- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 1
- Soundness
- 1
Read review
Paper summary
This paper proposes SYNC, a training-free framework for communication between heterogeneous language models. The sender selects symbols from a shared inventory, and the receiver writes its own word directions into its hidden states. The authors evaluate the method across different sender—receiver pairs and write depths, and use a “probe” to predict usable write depths.
Critical strengths
The paper conducts a wide range of experiments to evaluate the proposed method, varying senders, receivers, and write depths. The experiments also include several control conditions that help characterize the behavior of the proposed method.
Critical weaknesses
-
The paper does not clearly position its contribution. As the authors acknowledge that internal writing shows no consistent advantage for facts with equal weights, this paper seems more like a proof of concept, which should be made explicit in the introduction. While the authors say that “advantage is with graded weights”, it is unclear under in what practical scenarios such weights would be useful.
-
The method relies on a pre-defined list of symbols shared by senders and receivers, which is difficult to maintain in real-world applications. More importantly, real-world communication involves much more than simple object—attribute relationships (also see question 3), making the applicability of the proposed method rather limited.
-
The presentation lacks clarity and should be improved. Several important experimental settings and methodological details are insufficiently explained in the main text or deferred to the appendix. For example, the “binding probe” in Section 4.3 (Appendix B), despite being presented as a core contribution. The literal prompts are also important, but they are provided only in Appendix H.4, with no direct reference in the main text. While the paragraph in line 305 titles “Why does late writing fail”, the content is largely descriptive, and all explanations are in Appendix G. Some terminology is also confusing, e.g., the “probe” in Section 4.3 differs from its common usage in interpretability research.
-
Another minor problem: in introduction, the second point of contribution has two “§”.
Questions for authors
-
In practical applications, is it realistic to assume that senders and receivers share a predefined symbol inventory?
-
The sender only extracts symbols from private descriptions and does not transmit any model-specific representations. If my understanding is correct, this step could be replaced by a simple rule-based method, as all we need is to transmit symbol indices, especially for the object—attribute tasks considered in this paper. If so, what is the specific contribution of using an LLM sender to cross-model communication?
-
Real-life cross-model communication can involve much more complex relationships than object—attribute relationship. For example, the same symbols "green" and "sweet" could represent either "the green candy is sweet" or "the green candy is not sweet." How would the proposed method handle such cases?
-
How was the 221-symbol inventory (Table 16) constructed, and what criteria were used to select these words? In particular, Table 16 includes not only nouns, but also words such as “not”, “if”, “unless”. This raises two questions: (1) Why do you call these words as "symbols"? Is it consistent with the common usage of the term "symbol" in the literature? (2) If the sender assigns “green,” “sweet,” and “not” to the same object, how can the receiver distinguish between “the green candy is not sweet” and “the candy is not green but is sweet”? This question also related to Q3.
-
If the goal is to achieve good four-answer success, why not simply use a fixed early write depth? If the goal is to select the optimal write depth, why does the probe only predict whether four-answer success exceeds 50%, rather than ranking candidate depths? More importantly, the “binding probe” in Section 4.3 appears structurally similar to the four-answer success evaluation, as both test whether the receiver could associate the object with an attribute, so the high correlation between them is not surprising. To what extent does this reflect the probe's predictive power?
8 comments
Reply to Reviewer 9qmD (2/3)
W2. "[...] real-world communication involves much more than simple object—attribute relationships [...]"
Beyond object attributes, the same internal write carries directed relations between objects and comparative words, where 6 of 6 and 5 of 6 receivers meet our prespecified success rule (Appendices P.2 and P.3). Thank you for this; it is why we added new kinds of fact rather than more scenes of the same kind. These two tasks also preserve slot order for each object's symbols, extending the submitted format.
W3 and W4. "The presentation lacks clarity and should be improved."
Thank you for these; every point you listed under them produced a change in the manuscript.
The revision updates the main text on each point that you raised. (1) Binding probe, Section 4.3: "We call this test a probe: it fits nothing, and it differs from the trained classifiers that the term often denotes in interpretability." (2) Literal prompts: Section 2.1 now cites them directly as "(literal prompts: Appendix H.4)".
