From 4011ba9b2ab4f408e8c8ec54be79a3fec473bea0 Mon Sep 17 00:00:00 2001
From: Xin Wang <wangx@macmini.com>
Date: Fri, 28 Aug 2026 11:15:09 +0800
Subject: [PATCH] Record EasyCom encoder-only results and horizon analysis

---
 README.md                                     |    6 +
 docs/easycom_encoder_only.md                  |    8 +
 .../README.md                                 |  242 +++
 .../raw_encoder_audio_seed0.txt               |   40 +
 .../raw_encoder_av_seed0.txt                  |   39 +
 .../raw_encoder_visual_seed0.txt              |   40 +
 .../raw_previous_thinker_audio_seed0.txt      |   36 +
 .../raw_previous_thinker_av_seed0.txt         |   37 +
 .../reported_causal_summary.txt               |   15 +
 .../summary.json                              | 1318 +++++++++++++++++
 10 files changed, 1781 insertions(+)
 create mode 100644 results/easycom_encoder_only_1s_4ep_2026-08-28/README.md
 create mode 100644 results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_audio_seed0.txt
 create mode 100644 results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_av_seed0.txt
 create mode 100644 results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_visual_seed0.txt
 create mode 100644 results/easycom_encoder_only_1s_4ep_2026-08-28/raw_previous_thinker_audio_seed0.txt
 create mode 100644 results/easycom_encoder_only_1s_4ep_2026-08-28/raw_previous_thinker_av_seed0.txt
 create mode 100644 results/easycom_encoder_only_1s_4ep_2026-08-28/reported_causal_summary.txt
 create mode 100644 results/easycom_encoder_only_1s_4ep_2026-08-28/summary.json

diff --git a/README.md b/README.md
index 85d59b8..cb12c46 100644
--- a/README.md
+++ b/README.md
@@ -16,6 +16,12 @@ For **EasyCom encoder-only A / V / AV** with no Thinker states, start with
 They reuse the verified local sidecars, train on Session 4–12, test on 3/2/1,
 and keep four epochs with no dev set.
 
+The user-reported EasyCom encoder-only seed-0 results and complete supplied
+horizon curves are archived in
+[the 2026-08-28 result record](results/easycom_encoder_only_1s_4ep_2026-08-28/README.md).
+It keeps causal-valid summaries separate from full-timeline curves and records
+the head-capacity/context caveats for comparisons with G+A and G+A+V.
+
 The main experiment is deliberately simple:
 
 ```text
diff --git a/docs/easycom_encoder_only.md b/docs/easycom_encoder_only.md
index 8f79ac4..7c90d2e 100644
--- a/docs/easycom_encoder_only.md
+++ b/docs/easycom_encoder_only.md
@@ -7,8 +7,16 @@
 固定训练 Session **4–12（升序，与原 B1 一致）**，测试 **3/2/1**；4 epochs、无 dev，
 评测最后一轮。沿用原来的 5Hz 行索引、十个 horizon、LSTR-MCE 和 AP 协议，不改变标签。
 
+已收到的 seed 0 结果与完整日志见
+[2026-08-28 实验记录](../results/easycom_encoder_only_1s_4ep_2026-08-28/README.md)。
+其中完整曲线为 `egospeak_full_timeline`，此前主表为 `causal_valid_only`，两者分开记录。
+
 ## 1. 确定已经验证过的局部特征目录
 
+用户已确认本次使用的实际目录是
+`/data1/yfzx/wangx/Datasets/easycom/omni_features/local_qwen_1s_encoded_tf4576`，
+不是早期默认的 `local_qwen_1s`；以下 checkpoint 读取方法可用于再次核对。
+
 在 L20 机器更新本次代码后执行。下面从你之前跑完的 `av_seed0.pth` 读取实际路径，
 避免猜错目录或混入 bbox 修正前的旧视觉特征；checkpoint 仅用于找路径，不作为新模型输入。
 
diff --git a/results/easycom_encoder_only_1s_4ep_2026-08-28/README.md b/results/easycom_encoder_only_1s_4ep_2026-08-28/README.md
new file mode 100644
index 0000000..b5358c3
--- /dev/null
+++ b/results/easycom_encoder_only_1s_4ep_2026-08-28/README.md
@@ -0,0 +1,242 @@
+# EasyCom encoder-only 1s：seed 0 完整曲线记录与分析
+
+记录日期：2026-08-28。来源是用户在 L20 远程机器运行后提供的控制台输出；
+本地没有重新训练或执行真实数据推理。本记录仍是**单种子初步结果**，标准差为 n/a，
+不是 0；不能据此报告统计显著性。
+
+## 1. 收到了什么，尚未收到什么
+
+- 本次三份附件：encoder A、V、AV 的 `egospeak_full_timeline` 十个 horizon 的三类 AP、mAP 和均值。
+- 此前两份附件：G+A、G+A+V 同协议的完整曲线；在本目录一并归档用于同协议比较。
+- 此前用户汇总表：七组 `causal_valid_only` 的标称 0.2s 三类 AP/mAP 和十-horizon 平均。
+- 本地没有收到原始 `metrics.json`、预测数组、checkpoint 内容或训练摘要 JSON。
+  附件中的 `training summary: ...pth.json` 只是路径，不是文件内容；没有训练曲线。
+- 本次附件在 full-timeline 均值之后结束，**没有包含 causal-valid 的完整曲线**。
+  不将前者改名为后者，也不根据均值补造未提供的 horizon。
+- 以下四轮训练、训练 split、窗口和 head 设置来自已约定命令与代码；
+  尚未从远程 checkpoint 元数据独立核验 epoch、seed、参数量或运行 commit。
+
+G = 全局 Thinker 特征；A = 近期混合音频编码器向量；V = 逐人局部视觉编码器向量的池化。
+旧 B1 中名为 `audio / visual / av` 的运行实际输入为 G+A / G+V / G+A+V，
+不是本次 encoder-only 的 A / V / AV。
+
+## 2. 实验配置与来源路径
+
+约定配置：
+
+- EasyCom：训练 Session 4–12 升序、无 dev、测试 Session 3/2/1。
+- 冻结 Qwen2.5-Omni-3B 编码器；本次实验名为 seed 0、4 epochs、最后一轮评测。
+- A/V 原生各 2048D；A 是最近 1s 混合音频的有效编码器 token pooling。
+- V 使用提供的人物框；2FPS 空间 token mean → 同帧逐人等权 mean → 最近 1s 有效帧 mean。
+  这不是全帧视觉，不是逐人 attention，也没有评测在线人脸检测器。
+- 每 200ms 一行，输入窗口按 `[max(0,t-1),t)` 约定，标签与原实验相同。
+- encoder-only 不输入 Thinker 状态；将选定原生向量和 validity 位拼接后输入 MLP。
+- 对照 G：8s open-stream-replay、2FPS、Thinker last4 stack + concat-MLP。
+- LSTR-MCE（三路 softmax），batch 512、lr 7e-5、weight decay 5e-5、
+  warmup 1.6 epochs、warmup factor 0.3、dropout 0.2、无 transition objective。
+- 局部提取版本按先前实验报告为 Transformers 4.57.6；新附件不是编码器提取日志，
+  没有再次核验所有 sidecar 的实际 config。
+
+用户已通过目录列表确认局部源特征在：
+
+```text
+/data1/yfzx/wangx/Datasets/easycom/omni_features/local_qwen_1s_encoded_tf4576
+```
+
+新导出目录为同一 `omni_features` 下的
+`encoder_only_audio_1s`、`encoder_only_visual_1s`、`encoder_only_av_1s`。
+这些是已有向量的重新打包，不是重新运行 Qwen。
+
+其他路径：
+
+- 项目：`/data1/yfzx/wangx/Code/icassp2027-experiment`
+- target：`/data1/yfzx/wangx/Datasets/easycom/easycom_target_frame`
+- 新 checkpoint：`checkpoints/easycom_encoder_only_1s/encoder_<modality>_seed0.pth`
+- 新评测：`results/easycom_encoder_only_1s/encoder_<modality>_seed0/metrics.json`
+- 旧 B1 checkpoint：`checkpoints/easycom_local_b1_encoded_tf4576/{audio,av}_seed0.pth`
+- 实验入口与命令：[EasyCom encoder-only 指南](../../docs/easycom_encoder_only.md)；
+  本地入口版本为 `321238a`，不是对远程运行版本的证明。
+
+## 3. 两种评测协议必须分开
+
+三个新运行及两个旧对照都打印了相同预测形状：
+Session_3 = (8400,3,10)、Session_2 = (7706,3,10)、Session_1 = (6942,3,10)，
+总计 23048 行。
+
+`egospeak_full_timeline` 每个 horizon 都评测 23048 行，
+包含每个 session 开头没有历史预测的 h 行，数组中这些分数初始化为零。
+`causal_valid_only` 排除这部分行，因此标称 0.2s 为 23045 行，
+依次减少，到 2.0s 为 23018 行。这里 h 指 1–10 的 horizon 索引。
+
+新日志和此前汇总值的小差异与这两种协议一致，不应当成新的训练结果或抄录误差：
+
+| 输入 | causal − full：0.2s mAP | causal − full：十-horizon 均值 |
+|---|---:|---:|
+| V | +0.00108 | +0.00646 |
+| A+V | +0.00236 | +0.01176 |
+| A | +0.00240 | +0.01154 |
+
+数值是 AP/mAP 百分数；差值单位是百分点。代码依据：
+[`infer_turn_head.py`](../../infer_turn_head.py) 的 `_align_to_egospeak_timeline` 与 `_evaluate`。
+`causal_valid_only` 是有效预测行的评测掩码，**不是原始输入没有泄漏的独立证据**。
+本次没有重新审计标签时间戳、原视频时钟或在线延迟；表中 0.2–2.0s 一律指协议标称 horizon，
+不直接当作已证实的物理提前量。
+
+## 4. 主协议汇总：causal_valid_only，seed 0
+
+来自此前用户直接提供的汇总，不是由 full-timeline 反推。仅有首个 horizon 和均值。
+
+| 输入 | 标称 0.2s mAP | BG AP | wearer AP | others AP | 十-horizon 平均 mAP |
+|---|---:|---:|---:|---:|---:|
+| G | 73.44637 | 56.76438 | 77.86650 | 85.70824 | 60.84861 |
+| A | 79.69221 | 63.69477 | 84.09982 | 91.28205 | 60.50771 |
+| V | 48.38434 | 30.16284 | 41.68834 | 73.30186 | 46.25250 |
+| A+V | 79.00727 | 63.67397 | 82.69832 | 90.64952 | 60.76303 |
+| G+A | 81.72257 | 68.15248 | 85.06554 | 91.94969 | 65.10895 |
