Paper Citation Record · LEDGER

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training

As of 22 August 2026, this Paper Citation Record lists 21 of 21 outbound references and 0 inbound Pith citation observations for arXiv:2603.06642.

A citation records a reference. It does not transfer a finding from one paper to another.

pith.paper-citation-record.v1
2603.06642 v2

Coverage vector

measured 21 of 21 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-02T20:43:37.635374Z

measured 21 of 21 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-22T06:32:14.747728+00:00

measured 0 of 0 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links

measured 0 of 1 external citation measurements

A source-named dated measurement, never combined with another source.

Source: cited_works

Reference resolution

21 of 21 outbound references displayed

verified exact0
verified fuzzy0
unresolved21
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 75694d6f-4625-4823-8e68-928d4a3e55d6 · outbound

This paper cites (2017).Attention Is All You Need.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2017).Attention Is All You Need

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:35.881099Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3422a72c-4495-4ace-94be-257c14891e61 · outbound

This paper cites (2020).Test-Time Training with Self-Supervision for Generalization under Distribution Shifts.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2020).Test-Time Training with Self-Supervision for Generalization under Distribution Shifts

Reference 2

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:35.950613Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d1bbf254-a46d-4612-8e47-76ae9f9f50aa · outbound

This paper cites (2024).Learning to (Learn at Test Time): RNNs with Expressive Hidden States.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2024).Learning to (Learn at Test Time): RNNs with Expressive Hidden States

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:36.021379Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6dec8a8c-1dde-48dc-8d60-410414e3e872 · outbound

This paper cites Test-Time Training Done Right.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training Test-Time Training Done Right

Reference 4

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:36.082526Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3deafdcc-16e1-403b-83bb-0c50dcd08f2f · outbound

This paper cites (2026).Gated Differentiable Working Memory (GDWM) for Long-Context Language Modeling.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2026).Gated Differentiable Working Memory (GDWM) for Long-Context Language Modeling

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:36.191586Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fbd984b9-f0c5-4bd9-8884-6e27320563b0 · outbound

This paper cites F., et al.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training F., et al

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:36.359871Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8036a418-5918-4475-b936-76afb824ee1a · outbound

This paper cites (2023).Needle In A Haystack – Pressure Testing LLMs.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2023).Needle In A Haystack – Pressure Testing LLMs

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:36.426811Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation cb23690b-88a9-4288-a9e0-c74fb80aa210 · outbound

This paper cites Mamba: Linear-Time Sequence Modeling with Selective State Spaces.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training Mamba: Linear-Time Sequence Modeling with Selective State Spaces

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:36.501176Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c9b257b6-f772-4e0b-acf5-27871be06db9 · outbound

This paper cites (2023).RWKV: Reinventing RNNs for the Transformer Era.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2023).RWKV: Reinventing RNNs for the Transformer Era

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:36.612867Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0989ade6-e56f-49bd-a9cd-9f117b0b9eff · outbound

This paper cites Titans: Learning to Memorize at Test Time.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training Titans: Learning to Memorize at Test Time

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:36.742463Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f3edf507-b039-4c99-a4c5-eab6352b839c · outbound

This paper cites (2024).Simple Linear Attention Language Models Balance the Recall-Throughput Tradeoff.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2024).Simple Linear Attention Language Models Balance the Recall-Throughput Tradeoff

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:36.817413Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9e0559d0-6000-4a9c-ad3e-677e2188302a · outbound

This paper cites (2024).Zoology: Measuring and Improving Recall in Efficient Language Models.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2024).Zoology: Measuring and Improving Recall in Efficient Language Models

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:36.868262Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8dcd8aa9-e883-4b6c-adb1-558b0d38a0c6 · outbound

This paper cites (2024).Get More with LESS: Synthesizing Recurrence with KV Cache Compression for Efficient LLM Inference.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2024).Get More with LESS: Synthesizing Recurrence with KV Cache Compression for Efficient LLM Inference

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:36.942675Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0186dea8-15c6-4ffc-aefd-e06921973236 · outbound

This paper cites (2025).Cache What Lasts: Token Retention for Memory-Bounded KV Cache in LLMs.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2025).Cache What Lasts: Token Retention for Memory-Bounded KV Cache in LLMs

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:37.044877Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c6d67a6f-192d-4ea5-b37f-4d977bb567e5 · outbound

This paper cites (2024).MInference 1.0: Accelerating Pre-filling for Long-Context LLMs via Dynamic Sparse Attention.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2024).MInference 1.0: Accelerating Pre-filling for Long-Context LLMs via Dynamic Sparse Attention

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:37.166781Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d8e2c0b5-458f-4baf-b766-d3d418fc5841 · outbound

This paper cites KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization

Reference 16

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:37.256381Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 27751f3e-0844-40e9-8b11-95a7306c453b · outbound

This paper cites (2024).H 2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2024).H 2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models

Reference 17

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:37.378095Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fb62226e-15c4-4ea8-8d94-79f5bc76a46e · outbound

This paper cites (2024).Dynamic Memory Compression: Retrofitting LLMs for Accelerated Inference.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training (2024).Dynamic Memory Compression: Retrofitting LLMs for Accelerated Inference

Reference 18

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:37.467237Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation bb4bc3da-3819-4872-af09-d018dbcf2a08 · outbound

This paper cites TinyStories: How Small Can Language Models Be and Still Speak Coherent English?.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training TinyStories: How Small Can Language Models Be and Still Speak Coherent English?

Reference 19

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:37.517425Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 57b77d9f-222a-421a-8a8a-f7d04c599a57 · outbound

This paper cites RoFormer: Enhanced Transformer with Rotary Position Embedding.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training RoFormer: Enhanced Transformer with Rotary Position Embedding

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:37.565942Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ee5d69d5-004b-496d-9762-d59351477fe4 · outbound

This paper cites YaRN: Efficient Context Window Extension of Large Language Models.

SR-TTT Does Not Learn Retrieval: A Correction and Mechanistic Post-Mortem of Surprisal-Aware Residual Test-Time Training YaRN: Efficient Context Window Extension of Large Language Models

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-02T20:43:37.635374Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

No inbound Pith citation observations are available.