Evaluating Cognitive Maps and Planning in Large Language Models with CogEval

Recently an influx of studies claim emergent cognitive abilities in large language models (LLMs). Yet, most rely on anecdotes, overlook contamination of training sets, or lack systematic Evaluation involving multiple tasks, control conditions, multiple iterations, and statistical robustness tests. Here we make two major contributions. First, we propose CogEval, a cognitive science-inspired protocol for the systematic evaluation of cognitive capacities in Large Language Models. The CogEval protocol can be followed for the evaluation of various abilities. Second, here we follow CogEval to systematically evaluate cognitive maps and planning ability across eight LLMs (OpenAI GPT-4, GPT-3.5-turbo-175B, davinci-003-175B, Google Bard, Cohere-xlarge-52.4B, Anthropic Claude-1-52B, LLaMA-13B, and Alpaca-7B). We base our task prompts on human experiments, which offer both established construct validity for evaluating planning, and are absent from LLM training sets. We find that, while LLMs show apparent competence in a few planning tasks with simpler structures, systematic evaluation reveals striking failure modes in planning tasks, including hallucinations of invalid trajectories and getting trapped in loops. These findings do not support the idea of emergent out-of-the-box planning ability in LLMs. This could be because LLMs do not understand the latent relational structures underlying planning problems, known as cognitive maps, and fail at unrolling goal-directed trajectories based on the underlying structure. Implications for application and future directions are discussed.

Paper

References (69)

Scroll for more · 38 remaining

Similar papers

Peer review

Reviewer xxkR5/10 · confidence 3/52023-06-20

Summary

This paper describes a test battery to test for the emergence of cognitive maps and planning abilities in LLM. The tests are based on existing cogsci experiments converted into text prompts. For example, to test for a planning ability, the prompt first describes an apartment layout, and then asks the model to plan a path through several rooms to retrieve something. The authors show that LLM are poor at completing such tasks.

Strengths

I think the adaptation of cogsci planning experiments to text prompts is a great idea. There needs to be more tests like this to map the scope of LLM capabilities. The authors evaluate multiple prompts from the same generative model and compute statistics of LLM responses. This approach stands in contrast to many arguments for emergent intelligence in LLM, which derive from anecdotal examples. It is important to do actual experiments with LLM.

Weaknesses

This is a presentation issue, but it affects my ability to understand and evaluate the paper. The writing is unedited. There are a lot of unnecessarily long sentences that can be condensed for pragmatics and readability. The grammar and punctuation are weird, like this might be an initial draft. I had to read the same content multiple times, and missed important details throughout the paper. Figures are odd, seem to be just thrown in together from unrelated bits. For example Figure 1 shows examples of graphs A-E. Are these all the graphs used in Experiment 1? If so, this is not stated. Weirdly, each graph is shown in a different pictorial style. I'm sorry, but this looks like someone just downloaded different types of graphs from google search, panel E has a random red outline around it. Experiment design is not given. I can not check the statistics without the experiment design. What are these degrees of freedom? It looks like multiple regression models were built? I think the intention is a good one, but it is not clear what was done

Questions

1. " We evaluated the robustness of our findings by regenerating the results for each combination of factors and parameters multiple times and applying a statistical model of how each of these factors contribute to variance in LLM performance. " What does this mean? Where is the exact experimental design? How many trials were simulated? Which conditions were evaluated? 2. Where is Experiment design for Experiment 1? Where are the graphs? 3. "198 For the graph community block model, example graphs are shown in Figure 2 ..." What does this mean? Does Figure 2 show all graphs, or a subset of them? If this is a subset, then how many graphs of each kind were shown? Where is the "community block model" described?

Rating

5: Borderline accept: Technically solid paper where reasons to accept outweigh reasons to reject, e.g., limited evaluation. Please use sparingly.

Confidence

3: You are fairly confident in your assessment. It is possible that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked.

Soundness

2 fair

Presentation

1 poor

Contribution

3 good

Limitations

I can not make out what was done exactly, but am happy to read the next draft. The experiments and statistical analysis need to be described so that people could reproduce them.

Reviewer Vmb55/10 · confidence 4/52023-06-26

Summary

This paper evaluates LLMs on a set of tasks that could be solved by cognitive maps, such as goal-oriented planning, or incorporating shortcuts. The work finds that LLMs generally perform poorly at these tasks, and their performance is affected by features such as graph sparsity.

