Pith. sign in

REVIEW 3 major objections 5 minor 25 references

For AI-native 6G networks, semantic reasoning reliability scales non-uniformly: deterministic accuracy jumps sharply between 1B and 1.5B parameters and then plateaus, making 1.5–3B models the best edge-deployment balance.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review

2026-08-02 19:25 UTC pith:ZF5UDQDI

load-bearing objection Useful raw pass@1 measurements, but the stability threshold and edge-efficiency conclusions are confounded by model-family comparisons and metric inconsistencies. the 3 major comments →

arxiv 2603.02156 v2 pith:ZF5UDQDI submitted 2026-03-02 cs.NI cs.AI

How Small Can 6G Reason? Scaling Tiny-to-Small Language Models for AI-Native Networks

Mohamed Amine Ferrag , Abderrahmane Lakas , Merouane Debbah This is my paper
classification cs.NI cs.AI
keywords 6Gsmall language modelssemantic reasoningscaling lawsedge deploymentinstability gapAI-native networksreasoning stability
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tries to establish that the right size for an AI reasoning layer in a 6G network is not the largest model that fits, but the smallest model that reasons stably. Evaluating ten instruction-tuned language models from 135 million to 7 billion parameters on 6G-Bench's 30 decision-making tasks, it finds deterministic accuracy rises from 0.224 to 0.707, but the gains are concentrated between 1B and 1.5B, where accuracy jumps from 0.373 to 0.531 and the instability gap collapses. Beyond 3B, improvements shrink to +0.064. Using an Edge Score that divides accuracy by latency and memory, the paper argues that mid-scale models around 1.5–3B give the best reliability per unit edge resource, while tiny models serve only low-risk localized decisions.

Core claim

On the paper's own terms: there is an empirical scaling regime for network-level semantic reasoning in compact language models. Deterministic single-shot accuracy grows monotonically from 0.224 at 135M parameters to 0.707 at 7B, but the growth is not smooth; the 1B-to-1.5B step produces the largest gain, and the instability gap Δ5, defined as pass@5 minus pass@1, contracts from roughly 0.36 in the sub-1B regime to 0.031 at 7B. This contraction is interpreted as convergence between stochastic exploration and deterministic inference, a prerequisite for safety-critical control. Additionally, semantic reliability per unit edge resource, as measured by the Edge Score, does not scale with paramete

What carries the argument

The central objects are (1) 6G-Bench, a standardization-aligned benchmark of 30 tasks and 3,722 multiple-choice questions spanning intent and policy reasoning, network slicing, trust and security, agentic control, and distributed intelligence; (2) three metrics: pass@1 for deterministic accuracy, pass@k for stochastic robustness, and the instability gap Δk = Ak − A1; and (3) the Edge Score ES = A1/(L·M), which normalizes accuracy by inference latency and peak VRAM. The argument runs by comparing these quantities across ten models from 135M to 7B parameters and fitting a log-linear scaling curve A(N) = α log(1+N) + β, whose residuals and domain-level sensitivity coefficients reveal that scali

Load-bearing premise

The central claims rest on two measurements: the instability gap Δ5 is computed from pass@5 on only 7 of 30 tasks while pass@1 covers all 3,722 questions, and the latency/VRAM profiles come from single-GPU bf16 inference that may not reflect real edge costs; if either is off, the 1–1.5B stability threshold and the Edge Score ranking shift.