+| G+V | 73.47412 | 56.67089 | 78.60954 | 85.14194 | 60.99238 |
+| G+A+V | 82.02661 | 68.93423 | 85.08583 | 92.05978 | 65.39978 |
+
+来源转录：[reported_causal_summary.txt](reported_causal_summary.txt)。
+不要将这里的 G 行与下一节 full-timeline 的逐 horizon 行混算差值。
+
+## 5. 完整 mAP 曲线：egospeak_full_timeline，seed 0
+
+A/V/AV 为本次三份附件；G+A/G+A+V 为此前两份附件。所有列是同一协议。
+
+| 标称 horizon / s | A | V | A+V | G+A | G+A+V |
+|---|---:|---:|---:|---:|---:|
+| 0.2 | 79.68981 | 48.38326 | 79.00491 | 81.72011 | 82.02415 |
+| 0.4 | 72.41576 | 47.68478 | 72.06793 | 75.91398 | 75.99169 |
+| 0.6 | 66.33666 | 47.26191 | 66.16179 | 70.37189 | 70.92665 |
+| 0.8 | 61.60584 | 46.62600 | 61.62137 | 66.45047 | 66.79615 |
+| 1.0 | 58.28586 | 46.00582 | 58.49237 | 63.79981 | 63.86856 |
+| 1.2 | 55.93708 | 45.75002 | 56.07422 | 61.35120 | 61.56280 |
+| 1.4 | 54.22585 | 45.41812 | 54.61933 | 59.98279 | 60.34070 |
+| 1.6 | 52.83014 | 45.56259 | 53.78180 | 58.29734 | 58.83407 |
+| 1.8 | 51.97218 | 44.79454 | 52.91556 | 56.93221 | 57.39446 |
+| 2.0 | 51.66256 | 44.97339 | 52.77341 | 56.13656 | 56.12449 |
+| 十个 horizon 平均 | 60.49617 | 46.24604 | 60.75127 | 65.09564 | 65.38637 |
+
+所有 50 个 horizon 的三类 AP、样本数和 mAP 都保存在
+[summary.json](summary.json) 的 `egospeak_full_timeline.runs`；
+不只保存表格中的 mAP。
+
+## 6. 视觉不是“全程无效”：需要分 horizon、分类别看
+
+下表是 **A+V − A**，全部使用 full-timeline，单位为百分点：
+
+| 标称 horizon / s | ΔmAP | ΔBG AP | Δwearer AP | Δothers AP |
+|---|---:|---:|---:|---:|
+| 0.2 | -0.68490 | -0.02080 | -1.40150 | -0.63239 |
+| 0.4 | -0.34783 | +0.70624 | -1.59157 | -0.15812 |
+| 0.6 | -0.17487 | +0.82855 | -1.58313 | +0.22998 |
+| 0.8 | +0.01553 | +1.44947 | -1.66891 | +0.26602 |
+| 1.0 | +0.20651 | +1.56046 | -1.51747 | +0.57655 |
+| 1.2 | +0.13714 | +1.00313 | -1.31375 | +0.72202 |
+| 1.4 | +0.39348 | +1.42271 | -1.10816 | +0.86587 |
+| 1.6 | +0.95166 | +1.90632 | -0.41886 | +1.36752 |
+| 1.8 | +0.94338 | +1.81173 | -0.05834 | +1.07678 |
+| 2.0 | +1.11085 | +1.66905 | +0.66242 | +1.00108 |
+
+观察：
+
+- mAP 在 0.2/0.4/0.6s 下降；0.8s 的 +0.01553 只是很小的数值差。
+  1.6/1.8/2.0s 分别 +0.95166/+0.94338/+1.11085，十-horizon 均值为 +0.25510。
+- wearer AP 在前 9 个 horizon 均下降，直到 2.0s 才提高。
+  十-horizon 的 BG、wearer、others AP 平均变化分别为
+  **+1.23369 / −0.99993 / +0.53153**。
+- 因此“AV 的平均 mAP 略好”掩盖了 wearer 分支的退步；也不能再笼统说视觉在所有 horizon 都有害。
+- 这些 horizon 来自同一 seed、同一批序列，彼此相关；7/10 个正差不能当成七次独立复现实验。
+  正负号是描述，不是显著性检验。
+
+V-only 的 mAP 从 48.38326 到 44.97339，整体较低；但 AP 不是分类准确率，
+未给出测试标签 prevalence 或随机排序基准，不能仅凭 48% 宣称“随机水平”。
+可见人物框不直接提供 wearer 的脸，且当前逐人／时间平均不保留显式交互关系；
+这可作为解释假设，尚不能证明退步确由视觉信息不足或池化导致。
+
+## 7. 加入 G 的组合增益：在完整曲线上不只是均值现象
+
+以下仍是 full-timeline。各类别列为十个 horizon 的 AP 差值平均：
+
+| 对照 | Δ十-horizon mAP | ΔBG AP 均值 | Δwearer AP 均值 | Δothers AP 均值 | mAP 正差 horizon |
+|---|---:|---:|---:|---:|---:|
+| G+A − A | +4.59947 | +8.71604 | +2.42825 | +2.65410 | 10/10 |
+| G+A+V − A+V | +4.63510 | +8.16923 | +3.48897 | +2.24711 | 10/10 |
+| G+A+V − G+A | +0.29073 | +0.68687 | +0.06078 | +0.12454 | 9/10 |
+
+- **G+A 相比 A 在全部 10 个 horizon、每个类别上都更高。**
+  mAP 的增益从 0.2s 的 +2.03030 增至 1.4s 的 +5.75694，
+  在 2.0s 仍为 +4.47400。完整曲线支持组合方案具有互补价值，
+  不是仅由某个 horizon 抬高均值。
+- **G+A+V 相比 A+V 同样在所有 horizon、所有类别上更高。**
+  十-horizon mAP 为 +4.63510。
+- **G+A+V 相比 G+A 的收益仍小且偏向 BG。**
+  十-horizon mAP +0.29073；wearer AP 均值仅 +0.06078；
+  0.2s wearer AP 仅 +0.02029，2.0s 的 mAP 反而 −0.01207。
+  因此不能把整体增益描述为已解决 wearer 发言预测或多说话人视觉理解。
+- 主协议对应的均值差是 G+A − A = +4.60124、
+  G+A+V − A+V = +4.63675、G+A+V − G+A = +0.29083。
+  这些数值与 full-timeline 接近，但仍分别标注，不互相替代。
+
+## 8. 归因限制：这还不是等容量、等上下文的 decoder 消融
+
+按当前 `TurnPredictionHead` 和约定维度计算的可训练参数量：
+
+| 输入 | 参数量 |
+|---|---:|
+| A 或 V | 2,134,048 |
+| A+V | 8,468,547 |
+| G | 18,928,670 |
+| G+A 或 G+V | 19,720,992 |
+| G+A+V | 20,513,314 |
+
+来源是本地代码实例化计算，不是远程 checkpoint 读取。A/V 单模态原生向量＋有效性位
+为 2049D；AV 为 4098D，MLP 隐藏宽度分别 1024 和 2049。
+G 使用 last4 concat-MLP；G+A/G+V/G+AV 还使用各路 256D 投影。
+因此 A→AV 同时改变参数量和优化问题；G+A→A 还改变了 history、视野和融合结构。
+
+这组结果不能单独证明：
+
+- Thinker 的语言推理或长时记忆是增益的唯一原因；
+- 模型已学到多人互动、speaker identity 或 turn transition；
+- AV 退步一定是过拟合，或一定由跨人／跨时间 pooling 引起；
+- 同为四轮就代表不同模型同样充分收敛；
+- encoder-only 或某一模型端到端更快。附件的 batch/s 不包含特征提取，
+  也没有同步计时、预热和统一基准，不能当作系统速度对照。
+
+## 9. 下一步实验优先级与论文表述
+
+1. **先补证据**：收集三组原始 `metrics.json`、checkpoint epoch/head 配置和训练摘要，
+   补全 causal-valid 曲线；在固定四轮／无 dev 的既定协议下补 seed 1/2。
+   不根据这份测试曲线重新选 epoch、阈值或只挑有利 horizon。
+2. **先做受控融合对照**：固定分支投影与后端 MLP 架构，以明确的模态屏蔽比较 A/AV，
+   并报告总参数和实际参与学习的参数；用预先定义的相同原始可见时间范围控制 history。
+   再比较 G 的增量，避免把更大 head 或更长历史完全归给 decoder。
+3. **再测试多人视觉假设**：优先复用已经保存的逐人向量，比较平均池化与音频条件的逐人加权，
+   保留音频分支作为对照。事先定义多可见人物、重叠说话等分析子集及样本量，
+   同时报告 BG/wearer/others，而不只看总体 mAP。提供框的结果与在线检测器场景分开说明。
+
+当前可以写成：
+
+> 在本次单种子设置中，近期音频编码器提供强近时预测信号，加入全局 Thinker 表示的组合
+> 在所有评测 horizon 上优于对应的 encoder-only 方案。局部视觉的影响依赖类别和 horizon：
+> 部分较后 horizon 的宏平均指标改善，但 wearer AP 多数下降。等容量、等上下文控制和
+> 多种子复现仍是隔离各组件贡献的必要步骤。
+
+这支持继续研究“近期感知与上下文表示如何互补、多人视觉如何有效融入”，
+但尚不足以把“多人视觉调优成功”写成已经证实的贡献。
+
+## 10. 归档文件与校验范围
+
+- [summary.json](summary.json)：两种协议分区、完整 AP 曲线、计算差值、配置来源与证据边界。
+- [reported_causal_summary.txt](reported_causal_summary.txt)：此前主协议汇总的转录。
+- [raw_encoder_audio_seed0.txt](raw_encoder_audio_seed0.txt)、
+  [raw_encoder_visual_seed0.txt](raw_encoder_visual_seed0.txt)、
+  [raw_encoder_av_seed0.txt](raw_encoder_av_seed0.txt)：本次三份控制台附件。
+- [raw_previous_thinker_audio_seed0.txt](raw_previous_thinker_audio_seed0.txt)、
+  [raw_previous_thinker_av_seed0.txt](raw_previous_thinker_av_seed0.txt)：此前两个对照附件。
+
+控制台归档仅移除了行尾水平空白、统一最后的换行；保留换行折断的路径和全部数值。
+本地核对了 50 个 horizon 的 AP→mAP、十-horizon 均值、样本数、主协议转录与差值；
+四舍五入允许 1e-5 的百分数误差。没有从原始预测重新计算 AP，
+这些文件不是对远程 `metrics.json` 或真实训练产物的替代。
diff --git a/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_audio_seed0.txt b/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_audio_seed0.txt
new file mode 100644
index 0000000..c490b24
--- /dev/null
+++ b/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_audio_seed0.txt
@@ -0,0 +1,40 @@
+checkpoint: /data1/yfzx/wangx/Code/icassp2027-experiment/checkpoints/easycom_encoder_only_1s/encoder_audio_seed0.pth
+training summary: /data1/yfzx/wangx/Code/icassp2027-experiment/checkpoints/easycom_encoder_only_1s/encoder_audio_seed0.pth.json
+/usr/local/bin/python /data1/yfzx/wangx/Code/icassp2027-experiment/infer_turn_head.py --dataset easycom --feature-root /data1/yfzx/wangx/Datasets/easycom/omni_features/encoder_only_audio_1s --target-root /data1/yfzx/wangx/Datasets/easyco
+m/easycom_target_frame --expected-mode encoder-only-audio --device cuda:0 --checkpoint /data1/yfzx/wangx/Code/icassp2027-experiment/checkpoints/easycom_encoder_only_1s/encoder_audio_seed0.pth --output-dir /data1/yfzx/wangx/Code/icassp202
+7-experiment/results/easycom_encoder_only_1s/encoder_audio_seed0 --sessions Session_3 Session_2 Session_1
+/usr/local/lib/python3.11/site-packages/torch/cuda/__init__.py:63: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the
+ package that installed pynvml for you.