Strengths

* The paper is admirably thorough with experiments and analyses: - Assessing a range of LLMs (including varying parameters such as temperature). - Evaluating across a range of different graph structures, task paradigms, etc. - Creating new stimuli to avoid dataset contamination. - Performing regression analyses to determine how different features contribute to model performance, and describing the failure modes observed. - These thorough results and analyses suggest that the conclusions are likely to roughly generalize to some extent, and help the reader to understand which features will affect performance. * The results are interesting, there are a variety of patterns that could be investigated further.

Weaknesses

My primary concern is that the overall framing of the paper is misleading. In particular, the work is motivated with references to the cognitive and neuroscience literature on cognitive maps. This work performs draw strong conclusions from its experiments such as "no evidence for understanding cognitive maps or planning." Are conclusions such as these justified? A key issue in analyzing AI in comparison to human or animal capabilities is determining where a performance failure originates: is it a lack of an underlying capability (such as the ability to form a cognitive map), or a more superficial performance issue? Several recent papers have emphasized this point from a cognitive science perspective, and argued that it is essential to ensure fair comparisons between AI and natural intelligence to draw accurate conclusions (https://www.pnas.org/doi/abs/10.1073/pnas.1905334117; for LLMs specifically see: https://arxiv.org/abs/2210.15303). In that context, it's worth noting that the animal and human experiments cited involved a great deal more experience before the map was tested than the present experiments do. For example, Tolman's latent learning experiments involved the rats fully exploring the maze for multiple days before they were tested with a food reward at the end (and even then, performance continued to improve well after the rewards were first introduced). Or Schapiro's temporal community structure paper involved half an hour of exposure to transitions from the graph; that is, thousands of transitions from a graph with only 15 nodes. This a much denser sampling of experience than the current LLM experiments afford, and it is quite possible that some degree of repeated experience contributes to the ability of natural intelligences to form a cognitive map. The difference in learning conditions is briefly mentioned to in the limitations, but is quickly dismissed; however, the difference in experimental conditions is a fundamental challenge to concluding that LLMs are failing to form cognitive maps like animals/humans do. Likewise, it is typical in cognitive science to report comparisons to chance-level performance, and an underlying ability is usually inferred from better-than-chance performance, even if that performance is imperfect. For example, in Tolman & colleague's studies, the rats continued to improve for several days after the reward was introduced (that is, their paths were not optimal on the first test), and the rats were clearly stitching together trajectories that they had observed (since there was only a single route through the maze, everything else lead to dead ends), but we still interpret their performance as showing latent learning. It would be useful to report chance-level performance across all conditions, and, in the case that a model performs better than chance across a broad range of conditions, that would suggest some underlying ability, even if it is imperfect. For example, GPT-4's performance seems reasonably high across most conditions in Table 2 (though certainly imperfect). In addition, explicit comparisons to human performance on these tasks (presented exactly as they are presented to the language model), would strengthen the claim that language models are failing in a fundamental way that humans or animals would not. These points seem critical to the overall framing of the paper, and also to much of the discussion. I therefore think the paper would be substantially improved by: 1) providing a more nuanced framing of the very interesting results, that suggests they emphasize some limitations of planning in LLMs, without making overly strong claims such as "no emergent planning" or "no evidence of cognitive maps" 2) Providing explicit comparisons to chance-level (and ideally human) performance to help contextualize the results.

Questions

* In the discussion, what explicit experiment is referred to by "We observe that LLMs do better in problems where the entire trajectories are explicitly available in the text prompts, and they only need to piece together partial changes"?

Rating

5: Borderline accept: Technically solid paper where reasons to accept outweigh reasons to reject, e.g., limited evaluation. Please use sparingly.

Confidence

4: You are confident in your assessment, but not absolutely certain. It is unlikely, but not impossible, that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work.

Soundness

2 fair

Presentation

1 poor

Contribution

3 good

Limitations

See weaknesses; I believe that the limitations of the experimental paradigm not matching the inspiration are not fully discussed, and that more generally the conclusions are not fully supported by the experiments presented.