What would settle it

Recompute A1 and A5 on the same 7-task subset used for pass@k and on the full 30-task set; if the 1B-to-1.5B Δ5 collapse disappears on the full set, the stability transition is a subset artifact. Separately, rerun the Edge Score with memory footprint measured at matched sequence lengths and with quantization enabled, and check whether mid-scale models still dominate the frontier.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • If the 1–1.5B stability transition is real, edge deployments should target 1.5–3B models for stability-critical intent translation, agentic coordination, and cross-slice arbitration rather than the largest model that fits.
  • Ultra-compact models (≤350M) remain useful only for low-risk, localized decisions; their large instability gap disqualifies them from safety-critical control loops.
  • The diminishing returns beyond 3B (+0.064 from 3B to 7B) imply that 7B-class models should be reserved for centralized orchestration where marginal robustness justifies higher latency and memory cost.
  • Domain-level differences mean trust and security tasks saturate earlier and can be served by smaller models, while intent and policy reasoning and distributed-intelligence tasks need more capacity.
  • The Edge Score ranking suggests a tiered deployment hierarchy, placing different model sizes at different points between radio/edge nodes and centralized control.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The Δ5 contraction may be overstated because pass@5 is measured on only 7 of the 30 tasks while pass@1 is averaged over all 3,722 questions; recomputing both metrics on the same task set could move or erase the reported 1B–1.5B threshold.
  • The Edge Score ranking is a snapshot of bf16, single-GPU, zero-shot inference; under quantization, batched serving, or lower-precision edge hardware, the memory and latency profiles could change which model class wins.
  • If the non-uniform scaling curve generalizes, the search for 'minimal stable capacity' could become a standard design step for safety-critical LLM deployments beyond 6G, such as autonomous vehicles or medical triage.
  • The results are zero-shot; a testable extension is whether the 1B–1.5B transition persists under few-shot prompting or chain-of-thought, where smaller models sometimes close the gap to larger ones.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper reports an empirical scaling study of ten instruction-tuned small language models (135M–7B parameters) on 6G-Bench, the authors' 30-task / 3,722-MCQ benchmark for network-level semantic reasoning in AI-native 6G. It measures deterministic accuracy pass@1, stochastic pass@3/pass@5, and an instability gap Δ5 = A5 − A1; fits a log-linear accuracy-vs-parameter curve; defines group-level scaling sensitivity; profiles single-query latency and peak VRAM; and introduces an Edge Score ES = A1/(L·M). The headline claims are that deterministic accuracy rises from 0.224 (SmolLM2-135M) to 0.707 (Qwen2.5-7B), that a stability transition occurs in the 1–1.5B range, and that mid-scale 1.5–3B models achieve the most favorable balance between deterministic stability and edge efficiency.

Significance. If the empirical regularities hold, the deployment-tier guidance would be a useful input to 6G architecture discussions: ultra-compact models for localized decisions, 1.5–3B for stability-critical edge reasoning, and 7B for centralized orchestration. The paper's strengths are empirical: a reproducible public benchmark, local evaluation of all checkpoints under a unified prompting/extraction protocol, explicit binomial CIs for pass@1, and an honest limitations section. It does not propose new theory; the scaling fit is descriptive. The main risks are that the headline '1–1.5B stability threshold' conflates model identity with parameter count, and that the Edge Score ranking is sensitive to profiling artifacts. Because the underlying data and code are public, these issues are fixable by reframing the claims and adding the missing uncertainty/sensitivity analysis; the study remains a potentially valuable empirical contribution.

