Number AFFECTOR Markdown figures and references

SHA256SUMS.txt

@@ -1,7 +1,7 @@
 321cd041f8060331e034b2a72578f57cd3e498883e660016cf73f0a8ba69a9a6  README.md
 e059179343c7acd54f1d6b8da8ba36cdb0fe941aca291f9be6d9bf4aa5a9dd0b  docs/AFFECTOR-Technical-Paper.pdf
-1618236f108a297a7b545549670729469e970f3c0bee4f46d44405cd24966db8  docs/TECHNICAL_PAPER.md
-1618236f108a297a7b545549670729469e970f3c0bee4f46d44405cd24966db8  docs/TECHNICAL_REPORT.md
+1f66c1e5ebf5db7e7de12c921f2a242af266a72508c01394c0123bfac89eff5e  docs/TECHNICAL_PAPER.md
+1f66c1e5ebf5db7e7de12c921f2a242af266a72508c01394c0123bfac89eff5e  docs/TECHNICAL_REPORT.md
 ed8fa321af4fc4e4703efcc22037e244fd072c936d11afc566b8eb862f3f0b1b  downloads/AFFECTOR-macOS-AppleSilicon.zip
 d7621338f71fdc31ce9b6e1d53f15d61d6a76344f2111520fb62f9b00293853b  downloads/AFFECTOR-source.zip
 585966103f53f58a9983bba63e3d27dcb2020033a2f64c6dc63684afe418ae14  paper/AFFECTOR-ACM-source.zip
@@ -18,7 +18,7 @@ e2c0590754823491af4657a764ab6682e27c5eaf0173fc9f5f03a62e84ccb308  paper/ILLUSTRA
 d14edda139a36c0bac8b56d02e5cda63ab7558c6db1424f1087189a9429bfaf0  paper/LITERATURE_SOURCES.md
 578a471b301cdf67305c2b997fce39494c2fd0d15eaf56836379943b3d10f90c  paper/PAPER-SHA256SUMS.txt
 22f8f7bf6992778eb1190ee02b6e83dacbd54ff34af7777a18b4e27a195309a3  paper/README.md
-0f07bb640c2f3d79fe21930a2cd3b12f013cea815c7e697394a7cfc29b69ea15  paper/TECHNICAL_PAPER.md
+be5b286160b97b1bc5af382da1a5ccdea50743f53ece5acaaa532c44767378cd  paper/TECHNICAL_PAPER.md
 4bd1a1b7f6af31b12b9c92bcd978fe116482892b3463f8ec9c922bc1f636c7a0  paper/figure-specifications.json
 b11798a28d94d60110a7b19f58107d5e60926b68265f64e62f59e9230a8f500d  paper/figures/illustration-01.pdf
 d2ae42a17ff3f16b89a7cb85d21ac1f3c7a19c2bc62d61f1cc581c2ca6d3e596  paper/figures/illustration-01.png
@@ -32,4 +32,4 @@ a6a497a38f4ba9de8088f410535bee13dd862d1a97f4175d205df768e1d2c1c9  paper/figures/
 7d8f30b829993326e8d764ae18b3974acddbd86b760d4071bdff4edfcb9b299a  paper/figures/illustration-05.png
 9257f38423bd7e496f1a0916c0fbaefbfb32ab544a684ca9a411508201f0535e  paper/references.bib
 eed8b979a5a299b42e925d9027332e7df876400dcef046f2cfe26beef91d1189  paper/render_figures.py
-0f07bb640c2f3d79fe21930a2cd3b12f013cea815c7e697394a7cfc29b69ea15  paper/source.md
+be5b286160b97b1bc5af382da1a5ccdea50743f53ece5acaaa532c44767378cd  paper/source.md

docs/TECHNICAL_PAPER.md

@@ -17,9 +17,9 @@ A document conveys more than propositions. Its lexical choices, rhythm, imagery,
 
 Activation steering offers a way to connect this target to the process of generation. Turner and colleagues construct intervention vectors from contrasting prompts, while Panickssery and colleagues average activation differences across positive and negative behavioral examples ([Turner et al., 2023](https://arxiv.org/abs/2308.10248v1); [Panickssery et al., 2024](https://arxiv.org/abs/2312.06681v4)). Building on this contrastive approach, AFFECTOR maps emotional coordinates into changes to a frozen model's internal states during local inference. The application also provides an API route, in which the coordinates are expressed as instructions to a hosted model. Together, these routes make it possible to investigate numerical intervention alongside prompt-based control, although their settings and outcomes are not directly interchangeable.
 
-![Document workspace with an example input, rewrite controls, and an empty output pane before generation.](../paper/figures/screenshot-01.png)
+![Figure 1. Document workspace with an example input, rewrite controls, and an empty output pane before generation.](../paper/figures/screenshot-01.png)
 
-*Document workspace with an example input, rewrite controls, and an empty output pane before generation.*
+*Figure 1. Document workspace with an example input, rewrite controls, and an empty output pane before generation.*
 
 For either route, the usefulness of the coordinates depends on their relationship to human interpretations of the output. Work by Wu and colleagues connects model emotion spaces with human valence/arousal ratings and reports concept-based steering across linguistic-cultural settings ([Wu et al., 2025](https://arxiv.org/abs/2506.13978v1)). Diallo and colleagues extend this evaluative perspective by examining emotional intensity and text quality with human raters, finding effects that vary with the emotion and steering strength ([Diallo et al., 2026](https://arxiv.org/abs/2601.21505v1)). These findings motivate assessing emotional movement and readable expression separately. They also clarify what remains to be established for AFFECTOR: neither its inward/outward axis nor the correspondence between its RMS-scaled controls and a reader's emotional judgment has been validated by those studies.
 
@@ -33,9 +33,9 @@ Taken together, these considerations define the scope of the present report: the
 
 A geometry consists of ordered points `(t, v, d, i)`. Position `t` is in `[0,1]`; valence `v` and direction `d` are in `[-1,1]`; intensity `i` is in `[0,1]`. Negative valence denotes unpleasant affect and positive valence pleasant affect. Negative direction denotes self-focused or inward-directed emotion; positive direction denotes outward focus. Direction is independent of valence: anger toward oneself can be inward and sadness for another person outward. Intensity describes strength of affect, not pleasantness.
 
-![Geometry editor showing the selected three-point trajectory and its steering controls.](../paper/figures/screenshot-02.png)
+![Figure 2. Geometry editor showing the selected three-point trajectory and its steering controls.](../paper/figures/screenshot-02.png)
 
-*Geometry editor showing the selected three-point trajectory and its steering controls.*
+*Figure 2. Geometry editor showing the selected three-point trajectory and its steering controls.*
 For intervention, intensity is centered:
 
 `c(t) = (v(t), d(t), 2i(t) - 1)`.
@@ -46,17 +46,17 @@ The target document is divided into chunks of at most 700 characters, preferring
 
 This normalized alignment compresses or stretches a source trajectory to the target length. It samples only as many positions as there are chunks. It does not preserve all events or oscillations of a much longer curve. Changes to segmentation can change emotional targets even when the same geometry is selected.
 
-![One continuous path, a finite set of targets](../paper/figures/illustration-01.png)
+![Figure 3. One continuous path, a finite set of targets](../paper/figures/illustration-01.png)
 
-*Illustration 1. Normalized alignment schematic. Target chunks sample a continuous source path at original character midpoints; a short target can miss narrow changes in a long source trajectory.*
+*Figure 3. Normalized alignment schematic. Target chunks sample a continuous source path at original character midpoints; a short target can miss narrow changes in a long source trajectory.*
 
 ## Frozen model and contrastive calibration
 
 The local backend loads Qwen3.5-0.8B at revision `2fc06364715b967f1860aea9cf38778875588b17`. The implementation operates on its text stack of 24 blocks with hidden width 1,024. Base parameters are frozen; inference uses float32 and eager attention. Apple MPS is selected when available, otherwise CPU. The chat template disables thinking output. The package contains Python 3.14, PyTorch 2.15.0 and Transformers 5.22.2; exact dependency pins accompany the bundle.
 