+  import pynvml  # type: ignore[import]
+device: cuda:0
+loss: lstr-mce; output_space: three-class-softmax
+sessions: Session_3, Session_2, Session_1
+infer Session_3: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 9/9 [00:00<00:00, 21.47batch/s]
+Session_3: saved pred (8400, 3, 10), causal-valid counts=[8399, 8398, 8397, 8396, 8395, 8394, 8393, 8392, 8391, 8390]
+infer Session_2: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 8/8 [00:00<00:00, 1441.65batch/s]
+Session_2: saved pred (7706, 3, 10), causal-valid counts=[7705, 7704, 7703, 7702, 7701, 7700, 7699, 7698, 7697, 7696]
+infer Session_1: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:00<00:00, 1407.96batch/s]
+Session_1: saved pred (6942, 3, 10), causal-valid counts=[6941, 6940, 6939, 6938, 6937, 6936, 6935, 6934, 6933, 6932]
+
+egospeak_full_timeline:
+Action anticipation (0.20s) perframe mAP: 79.68981
+  AP: background=63.69477, camera_wearer_speaking=84.09982, other_people_speaking=91.27483; n=23048
+Action anticipation (0.40s) perframe mAP: 72.41576
+  AP: background=52.47780, camera_wearer_speaking=77.43606, other_people_speaking=87.33340; n=23048
+Action anticipation (0.60s) perframe mAP: 66.33666
+  AP: background=44.20066, camera_wearer_speaking=71.41221, other_people_speaking=83.39710; n=23048
+Action anticipation (0.80s) perframe mAP: 61.60584
+  AP: background=38.60179, camera_wearer_speaking=66.08811, other_people_speaking=80.12762; n=23048
+Action anticipation (1.00s) perframe mAP: 58.28586
+  AP: background=34.75843, camera_wearer_speaking=62.53734, other_people_speaking=77.56181; n=23048
+Action anticipation (1.20s) perframe mAP: 55.93708
+  AP: background=32.93325, camera_wearer_speaking=59.29096, other_people_speaking=75.58705; n=23048
+Action anticipation (1.40s) perframe mAP: 54.22585
+  AP: background=31.56977, camera_wearer_speaking=56.79707, other_people_speaking=74.31072; n=23048
+Action anticipation (1.60s) perframe mAP: 52.83014
+  AP: background=30.67884, camera_wearer_speaking=54.61175, other_people_speaking=73.19983; n=23048
+Action anticipation (1.80s) perframe mAP: 51.97218
+  AP: background=30.37030, camera_wearer_speaking=52.80894, other_people_speaking=72.73728; n=23048
+Action anticipation (2.00s) perframe mAP: 51.66256
+  AP: background=30.86998, camera_wearer_speaking=51.49671, other_people_speaking=72.62098; n=23048
+Action anticipation (mean) perframe mAP: 60.49617
diff --git a/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_av_seed0.txt b/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_av_seed0.txt
new file mode 100644
index 0000000..cb129ec
--- /dev/null
+++ b/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_av_seed0.txt
@@ -0,0 +1,39 @@
+training summary: /data1/yfzx/wangx/Code/icassp2027-experiment/checkpoints/easycom_encoder_only_1s/encoder_av_seed0.pth.json
+/usr/local/bin/python /data1/yfzx/wangx/Code/icassp2027-experiment/infer_turn_head.py --dataset easycom --feature-root /data1/yfzx/wangx/Datasets/easycom/omni_features/encoder_only_av_1s --target-root /data1/yfzx/wangx/Datasets/easycom/e
+asycom_target_frame --expected-mode encoder-only-av --device cuda:0 --checkpoint /data1/yfzx/wangx/Code/icassp2027-experiment/checkpoints/easycom_encoder_only_1s/encoder_av_seed0.pth --output-dir /data1/yfzx/wangx/Code/icassp2027-experim
+ent/results/easycom_encoder_only_1s/encoder_av_seed0 --sessions Session_3 Session_2 Session_1
+/usr/local/lib/python3.11/site-packages/torch/cuda/__init__.py:63: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the
+ package that installed pynvml for you.
+  import pynvml  # type: ignore[import]
+device: cuda:0
+loss: lstr-mce; output_space: three-class-softmax
+sessions: Session_3, Session_2, Session_1
+infer Session_3: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 9/9 [00:00<00:00, 69.00batch/s]
+Session_3: saved pred (8400, 3, 10), causal-valid counts=[8399, 8398, 8397, 8396, 8395, 8394, 8393, 8392, 8391, 8390]
+infer Session_2: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 8/8 [00:00<00:00, 683.75batch/s]
+Session_2: saved pred (7706, 3, 10), causal-valid counts=[7705, 7704, 7703, 7702, 7701, 7700, 7699, 7698, 7697, 7696]
+infer Session_1: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:00<00:00, 634.13batch/s]
+Session_1: saved pred (6942, 3, 10), causal-valid counts=[6941, 6940, 6939, 6938, 6937, 6936, 6935, 6934, 6933, 6932]
+
+egospeak_full_timeline:
+Action anticipation (0.20s) perframe mAP: 79.00491
+  AP: background=63.67397, camera_wearer_speaking=82.69832, other_people_speaking=90.64244; n=23048
+Action anticipation (0.40s) perframe mAP: 72.06793
+  AP: background=53.18404, camera_wearer_speaking=75.84449, other_people_speaking=87.17528; n=23048
+Action anticipation (0.60s) perframe mAP: 66.16179
+  AP: background=45.02921, camera_wearer_speaking=69.82908, other_people_speaking=83.62708; n=23048
+Action anticipation (0.80s) perframe mAP: 61.62137
+  AP: background=40.05126, camera_wearer_speaking=64.41920, other_people_speaking=80.39364; n=23048
+Action anticipation (1.00s) perframe mAP: 58.49237
+  AP: background=36.31889, camera_wearer_speaking=61.01987, other_people_speaking=78.13836; n=23048
+Action anticipation (1.20s) perframe mAP: 56.07422
+  AP: background=33.93638, camera_wearer_speaking=57.97721, other_people_speaking=76.30907; n=23048
+Action anticipation (1.40s) perframe mAP: 54.61933
+  AP: background=32.99248, camera_wearer_speaking=55.68891, other_people_speaking=75.17659; n=23048
+Action anticipation (1.60s) perframe mAP: 53.78180
+  AP: background=32.58516, camera_wearer_speaking=54.19289, other_people_speaking=74.56735; n=23048
+Action anticipation (1.80s) perframe mAP: 52.91556
+  AP: background=32.18203, camera_wearer_speaking=52.75060, other_people_speaking=73.81406; n=23048
+Action anticipation (2.00s) perframe mAP: 52.77341
+  AP: background=32.53903, camera_wearer_speaking=52.15913, other_people_speaking=73.62206; n=23048
+Action anticipation (mean) perframe mAP: 60.75127
diff --git a/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_visual_seed0.txt b/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_visual_seed0.txt
new file mode 100644
index 0000000..1d6a4cf
--- /dev/null
+++ b/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_encoder_visual_seed0.txt
@@ -0,0 +1,40 @@
+checkpoint: /data1/yfzx/wangx/Code/icassp2027-experiment/checkpoints/easycom_encoder_only_1s/encoder_visual_seed0.pth
+training summary: /data1/yfzx/wangx/Code/icassp2027-experiment/checkpoints/easycom_encoder_only_1s/encoder_visual_seed0.pth.json
+/usr/local/bin/python /data1/yfzx/wangx/Code/icassp2027-experiment/infer_turn_head.py --dataset easycom --feature-root /data1/yfzx/wangx/Datasets/easycom/omni_features/encoder_only_visual_1s --target-root /data1/yfzx/wangx/Datasets/easyc
+om/easycom_target_frame --expected-mode encoder-only-visual --device cuda:0 --checkpoint /data1/yfzx/wangx/Code/icassp2027-experiment/checkpoints/easycom_encoder_only_1s/encoder_visual_seed0.pth --output-dir /data1/yfzx/wangx/Code/icassp
+2027-experiment/results/easycom_encoder_only_1s/encoder_visual_seed0 --sessions Session_3 Session_2 Session_1
+/usr/local/lib/python3.11/site-packages/torch/cuda/__init__.py:63: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the
+ package that installed pynvml for you.