major comments (3)
  1. [§III-D, Eq. (3); Table III vs Table IV] Table III states that pass@k is evaluated only on 7 tasks (T2, T9, T12, T19, T20, T26, T30), while Table IV reports Δ5 = A5 − A1 with 95% CIs computed by variance summation of A1 and A5, as if both terms cover the full M=3,722 questions. Eq. (3) defines A_k over the whole task set T. Thus A5 and A1 are not commensurate: the gap can shrink merely because the 7-task subset is easier under greedy decoding. The paper never states this restriction in Table IV or in the abstract. This invalidates the reported contraction of Δ5 from 0.365 to 0.031 as a measure of reasoning instability and undermines the stability-transition conclusion.
  2. [§IV-A, Table IV; abstract] The '1–1.5B stability transition' is inferred from a single pair of checkpoints, Llama-3.2-1B versus Qwen2.5-1.5B. Parameter count is fully confounded with pretraining data, alignment, and architecture. Within the same nominal N≈1B, Granite-1B reaches A1=0.559 and Δ5=0.217 while Llama-1B reaches A1=0.373 and Δ5=0.356; the within-1B spread is comparable to or larger than the claimed cross-scale jump, and taking the best model at each size, A1 decreases from Granite-1B (0.559) to Qwen-1.5B (0.531). The paper's own residuals show architecture-dependent deviations of ±0.1. A z-test between two specific checkpoints is not evidence of a parameter threshold. The central 'mid-scale stability tier' therefore does not follow from Table IV.
  3. [§III-D.3, Eq. (8); Table VI; Fig. 7] Edge Score = A1/(L·M) is only as credible as the latency and memory profiles. Table VI reports Granite-350M with 16.194 GB peak VRAM and 1814.7±128.7 ms latency, and Granite-1B with 20.371 GB and 3161.3±63.3 ms — values an order of magnitude worse than the 7B models (≈1 GB, ≈340–390 ms). This is not a plausible scaling of computational cost and suggests implementation-dependent artifacts in the inference harness. Since ES is a ratio, Figure 7's ranking is dominated by these artifacts. No uncertainty is propagated to ES, and the deployment-tier recommendations in §IV-D rest on this unvalidated ranking.
minor comments (5)
  1. [Table III vs Table IV] Table III lists 11 models, but Table IV and the abstract evaluate 10; LFM2-8B appears only in Table VI. Please clarify the model count and the role of LFM2-8B.
  2. [Table IV caption] The caption should explicitly state that A3, A5, and Δ5 are computed on the 7-task subset; currently the notation implies the full 3,722 MCQs.
  3. [Eq. (4) vs Table IV residual] Eq. (4) defines A(N)=α log(1+N)+β, but Table IV's residual line says A1−(α log N+β). Align the two formulations.
  4. [Fig. 7] Add error bars or a note that Edge Score does not include uncertainty; currently the bars imply precision that the underlying measurements do not support.
  5. [Section IV-E] The limitations paragraph is welcome, but it is not connected back to the headline claims. Add a sentence stating which conclusions are robust to model-family effects and which are not.

Circularity Check

0 steps flagged

No significant circularity: the scaling results are empirical measurements; self-referential benchmark and metric do not force the conclusions.

full rationale

The paper's claimed derivation chain is primarily an empirical measurement loop, not a mathematical derivation. The central quantities (A1, A3, A5, Delta_5, Edge Score) are explicitly defined in Eqs. (1)-(3) and (8), and their values are reported in Tables IV and VI. The log-linear model in Eq. (4) is descriptive: the paper fits alpha and beta to all evaluated models and then examines residuals; it does not use the fit to produce an independent prediction that is then presented as a scaling law. No fitted constant is renamed as a predicted result. The only self-referential elements are (i) the evaluation environment 6G-Bench [17], authored by the same team, and (ii) the author-defined Edge Score. These do not make the central claims true by construction: 6G-Bench supplies measured MCQ responses, and the Edge Score ranking depends on measured latency/VRAM values reported in Table VI. The paper's own Limitations section (Sec. IV-E) acknowledges architecture-dependent deviations from pure log-linear scaling, which is a confounding issue for the '1-1.5B stability transition' but not a circularity. I also flag two validity threats that are not circularity: Table III restricts pass@k evaluation to 7 tasks while Table IV reports Delta_5 against the all-task A1, mixing different question sets; and the abstract's 1B-to-1.5B transition compares different model families (Llama-3.2-1B vs Qwen2.5-1.5B) rather than isolating parameter count. These affect whether the conclusions are sound, but they are not cases where a prediction is equivalent to its inputs by definition. Accordingly, no circular step meets the evidentiary bar, and the score is 2 to acknowledge the self-referential benchmark/metric provenance without treating it as load-bearing circularity.

Axiom & Free-Parameter Ledger

2 free parameters · 4 axioms · 0 invented entities

Central empirical claims rest on the validity of 6G-Bench's task design, on representative checkpoints, and on correct latency/memory profiling. 6G-Bench is self-authored, so external validity is assumed. The paper's own metric definitions contain an unflagged mismatch (A5 on a 7-task subset vs A1 on all MCQs), and the profiling table contains implausible values for Granite models. The fitted log-linear α and β are free parameters used only for residual commentary.