(3) Why late writing fails, Section 4.2: "A write at layer ℓ can change those keys and values only above ℓ, so the layers below stay neutral. If later tokens read the facts below ℓ, the write cannot reach them." Appendix G presents the diagnostic that tests this mechanism. (4) Notation: "§§" is now written out as "Sections 4.2 to 4.3". Finally, Table 3 (Section 4.4) now reports relational facts, comparative words and negation scope across six receivers.
Q1. "In practical applications, is it realistic to assume that senders and receivers share a predefined symbol inventory?"
Thank you for asking whether a shared inventory is “realistic”; that is the assumption we most wanted to test.
Sharing the paper's 221-word inventory is not required. In our open-vocabulary experiment, none of the 288 words appear in the paper's inventories, and each receiver fits a direction the first time that it encounters a word. Any word on a public word list is sent as its index, while any other word is sent by its spelling. With complete facts, the internal write reaches 79.9 to 100.0% four-answer success, significantly exceeding a control that writes directions of unrelated words on 6 of 6 receivers (Appendix O).
Q2. "[...] what is the specific contribution of using an LLM sender to cross-model communication?"
On the submitted bank a rule can produce the same indices, because the descriptions state the symbols literally. Thank you for this question about the sender specifically. The language-model senders show that models from other families can generate usable messages without tuning. When a Gemma-4 sender receives no word list and chooses its own words, the internal write reaches 83.3 to 98.3% (Appendix O). When each scene's evidence is split between two senders from different families, internal delivery reaches 70.1 to 94.4% across thirteen receivers (Appendix P.5). That is by design: the sender is replaceable, and the receiver requires nothing model-specific from it.
Q3 and Q4(2). "[...] how can the receiver distinguish between 'the green candy is not sweet' and 'the candy is not green but is sweet'?"
Thank you for the worked example; we answer with your own pair of sentences.
A receiver can distinguish "green, not sweet" from "not green, sweet". We write the negated word with a negative weight in its own slot, so the sign of the weight marks the negation scope. Those two meanings produce distinct messages—green (+), sweet (−) versus green (−), sweet (+)—and an audit finds no two meanings that share a message among 888 signed messages (Table R29).
All six receivers then meet our prespecified success rule, reaching 66.0 to 99.3% four-answer success, compared with 0.2 to 3.9% when directions of other words are written at the same edit size. Moving the negative sign to the object's other word flips the receiver's answers: on the questions whose answers change, the internal write leads this control by 99.3 to 100.0 points.
Plain negation meets the rule on 4 of 6 receivers, and temporal order meets it on 4 of 6 (Appendix P.4 reports every receiver and format). We added the scope, negation and temporal-order tasks in response to the reviews. Because the submitted format sorts each object's indices and therefore cannot mark scope, these tasks extend the format; the results hold for the extended format only, using messages generated by rule. When a Gemma-4 sender generates the negation messages from the scene descriptions instead, 4 of 6 receivers meet our prespecified success rule, reaching 73.6 to 81.9% on those four (Appendix P.16).
In an additional test, a Qwen3.6-27B sender generates the scope messages from the scene descriptions, and 5 of 6 receivers meet the same prespecified criteria, reaching 56.2 to 98.6% on those five (Appendix P.20).
Reply to Reviewer 9qmD (2/3) — tables
The result tables for the answer above.
Table R6 — negation, scope and temporal order.