Reviewer ydCJ7/10 · confidence 4/52023-06-29

Summary

The paper presents CogEval, a set of best practices ported from cognitive science on how to do behavioral evaluations. The authors also transcribe new tasks from human reinforcement learning and planning into text, such that LLMs can be tested on them. On these tasks, the authors do not observe evidence for an emergent capability for planning in large language models.

Strengths

- The presentation of CogEval is clear and potentially useful for the ML community. I’m excited to use CogEval in my own work. - Thorough and thoughtful discussion. - Exhaustive experimental setup—a feature perhaps enabled by the principled CogEval framework! - The paper is lucid overall and easy to read.

Weaknesses

There’s some mild overselling in the paper, given the empirical results, which only demonstrate the failure of an existence proof. Defining the conditions under which we can declare definitively that there is no emergent planning. However, given that we have not rigorously defined these conditions, for example, “No Emergent Planning” in the title seems too strong.

Questions

- Are you planning to release these tasks transcribed to text as a new testbed? - Does transcribing these tasks to text make them harder? Perhaps a vision-and-language model like GPT-4 or Flamingo might perform better on pictures of the task, in which case, perhaps your conclusions would be different. - Why is “Measurement and Evaluation for Large Language Models” capitalized in line 6? Nits: - Line 14: Stay in active voice here for consistency? - Figure 3: Can you make this less blurry?

Rating

7: Accept: Technically solid paper, with high impact on at least one sub-area, or moderate-to-high impact on more than one areas, with good-to-excellent evaluation, resources, reproducibility, and no unaddressed ethical considerations.

Confidence

4: You are confident in your assessment, but not absolutely certain. It is unlikely, but not impossible, that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work.

Soundness

3 good

Presentation

4 excellent

Contribution

4 excellent

Limitations

The authors have adequately addressed limitations.

Reviewer UmDW7/10 · confidence 3/52023-07-07

Summary

This paper proposes an evaluation of large language models with respect to their ability to solve problems that require use of latent cognitive maps. Evaluation focuses on different underlying graph structures, and the influence of chain-of-thought inference on performance.

Strengths

* I really liked the very clear outline of the motivation of the research design in the introduction of Section 2. * The experiments are extremely thorough, and the goal of designing experiments with statistical robustness in mind is good. * Good discussion around the capabilities and limitations of LLMs

Weaknesses

Some of the presentation could be refined: * The description of Figure 1 is somewhat difficult to understand. References to future aspects of the paper (e.g., "Experiment 3") are undefined, which makes it more difficult to understand. * The figures / tables should appear closer to where they are referenced in the text. * I'd suggest reordering 2.1, so that the tasks (i.e., maze learning) are described before the experimental setup. * Some details could use more context. e.g., what is temperature? What are the graph structures A/B/C... etc? * The discussion on BFS/DFS prompting should be moved to the experimental setup section (2.3). I also didn't quite understand the distinction between these two; an example would help. * Formatting of Figure 3 can be improved * What is "dialogue" referring to throughout the paper? I don't believe an actual multi-turn dialogue is taking place during the evaluation * Text of Table 2 is really tiny

Questions

* I am not completely familiar with the term "cognitive maps", and I'm not sure if this more of a metaphor, or usually applied to actual spatial reasoning tasks (hence "map"). Since the examples in the paper are about navigation, I am wondering if this means the domains studied are mostly about literal maps, or if there are other domains usually studied with the framework of cognitive maps. Are there other domains covered in this study? If so, what are they? * If there are different domains, how does performance vary across them?

Rating

7: Accept: Technically solid paper, with high impact on at least one sub-area, or moderate-to-high impact on more than one areas, with good-to-excellent evaluation, resources, reproducibility, and no unaddressed ethical considerations.

Confidence

3: You are fairly confident in your assessment. It is possible that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked.