+  import pynvml  # type: ignore[import]
+device: cuda:0
+loss: lstr-mce; output_space: three-class-softmax
+sessions: Session_3, Session_2, Session_1
+infer Session_3: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 9/9 [00:00<00:00, 71.22batch/s]
+Session_3: saved pred (8400, 3, 10), causal-valid counts=[8399, 8398, 8397, 8396, 8395, 8394, 8393, 8392, 8391, 8390]
+infer Session_2: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 8/8 [00:00<00:00, 1487.47batch/s]
+Session_2: saved pred (7706, 3, 10), causal-valid counts=[7705, 7704, 7703, 7702, 7701, 7700, 7699, 7698, 7697, 7696]
+infer Session_1: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:00<00:00, 1409.24batch/s]
+Session_1: saved pred (6942, 3, 10), causal-valid counts=[6941, 6940, 6939, 6938, 6937, 6936, 6935, 6934, 6933, 6932]
+
+egospeak_full_timeline:
+Action anticipation (0.20s) perframe mAP: 48.38326
+  AP: background=30.16284, camera_wearer_speaking=41.68834, other_people_speaking=73.29862; n=23048
+Action anticipation (0.40s) perframe mAP: 47.68478
+  AP: background=29.82215, camera_wearer_speaking=40.98323, other_people_speaking=72.24896; n=23048
+Action anticipation (0.60s) perframe mAP: 47.26191
+  AP: background=29.28129, camera_wearer_speaking=40.74974, other_people_speaking=71.75469; n=23048
+Action anticipation (0.80s) perframe mAP: 46.62600
+  AP: background=28.24103, camera_wearer_speaking=40.47370, other_people_speaking=71.16328; n=23048
+Action anticipation (1.00s) perframe mAP: 46.00582
+  AP: background=27.73638, camera_wearer_speaking=39.50369, other_people_speaking=70.77740; n=23048
+Action anticipation (1.20s) perframe mAP: 45.75002
+  AP: background=27.39577, camera_wearer_speaking=39.53354, other_people_speaking=70.32075; n=23048
+Action anticipation (1.40s) perframe mAP: 45.41812
+  AP: background=27.04920, camera_wearer_speaking=39.35560, other_people_speaking=69.84957; n=23048
+Action anticipation (1.60s) perframe mAP: 45.56259
+  AP: background=27.36515, camera_wearer_speaking=39.16056, other_people_speaking=70.16205; n=23048
+Action anticipation (1.80s) perframe mAP: 44.79454
+  AP: background=26.64061, camera_wearer_speaking=38.50155, other_people_speaking=69.24145; n=23048
+Action anticipation (2.00s) perframe mAP: 44.97339
+  AP: background=27.34810, camera_wearer_speaking=37.96646, other_people_speaking=69.60560; n=23048
+Action anticipation (mean) perframe mAP: 46.24604
diff --git a/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_previous_thinker_audio_seed0.txt b/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_previous_thinker_audio_seed0.txt
new file mode 100644
index 0000000..7b11610
--- /dev/null
+++ b/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_previous_thinker_audio_seed0.txt
@@ -0,0 +1,36 @@
+training summary: checkpoints/easycom_local_b1_encoded_tf4576/audio_seed0.pth.json
+/usr/local/lib/python3.11/site-packages/torch/cuda/__init__.py:63: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml d
+irectly, please report this to the maintainers of the package that installed pynvml for you.
+  import pynvml  # type: ignore[import]
+device: cuda:0
+loss: lstr-mce; output_space: three-class-softmax
+sessions: Session_3, Session_2, Session_1
+infer Session_3: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 9/9 [00:00<00:00, 39.03batch/s]
+Session_3: saved pred (8400, 3, 10), causal-valid counts=[8399, 8398, 8397, 8396, 8395, 8394, 8393, 8392, 8391, 8390]
+infer Session_2: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 8/8 [00:00<00:00, 142.52batch/s]
+Session_2: saved pred (7706, 3, 10), causal-valid counts=[7705, 7704, 7703, 7702, 7701, 7700, 7699, 7698, 7697, 7696]
+infer Session_1: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:00<00:00, 144.02batch/s]
+Session_1: saved pred (6942, 3, 10), causal-valid counts=[6941, 6940, 6939, 6938, 6937, 6936, 6935, 6934, 6933, 6932]
+
+egospeak_full_timeline:
+Action anticipation (0.20s) perframe mAP: 81.72011
+  AP: background=68.15248, camera_wearer_speaking=85.06554, other_people_speaking=91.94232; n=23048
+Action anticipation (0.40s) perframe mAP: 75.91398
+  AP: background=59.90715, camera_wearer_speaking=79.22465, other_people_speaking=88.61015; n=23048
+Action anticipation (0.60s) perframe mAP: 70.37189
+  AP: background=52.48480, camera_wearer_speaking=73.21461, other_people_speaking=85.41626; n=23048
+Action anticipation (0.80s) perframe mAP: 66.45047
+  AP: background=47.92066, camera_wearer_speaking=68.77900, other_people_speaking=82.65175; n=23048
+Action anticipation (1.00s) perframe mAP: 63.79981
+  AP: background=45.59262, camera_wearer_speaking=65.09890, other_people_speaking=80.70791; n=23048
+Action anticipation (1.20s) perframe mAP: 61.35120
+  AP: background=43.15384, camera_wearer_speaking=61.86698, other_people_speaking=79.03278; n=23048
+Action anticipation (1.40s) perframe mAP: 59.98279
+  AP: background=41.83907, camera_wearer_speaking=60.22346, other_people_speaking=77.88584; n=23048
+Action anticipation (1.60s) perframe mAP: 58.29734
+  AP: background=40.44538, camera_wearer_speaking=57.66189, other_people_speaking=76.78474; n=23048
+Action anticipation (1.80s) perframe mAP: 56.93221
+  AP: background=39.04677, camera_wearer_speaking=55.70067, other_people_speaking=76.04919; n=23048
+Action anticipation (2.00s) perframe mAP: 56.13656
+  AP: background=38.77318, camera_wearer_speaking=54.02582, other_people_speaking=75.61069; n=23048
+Action anticipation (mean) perframe mAP: 65.09564
diff --git a/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_previous_thinker_av_seed0.txt b/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_previous_thinker_av_seed0.txt
new file mode 100644
index 0000000..ff9eaa8
--- /dev/null
+++ b/results/easycom_encoder_only_1s_4ep_2026-08-28/raw_previous_thinker_av_seed0.txt
@@ -0,0 +1,37 @@
+checkpoint: checkpoints/easycom_local_b1_encoded_tf4576/av_seed0.pth
+training summary: checkpoints/easycom_local_b1_encoded_tf4576/av_seed0.pth.json
+/usr/local/lib/python3.11/site-packages/torch/cuda/__init__.py:63: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml d
+irectly, please report this to the maintainers of the package that installed pynvml for you.
+  import pynvml  # type: ignore[import]
+device: cuda:0
+loss: lstr-mce; output_space: three-class-softmax
+sessions: Session_3, Session_2, Session_1
+infer Session_3: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 9/9 [00:00<00:00, 40.61batch/s]
+Session_3: saved pred (8400, 3, 10), causal-valid counts=[8399, 8398, 8397, 8396, 8395, 8394, 8393, 8392, 8391, 8390]
+infer Session_2: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 8/8 [00:00<00:00, 143.92batch/s]
+Session_2: saved pred (7706, 3, 10), causal-valid counts=[7705, 7704, 7703, 7702, 7701, 7700, 7699, 7698, 7697, 7696]
+infer Session_1: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:00<00:00, 160.27batch/s]
+Session_1: saved pred (6942, 3, 10), causal-valid counts=[6941, 6940, 6939, 6938, 6937, 6936, 6935, 6934, 6933, 6932]
+
+egospeak_full_timeline:
+Action anticipation (0.20s) perframe mAP: 82.02415
+  AP: background=68.93423, camera_wearer_speaking=85.08583, other_people_speaking=92.05239; n=23048
+Action anticipation (0.40s) perframe mAP: 75.99169
+  AP: background=60.58903, camera_wearer_speaking=78.83427, other_people_speaking=88.55177; n=23048
+Action anticipation (0.60s) perframe mAP: 70.92665
+  AP: background=53.84208, camera_wearer_speaking=73.45503, other_people_speaking=85.48283; n=23048
+Action anticipation (0.80s) perframe mAP: 66.79615
+  AP: background=49.21382, camera_wearer_speaking=68.56072, other_people_speaking=82.61390; n=23048
+Action anticipation (1.00s) perframe mAP: 63.86856
+  AP: background=45.62073, camera_wearer_speaking=65.27034, other_people_speaking=80.71461; n=23048
+Action anticipation (1.20s) perframe mAP: 61.56280
+  AP: background=43.46670, camera_wearer_speaking=62.05277, other_people_speaking=79.16893; n=23048
+Action anticipation (1.40s) perframe mAP: 60.34070
+  AP: background=42.54682, camera_wearer_speaking=60.20540, other_people_speaking=78.26988; n=23048
+Action anticipation (1.60s) perframe mAP: 58.83407
+  AP: background=41.15952, camera_wearer_speaking=58.18522, other_people_speaking=77.15747; n=23048
+Action anticipation (1.80s) perframe mAP: 57.39446
+  AP: background=39.94176, camera_wearer_speaking=55.99271, other_people_speaking=76.24890; n=23048
+Action anticipation (2.00s) perframe mAP: 56.12449
+  AP: background=38.87001, camera_wearer_speaking=53.82707, other_people_speaking=75.67639; n=23048
+Action anticipation (mean) perframe mAP: 65.38637
diff --git a/results/easycom_encoder_only_1s_4ep_2026-08-28/reported_causal_summary.txt b/results/easycom_encoder_only_1s_4ep_2026-08-28/reported_causal_summary.txt
new file mode 100644
index 0000000..5a28384
--- /dev/null
+++ b/results/easycom_encoder_only_1s_4ep_2026-08-28/reported_causal_summary.txt
@@ -0,0 +1,15 @@
+User-reported causal_valid_only summaries from earlier messages in this thread.
+Not a dump of remote metrics.json; only the first horizon and ten-horizon mean were supplied.
+G=Thinker; A/V=local encoders. All runs below are seed 0. AP/mAP values are percent.
+
+variant,input,seed,nominal_0.2s_map,background_ap,wearer_ap,others_ap,ten_horizon_mean_map
+thinker,G,0,73.44637,56.76438,77.86650,85.70824,60.84861
+encoder_audio,A,0,79.69221,63.69477,84.09982,91.28205,60.50771
+encoder_visual,V,0,48.38434,30.16284,41.68834,73.30186,46.25250
+encoder_av,A+V,0,79.00727,63.67397,82.69832,90.64952,60.76303
+thinker_audio,G+A,0,81.72257,68.15248,85.06554,91.94969,65.10895
+thinker_visual,G+V,0,73.47412,56.67089,78.60954,85.14194,60.99238
+thinker_av,G+A+V,0,82.02661,68.93423,85.08583,92.05978,65.39978
+
+The newly supplied full logs use egospeak_full_timeline, not this protocol.