-![Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.](../paper/figures/screenshot-03.png)
+![Figure 4. Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.](../paper/figures/screenshot-03.png)
 
-*Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.*
+*Figure 4. Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.*
 Before steering, the application obtains three directions per selected layer from twelve manually authored English contrast pairs, four per axis. For example, valence pairs contrast despair with joy, and intensity pairs contrast mild unease with overwhelming terror. Both members of every pair are processed under a continuation instruction. Thus calibration requires 24 forward passes.
 
 The capture site is the final chat-formatted prompt position immediately before the post-attention normalization that feeds the feed-forward sublayer. It is not a mean over all content tokens. For layer `l` and axis `j`, the mean contrast is:
@@ -75,9 +75,9 @@ There is no gradient training, score-supervised fitting, orthogonalization, or g
 
 Axis temperature `T` transforms each coordinate magnitude while retaining its sign:
 
-![Expanded numerical weights for the selected geometry, including the local calibration metadata.](../paper/figures/screenshot-04.png)
+![Figure 5. Expanded numerical weights for the selected geometry, including the local calibration metadata.](../paper/figures/screenshot-04.png)
 
-*Expanded numerical weights for the selected geometry, including the local calibration metadata.*
+*Figure 5. Expanded numerical weights for the selected geometry, including the local calibration metadata.*
 `q[j] = sign(c[j]) * abs(c[j])^(1/T)`.
 
 The implementation ranks axes by magnitude, retains at most axis top-k, and stops adding axes once cumulative magnitude reaches the configured mass fraction. Surviving coefficients are not renormalized. These operations act on three emotional coefficients, not on the model's vocabulary distribution.
@@ -181,29 +181,29 @@ The equations and implementation descriptions above are derived from the include
 
 Version-specific arXiv records are linked below. Years distinguish original release from later revision where applicable. These research references supplement the implementation and infrastructure sources elsewhere in this report. Reusable citations: [BibTeX](references.bib). Verification details: [source-check notes](LITERATURE_SOURCES.md).
 
-**R1.** Alex Turner; Lisa Thiergart; David Udell; Gavin Leech; Ulisse Mini; Monte MacDiarmid. 2023. Activation Addition: Steering Language Models Without Optimization. [arXiv:2308.10248v1](https://arxiv.org/abs/2308.10248v1). Cited version dated 2023-08-20.
+1. Alex Turner; Lisa Thiergart; David Udell; Gavin Leech; Ulisse Mini; Monte MacDiarmid. 2023. Activation Addition: Steering Language Models Without Optimization. [arXiv:2308.10248v1](https://arxiv.org/abs/2308.10248v1). Cited version dated 2023-08-20.
 
-**R2.** Nina Panickssery; Nick Gabrieli; Julian Schulz; Meg Tong; Evan Hubinger; Alexander Matt Turner. 2024. Steering Llama 2 via Contrastive Activation Addition. [arXiv:2312.06681v4](https://arxiv.org/abs/2312.06681v4). Cited version dated 2024-07-05.
+2. Nina Panickssery; Nick Gabrieli; Julian Schulz; Meg Tong; Evan Hubinger; Alexander Matt Turner. 2024. Steering Llama 2 via Contrastive Activation Addition. [arXiv:2312.06681v4](https://arxiv.org/abs/2312.06681v4). Cited version dated 2024-07-05.
 
-**R3.** Xiuwen Wu; Hao Wang; Zhiang Yan; Xiaohan Tang; Pengfei Xu; Wai-Ting Siok; Ping Li; Jia-Hong Gao; Bingjiang Lyu; Lang Qin. 2025. AI shares emotion with humans across languages and cultures. [arXiv:2506.13978v1](https://arxiv.org/abs/2506.13978v1). Cited version dated 2025-06-11.
+3. Xiuwen Wu; Hao Wang; Zhiang Yan; Xiaohan Tang; Pengfei Xu; Wai-Ting Siok; Ping Li; Jia-Hong Gao; Bingjiang Lyu; Lang Qin. 2025. AI shares emotion with humans across languages and cultures. [arXiv:2506.13978v1](https://arxiv.org/abs/2506.13978v1). Cited version dated 2025-06-11.
 
-**R4.** Diaoulé Diallo; Katharina Dworatzyk; Sophie Jentzsch; Peer Schütt; Sabine Theis; Tobias Hecking. 2026. The Effectiveness of Style Vectors for Steering Large Language Models: A Human Evaluation. [arXiv:2601.21505v1](https://arxiv.org/abs/2601.21505v1). Cited version dated 2026-01-29. ArXiv record lists IEEE Access 13 (2025), 191443-191457; DOI: 10.1109/ACCESS.2025.3628500.
+4. Diaoulé Diallo; Katharina Dworatzyk; Sophie Jentzsch; Peer Schütt; Sabine Theis; Tobias Hecking. 2026. The Effectiveness of Style Vectors for Steering Large Language Models: A Human Evaluation. [arXiv:2601.21505v1](https://arxiv.org/abs/2601.21505v1). Cited version dated 2026-01-29. ArXiv record lists IEEE Access 13 (2025), 191443-191457; DOI: 10.1109/ACCESS.2025.3628500.
 
-**R5.** Moran Sun; Tianlin Li; Yuwei Zheng; Zhenhong Zhou; Aishan Liu; Xianglong Liu; Yang Liu. 2026. How Emotion Shapes the Behavior of LLMs and Agents: A Mechanistic Study. [arXiv:2604.00005v1](https://arxiv.org/abs/2604.00005v1). Cited version dated 2026-03-09.
+5. Moran Sun; Tianlin Li; Yuwei Zheng; Zhenhong Zhou; Aishan Liu; Xianglong Liu; Yang Liu. 2026. How Emotion Shapes the Behavior of LLMs and Agents: A Mechanistic Study. [arXiv:2604.00005v1](https://arxiv.org/abs/2604.00005v1). Cited version dated 2026-03-09.
 
-**R6.** Jihoon Jeong. 2026. Extracting and Steering Emotion Representations in Small Language Models: A Methodological Comparison. [arXiv:2604.04064v1](https://arxiv.org/abs/2604.04064v1). Cited version dated 2026-04-05.
+6. Jihoon Jeong. 2026. Extracting and Steering Emotion Representations in Small Language Models: A Methodological Comparison. [arXiv:2604.04064v1](https://arxiv.org/abs/2604.04064v1). Cited version dated 2026-04-05.
 
-**R7.** Bangzhao Shu; Arinjay Singh; Mai ElSherief. 2026. Why Are Some Emotions Harder for LLMs? Uncovering the Causal Mechanisms of Emotion Inference via Sparse Autoencoders. [arXiv:2604.25866v3](https://arxiv.org/abs/2604.25866v3). Cited version dated 2026-10-24.
+7. Bangzhao Shu; Arinjay Singh; Mai ElSherief. 2026. Why Are Some Emotions Harder for LLMs? Uncovering the Causal Mechanisms of Emotion Inference via Sparse Autoencoders. [arXiv:2604.25866v3](https://arxiv.org/abs/2604.25866v3). Cited version dated 2026-10-24.
 
-**R8.** Yibo Wang; Jisheng Dang; Bimei Wang; Yitao Wu; Wencan Zhang; Hong Peng; Jizhao Liu; Bin Hu; Qi Tian; Tat-Seng Chua. 2026. Read-Best Is Not Steer-Best: A Probing--Steering Layer Dissociation in Omni-Modal Large Language Models. [arXiv:2609.22135v1](https://arxiv.org/abs/2609.22135v1). Cited version dated 2026-08-24.
+8. Yibo Wang; Jisheng Dang; Bimei Wang; Yitao Wu; Wencan Zhang; Hong Peng; Jizhao Liu; Bin Hu; Qi Tian; Tat-Seng Chua. 2026. Read-Best Is Not Steer-Best: A Probing--Steering Layer Dissociation in Omni-Modal Large Language Models. [arXiv:2609.22135v1](https://arxiv.org/abs/2609.22135v1). Cited version dated 2026-08-24.
 