Soundness

3 good

Presentation

2 fair

Contribution

3 good

Limitations

Yes

Reviewer Vmb52023-08-11

Thanks for the improvements, and some follow up thoughts

Thanks to the authors for their thoughtful response. I've updated my score accordingly. Some follow-up thoughts below: 1. I believe this reframing will improve the paper. 2. In the context of the above reframing, this issue will likely be improved. However, I have some lingering concerns about the comparison, that depend on precisely how the paper is reframed. For example, if the authors still reference rodent + human studies in motivating the cognitive maps (which I think is a good thing!), then it would still be useful to highlight the distinctions between the experimental methods in that work (e.g. substantial experience, learning as the authors point out) and the zero-shot methods in this work. I understand that the goal is not to make direct comparisons to human behavior, but I think it will be hard to write a paper talking about cognitive maps and citing the prior literature without (implicitly, at least) drawing that comparison. Thus, I hope the authors will highlight these discrepancies in the paper. 3. I agree that chance level performance can be tricky to determine. Nevertheless, I think that including some such comparisons would help to situate the results in context. - From my perspective, any chance metric that takes account of the graph structure (such as random walks on the graph) is not really the most appropriate chance-level comparison for these experiments, because it effectively presumes some kind of cognitive map (that is, if the model sampled from these distributions, it would thereby be fully respecting the constraints imposed by the graph structure). In a fully no-cognitive-map baseline I would expect chance to be sampling paths truly at random from the set of possible nodes, without respect to any spatial constraints (perhaps sampling without replacement, so that the set of paths is finite). - Alternatively, the authors might suggest that recognizing pairwise constraints would be possible just from the experienced paths, and so an additional chance level baseline would be sampling from transitions observed in the prompt. - I think comparing the models' ability to respect local dependecies to the above two chance level baselines would help to elucidate the extent to which the issues the model is facing stem from lack of *any* cognitive map (i.e. not respecting local dependency constraints, or only respecting observed ones) vs. inability to plan over longer distances. I think that such a comparison would help to quantitatively clarify some of the qualitative claims in the current discussion. - I do think it would be reasonable to also compare to a random walk over the graph with no repeats baseline; that would be more of a "cognitive-map-but-no-planning" comparison, which would also be useful (but I'd see the above as more valuable). Questions. Thanks, this clarifies things.

Authorsrebuttal2023-08-19

Thank you and following up

1- We appreciate it. 2- We agree with the reviewer that the difference between human experiments and our prompts for LLMs is important. We already noted this in the submitted manuscript under limitations, lines 298-304: “in the human experiments that influenced our prompts, participants learn gradually, experiencing states one-by-one, but were only tested after they showed signs of learning, similar to a model-based RL agent having the transition structure and using it for inference and planning. To address this difference, we present the environment’s structure in linguistic format. In all cases, the participant or model had to identify the goal location based on instructions and infer the policy towards the goal, which is the room with the maximum reward.” Second, on lines 318-319 we explicitly said that we use a functionalist notion of cognitive maps and planning and not a “human-like” notion: “we evaluated emergent cognitive capacities in LLMs in a functionalist and multiple-realizablity sense rather than requiring any assumptions of them being "human-like" [31]” Even Tollman’s original 1948 paper imbues cognitive maps with far more abstract & general intentions. Behrens et al. 2018's “What is a cognitive map?” defines cognitive maps as relational structures of knowledge, & Epstein et al. 2017 as “a unified representation of the environment to support memory and guide future action”. Consistently, we functionally test whether LLM behavior in planning tasks is consistent with having a unified representation of environment that can be accurately recalled for planning. We described the structure of the env & asked questions to test whether LLMs can *extract a unified representation (cognitive map) and accurately recall it for flexible planning* & found that while some LLMs can list state-state transitions, when it comes to using this knowledge for planning they *hallucinate edges that don’t exist* (GPT-4: 25.57% responses on Schapiro graph were wrong due to hallucinations), *fall into loops*, or say irrelevant things. In a functionalist spirit we a) report this performance & classify failure modes, b) suggest that these failures are inconsistent with LLMs accurately using a unified representation of the environment (cognitive map) for planning. We agree with the reviewer’s broad comments & we've touched on them in the MS but can further clarify. 3- We appreciate the reviewer noting sampling from observed transitions in the prompt to check if they’re from the set. As shown in Supplementary Figure 2, now in the main MS, we visualize 3 failure modes of GPT-4. Among them hallucinating edges that don’t exist speaks to this. Namely, for Schapiro graph 25.57% of GPT-4 responses were incorrect because of hallucination. This is despite being able to list the tuples when asked as opposed to planning tasks. This speaks to a failure of “a unified representation of the environment for memory & planning”, or cognitive map, revealing planning failure due to inaccurate recall of transition structures. The reviewer’s point is thus relevant & helpful: “not respecting local dependency constraints” and inspired us to compute the %hallucination. If the reviewer agrees, we will add this to the paper. We kindly note that many studies in the cognitive map literature do not use a chance level but a task condition as baseline (Garvert et al 2017; Momennejad et al 2017). Even Tolman & many rodent experiments don’t focus on chance levels but a condition as baseline (e.g., latent learning contrasts having vs. not having explored the environment before rewards are introduced). Similar to Tolman’s latent learning, a given LLM or condition can serve as a baseline in statistical analyses to Compare: A) tasks with on (traversal) as baseline B) LLMs (odds of success are 6.1464X higher for GPT-4 > alpaca-7b) C) robustness to domain (baseline: spatial) D) robustness to local changes (reward reval, detour, etc) & temperatures Such analyses revealed how different factors (model engine, graph type, domain, task) & their interactions affect the odds of success for LLMs on each task. We agree with the reviewer that random walks do not offer a satisfactory chance level. We believe that given the open multistep responses our design is not comparable to classification or experiments that require chance. In our opinion, artificially imposing nontrivial chance metrics might unfairly bias the interpretation to make it seem like LLMs do better. Moreover, LLMs are explicitly trained on next word prediction so randomly sampling any tuples does not seem like an appropriate baseline. Thus, we respectfully do not believe that random sampling is a fair comparison. We hope our responses, rooted in over a decade of experience with cognitive maps, have addressed the reviewer’s concerns & are happy to clarify further. We hope our difference in optimism about LLM abilities does not impede the reviewer from raising the score!