+The reported sample standard deviation for each encoder-only run was n/a, not zero.
diff --git a/results/easycom_encoder_only_1s_4ep_2026-08-28/summary.json b/results/easycom_encoder_only_1s_4ep_2026-08-28/summary.json
new file mode 100644
index 0000000..953affd
--- /dev/null
+++ b/results/easycom_encoder_only_1s_4ep_2026-08-28/summary.json
@@ -0,0 +1,1318 @@
+{
+  "schema_version": 1,
+  "recorded_date": "2026-08-28",
+  "dataset": "easycom",
+  "status": "preliminary_single_seed_user_reported_remote_evaluation",
+  "units": {
+    "ap_and_map": "percent",
+    "deltas": "percentage_points",
+    "printed_precision_decimals": 5
+  },
+  "primary_protocol": "causal_valid_only",
+  "full_horizon_protocol_received": "egospeak_full_timeline",
+  "evidence": {
+    "current_user_attachments": 3,
+    "earlier_user_attachments": 2,
+    "full_horizon_records_received": 50,
+    "remote_metrics_json_received": false,
+    "remote_predictions_received": false,
+    "remote_checkpoint_payload_received": false,
+    "training_summary_json_received": false,
+    "training_curves_received": false,
+    "checkpoint_epoch_independently_verified": false,
+    "raw_media_or_feature_causality_reaudited": false,
+    "remote_run_git_commit_verified": false,
+    "seed_variance_available": false,
+    "note": "Attachments show inference excerpts and paths to checkpoints/training summaries, not their contents. No experiment was re-run locally."
+  },
+  "experiment": {
+    "new_variants": [
+      "encoder_audio",
+      "encoder_visual",
+      "encoder_av"
+    ],
+    "reported_seed": 0,
+    "missing_seeds": [
+      1,
+      2
+    ],
+    "planned_epochs": 4,
+    "train_sessions": [
+      4,
+      5,
+      6,
+      7,
+      8,
+      9,
+      10,
+      11,
+      12
+    ],
+    "dev_sessions": [],
+    "test_sessions_observed": [
+      "Session_3",
+      "Session_2",
+      "Session_1"
+    ],
+    "split_and_training_budget_source": "User run instructions and local runner at 321238a; not remote checkpoint inspection.",
+    "loss_observed": "lstr-mce",
+    "output_space_observed": "three-class-softmax",
+    "configured_training": {
+      "batch_size": 512,
+      "learning_rate": 0.00007,
+      "weight_decay": 0.00005,
+      "warmup_epochs": 1.6,
+      "warmup_factor": 0.3,
+      "dropout": 0.2,
+      "transition_objective": "none",
+      "selection": "final_epoch_no_dev"
+    },
+    "configured_encoder": {
+      "checkpoint": "Qwen2.5-Omni-3B",
+      "frozen": true,
+      "reported_transformers": "4.57.6",
+      "audio_window_seconds": 1,
+      "visual_window_seconds": 1,
+      "video_fps": 2,
+      "state_hz": 5,
+      "native_audio_dim": 2048,
+      "native_visual_dim": 2048,
+      "visual_readout": "mean tokens, equal-person mean, then trailing-window mean over valid frames",
+      "audio_readout": "pooled valid encoder outputs from half-open trailing mixed-audio window",
+      "provided_person_boxes": true,
+      "online_detector_evaluated": false,
+      "uses_thinker_hidden_states": false
+    },
+    "prior_G_definition": "Frozen 3B Thinker, open-stream-replay, 8s memory, 2fps, last-four-layer stack with concat-MLP",
+    "remote_paths": {
+      "project": "/data1/yfzx/wangx/Code/icassp2027-experiment",
+      "source_local_features_confirmed_by_user": "/data1/yfzx/wangx/Datasets/easycom/omni_features/local_qwen_1s_encoded_tf4576",
+      "target_root": "/data1/yfzx/wangx/Datasets/easycom/easycom_target_frame",
+      "exported_features_template": "/data1/yfzx/wangx/Datasets/easycom/omni_features/encoder_only_{audio,av,visual}_1s",
+      "checkpoint_root": "checkpoints/easycom_encoder_only_1s",
+      "result_root": "results/easycom_encoder_only_1s",
+      "prior_checkpoint_root": "checkpoints/easycom_local_b1_encoded_tf4576"
+    },
+    "trainable_parameter_counts_from_local_code": {
+      "assumption": "Native A/V=2048D; G=4x2048 with concat-MLP; 256D B1 local projections; LSTR-MCE; no transition head. Counts not read from remote checkpoints.",
+      "source": "easycom_head.TurnPredictionHead at 321238a",
+      "counts": {
+        "encoder_audio": 2134048,
+        "encoder_visual": 2134048,
+        "encoder_av": 8468547,
+        "thinker": 18928670,
+        "thinker_audio": 19720992,
+        "thinker_visual": 19720992,
+        "thinker_av": 20513314
+      }
+    }
+  },
+  "evaluation": {
+    "horizon_semantics": "Nominal EgoSpeak labels, 0.2 to 2.0 seconds. Physical input-to-target lead and online latency are not inferred from metric names.",
+    "full_timeline_note": "Includes unavailable initial h target rows per session as zero-score predictions.",
+    "causal_valid_only_note": "Excludes the first h target rows per session for horizon h; not an independent audit of raw input causality.",
+    "session_outputs_observed": [
+      {
+        "sample_id": "Session_3",
+        "shape": [
+          8400,
+          3,
+          10
+        ],
+        "causal_valid_counts": [
+          8399,
+          8398,
+          8397,
+          8396,
+          8395,
+          8394,
+          8393,
+          8392,
+          8391,
+          8390
+        ]
+      },
+      {
+        "sample_id": "Session_2",
+        "shape": [
+          7706,
+          3,
+          10
+        ],
+        "causal_valid_counts": [
+          7705,
+          7704,
+          7703,
+          7702,
+          7701,
+          7700,
+          7699,
+          7698,
+          7697,
+          7696
+        ]
+      },
+      {
+        "sample_id": "Session_1",
+        "shape": [
+          6942,
+          3,
+          10
+        ],
+        "causal_valid_counts": [
+          6941,
+          6940,
+          6939,
+          6938,
+          6937,
+          6936,
+          6935,
+          6934,
+          6933,
+          6932
+        ]
+      }
+    ],
+    "full_timeline_num_samples_each_horizon": 23048,
+    "causal_valid_num_samples_each_horizon": [
+      23045,
+      23042,
+      23039,
+      23036,
+      23033,
+      23030,
+      23027,
+      23024,
+      23021,
+      23018
+    ]
+  },
+  "causal_valid_only": {
+    "availability": "Only the earlier reported first-horizon class AP/mAP and ten-horizon mean; full curves not received.",
+    "runs": [
+      {
+        "variant": "thinker",
+        "input": "G",
+        "seed": 0,
+        "horizon_0_2": {
+          "nominal_seconds": 0.2,
+          "map_percent": 73.44637,
+          "background_ap_percent": 56.76438,
+          "wearer_ap_percent": 77.8665,
+          "others_ap_percent": 85.70824
+        },
+        "ten_horizon_mean_map_percent": 60.84861,
+        "full_horizon_metrics_available": false,
+        "source_file": "reported_causal_summary.txt"
+      },
+      {
+        "variant": "encoder_audio",
+        "input": "A",
+        "seed": 0,
+        "horizon_0_2": {
+          "nominal_seconds": 0.2,
+          "map_percent": 79.69221,
+          "background_ap_percent": 63.69477,
+          "wearer_ap_percent": 84.09982,
+          "others_ap_percent": 91.28205
+        },
+        "ten_horizon_mean_map_percent": 60.50771,
+        "full_horizon_metrics_available": false,
+        "source_file": "reported_causal_summary.txt"
+      },
+      {
+        "variant": "encoder_visual",
+        "input": "V",
+        "seed": 0,
+        "horizon_0_2": {
+          "nominal_seconds": 0.2,
+          "map_percent": 48.38434,
+          "background_ap_percent": 30.16284,
+          "wearer_ap_percent": 41.68834,
+          "others_ap_percent": 73.30186
+        },
+        "ten_horizon_mean_map_percent": 46.2525,
+        "full_horizon_metrics_available": false,
+        "source_file": "reported_causal_summary.txt"
+      },
+      {
+        "variant": "encoder_av",
+        "input": "A+V",
+        "seed": 0,
+        "horizon_0_2": {
+          "nominal_seconds": 0.2,
+          "map_percent": 79.00727,
+          "background_ap_percent": 63.67397,
+          "wearer_ap_percent": 82.69832,
+          "others_ap_percent": 90.64952
+        },
+        "ten_horizon_mean_map_percent": 60.76303,
+        "full_horizon_metrics_available": false,
+        "source_file": "reported_causal_summary.txt"
+      },
+      {
+        "variant": "thinker_audio",
+        "input": "G+A",
+        "seed": 0,
+        "horizon_0_2": {
+          "nominal_seconds": 0.2,
+          "map_percent": 81.72257,
+          "background_ap_percent": 68.15248,
+          "wearer_ap_percent": 85.06554,
+          "others_ap_percent": 91.94969
+        },
+        "ten_horizon_mean_map_percent": 65.10895,
+        "full_horizon_metrics_available": false,
+        "source_file": "reported_causal_summary.txt"
+      },
+      {
+        "variant": "thinker_visual",
+        "input": "G+V",
+        "seed": 0,
+        "horizon_0_2": {
+          "nominal_seconds": 0.2,
+          "map_percent": 73.47412,
+          "background_ap_percent": 56.67089,
+          "wearer_ap_percent": 78.60954,
+          "others_ap_percent": 85.14194
+        },
+        "ten_horizon_mean_map_percent": 60.99238,
+        "full_horizon_metrics_available": false,
+        "source_file": "reported_causal_summary.txt"
+      },
+      {
+        "variant": "thinker_av",
+        "input": "G+A+V",
+        "seed": 0,
+        "horizon_0_2": {
+          "nominal_seconds": 0.2,
+          "map_percent": 82.02661,
+          "background_ap_percent": 68.93423,
+          "wearer_ap_percent": 85.08583,
+          "others_ap_percent": 92.05978
+        },
+        "ten_horizon_mean_map_percent": 65.39978,
+        "full_horizon_metrics_available": false,
+        "source_file": "reported_causal_summary.txt"
+      }
+    ],
+    "comparisons": [
+      {
+        "contrast": "encoder_audio_minus_thinker",
+        "protocol": "causal_valid_only",
+        "horizon_0_2_delta_pp": {
+          "map": 6.24584,
+          "background_ap": 6.93039,
+          "wearer_ap": 6.23332,
+          "others_ap": 5.57381
+        },
+        "ten_horizon_mean_delta_pp": -0.3409
+      },
+      {
+        "contrast": "encoder_av_minus_encoder_audio",
+        "protocol": "causal_valid_only",
+        "horizon_0_2_delta_pp": {
+          "map": -0.68494,
+          "background_ap": -0.0208,
+          "wearer_ap": -1.4015,
+          "others_ap": -0.63253
+        },
+        "ten_horizon_mean_delta_pp": 0.25532
+      },
+      {
+        "contrast": "thinker_audio_minus_encoder_audio",
+        "protocol": "causal_valid_only",
+        "horizon_0_2_delta_pp": {
+          "map": 2.03036,