-**R9.** JuHeon Ha; Byounghan Lee; Yunseo Choi; Kyung-Ah Sohn. 2026. Empathy Is Steerable but Multi-Axial: Mechanism Geometry and Persona Effects in LLMs. [arXiv:2609.15654v1](https://arxiv.org/abs/2609.15654v1). Cited version dated 2026-09-14. ArXiv comments report acceptance to the EMNLP 2026 main conference.
+9. JuHeon Ha; Byounghan Lee; Yunseo Choi; Kyung-Ah Sohn. 2026. Empathy Is Steerable but Multi-Axial: Mechanism Geometry and Persona Effects in LLMs. [arXiv:2609.15654v1](https://arxiv.org/abs/2609.15654v1). Cited version dated 2026-09-14. ArXiv comments report acceptance to the EMNLP 2026 main conference.
 
-**R10.** Sasha Cui; Zhongren Chen. 2025. Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models. [arXiv:2509.22739v3](https://arxiv.org/abs/2509.22739v3). Cited version dated 2026-05-15.
+10. Sasha Cui; Zhongren Chen. 2025. Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models. [arXiv:2509.22739v3](https://arxiv.org/abs/2509.22739v3). Cited version dated 2026-05-15.
 
-**R11.** Alex Ferrando; Xavier Suau; Jordi Gonzàlez; Pau Rodriguez. 2025. Dynamically Scaled Activation Steering. [arXiv:2512.03661v2](https://arxiv.org/abs/2512.03661v2). Cited version dated 2026-07-30.
+11. Alex Ferrando; Xavier Suau; Jordi Gonzàlez; Pau Rodriguez. 2025. Dynamically Scaled Activation Steering. [arXiv:2512.03661v2](https://arxiv.org/abs/2512.03661v2). Cited version dated 2026-07-30.
 
-**R12.** Haotian Yang; Huikang Jiang; Yucheng Wu; Wen-Jie Jiang; Chenpeng Wang; Yibin Lou; Liangming Pan. 2026. Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods. [arXiv:2610.07722v1](https://arxiv.org/abs/2610.07722v1). Cited version dated 2026-10-06.
+12. Haotian Yang; Huikang Jiang; Yucheng Wu; Wen-Jie Jiang; Chenpeng Wang; Yibin Lou; Liangming Pan. 2026. Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods. [arXiv:2610.07722v1](https://arxiv.org/abs/2610.07722v1). Cited version dated 2026-10-06.
 
 ## Suggested Citation
 

docs/TECHNICAL_REPORT.md

@@ -17,9 +17,9 @@ A document conveys more than propositions. Its lexical choices, rhythm, imagery,
 
 Activation steering offers a way to connect this target to the process of generation. Turner and colleagues construct intervention vectors from contrasting prompts, while Panickssery and colleagues average activation differences across positive and negative behavioral examples ([Turner et al., 2023](https://arxiv.org/abs/2308.10248v1); [Panickssery et al., 2024](https://arxiv.org/abs/2312.06681v4)). Building on this contrastive approach, AFFECTOR maps emotional coordinates into changes to a frozen model's internal states during local inference. The application also provides an API route, in which the coordinates are expressed as instructions to a hosted model. Together, these routes make it possible to investigate numerical intervention alongside prompt-based control, although their settings and outcomes are not directly interchangeable.
 
-![Document workspace with an example input, rewrite controls, and an empty output pane before generation.](../paper/figures/screenshot-01.png)
+![Figure 1. Document workspace with an example input, rewrite controls, and an empty output pane before generation.](../paper/figures/screenshot-01.png)
 
-*Document workspace with an example input, rewrite controls, and an empty output pane before generation.*
+*Figure 1. Document workspace with an example input, rewrite controls, and an empty output pane before generation.*
 
 For either route, the usefulness of the coordinates depends on their relationship to human interpretations of the output. Work by Wu and colleagues connects model emotion spaces with human valence/arousal ratings and reports concept-based steering across linguistic-cultural settings ([Wu et al., 2025](https://arxiv.org/abs/2506.13978v1)). Diallo and colleagues extend this evaluative perspective by examining emotional intensity and text quality with human raters, finding effects that vary with the emotion and steering strength ([Diallo et al., 2026](https://arxiv.org/abs/2601.21505v1)). These findings motivate assessing emotional movement and readable expression separately. They also clarify what remains to be established for AFFECTOR: neither its inward/outward axis nor the correspondence between its RMS-scaled controls and a reader's emotional judgment has been validated by those studies.
 
@@ -33,9 +33,9 @@ Taken together, these considerations define the scope of the present report: the
 
 A geometry consists of ordered points `(t, v, d, i)`. Position `t` is in `[0,1]`; valence `v` and direction `d` are in `[-1,1]`; intensity `i` is in `[0,1]`. Negative valence denotes unpleasant affect and positive valence pleasant affect. Negative direction denotes self-focused or inward-directed emotion; positive direction denotes outward focus. Direction is independent of valence: anger toward oneself can be inward and sadness for another person outward. Intensity describes strength of affect, not pleasantness.
 
-![Geometry editor showing the selected three-point trajectory and its steering controls.](../paper/figures/screenshot-02.png)
+![Figure 2. Geometry editor showing the selected three-point trajectory and its steering controls.](../paper/figures/screenshot-02.png)
 
-*Geometry editor showing the selected three-point trajectory and its steering controls.*
+*Figure 2. Geometry editor showing the selected three-point trajectory and its steering controls.*
 For intervention, intensity is centered:
 
 `c(t) = (v(t), d(t), 2i(t) - 1)`.
@@ -46,17 +46,17 @@ The target document is divided into chunks of at most 700 characters, preferring
 
 This normalized alignment compresses or stretches a source trajectory to the target length. It samples only as many positions as there are chunks. It does not preserve all events or oscillations of a much longer curve. Changes to segmentation can change emotional targets even when the same geometry is selected.
 
-![One continuous path, a finite set of targets](../paper/figures/illustration-01.png)
+![Figure 3. One continuous path, a finite set of targets](../paper/figures/illustration-01.png)
 
-*Illustration 1. Normalized alignment schematic. Target chunks sample a continuous source path at original character midpoints; a short target can miss narrow changes in a long source trajectory.*
+*Figure 3. Normalized alignment schematic. Target chunks sample a continuous source path at original character midpoints; a short target can miss narrow changes in a long source trajectory.*
 
 ## Frozen model and contrastive calibration
 
 The local backend loads Qwen3.5-0.8B at revision `2fc06364715b967f1860aea9cf38778875588b17`. The implementation operates on its text stack of 24 blocks with hidden width 1,024. Base parameters are frozen; inference uses float32 and eager attention. Apple MPS is selected when available, otherwise CPU. The chat template disables thinking output. The package contains Python 3.14, PyTorch 2.15.0 and Transformers 5.22.2; exact dependency pins accompany the bundle.
 
-![Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.](../paper/figures/screenshot-03.png)
+![Figure 4. Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.](../paper/figures/screenshot-03.png)
 
-*Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.*
+*Figure 4. Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.*
 Before steering, the application obtains three directions per selected layer from twelve manually authored English contrast pairs, four per axis. For example, valence pairs contrast despair with joy, and intensity pairs contrast mild unease with overwhelming terror. Both members of every pair are processed under a continuation instruction. Thus calibration requires 24 forward passes.
 
 The capture site is the final chat-formatted prompt position immediately before the post-attention normalization that feeds the feed-forward sublayer. It is not a mean over all content tokens. For layer `l` and axis `j`, the mean contrast is:
@@ -75,9 +75,9 @@ There is no gradient training, score-supervised fitting, orthogonalization, or g
 
 Axis temperature `T` transforms each coordinate magnitude while retaining its sign:
 
-![Expanded numerical weights for the selected geometry, including the local calibration metadata.](../paper/figures/screenshot-04.png)
+![Figure 5. Expanded numerical weights for the selected geometry, including the local calibration metadata.](../paper/figures/screenshot-04.png)
 
-*Expanded numerical weights for the selected geometry, including the local calibration metadata.*
+*Figure 5. Expanded numerical weights for the selected geometry, including the local calibration metadata.*
 `q[j] = sign(c[j]) * abs(c[j])^(1/T)`.
 
 The implementation ranks axes by magnitude, retains at most axis top-k, and stops adding axes once cumulative magnitude reaches the configured mass fraction. Surviving coefficients are not renormalized. These operations act on three emotional coefficients, not on the model's vocabulary distribution.
@@ -181,29 +181,29 @@ The equations and implementation descriptions above are derived from the include
 
 Version-specific arXiv records are linked below. Years distinguish original release from later revision where applicable. These research references supplement the implementation and infrastructure sources elsewhere in this report. Reusable citations: [BibTeX](references.bib). Verification details: [source-check notes](LITERATURE_SOURCES.md).
 