free parameters (2)
  • log-linear slope α = 0.115
    OLS fit of A1 vs log(1+N) over 10 models; used for residual analysis and scaling commentary, not for the central claim.
  • log-linear intercept β = 0.478
    OLS intercept in the same fit; not independently fixed by theory.
axioms (4)
  • domain assumption 6G-Bench tasks (T1–T30) faithfully represent AI-native 6G semantic control reasoning
    The paper derives the taxonomy from standards bodies, but the benchmark itself is self-authored [17]; external validity rests on this assumption.
  • ad hoc to paper Δ5 = A5 − A1 with A5 on a 7-task subset is a valid instability measure
    Table III limits pass@k to 7 tasks while Table IV computes Δ5 without disclosing the subset; if the subset is not representative, the stability-transition claim is unsupported.
  • standard math Normal-approximation binomial confidence intervals are adequate for M=3,722 MCQs
    Used for A1 and Δ5 CIs; conventional for large binomial samples.
  • domain assumption Single-GPU bf16 profiling approximates real edge deployment costs
    Table VI feeds the Edge Score; Granite VRAM/latency outliers suggest the profiling setup does not reflect typical edge conditions.

reviewed 2026-08-02 · how reviews work

0 comments
Cite this review

Pith. "Pith review of How Small Can 6G Reason? Scaling Tiny-to-Small Language Models for AI-Native Networks." pith.science (2026). https://pith.science/paper/ZF5UDQDI

@misc{pith2026260302156,
  author       = {Pith},
  title        = {Pith review of: How Small Can 6G Reason? Scaling Tiny-to-Small Language Models for AI-Native Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZF5UDQDI}},
  note         = {Machine review of arXiv:2603.02156}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Emerging 6G visions, reflected in ongoing standardization efforts within 3GPP, IETF, ETSI, ITU-T, and the O-RAN Alliance, increasingly characterize networks as AI-native systems in which high-level semantic reasoning layers operate above standardized control and data-plane functions. Although frontier-scale large language models (LLMs) such as Qwen2.5-7B and Olmo-3-7B demonstrate strong reasoning capability, their computational footprint limits deployment in latency-sensitive, edge-native infrastructures. This paper presents a systematic empirical study of the scaling behavior and deployment efficiency of compact language models for network-level semantic reasoning in AI-native 6G systems. Using 6G-Bench, a standardization-aligned benchmark comprising 30 decision-making tasks across five capability domains, we evaluate models ranging from 135M (SmolLM2-135M) to 7B parameters (Qwen2.5-7B), including mid-scale architectures such as Llama-3.2-1B, Granite-1B, and Qwen2.5-3B. Deterministic accuracy (pass@1) increases from 0.224 at 135M to 0.707 at 7B, but scaling gains are highly non-uniform. A pronounced stability transition occurs in the 1 to 1.5B range, where accuracy rises from 0.373 (Llama-3.2-1B) to 0.531 (Qwen2.5-1.5B) and the instability gap Delta_5 contracts from 0.356 to 0.138. Beyond 3B parameters, improvements diminish (+0.064 from 3B to 7B). Through single-query inference profiling and an Edge Score metric that normalizes accuracy by latency and memory footprint, we show that semantic reliability per unit edge resource does not scale monotonically with parameter count. Instead, mid-scale models (approximately 1.5 to 3B) achieve the most favorable balance between deterministic stability and computational efficiency, providing deployment-relevant guidance for AI-native 6G architectures. All scripts and results are publicly available at https://github.com/maferrag/6G-Bench

Figures

Figures reproduced from arXiv: 2603.02156 by Abderrahmane Lakas, Merouane Debbah, Mohamed Amine Ferrag.

Figure 1
Figure 1. Figure 1: End-to-End Methodological Framework for Evaluating Parameter Scaling and Semantic Reasoning Stability in 6G-Bench. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Group G1 (Intent & Policy Reasoning): pass@1 accuracy across model [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 6
Figure 6. Figure 6: Group G5 (Distributed Intelligence & Emerging 6G Use Cases): [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 5
Figure 5. Figure 5: Group G4 (AI-Native Networking & Agentic Control): pass@1 [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: Edge Score ranking of evaluated LLMs on 6G-Bench. Edge Score [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

25 extracted references · 8 linked inside Pith

  1. [1]