| Receiver | Internal | Unsigned text | No message | Reassign. | Lead | Low | Confirms |
|---|---|---|---|---|---|---|---|
| Negation: 4 of 6 receivers confirm | |||||||
| Qwen3-32B | 62.5 | 63.2 | 0.0 | 0.9 | 98.6 | 40.3 | no |
| Qwen3.5-27B | 83.3 | 18.8 | 0.0 | 1.6 | 100.0 | 68.1 | yes |
| Qwen3.6-27B | 26.4 | 14.6 | 0.0 | 2.3 | 100.0 | 15.3 | no |
| Mistral-Small-3.2-24B | 68.1 | 4.2 | 0.0 | 1.6 | 82.6 | 50.0 | yes |
| Qwen3.5-9B | 72.9 | 22.9 | 0.0 | 1.2 | 97.9 | 63.9 | yes |
| Granite-4.1-8B | 84.0 | 32.6 | 0.0 | 0.0 | 95.8 | 81.9 | yes |
| Negation scope: 6 of 6 receivers confirm | |||||||
| Qwen3-32B | 78.5 | 20.1 | 0.0 | 3.2 | 99.3 | 72.2 | yes |
| Qwen3.5-27B | 88.2 | 0.0 | 0.0 | 3.9 | 100.0 | 86.1 | yes |
| Qwen3.6-27B | 66.0 | 9.7 | 0.0 | 3.7 | 100.0 | 62.5 | yes |
| Mistral-Small-3.2-24B | 80.6 | 2.1 | 0.0 | 3.7 | 100.0 | 79.2 | yes |
| Qwen3.5-9B | 89.6 | 0.0 | 0.0 | 3.7 | 100.0 | 87.5 | yes |
| Granite-4.1-8B | 99.3 | 0.7 | 0.0 | 0.2 | 99.3 | 98.6 | yes |
| Receiver | Internal | Text | No message | Reassign. | Lead | Low | Confirms |
| Temporal order: 4 of 6 receivers confirm | |||||||
| Qwen3-32B | 83.3 | 86.8 | 0.0 | 0.0 | 72.9 | 80.6 | yes |
| Qwen3.5-27B | 77.8 | 91.0 | 0.0 | 0.7 | 59.7 | 76.4 | yes |
| Qwen3.6-27B | 87.5 | 99.3 | 0.0 | 0.0 | 63.9 | 84.7 | yes |
| Mistral-Small-3.2-24B | 8.3 | 93.8 | 0.0 | 0.0 | -0.7 | 4.2 | no |
| Qwen3.5-9B | 26.4 | 43.8 | 0.0 | 0.9 | 11.1 | 26.4 | no |
| Granite-4.1-8B | 74.3 | 66.0 | 0.0 | 0.0 | 63.2 | 69.4 | yes |
Negation, scope and temporal order. Four-answer success (%) on 144 scenes per receiver; Reassign. is the mean of three reassignments, Lead is the internal write minus its structural control on the rate at which both changed questions are correct (points), and Low is the weakest registered subclass. Unsigned text prints the word list with "not" as a word and is not a matched comparison (text arms on new banks: the negation and scope appendix). The appendix keeps all of its columns.
Table R7 — two senders, thirteen receivers.
| Receiver | Four-answer success Internal | Four-answer success Text | Four-answer success Reassign. | Four-answer success No message | Internal minus Reassign. | Internal minus Text |
|---|---|---|---|---|---|---|
| Six receivers of the open-vocabulary experiment | ||||||
| Qwen3-32B | 91.7 | 89.6 | 1.4 | 0.0 | +90.3 [+81.2, +97.9] | +2.1 [0.0, +4.9] |
| Qwen3.5-27B | 94.4 | 94.4 | 1.4 | 0.0 | +93.1 [+86.1, +98.6] | 0.0 [0.0, 0.0] |
| Qwen3.6-27B | 93.8 | 95.1 | 2.8 | 0.0 | +91.0 [+83.3, +97.9] | -1.4 [-3.5, 0.0] |
| Mistral-Small-3.2-24B | 93.1 | 93.1 | 2.1 | 0.0 | +91.0 [+85.4, +95.8] | 0.0 [-3.5, +3.5] |
| Qwen3.5-9B | 89.6 | 91.0 | 3.5 | 0.0 | +86.1 [+75.0, +94.5] | -1.4 [-5.6, +2.1] |
| Granite-4.1-8B | 88.9 | 90.3 | 0.0 | 0.0 | +88.9 [+79.2, +97.2] | -1.4 [-3.5, 0.0] |
| Seven further receivers | ||||||
| Gemma-3-27B-IT | 93.1 | 93.8 | 0.0 | 0.0 | +93.1 [+85.4, +100.0] | -0.7 [-2.1, 0.0] |
| GLM-4-9B | 74.3 | 81.9 | 0.0 | 0.0 | +74.3 [+62.5, +85.4] | -7.6 [-15.3, +1.4] |
| GPT-OSS-20B | 79.9 | 93.8 | 0.7 | 0.0 | +79.2 [+68.7, +88.9] | -13.9 [-23.6, -5.6] |
| OLMo-3-7B-Instruct | 70.1 | 77.1 | 1.4 | 0.0 | +68.8 [+54.9, +81.9] | -6.9 [-15.3, +1.4] |
| Qwen3-14B-Base | 93.1 | 93.1 | 2.1 | 0.0 | +91.0 [+84.0, +96.5] | 0.0 [-2.1, +2.1] |
| Qwen3.5-2B | 80.6 | 83.3 | 2.8 | 0.0 | +77.8 [+71.5, +83.3] | -2.8 [-7.6, +2.1] |
| Qwen3.5-4B | 92.4 | 93.8 | 2.8 | 0.0 | +89.6 [+84.0, +94.4] | -1.4 [-5.6, +2.1] |
Two senders, thirteen receivers. Four-answer success (%) on 144 scenes, and differences in points with 95% intervals. Reassignment is one reassignment of unrelated words at the same edit size. The appendix gives it with every column.