-**R1.** Alex Turner; Lisa Thiergart; David Udell; Gavin Leech; Ulisse Mini; Monte MacDiarmid. 2023. Activation Addition: Steering Language Models Without Optimization. [arXiv:2308.10248v1](https://arxiv.org/abs/2308.10248v1). Cited version dated 2023-08-20.
+1. Alex Turner; Lisa Thiergart; David Udell; Gavin Leech; Ulisse Mini; Monte MacDiarmid. 2023. Activation Addition: Steering Language Models Without Optimization. [arXiv:2308.10248v1](https://arxiv.org/abs/2308.10248v1). Cited version dated 2023-08-20.
 
-**R2.** Nina Panickssery; Nick Gabrieli; Julian Schulz; Meg Tong; Evan Hubinger; Alexander Matt Turner. 2024. Steering Llama 2 via Contrastive Activation Addition. [arXiv:2312.06681v4](https://arxiv.org/abs/2312.06681v4). Cited version dated 2024-07-05.
+2. Nina Panickssery; Nick Gabrieli; Julian Schulz; Meg Tong; Evan Hubinger; Alexander Matt Turner. 2024. Steering Llama 2 via Contrastive Activation Addition. [arXiv:2312.06681v4](https://arxiv.org/abs/2312.06681v4). Cited version dated 2024-07-05.
 
-**R3.** Xiuwen Wu; Hao Wang; Zhiang Yan; Xiaohan Tang; Pengfei Xu; Wai-Ting Siok; Ping Li; Jia-Hong Gao; Bingjiang Lyu; Lang Qin. 2025. AI shares emotion with humans across languages and cultures. [arXiv:2506.13978v1](https://arxiv.org/abs/2506.13978v1). Cited version dated 2025-06-11.
+3. Xiuwen Wu; Hao Wang; Zhiang Yan; Xiaohan Tang; Pengfei Xu; Wai-Ting Siok; Ping Li; Jia-Hong Gao; Bingjiang Lyu; Lang Qin. 2025. AI shares emotion with humans across languages and cultures. [arXiv:2506.13978v1](https://arxiv.org/abs/2506.13978v1). Cited version dated 2025-06-11.
 
-**R4.** Diaoulé Diallo; Katharina Dworatzyk; Sophie Jentzsch; Peer Schütt; Sabine Theis; Tobias Hecking. 2026. The Effectiveness of Style Vectors for Steering Large Language Models: A Human Evaluation. [arXiv:2601.21505v1](https://arxiv.org/abs/2601.21505v1). Cited version dated 2026-01-29. ArXiv record lists IEEE Access 13 (2025), 191443-191457; DOI: 10.1109/ACCESS.2025.3628500.
+4. Diaoulé Diallo; Katharina Dworatzyk; Sophie Jentzsch; Peer Schütt; Sabine Theis; Tobias Hecking. 2026. The Effectiveness of Style Vectors for Steering Large Language Models: A Human Evaluation. [arXiv:2601.21505v1](https://arxiv.org/abs/2601.21505v1). Cited version dated 2026-01-29. ArXiv record lists IEEE Access 13 (2025), 191443-191457; DOI: 10.1109/ACCESS.2025.3628500.
 
-**R5.** Moran Sun; Tianlin Li; Yuwei Zheng; Zhenhong Zhou; Aishan Liu; Xianglong Liu; Yang Liu. 2026. How Emotion Shapes the Behavior of LLMs and Agents: A Mechanistic Study. [arXiv:2604.00005v1](https://arxiv.org/abs/2604.00005v1). Cited version dated 2026-03-09.
+5. Moran Sun; Tianlin Li; Yuwei Zheng; Zhenhong Zhou; Aishan Liu; Xianglong Liu; Yang Liu. 2026. How Emotion Shapes the Behavior of LLMs and Agents: A Mechanistic Study. [arXiv:2604.00005v1](https://arxiv.org/abs/2604.00005v1). Cited version dated 2026-03-09.
 
-**R6.** Jihoon Jeong. 2026. Extracting and Steering Emotion Representations in Small Language Models: A Methodological Comparison. [arXiv:2604.04064v1](https://arxiv.org/abs/2604.04064v1). Cited version dated 2026-04-05.
+6. Jihoon Jeong. 2026. Extracting and Steering Emotion Representations in Small Language Models: A Methodological Comparison. [arXiv:2604.04064v1](https://arxiv.org/abs/2604.04064v1). Cited version dated 2026-04-05.
 
-**R7.** Bangzhao Shu; Arinjay Singh; Mai ElSherief. 2026. Why Are Some Emotions Harder for LLMs? Uncovering the Causal Mechanisms of Emotion Inference via Sparse Autoencoders. [arXiv:2604.25866v3](https://arxiv.org/abs/2604.25866v3). Cited version dated 2026-10-24.
+7. Bangzhao Shu; Arinjay Singh; Mai ElSherief. 2026. Why Are Some Emotions Harder for LLMs? Uncovering the Causal Mechanisms of Emotion Inference via Sparse Autoencoders. [arXiv:2604.25866v3](https://arxiv.org/abs/2604.25866v3). Cited version dated 2026-10-24.
 
-**R8.** Yibo Wang; Jisheng Dang; Bimei Wang; Yitao Wu; Wencan Zhang; Hong Peng; Jizhao Liu; Bin Hu; Qi Tian; Tat-Seng Chua. 2026. Read-Best Is Not Steer-Best: A Probing--Steering Layer Dissociation in Omni-Modal Large Language Models. [arXiv:2609.22135v1](https://arxiv.org/abs/2609.22135v1). Cited version dated 2026-08-24.
+8. Yibo Wang; Jisheng Dang; Bimei Wang; Yitao Wu; Wencan Zhang; Hong Peng; Jizhao Liu; Bin Hu; Qi Tian; Tat-Seng Chua. 2026. Read-Best Is Not Steer-Best: A Probing--Steering Layer Dissociation in Omni-Modal Large Language Models. [arXiv:2609.22135v1](https://arxiv.org/abs/2609.22135v1). Cited version dated 2026-08-24.
 
-**R9.** JuHeon Ha; Byounghan Lee; Yunseo Choi; Kyung-Ah Sohn. 2026. Empathy Is Steerable but Multi-Axial: Mechanism Geometry and Persona Effects in LLMs. [arXiv:2609.15654v1](https://arxiv.org/abs/2609.15654v1). Cited version dated 2026-09-14. ArXiv comments report acceptance to the EMNLP 2026 main conference.
+9. JuHeon Ha; Byounghan Lee; Yunseo Choi; Kyung-Ah Sohn. 2026. Empathy Is Steerable but Multi-Axial: Mechanism Geometry and Persona Effects in LLMs. [arXiv:2609.15654v1](https://arxiv.org/abs/2609.15654v1). Cited version dated 2026-09-14. ArXiv comments report acceptance to the EMNLP 2026 main conference.
 
-**R10.** Sasha Cui; Zhongren Chen. 2025. Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models. [arXiv:2509.22739v3](https://arxiv.org/abs/2509.22739v3). Cited version dated 2026-05-15.
+10. Sasha Cui; Zhongren Chen. 2025. Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models. [arXiv:2509.22739v3](https://arxiv.org/abs/2509.22739v3). Cited version dated 2026-05-15.
 
-**R11.** Alex Ferrando; Xavier Suau; Jordi Gonzàlez; Pau Rodriguez. 2025. Dynamically Scaled Activation Steering. [arXiv:2512.03661v2](https://arxiv.org/abs/2512.03661v2). Cited version dated 2026-07-30.
+11. Alex Ferrando; Xavier Suau; Jordi Gonzàlez; Pau Rodriguez. 2025. Dynamically Scaled Activation Steering. [arXiv:2512.03661v2](https://arxiv.org/abs/2512.03661v2). Cited version dated 2026-07-30.
 