+          "background_ap": 4.45771,
+          "wearer_ap": 0.96572,
+          "others_ap": 0.66764
+        },
+        "ten_horizon_mean_delta_pp": 4.60124
+      },
+      {
+        "contrast": "thinker_av_minus_encoder_av",
+        "protocol": "causal_valid_only",
+        "horizon_0_2_delta_pp": {
+          "map": 3.01934,
+          "background_ap": 5.26026,
+          "wearer_ap": 2.38751,
+          "others_ap": 1.41026
+        },
+        "ten_horizon_mean_delta_pp": 4.63675
+      },
+      {
+        "contrast": "thinker_av_minus_thinker_audio",
+        "protocol": "causal_valid_only",
+        "horizon_0_2_delta_pp": {
+          "map": 0.30404,
+          "background_ap": 0.78175,
+          "wearer_ap": 0.02029,
+          "others_ap": 0.11009
+        },
+        "ten_horizon_mean_delta_pp": 0.29083
+      }
+    ]
+  },
+  "egospeak_full_timeline": {
+    "availability": "Parsed from 3 current attachments and 2 earlier attachments; not original metrics.json.",
+    "runs": [
+      {
+        "variant": "encoder_visual",
+        "input": "V",
+        "seed": 0,
+        "source_file": "raw_encoder_visual_seed0.txt",
+        "source_kind": "current_user_attachment",
+        "horizons": [
+          {
+            "nominal_seconds": 0.2,
+            "map_percent": 48.38326,
+            "background_ap_percent": 30.16284,
+            "wearer_ap_percent": 41.68834,
+            "others_ap_percent": 73.29862,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.4,
+            "map_percent": 47.68478,
+            "background_ap_percent": 29.82215,
+            "wearer_ap_percent": 40.98323,
+            "others_ap_percent": 72.24896,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.6,
+            "map_percent": 47.26191,
+            "background_ap_percent": 29.28129,
+            "wearer_ap_percent": 40.74974,
+            "others_ap_percent": 71.75469,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.8,
+            "map_percent": 46.626,
+            "background_ap_percent": 28.24103,
+            "wearer_ap_percent": 40.4737,
+            "others_ap_percent": 71.16328,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1,
+            "map_percent": 46.00582,
+            "background_ap_percent": 27.73638,
+            "wearer_ap_percent": 39.50369,
+            "others_ap_percent": 70.7774,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.2,
+            "map_percent": 45.75002,
+            "background_ap_percent": 27.39577,
+            "wearer_ap_percent": 39.53354,
+            "others_ap_percent": 70.32075,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.4,
+            "map_percent": 45.41812,
+            "background_ap_percent": 27.0492,
+            "wearer_ap_percent": 39.3556,
+            "others_ap_percent": 69.84957,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.6,
+            "map_percent": 45.56259,
+            "background_ap_percent": 27.36515,
+            "wearer_ap_percent": 39.16056,
+            "others_ap_percent": 70.16205,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.8,
+            "map_percent": 44.79454,
+            "background_ap_percent": 26.64061,
+            "wearer_ap_percent": 38.50155,
+            "others_ap_percent": 69.24145,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 2,
+            "map_percent": 44.97339,
+            "background_ap_percent": 27.3481,
+            "wearer_ap_percent": 37.96646,
+            "others_ap_percent": 69.6056,
+            "num_samples": 23048
+          }
+        ],
+        "ten_horizon_mean_map_percent": 46.24604,
+        "ten_horizon_class_mean_ap_percent": {
+          "background_ap_percent": 28.10425,
+          "wearer_ap_percent": 39.79164,
+          "others_ap_percent": 70.84224
+        }
+      },
+      {
+        "variant": "encoder_av",
+        "input": "A+V",
+        "seed": 0,
+        "source_file": "raw_encoder_av_seed0.txt",
+        "source_kind": "current_user_attachment",
+        "horizons": [
+          {
+            "nominal_seconds": 0.2,
+            "map_percent": 79.00491,
+            "background_ap_percent": 63.67397,
+            "wearer_ap_percent": 82.69832,
+            "others_ap_percent": 90.64244,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.4,
+            "map_percent": 72.06793,
+            "background_ap_percent": 53.18404,
+            "wearer_ap_percent": 75.84449,
+            "others_ap_percent": 87.17528,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.6,
+            "map_percent": 66.16179,
+            "background_ap_percent": 45.02921,
+            "wearer_ap_percent": 69.82908,
+            "others_ap_percent": 83.62708,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.8,
+            "map_percent": 61.62137,
+            "background_ap_percent": 40.05126,
+            "wearer_ap_percent": 64.4192,
+            "others_ap_percent": 80.39364,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1,
+            "map_percent": 58.49237,
+            "background_ap_percent": 36.31889,
+            "wearer_ap_percent": 61.01987,
+            "others_ap_percent": 78.13836,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.2,
+            "map_percent": 56.07422,
+            "background_ap_percent": 33.93638,
+            "wearer_ap_percent": 57.97721,
+            "others_ap_percent": 76.30907,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.4,
+            "map_percent": 54.61933,
+            "background_ap_percent": 32.99248,
+            "wearer_ap_percent": 55.68891,
+            "others_ap_percent": 75.17659,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.6,
+            "map_percent": 53.7818,
+            "background_ap_percent": 32.58516,
+            "wearer_ap_percent": 54.19289,
+            "others_ap_percent": 74.56735,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.8,
+            "map_percent": 52.91556,
+            "background_ap_percent": 32.18203,
+            "wearer_ap_percent": 52.7506,
+            "others_ap_percent": 73.81406,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 2,
+            "map_percent": 52.77341,
+            "background_ap_percent": 32.53903,
+            "wearer_ap_percent": 52.15913,
+            "others_ap_percent": 73.62206,
+            "num_samples": 23048
+          }
+        ],
+        "ten_horizon_mean_map_percent": 60.75127,
+        "ten_horizon_class_mean_ap_percent": {
+          "background_ap_percent": 40.24924,
+          "wearer_ap_percent": 62.65797,
+          "others_ap_percent": 79.34659
+        }
+      },
+      {
+        "variant": "encoder_audio",
+        "input": "A",
+        "seed": 0,
+        "source_file": "raw_encoder_audio_seed0.txt",
+        "source_kind": "current_user_attachment",
+        "horizons": [
+          {
+            "nominal_seconds": 0.2,
+            "map_percent": 79.68981,
+            "background_ap_percent": 63.69477,
+            "wearer_ap_percent": 84.09982,
+            "others_ap_percent": 91.27483,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.4,
+            "map_percent": 72.41576,
+            "background_ap_percent": 52.4778,
+            "wearer_ap_percent": 77.43606,
+            "others_ap_percent": 87.3334,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.6,
+            "map_percent": 66.33666,
+            "background_ap_percent": 44.20066,
+            "wearer_ap_percent": 71.41221,
+            "others_ap_percent": 83.3971,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.8,
+            "map_percent": 61.60584,
+            "background_ap_percent": 38.60179,
+            "wearer_ap_percent": 66.08811,
+            "others_ap_percent": 80.12762,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1,
+            "map_percent": 58.28586,
+            "background_ap_percent": 34.75843,
+            "wearer_ap_percent": 62.53734,
+            "others_ap_percent": 77.56181,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.2,
+            "map_percent": 55.93708,
+            "background_ap_percent": 32.93325,
+            "wearer_ap_percent": 59.29096,
+            "others_ap_percent": 75.58705,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.4,
+            "map_percent": 54.22585,
+            "background_ap_percent": 31.56977,
+            "wearer_ap_percent": 56.79707,
+            "others_ap_percent": 74.31072,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.6,
+            "map_percent": 52.83014,
+            "background_ap_percent": 30.67884,
+            "wearer_ap_percent": 54.61175,
+            "others_ap_percent": 73.19983,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.8,
+            "map_percent": 51.97218,
+            "background_ap_percent": 30.3703,
+            "wearer_ap_percent": 52.80894,
+            "others_ap_percent": 72.73728,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 2,
+            "map_percent": 51.66256,
+            "background_ap_percent": 30.86998,
+            "wearer_ap_percent": 51.49671,
+            "others_ap_percent": 72.62098,
+            "num_samples": 23048
+          }
+        ],
+        "ten_horizon_mean_map_percent": 60.49617,
+        "ten_horizon_class_mean_ap_percent": {
+          "background_ap_percent": 39.01556,
+          "wearer_ap_percent": 63.6579,
+          "others_ap_percent": 78.81506
+        }
+      },
+      {
+        "variant": "thinker_av",
+        "input": "G+A+V",
+        "seed": 0,
+        "source_file": "raw_previous_thinker_av_seed0.txt",
+        "source_kind": "earlier_user_attachment",
+        "horizons": [
+          {
+            "nominal_seconds": 0.2,
+            "map_percent": 82.02415,
+            "background_ap_percent": 68.93423,
+            "wearer_ap_percent": 85.08583,
+            "others_ap_percent": 92.05239,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.4,
+            "map_percent": 75.99169,
+            "background_ap_percent": 60.58903,
+            "wearer_ap_percent": 78.83427,
+            "others_ap_percent": 88.55177,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.6,
+            "map_percent": 70.92665,
+            "background_ap_percent": 53.84208,
+            "wearer_ap_percent": 73.45503,
+            "others_ap_percent": 85.48283,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.8,
+            "map_percent": 66.79615,
+            "background_ap_percent": 49.21382,
+            "wearer_ap_percent": 68.56072,
+            "others_ap_percent": 82.6139,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1,