    F-iran: Performance analysis of 6g fog intelli- gent radio access network,

    E. A. Anowr, M. Nashaat, M. I. Ismail, M. A. Mohamed, M. M. Fouda, and H. M. Abdel-Atty, “F-iran: Performance analysis of 6g fog intelli- gent radio access network,”IEEE Open Journal of the Communications Society, 2025

    2025

  2. [2]

    Study on 6g use cases and service requirements,

    3GPP, “Study on 6g use cases and service requirements,” 3rd Generation Partnership Project (3GPP), Tech. Rep. TR 22.870, 2025

    2025

  3. [3]

    Generative ai use cases and requirements on 6g network,

    O-RAN Alliance e.V ., “Generative ai use cases and requirements on 6g network,” O-RAN Alliance, Research Report, Jan 2025, next Generation Research Group (nGRG). [Online]. Available: https://www.o-ran.org/research-reports/ generative-ai-use-cases-and-requirements-on-6g-network

    2025

  4. [4]

    GS MEC 003 V3.1.1, Mar

    ETSI,Multi-access Edge Computing (MEC); Framework and Reference Architecture, European Telecommunications Standards Institute (ETSI) Std. GS MEC 003 V3.1.1, Mar. 2022

    2022

  5. [5]

    Requirements and enabling technologies of agent protocols for 6g networks,

    C. Yang, H. Huang, A. Akhavain, F. Liu, X. An, W. Xing, J. Li, A. Wang, and W. Yang, “Requirements and enabling technologies of agent protocols for 6g networks,” Internet Engineering Task Force (IETF), Internet-Draft draft-hw-ai-agent-6g-00, Jul. 2025. [Online]. Available: https://www.ietf.org/archive/id/draft-hw-ai-agent-6g-00.html

    2025

  6. [6]

    6g cellular networks: Mapping the land- scape for the imt-2030 framework,

    E. Hossain and A. Vera-Rivera, “6g cellular networks: Mapping the land- scape for the imt-2030 framework,”IEEE Transactions on Technology and Society, 2025

    2030

  7. [7]

    Extremely large aperture array (elaa) communications: Foundations, research advances and challenges,

    S. Ye, M. Xiao, M.-W. Kwan, Z. Ma, Y . Huang, G. Karagiannidis, and P. Fan, “Extremely large aperture array (elaa) communications: Foundations, research advances and challenges,”IEEE Open Journal of the Communications Society, vol. 5, pp. 7075–7120, 2024

    2024

  8. [8]

    Energy-efficient ris-aided cell-free massive mimo systems: Application, opportunities, and challenges,

    Y . Lu, J. Zhang, E. Shi, P. Zhang, D. W. K. Ng, D. Niyato, and B. Ai, “Energy-efficient ris-aided cell-free massive mimo systems: Application, opportunities, and challenges,”IEEE Wireless Communications, 2025

    2025

  9. [9]

    6g phy: Insights from 6g-anna research initiative,

    B. Banerjeeet al., “6g phy: Insights from 6g-anna research initiative,” IEEE Open Journal of the Communications Society, 2026

    2026

  10. [10]

    Resource optimization for semantic communication in 6g networks: A survey,

    S. Sharif, F. Khandaker, M. Naeem, and W. Ejaz, “Resource optimization for semantic communication in 6g networks: A survey,”IEEE Open Journal of the Communications Society, 2026

    2026

  11. [11]

    Towards 6g authen- tication and key agreement protocol: A survey on hybrid post quantum cryptography,

    T. N. Turnip, B. Andersen, and C. Vargas-Rosales, “Towards 6g authen- tication and key agreement protocol: A survey on hybrid post quantum cryptography,”IEEE Communications Surveys & Tutorials, 2025

    2025

  12. [12]

    Llms on a budget: System-level approaches to power-efficient and scalable fine-tuning,

    K. Gogineni, A. Suvizi, and G. Venkataramani, “Llms on a budget: System-level approaches to power-efficient and scalable fine-tuning,” IEEE Open Journal of the Computer Society, 2025

    2025

  13. [13]

    α 3-bench: A unified bench- mark of safety, robustness, and efficiency for llm-based uav agents over 6g networks,