-**R12.** Haotian Yang; Huikang Jiang; Yucheng Wu; Wen-Jie Jiang; Chenpeng Wang; Yibin Lou; Liangming Pan. 2026. Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods. [arXiv:2610.07722v1](https://arxiv.org/abs/2610.07722v1). Cited version dated 2026-10-06.
+12. Haotian Yang; Huikang Jiang; Yucheng Wu; Wen-Jie Jiang; Chenpeng Wang; Yibin Lou; Liangming Pan. 2026. Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods. [arXiv:2610.07722v1](https://arxiv.org/abs/2610.07722v1). Cited version dated 2026-10-06.
 
 ## Suggested Citation
 

paper/AFFECTOR-source.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6ba7464f80ea5466e17ae6d7c7135cb579ab90c7fec9c9564c82fa0f928436cd
-size 4207675
+oid sha256:0090b72a4a5bc30bd103df9c5fc7b9671a9569e7f20174b539cec8c2f4ec3ace
+size 4207665

paper/PAPER-SHA256SUMS.txt

@@ -2,8 +2,8 @@
 d14edda139a36c0bac8b56d02e5cda63ab7558c6db1424f1087189a9429bfaf0  LITERATURE_SOURCES.md
 633baa667a0dc91028ef23e8483a1d57a6a7c32c89ffc3515441246cbad357b9  AFFECTOR.tex
 9257f38423bd7e496f1a0916c0fbaefbfb32ab544a684ca9a411508201f0535e  references.bib
-0f07bb640c2f3d79fe21930a2cd3b12f013cea815c7e697394a7cfc29b69ea15  source.md
-0f07bb640c2f3d79fe21930a2cd3b12f013cea815c7e697394a7cfc29b69ea15  TECHNICAL_PAPER.md
+be5b286160b97b1bc5af382da1a5ccdea50743f53ece5acaaa532c44767378cd  source.md
+be5b286160b97b1bc5af382da1a5ccdea50743f53ece5acaaa532c44767378cd  TECHNICAL_PAPER.md
 3125eeed5e35b5e01506ebe58adfd229b434521f8c1d02b62eee45f35c2a6e65  AFFECTOR.pdf
 d78577482f74cba4163db18d88af82786c923d72cd09607bde62cb747c645c5d  AFFECTOR.bbl
 eed8b979a5a299b42e925d9027332e7df876400dcef046f2cfe26beef91d1189  render_figures.py
@@ -24,4 +24,4 @@ a6a497a38f4ba9de8088f410535bee13dd862d1a97f4175d205df768e1d2c1c9  figures/illust
 8d267fd6d50e2146279f1f33e07f88a3cbf76d023dd9ee36b29e390ac0473cd1  figures/screenshot-02.png
 759dabd81a2ce67f01a3ec96a01b014468402baee5774792fccd5b7fe53af80a  figures/screenshot-03.png
 ff1ace53310f4f8ea57dedb7a202edd7404886091ed34b1bcc70cbb6c93b2779  figures/screenshot-01.png
-6ba7464f80ea5466e17ae6d7c7135cb579ab90c7fec9c9564c82fa0f928436cd  AFFECTOR-source.zip
+0090b72a4a5bc30bd103df9c5fc7b9671a9569e7f20174b539cec8c2f4ec3ace  AFFECTOR-source.zip

paper/TECHNICAL_PAPER.md

@@ -17,9 +17,9 @@ A document conveys more than propositions. Its lexical choices, rhythm, imagery,
 
 Activation steering offers a way to connect this target to the process of generation. Turner and colleagues construct intervention vectors from contrasting prompts, while Panickssery and colleagues average activation differences across positive and negative behavioral examples ([Turner et al., 2023](https://arxiv.org/abs/2308.10248v1); [Panickssery et al., 2024](https://arxiv.org/abs/2312.06681v4)). Building on this contrastive approach, AFFECTOR maps emotional coordinates into changes to a frozen model's internal states during local inference. The application also provides an API route, in which the coordinates are expressed as instructions to a hosted model. Together, these routes make it possible to investigate numerical intervention alongside prompt-based control, although their settings and outcomes are not directly interchangeable.
 
-![Document workspace with an example input, rewrite controls, and an empty output pane before generation.](figures/screenshot-01.png)
+![Figure 1. Document workspace with an example input, rewrite controls, and an empty output pane before generation.](figures/screenshot-01.png)
 
-*Document workspace with an example input, rewrite controls, and an empty output pane before generation.*
+*Figure 1. Document workspace with an example input, rewrite controls, and an empty output pane before generation.*
 
 For either route, the usefulness of the coordinates depends on their relationship to human interpretations of the output. Work by Wu and colleagues connects model emotion spaces with human valence/arousal ratings and reports concept-based steering across linguistic-cultural settings ([Wu et al., 2025](https://arxiv.org/abs/2506.13978v1)). Diallo and colleagues extend this evaluative perspective by examining emotional intensity and text quality with human raters, finding effects that vary with the emotion and steering strength ([Diallo et al., 2026](https://arxiv.org/abs/2601.21505v1)). These findings motivate assessing emotional movement and readable expression separately. They also clarify what remains to be established for AFFECTOR: neither its inward/outward axis nor the correspondence between its RMS-scaled controls and a reader's emotional judgment has been validated by those studies.
 
@@ -33,9 +33,9 @@ Taken together, these considerations define the scope of the present report: the
 
 A geometry consists of ordered points `(t, v, d, i)`. Position `t` is in `[0,1]`; valence `v` and direction `d` are in `[-1,1]`; intensity `i` is in `[0,1]`. Negative valence denotes unpleasant affect and positive valence pleasant affect. Negative direction denotes self-focused or inward-directed emotion; positive direction denotes outward focus. Direction is independent of valence: anger toward oneself can be inward and sadness for another person outward. Intensity describes strength of affect, not pleasantness.
 
-![Geometry editor showing the selected three-point trajectory and its steering controls.](figures/screenshot-02.png)
+![Figure 2. Geometry editor showing the selected three-point trajectory and its steering controls.](figures/screenshot-02.png)
 
-*Geometry editor showing the selected three-point trajectory and its steering controls.*
+*Figure 2. Geometry editor showing the selected three-point trajectory and its steering controls.*
 For intervention, intensity is centered:
 
 `c(t) = (v(t), d(t), 2i(t) - 1)`.
@@ -46,17 +46,17 @@ The target document is divided into chunks of at most 700 characters, preferring
 
 This normalized alignment compresses or stretches a source trajectory to the target length. It samples only as many positions as there are chunks. It does not preserve all events or oscillations of a much longer curve. Changes to segmentation can change emotional targets even when the same geometry is selected.
 
-![One continuous path, a finite set of targets](figures/illustration-01.png)
+![Figure 3. One continuous path, a finite set of targets](figures/illustration-01.png)
 
-*Illustration 1. Normalized alignment schematic. Target chunks sample a continuous source path at original character midpoints; a short target can miss narrow changes in a long source trajectory.*
+*Figure 3. Normalized alignment schematic. Target chunks sample a continuous source path at original character midpoints; a short target can miss narrow changes in a long source trajectory.*
 
 ## Frozen model and contrastive calibration
 
 The local backend loads Qwen3.5-0.8B at revision `2fc06364715b967f1860aea9cf38778875588b17`. The implementation operates on its text stack of 24 blocks with hidden width 1,024. Base parameters are frozen; inference uses float32 and eager attention. Apple MPS is selected when available, otherwise CPU. The chat template disables thinking output. The package contains Python 3.14, PyTorch 2.15.0 and Transformers 5.22.2; exact dependency pins accompany the bundle.
 
-![Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.](figures/screenshot-03.png)
+![Figure 4. Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.](figures/screenshot-03.png)
 
-*Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.*
+*Figure 4. Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.*
 Before steering, the application obtains three directions per selected layer from twelve manually authored English contrast pairs, four per axis. For example, valence pairs contrast despair with joy, and intensity pairs contrast mild unease with overwhelming terror. Both members of every pair are processed under a continuation instruction. Thus calibration requires 24 forward passes.
 