+            "map_percent": 63.86856,
+            "background_ap_percent": 45.62073,
+            "wearer_ap_percent": 65.27034,
+            "others_ap_percent": 80.71461,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.2,
+            "map_percent": 61.5628,
+            "background_ap_percent": 43.4667,
+            "wearer_ap_percent": 62.05277,
+            "others_ap_percent": 79.16893,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.4,
+            "map_percent": 60.3407,
+            "background_ap_percent": 42.54682,
+            "wearer_ap_percent": 60.2054,
+            "others_ap_percent": 78.26988,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.6,
+            "map_percent": 58.83407,
+            "background_ap_percent": 41.15952,
+            "wearer_ap_percent": 58.18522,
+            "others_ap_percent": 77.15747,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.8,
+            "map_percent": 57.39446,
+            "background_ap_percent": 39.94176,
+            "wearer_ap_percent": 55.99271,
+            "others_ap_percent": 76.2489,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 2,
+            "map_percent": 56.12449,
+            "background_ap_percent": 38.87001,
+            "wearer_ap_percent": 53.82707,
+            "others_ap_percent": 75.67639,
+            "num_samples": 23048
+          }
+        ],
+        "ten_horizon_mean_map_percent": 65.38637,
+        "ten_horizon_class_mean_ap_percent": {
+          "background_ap_percent": 48.41847,
+          "wearer_ap_percent": 66.14694,
+          "others_ap_percent": 81.59371
+        }
+      },
+      {
+        "variant": "thinker_audio",
+        "input": "G+A",
+        "seed": 0,
+        "source_file": "raw_previous_thinker_audio_seed0.txt",
+        "source_kind": "earlier_user_attachment",
+        "horizons": [
+          {
+            "nominal_seconds": 0.2,
+            "map_percent": 81.72011,
+            "background_ap_percent": 68.15248,
+            "wearer_ap_percent": 85.06554,
+            "others_ap_percent": 91.94232,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.4,
+            "map_percent": 75.91398,
+            "background_ap_percent": 59.90715,
+            "wearer_ap_percent": 79.22465,
+            "others_ap_percent": 88.61015,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.6,
+            "map_percent": 70.37189,
+            "background_ap_percent": 52.4848,
+            "wearer_ap_percent": 73.21461,
+            "others_ap_percent": 85.41626,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 0.8,
+            "map_percent": 66.45047,
+            "background_ap_percent": 47.92066,
+            "wearer_ap_percent": 68.779,
+            "others_ap_percent": 82.65175,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1,
+            "map_percent": 63.79981,
+            "background_ap_percent": 45.59262,
+            "wearer_ap_percent": 65.0989,
+            "others_ap_percent": 80.70791,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.2,
+            "map_percent": 61.3512,
+            "background_ap_percent": 43.15384,
+            "wearer_ap_percent": 61.86698,
+            "others_ap_percent": 79.03278,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.4,
+            "map_percent": 59.98279,
+            "background_ap_percent": 41.83907,
+            "wearer_ap_percent": 60.22346,
+            "others_ap_percent": 77.88584,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.6,
+            "map_percent": 58.29734,
+            "background_ap_percent": 40.44538,
+            "wearer_ap_percent": 57.66189,
+            "others_ap_percent": 76.78474,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 1.8,
+            "map_percent": 56.93221,
+            "background_ap_percent": 39.04677,
+            "wearer_ap_percent": 55.70067,
+            "others_ap_percent": 76.04919,
+            "num_samples": 23048
+          },
+          {
+            "nominal_seconds": 2,
+            "map_percent": 56.13656,
+            "background_ap_percent": 38.77318,
+            "wearer_ap_percent": 54.02582,
+            "others_ap_percent": 75.61069,
+            "num_samples": 23048
+          }
+        ],
+        "ten_horizon_mean_map_percent": 65.09564,
+        "ten_horizon_class_mean_ap_percent": {
+          "background_ap_percent": 47.73159,
+          "wearer_ap_percent": 66.08615,
+          "others_ap_percent": 81.46916
+        }
+      }
+    ],
+    "comparisons": [
+      {
+        "contrast": "encoder_av_minus_encoder_audio",
+        "protocol": "egospeak_full_timeline",
+        "ten_horizon_mean_delta_pp": 0.2551,
+        "per_horizon": [
+          {
+            "nominal_seconds": 0.2,
+            "map_delta_pp": -0.6849,
+            "background_ap_delta_pp": -0.0208,
+            "wearer_ap_delta_pp": -1.4015,
+            "others_ap_delta_pp": -0.63239
+          },
+          {
+            "nominal_seconds": 0.4,
+            "map_delta_pp": -0.34783,
+            "background_ap_delta_pp": 0.70624,
+            "wearer_ap_delta_pp": -1.59157,
+            "others_ap_delta_pp": -0.15812
+          },
+          {
+            "nominal_seconds": 0.6,
+            "map_delta_pp": -0.17487,
+            "background_ap_delta_pp": 0.82855,
+            "wearer_ap_delta_pp": -1.58313,
+            "others_ap_delta_pp": 0.22998
+          },
+          {
+            "nominal_seconds": 0.8,
+            "map_delta_pp": 0.01553,
+            "background_ap_delta_pp": 1.44947,
+            "wearer_ap_delta_pp": -1.66891,
+            "others_ap_delta_pp": 0.26602
+          },
+          {
+            "nominal_seconds": 1,
+            "map_delta_pp": 0.20651,
+            "background_ap_delta_pp": 1.56046,
+            "wearer_ap_delta_pp": -1.51747,
+            "others_ap_delta_pp": 0.57655
+          },
+          {
+            "nominal_seconds": 1.2,
+            "map_delta_pp": 0.13714,
+            "background_ap_delta_pp": 1.00313,
+            "wearer_ap_delta_pp": -1.31375,
+            "others_ap_delta_pp": 0.72202
+          },
+          {
+            "nominal_seconds": 1.4,
+            "map_delta_pp": 0.39348,
+            "background_ap_delta_pp": 1.42271,
+            "wearer_ap_delta_pp": -1.10816,
+            "others_ap_delta_pp": 0.86587
+          },
+          {
+            "nominal_seconds": 1.6,
+            "map_delta_pp": 0.95166,
+            "background_ap_delta_pp": 1.90632,
+            "wearer_ap_delta_pp": -0.41886,
+            "others_ap_delta_pp": 1.36752
+          },
+          {
+            "nominal_seconds": 1.8,
+            "map_delta_pp": 0.94338,
+            "background_ap_delta_pp": 1.81173,
+            "wearer_ap_delta_pp": -0.05834,
+            "others_ap_delta_pp": 1.07678
+          },
+          {
+            "nominal_seconds": 2,
+            "map_delta_pp": 1.11085,
+            "background_ap_delta_pp": 1.66905,
+            "wearer_ap_delta_pp": 0.66242,
+            "others_ap_delta_pp": 1.00108
+          }
+        ],
+        "positive_map_horizons": [
+          0.8,
+          1,
+          1.2,
+          1.4,
+          1.6,
+          1.8,
+          2
+        ],
+        "negative_wearer_horizons": [
+          0.2,
+          0.4,
+          0.6,
+          0.8,
+          1,
+          1.2,
+          1.4,
+          1.6,
+          1.8
+        ],
+        "ten_horizon_mean_class_delta_pp": {
+          "background_ap_delta_pp": 1.23369,
+          "wearer_ap_delta_pp": -0.99993,
+          "others_ap_delta_pp": 0.53153
+        }
+      },
+      {
+        "contrast": "thinker_audio_minus_encoder_audio",
+        "protocol": "egospeak_full_timeline",
+        "ten_horizon_mean_delta_pp": 4.59947,
+        "per_horizon": [
+          {
+            "nominal_seconds": 0.2,
+            "map_delta_pp": 2.0303,
+            "background_ap_delta_pp": 4.45771,
+            "wearer_ap_delta_pp": 0.96572,
+            "others_ap_delta_pp": 0.66749
+          },
+          {
+            "nominal_seconds": 0.4,
+            "map_delta_pp": 3.49822,
+            "background_ap_delta_pp": 7.42935,
+            "wearer_ap_delta_pp": 1.78859,
+            "others_ap_delta_pp": 1.27675
+          },
+          {
+            "nominal_seconds": 0.6,
+            "map_delta_pp": 4.03523,
+            "background_ap_delta_pp": 8.28414,
+            "wearer_ap_delta_pp": 1.8024,
+            "others_ap_delta_pp": 2.01916
+          },
+          {
+            "nominal_seconds": 0.8,
+            "map_delta_pp": 4.84463,
+            "background_ap_delta_pp": 9.31887,
+            "wearer_ap_delta_pp": 2.69089,
+            "others_ap_delta_pp": 2.52413
+          },
+          {
+            "nominal_seconds": 1,
+            "map_delta_pp": 5.51395,
+            "background_ap_delta_pp": 10.83419,
+            "wearer_ap_delta_pp": 2.56156,
+            "others_ap_delta_pp": 3.1461
+          },
+          {
+            "nominal_seconds": 1.2,
+            "map_delta_pp": 5.41412,
+            "background_ap_delta_pp": 10.22059,
+            "wearer_ap_delta_pp": 2.57602,
+            "others_ap_delta_pp": 3.44573
+          },
+          {
+            "nominal_seconds": 1.4,
+            "map_delta_pp": 5.75694,
+            "background_ap_delta_pp": 10.2693,
+            "wearer_ap_delta_pp": 3.42639,
+            "others_ap_delta_pp": 3.57512
+          },
+          {
+            "nominal_seconds": 1.6,
+            "map_delta_pp": 5.4672,
+            "background_ap_delta_pp": 9.76654,
+            "wearer_ap_delta_pp": 3.05014,
+            "others_ap_delta_pp": 3.58491
+          },
+          {
+            "nominal_seconds": 1.8,
+            "map_delta_pp": 4.96003,
+            "background_ap_delta_pp": 8.67647,
+            "wearer_ap_delta_pp": 2.89173,
+            "others_ap_delta_pp": 3.31191
+          },
+          {
+            "nominal_seconds": 2,
+            "map_delta_pp": 4.474,
+            "background_ap_delta_pp": 7.9032,
+            "wearer_ap_delta_pp": 2.52911,
+            "others_ap_delta_pp": 2.98971
+          }
+        ],
+        "positive_map_horizons": [
+          0.2,
+          0.4,
+          0.6,
+          0.8,
+          1,
+          1.2,
+          1.4,
+          1.6,
+          1.8,
+          2
+        ],
+        "negative_wearer_horizons": [],
+        "ten_horizon_mean_class_delta_pp": {
+          "background_ap_delta_pp": 8.71604,
+          "wearer_ap_delta_pp": 2.42825,
+          "others_ap_delta_pp": 2.6541