    M. A. Ferrag, A. Lakas, and M. Debbah, “α 3-bench: A unified bench- mark of safety, robustness, and efficiency for llm-based uav agents over 6g networks,”arXiv preprint arXiv:2601.03281, 2026

    arXiv 2026

  14. [14]

    Training compute-optimal large language models,

    J. Hoffmann, S. Borgeaud, A. Mensch, E. Buchatskaya, T. Cai, E. Rutherford, D. Casas, L. A. Hendricks, J. Welbl, A. Clarket al., “Training compute-optimal large language models,”arXiv preprint arXiv:2203.15556, vol. 10, 2022

    Pith/arXiv arXiv 2022

  15. [15]

    Scaling laws for neural language models,

    J. Kaplan, S. McCandlish, T. Henighan, T. B. Brown, B. Chess, R. Child, S. Gray, A. Radford, J. Wu, and D. Amodei, “Scaling laws for neural language models,”arXiv preprint arXiv:2001.08361, 2020

    Pith/arXiv arXiv 2001

  16. [16]

    Emergent abilities of large language models,

    J. Wei, Y . Tay, R. Bommasani, C. Raffel, B. Zoph, S. Borgeaud, D. Yogatama, M. Bosma, D. Zhou, D. Metzleret al., “Emergent abilities of large language models,”arXiv preprint arXiv:2206.07682, 2022

    Pith/arXiv arXiv 2022

  17. [17]

    6g-bench: An open benchmark for semantic communication and network-level reasoning with founda- tion models in ai-native 6g networks,

    M. A. Ferrag, A. Lakas, and M. Debbah, “6g-bench: An open benchmark for semantic communication and network-level reasoning with founda- tion models in ai-native 6g networks,”arXiv preprint arXiv:2602.08675, 2026

    arXiv 2026

  18. [18]

    Scaling language models: Methods, analysis & insights from training gopher,

    J. W. Rae, S. Borgeaud, T. Cai, K. Millican, J. Hoffmann, F. Song, J. Aslanides, S. Henderson, R. Ring, S. Younget al., “Scaling language models: Methods, analysis & insights from training gopher,”arXiv preprint arXiv:2112.11446, 2021

    Pith/arXiv arXiv 2021

  19. [19]

    Smollm2: When smol goes big–data-centric training of a small language model,

    L. B. Allal, A. Lozhkov, E. Bakouch, G. M. Bl ´azquez, G. Penedo, L. Tunstall, A. Marafioti, H. Kydl ´ıˇcek, A. P. Lajar ´ın, V . Srivastav et al., “Smollm2: When smol goes big–data-centric training of a small language model,”arXiv preprint arXiv:2502.02737, 2025

    Pith/arXiv arXiv 2025

  20. [20]

    Granite 4.0 nano language models,

    IBM Research, “Granite 4.0 nano language models,” https://github.com/ ibm-granite/granite-4.0-nano-language-models, 2025, accessed: 2026- 02-19

    2025

  21. [21]

    Lfm2 technical report,

    L. AI, “Lfm2 technical report,”arXiv preprint arXiv:2511.23404, 2025

    arXiv 2025

  22. [22]

    The llama 3 herd of models,

    A. Grattafiori, A. Dubey, A. Jauhri, A. Pandey, A. Kadian, A. Al-Dahle, A. Letman, A. Mathur, A. Schelten, A. Vaughanet al., “The llama 3 herd of models,”arXiv preprint arXiv:2407.21783, 2024

    Pith/arXiv arXiv 2024

  23. [23]

    Qwen2.5: A party of foundation models,

    Q. Team, “Qwen2.5: A party of foundation models,” September 2024. [Online]. Available: https://qwenlm.github.io/blog/qwen2.5/

    2024

  24. [24]

    Qwen2 technical report,

    A. Yanget al., “Qwen2 technical report,”arXiv preprint arXiv:2407.10671, 2024

    Pith/arXiv arXiv 2024

  25. [25]

    Olmoet al., “Olmo 3,”arXiv preprint arXiv:2512.13961, 2025

    T. Olmoet al., “Olmo 3,”arXiv preprint arXiv:2512.13961, 2025

    Pith/arXiv arXiv 2025

This paper was first reviewed by deepseek-v4-flash on August 2, 2026.