 The capture site is the final chat-formatted prompt position immediately before the post-attention normalization that feeds the feed-forward sublayer. It is not a mean over all content tokens. For layer `l` and axis `j`, the mean contrast is:
@@ -75,9 +75,9 @@ There is no gradient training, score-supervised fitting, orthogonalization, or g
 
 Axis temperature `T` transforms each coordinate magnitude while retaining its sign:
 
-![Expanded numerical weights for the selected geometry, including the local calibration metadata.](figures/screenshot-04.png)
+![Figure 5. Expanded numerical weights for the selected geometry, including the local calibration metadata.](figures/screenshot-04.png)
 
-*Expanded numerical weights for the selected geometry, including the local calibration metadata.*
+*Figure 5. Expanded numerical weights for the selected geometry, including the local calibration metadata.*
 `q[j] = sign(c[j]) * abs(c[j])^(1/T)`.
 
 The implementation ranks axes by magnitude, retains at most axis top-k, and stops adding axes once cumulative magnitude reaches the configured mass fraction. Surviving coefficients are not renormalized. These operations act on three emotional coefficients, not on the model's vocabulary distribution.
@@ -181,29 +181,29 @@ The equations and implementation descriptions above are derived from the include
 
 Version-specific arXiv records are linked below. Years distinguish original release from later revision where applicable. These research references supplement the implementation and infrastructure sources elsewhere in this report. Reusable citations: [BibTeX](references.bib). Verification details: [source-check notes](LITERATURE_SOURCES.md).
 
-**R1.** Alex Turner; Lisa Thiergart; David Udell; Gavin Leech; Ulisse Mini; Monte MacDiarmid. 2023. Activation Addition: Steering Language Models Without Optimization. [arXiv:2308.10248v1](https://arxiv.org/abs/2308.10248v1). Cited version dated 2023-08-20.
+1. Alex Turner; Lisa Thiergart; David Udell; Gavin Leech; Ulisse Mini; Monte MacDiarmid. 2023. Activation Addition: Steering Language Models Without Optimization. [arXiv:2308.10248v1](https://arxiv.org/abs/2308.10248v1). Cited version dated 2023-08-20.
 
-**R2.** Nina Panickssery; Nick Gabrieli; Julian Schulz; Meg Tong; Evan Hubinger; Alexander Matt Turner. 2024. Steering Llama 2 via Contrastive Activation Addition. [arXiv:2312.06681v4](https://arxiv.org/abs/2312.06681v4). Cited version dated 2024-07-05.
+2. Nina Panickssery; Nick Gabrieli; Julian Schulz; Meg Tong; Evan Hubinger; Alexander Matt Turner. 2024. Steering Llama 2 via Contrastive Activation Addition. [arXiv:2312.06681v4](https://arxiv.org/abs/2312.06681v4). Cited version dated 2024-07-05.
 
-**R3.** Xiuwen Wu; Hao Wang; Zhiang Yan; Xiaohan Tang; Pengfei Xu; Wai-Ting Siok; Ping Li; Jia-Hong Gao; Bingjiang Lyu; Lang Qin. 2025. AI shares emotion with humans across languages and cultures. [arXiv:2506.13978v1](https://arxiv.org/abs/2506.13978v1). Cited version dated 2025-06-11.
+3. Xiuwen Wu; Hao Wang; Zhiang Yan; Xiaohan Tang; Pengfei Xu; Wai-Ting Siok; Ping Li; Jia-Hong Gao; Bingjiang Lyu; Lang Qin. 2025. AI shares emotion with humans across languages and cultures. [arXiv:2506.13978v1](https://arxiv.org/abs/2506.13978v1). Cited version dated 2025-06-11.
 
-**R4.** Diaoulé Diallo; Katharina Dworatzyk; Sophie Jentzsch; Peer Schütt; Sabine Theis; Tobias Hecking. 2026. The Effectiveness of Style Vectors for Steering Large Language Models: A Human Evaluation. [arXiv:2601.21505v1](https://arxiv.org/abs/2601.21505v1). Cited version dated 2026-01-29. ArXiv record lists IEEE Access 13 (2025), 191443-191457; DOI: 10.1109/ACCESS.2025.3628500.
+4. Diaoulé Diallo; Katharina Dworatzyk; Sophie Jentzsch; Peer Schütt; Sabine Theis; Tobias Hecking. 2026. The Effectiveness of Style Vectors for Steering Large Language Models: A Human Evaluation. [arXiv:2601.21505v1](https://arxiv.org/abs/2601.21505v1). Cited version dated 2026-01-29. ArXiv record lists IEEE Access 13 (2025), 191443-191457; DOI: 10.1109/ACCESS.2025.3628500.
 
-**R5.** Moran Sun; Tianlin Li; Yuwei Zheng; Zhenhong Zhou; Aishan Liu; Xianglong Liu; Yang Liu. 2026. How Emotion Shapes the Behavior of LLMs and Agents: A Mechanistic Study. [arXiv:2604.00005v1](https://arxiv.org/abs/2604.00005v1). Cited version dated 2026-03-09.
+5. Moran Sun; Tianlin Li; Yuwei Zheng; Zhenhong Zhou; Aishan Liu; Xianglong Liu; Yang Liu. 2026. How Emotion Shapes the Behavior of LLMs and Agents: A Mechanistic Study. [arXiv:2604.00005v1](https://arxiv.org/abs/2604.00005v1). Cited version dated 2026-03-09.
 
-**R6.** Jihoon Jeong. 2026. Extracting and Steering Emotion Representations in Small Language Models: A Methodological Comparison. [arXiv:2604.04064v1](https://arxiv.org/abs/2604.04064v1). Cited version dated 2026-04-05.
+6. Jihoon Jeong. 2026. Extracting and Steering Emotion Representations in Small Language Models: A Methodological Comparison. [arXiv:2604.04064v1](https://arxiv.org/abs/2604.04064v1). Cited version dated 2026-04-05.
 
-**R7.** Bangzhao Shu; Arinjay Singh; Mai ElSherief. 2026. Why Are Some Emotions Harder for LLMs? Uncovering the Causal Mechanisms of Emotion Inference via Sparse Autoencoders. [arXiv:2604.25866v3](https://arxiv.org/abs/2604.25866v3). Cited version dated 2026-10-24.
+7. Bangzhao Shu; Arinjay Singh; Mai ElSherief. 2026. Why Are Some Emotions Harder for LLMs? Uncovering the Causal Mechanisms of Emotion Inference via Sparse Autoencoders. [arXiv:2604.25866v3](https://arxiv.org/abs/2604.25866v3). Cited version dated 2026-10-24.
 
-**R8.** Yibo Wang; Jisheng Dang; Bimei Wang; Yitao Wu; Wencan Zhang; Hong Peng; Jizhao Liu; Bin Hu; Qi Tian; Tat-Seng Chua. 2026. Read-Best Is Not Steer-Best: A Probing--Steering Layer Dissociation in Omni-Modal Large Language Models. [arXiv:2609.22135v1](https://arxiv.org/abs/2609.22135v1). Cited version dated 2026-08-24.
+8. Yibo Wang; Jisheng Dang; Bimei Wang; Yitao Wu; Wencan Zhang; Hong Peng; Jizhao Liu; Bin Hu; Qi Tian; Tat-Seng Chua. 2026. Read-Best Is Not Steer-Best: A Probing--Steering Layer Dissociation in Omni-Modal Large Language Models. [arXiv:2609.22135v1](https://arxiv.org/abs/2609.22135v1). Cited version dated 2026-08-24.
 
-**R9.** JuHeon Ha; Byounghan Lee; Yunseo Choi; Kyung-Ah Sohn. 2026. Empathy Is Steerable but Multi-Axial: Mechanism Geometry and Persona Effects in LLMs. [arXiv:2609.15654v1](https://arxiv.org/abs/2609.15654v1). Cited version dated 2026-09-14. ArXiv comments report acceptance to the EMNLP 2026 main conference.
+9. JuHeon Ha; Byounghan Lee; Yunseo Choi; Kyung-Ah Sohn. 2026. Empathy Is Steerable but Multi-Axial: Mechanism Geometry and Persona Effects in LLMs. [arXiv:2609.15654v1](https://arxiv.org/abs/2609.15654v1). Cited version dated 2026-09-14. ArXiv comments report acceptance to the EMNLP 2026 main conference.
 