+        }
+      },
+      {
+        "contrast": "thinker_av_minus_encoder_av",
+        "protocol": "egospeak_full_timeline",
+        "ten_horizon_mean_delta_pp": 4.6351,
+        "per_horizon": [
+          {
+            "nominal_seconds": 0.2,
+            "map_delta_pp": 3.01924,
+            "background_ap_delta_pp": 5.26026,
+            "wearer_ap_delta_pp": 2.38751,
+            "others_ap_delta_pp": 1.40995
+          },
+          {
+            "nominal_seconds": 0.4,
+            "map_delta_pp": 3.92376,
+            "background_ap_delta_pp": 7.40499,
+            "wearer_ap_delta_pp": 2.98978,
+            "others_ap_delta_pp": 1.37649
+          },
+          {
+            "nominal_seconds": 0.6,
+            "map_delta_pp": 4.76486,
+            "background_ap_delta_pp": 8.81287,
+            "wearer_ap_delta_pp": 3.62595,
+            "others_ap_delta_pp": 1.85575
+          },
+          {
+            "nominal_seconds": 0.8,
+            "map_delta_pp": 5.17478,
+            "background_ap_delta_pp": 9.16256,
+            "wearer_ap_delta_pp": 4.14152,
+            "others_ap_delta_pp": 2.22026
+          },
+          {
+            "nominal_seconds": 1,
+            "map_delta_pp": 5.37619,
+            "background_ap_delta_pp": 9.30184,
+            "wearer_ap_delta_pp": 4.25047,
+            "others_ap_delta_pp": 2.57625
+          },
+          {
+            "nominal_seconds": 1.2,
+            "map_delta_pp": 5.48858,
+            "background_ap_delta_pp": 9.53032,
+            "wearer_ap_delta_pp": 4.07556,
+            "others_ap_delta_pp": 2.85986
+          },
+          {
+            "nominal_seconds": 1.4,
+            "map_delta_pp": 5.72137,
+            "background_ap_delta_pp": 9.55434,
+            "wearer_ap_delta_pp": 4.51649,
+            "others_ap_delta_pp": 3.09329
+          },
+          {
+            "nominal_seconds": 1.6,
+            "map_delta_pp": 5.05227,
+            "background_ap_delta_pp": 8.57436,
+            "wearer_ap_delta_pp": 3.99233,
+            "others_ap_delta_pp": 2.59012
+          },
+          {
+            "nominal_seconds": 1.8,
+            "map_delta_pp": 4.4789,
+            "background_ap_delta_pp": 7.75973,
+            "wearer_ap_delta_pp": 3.24211,
+            "others_ap_delta_pp": 2.43484
+          },
+          {
+            "nominal_seconds": 2,
+            "map_delta_pp": 3.35108,
+            "background_ap_delta_pp": 6.33098,
+            "wearer_ap_delta_pp": 1.66794,
+            "others_ap_delta_pp": 2.05433
+          }
+        ],
+        "positive_map_horizons": [
+          0.2,
+          0.4,
+          0.6,
+          0.8,
+          1,
+          1.2,
+          1.4,
+          1.6,
+          1.8,
+          2
+        ],
+        "negative_wearer_horizons": [],
+        "ten_horizon_mean_class_delta_pp": {
+          "background_ap_delta_pp": 8.16923,
+          "wearer_ap_delta_pp": 3.48897,
+          "others_ap_delta_pp": 2.24711
+        }
+      },
+      {
+        "contrast": "thinker_av_minus_thinker_audio",
+        "protocol": "egospeak_full_timeline",
+        "ten_horizon_mean_delta_pp": 0.29073,
+        "per_horizon": [
+          {
+            "nominal_seconds": 0.2,
+            "map_delta_pp": 0.30404,
+            "background_ap_delta_pp": 0.78175,
+            "wearer_ap_delta_pp": 0.02029,
+            "others_ap_delta_pp": 0.11007
+          },
+          {
+            "nominal_seconds": 0.4,
+            "map_delta_pp": 0.07771,
+            "background_ap_delta_pp": 0.68188,
+            "wearer_ap_delta_pp": -0.39038,
+            "others_ap_delta_pp": -0.05838
+          },
+          {
+            "nominal_seconds": 0.6,
+            "map_delta_pp": 0.55476,
+            "background_ap_delta_pp": 1.35728,
+            "wearer_ap_delta_pp": 0.24042,
+            "others_ap_delta_pp": 0.06657
+          },
+          {
+            "nominal_seconds": 0.8,
+            "map_delta_pp": 0.34568,
+            "background_ap_delta_pp": 1.29316,
+            "wearer_ap_delta_pp": -0.21828,
+            "others_ap_delta_pp": -0.03785
+          },
+          {
+            "nominal_seconds": 1,
+            "map_delta_pp": 0.06875,
+            "background_ap_delta_pp": 0.02811,
+            "wearer_ap_delta_pp": 0.17144,
+            "others_ap_delta_pp": 0.0067
+          },
+          {
+            "nominal_seconds": 1.2,
+            "map_delta_pp": 0.2116,
+            "background_ap_delta_pp": 0.31286,
+            "wearer_ap_delta_pp": 0.18579,
+            "others_ap_delta_pp": 0.13615
+          },
+          {
+            "nominal_seconds": 1.4,
+            "map_delta_pp": 0.35791,
+            "background_ap_delta_pp": 0.70775,
+            "wearer_ap_delta_pp": -0.01806,
+            "others_ap_delta_pp": 0.38404
+          },
+          {
+            "nominal_seconds": 1.6,
+            "map_delta_pp": 0.53673,
+            "background_ap_delta_pp": 0.71414,
+            "wearer_ap_delta_pp": 0.52333,
+            "others_ap_delta_pp": 0.37273
+          },
+          {
+            "nominal_seconds": 1.8,
+            "map_delta_pp": 0.46225,
+            "background_ap_delta_pp": 0.89499,
+            "wearer_ap_delta_pp": 0.29204,
+            "others_ap_delta_pp": 0.19971
+          },
+          {
+            "nominal_seconds": 2,
+            "map_delta_pp": -0.01207,
+            "background_ap_delta_pp": 0.09683,
+            "wearer_ap_delta_pp": -0.19875,
+            "others_ap_delta_pp": 0.0657
+          }
+        ],
+        "positive_map_horizons": [
+          0.2,
+          0.4,
+          0.6,
+          0.8,
+          1,
+          1.2,
+          1.4,
+          1.6,
+          1.8
+        ],
+        "negative_wearer_horizons": [
+          0.4,
+          0.8,
+          1.4,
+          2
+        ],
+        "ten_horizon_mean_class_delta_pp": {
+          "background_ap_delta_pp": 0.68687,
+          "wearer_ap_delta_pp": 0.06078,
+          "others_ap_delta_pp": 0.12454
+        }
+      }
+    ]
+  },
+  "protocol_differences": [
+    {
+      "variant": "encoder_visual",
+      "causal_minus_full_horizon_0_2_map_pp": 0.00108,
+      "causal_minus_full_ten_horizon_mean_map_pp": 0.00646
+    },
+    {
+      "variant": "encoder_av",
+      "causal_minus_full_horizon_0_2_map_pp": 0.00236,
+      "causal_minus_full_ten_horizon_mean_map_pp": 0.01176
+    },
+    {
+      "variant": "encoder_audio",
+      "causal_minus_full_horizon_0_2_map_pp": 0.0024,
+      "causal_minus_full_ten_horizon_mean_map_pp": 0.01154
+    },
+    {
+      "variant": "thinker_av",
+      "causal_minus_full_horizon_0_2_map_pp": 0.00246,
+      "causal_minus_full_ten_horizon_mean_map_pp": 0.01341
+    },
+    {
+      "variant": "thinker_audio",
+      "causal_minus_full_horizon_0_2_map_pp": 0.00246,
+      "causal_minus_full_ten_horizon_mean_map_pp": 0.01331
+    }
+  ],
+  "interpretation": {
+    "observation_1": "In this seed, G+A beats A at every full-timeline horizon and class; G+AV likewise beats AV. This supports a useful combined system, not isolated decoder causation.",
+    "observation_2": "Encoder AV trails A in mAP at 0.2/0.4/0.6, turns slightly positive at 0.8, and gains about one pp at 1.6-2.0. Wearer AP declines at 9/10 horizons.",
+    "observation_3": "Adding local V to G+A yields small macro gains mainly from background; mean wearer AP change is only +0.06078 pp.",
+    "unproven_explanations": [
+      "Longer history contribution versus head capacity or nonlinear fusion",
+      "Whether person/time pooling suppresses speaker-specific information",
+      "Whether a different fusion or training budget would avoid negative wearer transfer"
+    ],
+    "limitations": [
+      "One reported seed",
+      "Different head capacities and visual fields/history budgets",
+      "No training curves, remote model metadata, or original prediction arrays",
+      "Overall AP does not establish multi-person interaction understanding or boundary anticipation",
+      "No significance from treating correlated horizons as independent replicates",
+      "No feature-extraction or end-to-end speed claim from inference batch/s"
+    ],
+    "next_steps": [
+      "Collect raw metrics.json and checkpoint/training metadata, plus causal-valid curves",
+      "Complete seeds 1/2 with fixed epoch-4/no-dev protocol",
+      "Run architecture/parameter and raw-visible-history controlled modality comparisons before attributing gains to Thinker",
+      "Only then test audio-conditioned per-person fusion and pre-specified multi-person/overlap reporting; avoid test-driven epoch/threshold selection"
+    ]
+  },
+  "provenance": {
+    "normalization": "Console archives preserve all lines and wrapped paths; trailing horizontal whitespace removed and one final newline added. Numeric values unchanged.",
+    "sources": [
+      {
+        "variant": "encoder_visual",
+        "source_kind": "current_user_attachment",
+        "original_attachment_path": "/Users/wx44wx/.codex/attachments/a58e2943-0d9a-4d93-8b20-94ecd405c632/pasted-text.txt",
+        "archived_file": "raw_encoder_visual_seed0.txt"
+      },
+      {
+        "variant": "encoder_av",
+        "source_kind": "current_user_attachment",
+        "original_attachment_path": "/Users/wx44wx/.codex/attachments/ce28f77e-e131-497e-b9aa-83d9b1dc9171/pasted-text.txt",
+        "archived_file": "raw_encoder_av_seed0.txt"
+      },
+      {
+        "variant": "encoder_audio",
+        "source_kind": "current_user_attachment",
+        "original_attachment_path": "/Users/wx44wx/.codex/attachments/094440a9-5a37-4737-a32b-e905cdefbf3d/pasted-text.txt",
+        "archived_file": "raw_encoder_audio_seed0.txt"
+      },
+      {
+        "variant": "thinker_av",
+        "source_kind": "earlier_user_attachment",
+        "original_attachment_path": "/Users/wx44wx/.codex/attachments/e3e61fca-d5d2-452f-b040-c156f3d415e2/pasted-text.txt",
+        "archived_file": "raw_previous_thinker_av_seed0.txt"
+      },
+      {
+        "variant": "thinker_audio",
+        "source_kind": "earlier_user_attachment",
+        "original_attachment_path": "/Users/wx44wx/.codex/attachments/4a1649ae-1881-4c79-a7c4-0b0f56140eb8/pasted-text.txt",
+        "archived_file": "raw_previous_thinker_audio_seed0.txt"
+      }
+    ],
+    "supplementary_source": "Earlier user summary tables and run commands in this thread, transcribed in reported_causal_summary.txt",
+    "locally_available_artifact": "Derived research record only, not a replacement for remote metrics.json"
+  }
+}