, and RL learns to reach answers in fewer \texttt{Search} calls than its SFT initialization.

2026-07-19

arXiv (preprint)

MobEvolve: An Agentic Self-Evolving Heuristic System for Interpretable Human Mobility Generation

Junlin He

Yihong Tang

Tong Nie

Ao Qu

Yuebing Liang

Hamzeh Alizadeh

Wei Ma

Lijun Sun

Human mobility generation aims to synthesize realistic trip chains for target populations based on individual features. Existing paradigms, … (see more)including deep generative models, LLM-based methods, and traditional heuristics, struggle to satisfy the complex demands of this task while simultaneously maintaining interpretability, behavioral plausibility, population-level distributional alignment, and inference efficiency. To bridge this gap, we introduce MobEvolve, the first agentic self-evolving heuristic framework for human mobility generation. MobEvolve initializes a behavior-inspired heuristic system and employs an LLM agent to iteratively evolve its internal logic. By diagnosing empirical misalignments and failure cases on a validation set, the agent proposes targeted updates and accumulates evolution memory for cumulative self-improvement. Extensive evaluations on the Singapore and Montreal benchmarks demonstrate that MobEvolve significantly outperforms state-of-the-art deep generative and LLM-based methods in individual trajectory fidelity, population-level distribution alignment, and behavioral plausibility, while preserving interpretability and high inference efficiency.

2026-05-31

arXiv (preprint)

A survey of agentic materials science and engineering: where are we and where are we going?

Jiayi Zhu

Longhan Zhang

Yizhang Zhu

Xiaotian Lin

Yifan Wu

Shimin Di

Yuyu Luo

Tongyi Zhang

Agents, primarily built upon large language models (LLMs) and equipped with planning, tool use, memory, and self-reﬂection capabilities, a… (see more)re revolutionizing all aspects of materials science and engineering (MSE), from materials design and experimental execution to industrial manufacturing and deployment, thereby opening the age of agentic MSE. Rather than functioning as isolated artificial intelligence (AI) predictive models, these agents coordinate multi-step scientiﬁc workﬂows by retrieving and structuring knowledge, proposing and reﬁning hypotheses, planning experiments, combining multimodal simulations and characterizations, and, when integrated with AI materials laboratories, closing the loop toward autonomous materials discovery. However, agentic systems exhibit varying degrees of autonomy, and their roles in materials research and development diﬀer accordingly. To systematically examine the landscape of agentic MSE, this survey proposes a six-level autonomy framework (Levels 0-5) that characterizes the progression from human-only workﬂows to fully autonomous scientiﬁc agents. The framework aligns with key task families across the entire MSE pipeline, including information retrieval, property prediction, simulation, synthesis, and characterization. By reviewing recent advances in agentic MSE, we reveal uneven progress across these domains. Knowledge-centric capabilities often remain at an early stage, while experimental orchestration and characterization are starting to explore higher-level agent behaviors. Importantly, achieving mature autonomy requires coordinating multiple tasks rather than optimizing any single task in isolation. Collectively, these insights provide a structured roadmap for advancing agentic MSE toward higher levels of autonomy.

2026-05-25

Journal of Materials Informatics (published)