-**R10.** Sasha Cui; Zhongren Chen. 2025. Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models. [arXiv:2509.22739v3](https://arxiv.org/abs/2509.22739v3). Cited version dated 2026-05-15.
+10. Sasha Cui; Zhongren Chen. 2025. Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models. [arXiv:2509.22739v3](https://arxiv.org/abs/2509.22739v3). Cited version dated 2026-05-15.
 
-**R11.** Alex Ferrando; Xavier Suau; Jordi Gonzàlez; Pau Rodriguez. 2025. Dynamically Scaled Activation Steering. [arXiv:2512.03661v2](https://arxiv.org/abs/2512.03661v2). Cited version dated 2026-07-30.
+11. Alex Ferrando; Xavier Suau; Jordi Gonzàlez; Pau Rodriguez. 2025. Dynamically Scaled Activation Steering. [arXiv:2512.03661v2](https://arxiv.org/abs/2512.03661v2). Cited version dated 2026-07-30.
 
-**R12.** Haotian Yang; Huikang Jiang; Yucheng Wu; Wen-Jie Jiang; Chenpeng Wang; Yibin Lou; Liangming Pan. 2026. Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods. [arXiv:2610.07722v1](https://arxiv.org/abs/2610.07722v1). Cited version dated 2026-10-06.
+12. Haotian Yang; Huikang Jiang; Yucheng Wu; Wen-Jie Jiang; Chenpeng Wang; Yibin Lou; Liangming Pan. 2026. Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods. [arXiv:2610.07722v1](https://arxiv.org/abs/2610.07722v1). Cited version dated 2026-10-06.
 
 ## Suggested Citation
 

paper/source.md

@@ -17,9 +17,9 @@ A document conveys more than propositions. Its lexical choices, rhythm, imagery,
 
 Activation steering offers a way to connect this target to the process of generation. Turner and colleagues construct intervention vectors from contrasting prompts, while Panickssery and colleagues average activation differences across positive and negative behavioral examples ([Turner et al., 2023](https://arxiv.org/abs/2308.10248v1); [Panickssery et al., 2024](https://arxiv.org/abs/2312.06681v4)). Building on this contrastive approach, AFFECTOR maps emotional coordinates into changes to a frozen model's internal states during local inference. The application also provides an API route, in which the coordinates are expressed as instructions to a hosted model. Together, these routes make it possible to investigate numerical intervention alongside prompt-based control, although their settings and outcomes are not directly interchangeable.
 
-![Document workspace with an example input, rewrite controls, and an empty output pane before generation.](figures/screenshot-01.png)
+![Figure 1. Document workspace with an example input, rewrite controls, and an empty output pane before generation.](figures/screenshot-01.png)
 
-*Document workspace with an example input, rewrite controls, and an empty output pane before generation.*
+*Figure 1. Document workspace with an example input, rewrite controls, and an empty output pane before generation.*
 
 For either route, the usefulness of the coordinates depends on their relationship to human interpretations of the output. Work by Wu and colleagues connects model emotion spaces with human valence/arousal ratings and reports concept-based steering across linguistic-cultural settings ([Wu et al., 2025](https://arxiv.org/abs/2506.13978v1)). Diallo and colleagues extend this evaluative perspective by examining emotional intensity and text quality with human raters, finding effects that vary with the emotion and steering strength ([Diallo et al., 2026](https://arxiv.org/abs/2601.21505v1)). These findings motivate assessing emotional movement and readable expression separately. They also clarify what remains to be established for AFFECTOR: neither its inward/outward axis nor the correspondence between its RMS-scaled controls and a reader's emotional judgment has been validated by those studies.
 
@@ -33,9 +33,9 @@ Taken together, these considerations define the scope of the present report: the
 
 A geometry consists of ordered points `(t, v, d, i)`. Position `t` is in `[0,1]`; valence `v` and direction `d` are in `[-1,1]`; intensity `i` is in `[0,1]`. Negative valence denotes unpleasant affect and positive valence pleasant affect. Negative direction denotes self-focused or inward-directed emotion; positive direction denotes outward focus. Direction is independent of valence: anger toward oneself can be inward and sadness for another person outward. Intensity describes strength of affect, not pleasantness.
 
-![Geometry editor showing the selected three-point trajectory and its steering controls.](figures/screenshot-02.png)
+![Figure 2. Geometry editor showing the selected three-point trajectory and its steering controls.](figures/screenshot-02.png)
 
-*Geometry editor showing the selected three-point trajectory and its steering controls.*
+*Figure 2. Geometry editor showing the selected three-point trajectory and its steering controls.*
 For intervention, intensity is centered:
 
 `c(t) = (v(t), d(t), 2i(t) - 1)`.
@@ -46,17 +46,17 @@ The target document is divided into chunks of at most 700 characters, preferring
 
 This normalized alignment compresses or stretches a source trajectory to the target length. It samples only as many positions as there are chunks. It does not preserve all events or oscillations of a much longer curve. Changes to segmentation can change emotional targets even when the same geometry is selected.
 
-![One continuous path, a finite set of targets](figures/illustration-01.png)
+![Figure 3. One continuous path, a finite set of targets](figures/illustration-01.png)
 
-*Illustration 1. Normalized alignment schematic. Target chunks sample a continuous source path at original character midpoints; a short target can miss narrow changes in a long source trajectory.*
+*Figure 3. Normalized alignment schematic. Target chunks sample a continuous source path at original character midpoints; a short target can miss narrow changes in a long source trajectory.*
 
 ## Frozen model and contrastive calibration
 
 The local backend loads Qwen3.5-0.8B at revision `2fc06364715b967f1860aea9cf38778875588b17`. The implementation operates on its text stack of 24 blocks with hidden width 1,024. Base parameters are frozen; inference uses float32 and eager attention. Apple MPS is selected when available, otherwise CPU. The chat template disables thinking output. The package contains Python 3.14, PyTorch 2.15.0 and Transformers 5.22.2; exact dependency pins accompany the bundle.
 
-![Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.](figures/screenshot-03.png)
+![Figure 4. Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.](figures/screenshot-03.png)
 
-*Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.*
+*Figure 4. Source-analysis dialog for generating a geometry and optionally calibrating a local activation basis.*
 Before steering, the application obtains three directions per selected layer from twelve manually authored English contrast pairs, four per axis. For example, valence pairs contrast despair with joy, and intensity pairs contrast mild unease with overwhelming terror. Both members of every pair are processed under a continuation instruction. Thus calibration requires 24 forward passes.
 
 The capture site is the final chat-formatted prompt position immediately before the post-attention normalization that feeds the feed-forward sublayer. It is not a mean over all content tokens. For layer `l` and axis `j`, the mean contrast is:
@@ -75,9 +75,9 @@ There is no gradient training, score-supervised fitting, orthogonalization, or g
 
 Axis temperature `T` transforms each coordinate magnitude while retaining its sign:
 
-![Expanded numerical weights for the selected geometry, including the local calibration metadata.](figures/screenshot-04.png)
+![Figure 5. Expanded numerical weights for the selected geometry, including the local calibration metadata.](figures/screenshot-04.png)
 
-*Expanded numerical weights for the selected geometry, including the local calibration metadata.*
+*Figure 5. Expanded numerical weights for the selected geometry, including the local calibration metadata.*
 `q[j] = sign(c[j]) * abs(c[j])^(1/T)`.
 
 The implementation ranks axes by magnitude, retains at most axis top-k, and stops adding axes once cumulative magnitude reaches the configured mass fraction. Surviving coefficients are not renormalized. These operations act on three emotional coefficients, not on the model's vocabulary distribution.
@@ -181,29 +181,29 @@ The equations and implementation descriptions above are derived from the include
 
 Version-specific arXiv records are linked below. Years distinguish original release from later revision where applicable. These research references supplement the implementation and infrastructure sources elsewhere in this report. Reusable citations: [BibTeX](references.bib). Verification details: [source-check notes](LITERATURE_SOURCES.md).
 
-**R1.** Alex Turner; Lisa Thiergart; David Udell; Gavin Leech; Ulisse Mini; Monte MacDiarmid. 2023. Activation Addition: Steering Language Models Without Optimization. [arXiv:2308.10248v1](https://arxiv.org/abs/2308.10248v1). Cited version dated 2023-08-20.
+1. Alex Turner; Lisa Thiergart; David Udell; Gavin Leech; Ulisse Mini; Monte MacDiarmid. 2023. Activation Addition: Steering Language Models Without Optimization. [arXiv:2308.10248v1](https://arxiv.org/abs/2308.10248v1). Cited version dated 2023-08-20.
 