Reply to Reviewer 9qmD (1/3)
Thank you for noting the wide range of experiments and the control conditions that characterize the method.
Our contribution is a training-free channel between unrelated models. Thank you for this; we state the contribution in one sentence before anything else. A receiver acts on facts written as its own word directions, selected via indices from a sender of another family. The depth of the write determines whether the receiver can use those facts, and a probe that fits nothing predicts which depths are usable. The same channel also conveys how strongly a fact is asserted. Before the contribution list, the introduction now states the kind of claim that the paper makes: "SYNC is a training-free interface that delivers facts through hidden states between models of different families, with messages of a few dozen bytes. This paper establishes that capability under controlled conditions, where success without a message is zero by construction." We have taken the scope question into the paper rather than into a definition: the sentence quoted above is now in Section 1, before the contribution list, and it states the conditions under which the capability is established.
On graded weights, the paper establishes a capability: a receiver can act on assertion strength transmitted through the channel. Internal delivery scores 36.1 and 33.8 points higher than printed weights on the two receivers in Table 10, and scores higher on seven of ten additional receivers. The use that we measure is conflict resolution by the receiver itself: when two competing values for one attribute arrive with weights of 255 and of 64 or 128, the receiver answers with the more strongly weighted one at 77.8 to 97.2% across the six receivers in Table R15. We also test signed weights: a fixed rule maps a sender's "not" to a negative weight, which represents a sign rather than a graded level (Appendix P.16). In our experiments these weights are supplied externally, just as a deployment would supply a retrieval score or per-source reliability.
Our experiments test the receiver's ability to act on them.
Across those ten plus the two in Table 10, the ten receivers where the comparison can be read, internal delivery outperforms printed weights on nine. On an additional set of 108 scenes, it outperforms printed weights on five of six receivers, and on three of the four receivers added to that set afterwards (Appendix P.13).
In the internal route the sender emits only a few symbol indices from a declared inventory, with their object assignments and an 8-bit weight each, and the text route spells the same symbols out. Only the sender reads the private descriptions, and it never sees the question or its answer, so the interface fits settings where that source text cannot be handed on, as in privacy-sensitive deployments.
Why not preprocess the text instead. Preprocessing moves the decision to the sender: it keeps each object's and attribute's highest-weighted value and hands the receiver the answer, reaching 100.0% on all six receivers of Table R15 by resolving the conflict before it sends. The arrangement we report is the other one, the two paragraphs above: the strengths travel and the receiver resolves them itself. Delivery is not the cost either: 0.99 to 1.02 times text once a receiver saves its neutral states (Section 6). The scenario is a receiver that has to act on strengths the sender has not already collapsed into an answer.
Table R12 — receiver-side latency with saved neutral states.
Receiver-side latency with saved neutral states. Warm median time (ms) per item and its ratio to the text time. Published is the ratio of the early-write and text medians in the appendix table of receiver-side delivery cost, given for the four development receivers.