Reviewer Vmb52023-08-20

Thanks, I would still prefer some more interpretable comparisons

Thanks for continuing to follow up. I think this most recent response has helped me to achieve a bit more clarity out of what I believe would improve the paper. I think fundamentally the issue is that performance on these tasks is hard to interpret without comparison to some baseline that is a "known quantity" that the reader will understand, and that supports the argument the authors are trying to make. That is, just as we measure distance with respect to some baseline unit, it is useful to contextualize performance on a task with respect to some well-understood comparison. While chance level is one such comparison (and RL research in AI often includes random baselines, even for multistep tasks), there are other approaches such as comparing to another well-known system or model. I will point to several approaches to this below. 1) The Tolman studies indeed use no-exposure as a baseline condition. This is a reasonably interpretable no-prior-experience baseline; and we interpret the rats as learning because they outperform it given prior experience. That is, this comparison supports Tolman's argument that latent learning occurs. I would be satisfied with such a comparison in this work; it would effectively fulfill the goal of what the chance level baseline was supposed to do, which is to provide a comparison of whether some sort of meaningful learning occurs. However, to my understanding the authors did not perform a no-exposure comparison (please correct me if I have misunderstood however). 2) many more recent computational studies make explicit comparisons to well-understood models. For example, if I recall correctly Momennejad et al 2017 compared human performance to a model-free baseline, a model-based planner, and SR. By making these comparisons to explicit computational models (in addition to the task condition comparisons), the authors more clearly highlighted what features humans seemed to use to solve their tasks. Again, these models provide benchmarks the reader can understand when assessing the performance of the system in question. Likewise, I think such baselines would be helpful for the corresponding tasks in this work; and more generally, comparing to well-understood models would help to provide a clearer benchmark for model performance. 3) alternatively, as I suggested originally, explicitly comparing to human performance can provide such a "known-quantity" baseline in some sense; while we don't understand human cognition in all its detail, it at least provides a familiar yardstick by which a system can be measured, and is a common standard in ML. By contrast, comparing to another LM or even another task condition is harder to interpret; there are many factors that could affect performance in each condition, so it's not entirely clear what the comparison means, other than to show that one model is better or one graph is harder. This issue is exacerbated because the comparisons are not clearly presented, at least in the original draft of the paper. For example, if I were interested in transition revaluation in Table 2, which other condition would be relevant to compare it to in order to isolate this ability? And even if this were clarified, the core point remains: these comparisons don't seem to communicate how the reader should understand the performance of the system overall. The authors note that "artificially imposing nontrivial chance metrics might unfairly bias the interpretation to make it seem like LLMs do better," and while I can understand this concern, I'm not sure that I agree it would bias the interpretation any more than leaving it out does. As highlighted in our discussion above, there are many versions of randomly acting baselines that could be used to compare different types of random behavior. What, precisely, is it that the authors think the models are doing that is better than these but doesn't count as some kind of real-but-imperfect planning? The authors could of course contextualize such results through many of the methods discussed above, such as comparisons to a well understood baseline model (such as a model-based planner, which presumably would be much better in many conditions), condition comparisons (e.g. "models only perform above random transition sampling on paths that have property X" or "as graphs get more complex, models get closer to chance"), or other metrics like quantifying % hallucinations (which I agree would be useful). In summary, I feel the current results are a bit hard to interpret without comparison to systems that can be more readily understood. The authors are, of course, also welcome to ignore these comments; I will re-emphasize that I find the experiments useful, and I have already increased my score to an accept (if a borderline one) in response to the reframing and improvements in the original response. Given that several other reviewers have given higher scores the paper may well be accepted regardless.