-**R2.** Nina Panickssery; Nick Gabrieli; Julian Schulz; Meg Tong; Evan Hubinger; Alexander Matt Turner. 2024. Steering Llama 2 via Contrastive Activation Addition. [arXiv:2312.06681v4](https://arxiv.org/abs/2312.06681v4). Cited version dated 2024-07-05.
+2. Nina Panickssery; Nick Gabrieli; Julian Schulz; Meg Tong; Evan Hubinger; Alexander Matt Turner. 2024. Steering Llama 2 via Contrastive Activation Addition. [arXiv:2312.06681v4](https://arxiv.org/abs/2312.06681v4). Cited version dated 2024-07-05.
 
-**R3.** Xiuwen Wu; Hao Wang; Zhiang Yan; Xiaohan Tang; Pengfei Xu; Wai-Ting Siok; Ping Li; Jia-Hong Gao; Bingjiang Lyu; Lang Qin. 2025. AI shares emotion with humans across languages and cultures. [arXiv:2506.13978v1](https://arxiv.org/abs/2506.13978v1). Cited version dated 2025-06-11.
+3. Xiuwen Wu; Hao Wang; Zhiang Yan; Xiaohan Tang; Pengfei Xu; Wai-Ting Siok; Ping Li; Jia-Hong Gao; Bingjiang Lyu; Lang Qin. 2025. AI shares emotion with humans across languages and cultures. [arXiv:2506.13978v1](https://arxiv.org/abs/2506.13978v1). Cited version dated 2025-06-11.
 
-**R4.** Diaoulé Diallo; Katharina Dworatzyk; Sophie Jentzsch; Peer Schütt; Sabine Theis; Tobias Hecking. 2026. The Effectiveness of Style Vectors for Steering Large Language Models: A Human Evaluation. [arXiv:2601.21505v1](https://arxiv.org/abs/2601.21505v1). Cited version dated 2026-01-29. ArXiv record lists IEEE Access 13 (2025), 191443-191457; DOI: 10.1109/ACCESS.2025.3628500.
+4. Diaoulé Diallo; Katharina Dworatzyk; Sophie Jentzsch; Peer Schütt; Sabine Theis; Tobias Hecking. 2026. The Effectiveness of Style Vectors for Steering Large Language Models: A Human Evaluation. [arXiv:2601.21505v1](https://arxiv.org/abs/2601.21505v1). Cited version dated 2026-01-29. ArXiv record lists IEEE Access 13 (2025), 191443-191457; DOI: 10.1109/ACCESS.2025.3628500.
 
-**R5.** Moran Sun; Tianlin Li; Yuwei Zheng; Zhenhong Zhou; Aishan Liu; Xianglong Liu; Yang Liu. 2026. How Emotion Shapes the Behavior of LLMs and Agents: A Mechanistic Study. [arXiv:2604.00005v1](https://arxiv.org/abs/2604.00005v1). Cited version dated 2026-03-09.
+5. Moran Sun; Tianlin Li; Yuwei Zheng; Zhenhong Zhou; Aishan Liu; Xianglong Liu; Yang Liu. 2026. How Emotion Shapes the Behavior of LLMs and Agents: A Mechanistic Study. [arXiv:2604.00005v1](https://arxiv.org/abs/2604.00005v1). Cited version dated 2026-03-09.
 
-**R6.** Jihoon Jeong. 2026. Extracting and Steering Emotion Representations in Small Language Models: A Methodological Comparison. [arXiv:2604.04064v1](https://arxiv.org/abs/2604.04064v1). Cited version dated 2026-04-05.
+6. Jihoon Jeong. 2026. Extracting and Steering Emotion Representations in Small Language Models: A Methodological Comparison. [arXiv:2604.04064v1](https://arxiv.org/abs/2604.04064v1). Cited version dated 2026-04-05.
 
-**R7.** Bangzhao Shu; Arinjay Singh; Mai ElSherief. 2026. Why Are Some Emotions Harder for LLMs? Uncovering the Causal Mechanisms of Emotion Inference via Sparse Autoencoders. [arXiv:2604.25866v3](https://arxiv.org/abs/2604.25866v3). Cited version dated 2026-10-24.
+7. Bangzhao Shu; Arinjay Singh; Mai ElSherief. 2026. Why Are Some Emotions Harder for LLMs? Uncovering the Causal Mechanisms of Emotion Inference via Sparse Autoencoders. [arXiv:2604.25866v3](https://arxiv.org/abs/2604.25866v3). Cited version dated 2026-10-24.
 
-**R8.** Yibo Wang; Jisheng Dang; Bimei Wang; Yitao Wu; Wencan Zhang; Hong Peng; Jizhao Liu; Bin Hu; Qi Tian; Tat-Seng Chua. 2026. Read-Best Is Not Steer-Best: A Probing--Steering Layer Dissociation in Omni-Modal Large Language Models. [arXiv:2609.22135v1](https://arxiv.org/abs/2609.22135v1). Cited version dated 2026-08-24.
+8. Yibo Wang; Jisheng Dang; Bimei Wang; Yitao Wu; Wencan Zhang; Hong Peng; Jizhao Liu; Bin Hu; Qi Tian; Tat-Seng Chua. 2026. Read-Best Is Not Steer-Best: A Probing--Steering Layer Dissociation in Omni-Modal Large Language Models. [arXiv:2609.22135v1](https://arxiv.org/abs/2609.22135v1). Cited version dated 2026-08-24.
 
-**R9.** JuHeon Ha; Byounghan Lee; Yunseo Choi; Kyung-Ah Sohn. 2026. Empathy Is Steerable but Multi-Axial: Mechanism Geometry and Persona Effects in LLMs. [arXiv:2609.15654v1](https://arxiv.org/abs/2609.15654v1). Cited version dated 2026-09-14. ArXiv comments report acceptance to the EMNLP 2026 main conference.
+9. JuHeon Ha; Byounghan Lee; Yunseo Choi; Kyung-Ah Sohn. 2026. Empathy Is Steerable but Multi-Axial: Mechanism Geometry and Persona Effects in LLMs. [arXiv:2609.15654v1](https://arxiv.org/abs/2609.15654v1). Cited version dated 2026-09-14. ArXiv comments report acceptance to the EMNLP 2026 main conference.
 
-**R10.** Sasha Cui; Zhongren Chen. 2025. Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models. [arXiv:2509.22739v3](https://arxiv.org/abs/2509.22739v3). Cited version dated 2026-05-15.
+10. Sasha Cui; Zhongren Chen. 2025. Painless Activation Steering: An Automated, Lightweight Approach for Post-Training Large Language Models. [arXiv:2509.22739v3](https://arxiv.org/abs/2509.22739v3). Cited version dated 2026-05-15.
 
-**R11.** Alex Ferrando; Xavier Suau; Jordi Gonzàlez; Pau Rodriguez. 2025. Dynamically Scaled Activation Steering. [arXiv:2512.03661v2](https://arxiv.org/abs/2512.03661v2). Cited version dated 2026-07-30.
+11. Alex Ferrando; Xavier Suau; Jordi Gonzàlez; Pau Rodriguez. 2025. Dynamically Scaled Activation Steering. [arXiv:2512.03661v2](https://arxiv.org/abs/2512.03661v2). Cited version dated 2026-07-30.
 
-**R12.** Haotian Yang; Huikang Jiang; Yucheng Wu; Wen-Jie Jiang; Chenpeng Wang; Yibin Lou; Liangming Pan. 2026. Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods. [arXiv:2610.07722v1](https://arxiv.org/abs/2610.07722v1). Cited version dated 2026-10-06.
+12. Haotian Yang; Huikang Jiang; Yucheng Wu; Wen-Jie Jiang; Chenpeng Wang; Yibin Lou; Liangming Pan. 2026. Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods. [arXiv:2610.07722v1](https://arxiv.org/abs/2610.07722v1). Cited version dated 2026-10-06.
 
 ## Suggested Citation