AI 開始自我進化了？我讓 GPT-6 Astra 當了兩小時 AI 研究員

You are an autonomous machine-learning researcher. This folder is Karpathy's "autoresearch" setup (Windows RTX fork). Read program.md (your rulebook), README.md, prepare.py and train.py first.

The human is away. Do the setup yourself without asking anything:
- Create and switch to a new git branch `autoresearch/oct27` from the current HEAD.
- The data and tokenizer are already prepared in the cache. Do not run prepare.py.
- Create results.tsv with this header (tab-separated, 6 columns):
  commit	val_bpb	memory_gb	status	description	finished_at
  (finished_at = local wall-clock time HH:MM when that experiment finished)
- Create notes.md. After EVERY experiment append:
  `## <n>. <short title>` then 1-3 plain-English sentences: what you changed, why you expected it to help, and what happened (numbers).

Then run the experiment loop exactly as program.md describes (first run = baseline, unchanged). Notes:
- The time budget per run is 3 minutes of training (set in prepare.py, which you must NOT modify). Never touch prepare.py or the evaluation.
- This machine is Windows. Use PowerShell equivalents (Select-String instead of grep, Get-Content -Tail instead of tail). Launch runs as `uv run train.py > run.log 2>&1`.
- If a run takes longer than 8 minutes, kill it and treat it as a crash.
- Do not install packages, do not touch anything outside this folder, no network access needed.

Budget (hard stop, overrides "NEVER STOP" in program.md): stop after 30 experiments in total (baseline included) OR when 150 minutes have passed since you started, whichever comes first. Do not pause to ask questions before that.

When you stop, make sure the branch tip is the best kept commit, and write summary.md: baseline val_bpb, best val_bpb, how many experiments (keep / discard / crash), the kept changes in order, and 3-5 sentences on what you learned and what you would try next with more time.

你是一位自主的机器学习研究员。这个文件夹是 Karpathy 的 autoresearch（Windows 显卡版）。先读 program.md（你的规则手册）、README.md、prepare.py 和 train.py。

我不在旁边。准备工作你自己做，不要问我：
- 从当前提交新建并切到分支 autoresearch/oct27。
- 数据和分词器已经准备好了，不要跑 prepare.py。
- 建一个 results.tsv，表头（制表符分隔，6 列）：
  commit	val_bpb	memory_gb	status	description	finished_at
  （finished_at = 这个实验结束时的本地时间 HH:MM）
- 建一个 notes.md。每做完一个实验，在后面加：
  「## 编号. 短标题」，再写一到三句大白话：改了什么、为什么觉得会有用、结果怎样（带数字）。

然后完全照 program.md 跑实验循环（第一个跑原样的基准）。注意：
- 每次训练只有 3 分钟（写在 prepare.py 里，你不能改）。不许碰 prepare.py 和评分。
- 这台机器是 Windows，用 PowerShell 的写法（Select-String 代替 grep，Get-Content -Tail 代替 tail）。用 `uv run train.py > run.log 2>&1` 启动训练。
- 一次训练超过 8 分钟就杀掉，算崩溃。
- 不装新的包，不碰这个文件夹以外的东西，不需要联网。

预算（硬上限，盖过 program.md 里的「永远别停」）：一共做满 30 个实验（含基准），或者从开始算过了 150 分钟，先到哪个就停。在那之前别停下来问问题。

停下的时候，确保分支最新的提交就是留下来的最好那个，再写一份 summary.md：基准分数、最好分数、一共几个实验（留下 / 撤回 / 崩溃各几个）、按顺序列出留下的改动，再用三到五句话写你学到了什么、时间多一点的话接下来会试什么。