Authorsrebuttal2023-08-21

Thank you and final comment

We sincerely thank the reviewer for their comments and engagement. We would kindly note the following as our final response: Re the following comments: "I think fundamentally the issue is that performance on these tasks is hard to interpret without comparison to some baseline that is a "known quantity" that the reader will understand, and that supports the argument the authors are trying to make. " " I feel the current results are a bit hard to interpret without comparison to systems that can be more readily understood." While we agree that comparison to other models is relevant (see 2 below), since model-based or SR agents would not hallucinate transitions that don't exist 25% of the times, we respectfully disagree that the results are hard to interpret. It seems clear that LLMs are failing at accurate recall and use of experienced edges for planning, and failing at having a unified representation of the environment that can be unrolled for planning paths that have not been directly spelled out. This is consistent with previous observations reported on LLM failures (e.g., see LLMs still can't plan, or work by Melanie Mitchell, infamous cases of failures due to hallucination, or reported failures in causal inference). Detailed responses below. 1- Traversal and value-based tasks are here used as the baseline task to compare, when something in the environment hasn't changed. If the reviewer deems necessary, we can add a condition where there's a second prompt rather than just the first prompt to test a "no change" condition, if what you're asking is to be comparable to the 2-prompt nature of conditions with local changes. Here's an example of what using a task or LLM as baseline would look like. Using the smallest LLM, i.e., replicate-alpaca-7b, as the baseline, the odds of gpt-4-32k success are approximately or 6.1464 times higher than the baseline, holding other factors constant. The same can be done for tasks, e.g., using traversal as baseline. 2- We agree about the relevance of comparison with other models, which is the reason we cite and mention this directly in the paper. Moreover, please note that a model-based or successor representation behavioral policy (compared in Momennejad et al. 2017) would never yield a 25% hallucination using tuples that were never experienced. So, we respectfully disagree that the results are hard to interpret. However, we understand that the results are not aligned with the reviewer's optimism about LLMs and respect the reviewer's different opinion. That said, we agree with the reviewer that comparison to other models, e.g., RL or dynamic programming, is helpful. We have already mentioned them in the paper, but we can certainly highlight it more or run them for the specific prompts. We thank the reviewer for the opportunity to spell this out. As noted earlier, we can compare the different LLMs too: using the smallest LLM, i.e., replicate-alpaca-7b, as the baseline, the odds of gpt-4-32k success are approximately or 6.1464 times higher than the baseline, holding other factors constant. 3- We thank the reviewer for this idea. We believe that direct comparison to human behavior is another worthwhile project that is outside the scope of this one paper (which we can barely fit in the 9-page limit of neurips). Overall, we are very grateful that the reviewer engaged deeply with our work. We can certainly clarify the baselines and statistical analysis based on the baseline, & comparison to other models. We agree that this is a novel approach, and it might be hard to classify it in a category of previous work comparing models to human behavior. We believe that as a part of a larger research program, this work can stand on its own - and then some, given the wide comparison with multiple LLMs. We believe reporting the present results is the responsible scientific duty, while future directions are both scientifically interesting and worthy. We thank the reviewer again for their kind engagement, which has allowed us to deepen our clarification of the work. We're happy to lengthen the sections on comparison to MBRL or SR agents and future directions. Thanks again!

Reviewer ydCJ2023-08-14

Thanks to the authors for acting on my few suggestions. The authors more accurately represent their results.

Program Chairsdecision2023-09-21

Decision

Accept (poster)

© 2026 NYSGPT2525 LLC