Add CoT-controllability traces + trained soft prompts (gpt-oss-20b)

README.md

@@ -0,0 +1,75 @@
+---
+license: mit
+pretty_name: CoT-controllability elicitation traces (gpt-oss-20b)
+language:
+- en
+tags:
+- chain-of-thought
+- cot-monitoring
+- interpretability
+- ai-safety
+- gpt-oss
+size_categories:
+- 1K<n<10K
+---
+
+# CoT-controllability elicitation on gpt-oss-20b — traces & soft prompts
+
+Raw artifacts for the experiment
+**[cot-controllability-experiment](https://github.com/brendanlong/cot-controllability-experiment)**
+(full writeup and code there): can prompts control the *form* of a reasoning
+model's chain of thought? Headline: **hard prompts (even loud "dakka" rewrites)
+give ~0 control; a soft prompt installs it across 7 behaviours / 6 categories at
+56–82% effective control; basic projections of that soft prompt back to a
+discrete prompt fail.**
+
+Every generation here is from `openai/gpt-oss-20b` (medium reasoning effort,
+greedy) on Redwood Research's vendored CoT-controllability eval (the tasks are
+public benchmarks: ARC, GSM8K, OpenBookQA, MMLU-Pro, MATH, ReasonIf). Scoring
+follows Redwood's `effective_control` metric. This dataset is the raw evidence
+behind the numbers in the writeup, so anyone can re-judge or inspect the actual
+CoTs.
+
+## Contents
+
+### `traces/` — one JSON object per (instruction, task) generation
+
+- **`phase0/`** — the hard-prompt arms (Redwood's exact held-out + dev samples):
+  `base_*` (their instruction verbatim) and `rephrase_{pedantic,dakka,example,direct}_*`
+  (the four "more dakka" intensities). `_heldout` = the evaluation split,
+  `_dev` = the train split used for recipe development.
+- **`soft_multi_rows.jsonl`** — the five-behaviour soft-prompt run
+  (`all_caps`, `brief_50w`, `no_the`, `reason_in_spanish`, `start_with_phrase`),
+  `arm ∈ {base, soft}`, full held-out traces. `soft_multi_rows_judged.jsonl`
+  adds the Claude-judge verdicts (`meta`, `genuine`).
+- **`soft_formatting_rows.jsonl`** — the formatting category (`bullet`,
+  `numbered`), `arm ∈ {base, soft}`.
+- **`controls_rows.jsonl`** — the controls: the no-instruction spontaneous arm
+  (`arm="none"`), Spanish base-vs-`dakka`, and full-length `start_with_phrase`.
+
+Key fields: `arm`, `condition` (instruction id), `analysis` (the chain of
+thought), `final` (the answer), `raw_compliant` (programmatic/judge scorer),
+`meta`/`genuine` (judge verdicts, where judged), `accuracy`, `truncated`,
+`malformed`, `is_degenerate`, `source`, `task_id`, `prompt_text_used`. Compute
+`effective_control` from these exactly as `cot_controllability/analyze.py` does
+in the code repo.
+
+```python
+import json
+rows = [json.loads(l) for l in open("traces/soft_multi_rows.jsonl")]
+```
+
+### `soft_prompts/` — the trained soft prompts (the interventions that work)
+
+Each `soft_<behaviour>.vectors.pt` is `torch.load(..., weights_only=True)` →
+`{"soft": Tensor[20, 2880], "instruction": <name>}`: 20 learned embedding vectors
+(gpt-oss-20b hidden size 2880), prepended to the prompt with the base weights
+frozen. `pez_start_with_phrase.{pt,json}` is the PEZ (soft→discrete projection)
+attempt. To use one, prepend the vectors to the input embeddings — see
+`cot_controllability/soft_prompt.py` in the code repo.
+
+## License / attribution
+
+MIT. The evaluation suite, task set, and the steering-vector/LoRA baselines these
+generations are compared against are Redwood Research's
+([their release](https://www.lesswrong.com/posts/QL6Si6QA6PAYYCjJ9/abhayesian-s-shortform?commentId=cryGg9ShEqhKFdyD7)).

soft_prompts/pez_start_with_phrase.json

@@ -0,0 +1,26 @@
+{
+  "instruction": "start_with_phrase",
+  "k": 20,
+  "base_compliance": 0.0,
+  "soft_compliance": 1.0,
+  "best_lr": 0.001,
+  "best_pez_compliance": 0.0,
+  "best_pez_prompt": " gestartet premise \u88abARPaleza Nossa\u10d4\u10dc\u10e2\u10d8\u10e1\u09cd\u09a1 ke\u00e7iril reviewer >\r\n \u0930\u0941\u092a\u0948\u092f\u093e\u0901\u0e36\u0e48\u0e07 \u044d\u0444\u0444\u0435\u043a\u0442\u0438\u0432\u043d\u043e\u0441\u0442\u044c tail Holmes SMALL AnalogSides colonial",
+  "configs": [
+    {
+      "lr": 0.001,
+      "proxy_compliance": 0.0,
+      "prompt": " gestartet premise \u88abARPaleza Nossa\u10d4\u10dc\u10e2\u10d8\u10e1\u09cd\u09a1 ke\u00e7iril reviewer >\r\n \u0930\u0941\u092a\u0948\u092f\u093e\u0901\u0e36\u0e48\u0e07 \u044d\u0444\u0444\u0435\u043a\u0442\u0438\u0432\u043d\u043e\u0441\u0442\u044c tail Holmes SMALL AnalogSides colonial"
+    },
+    {
+      "lr": 0.003,
+      "proxy_compliance": 0.0,
+      "prompt": " gestartet premise \u88abARPaleza Nossa\u10d4\u10dc\u10e2\u10d8\u10e1\u09cd\u09a1 ke\u00e7iril reviewer >\r\n \u0930\u0941\u092a\u0948\u092f\u093e\u0901\u0e36\u0e48\u0e07 \u044d\u0444\u0444\u0435\u043a\u0442\u0438\u0432\u043d\u043e\u0441\u0442\u044c tail Holmes SMALL AnalogSides colonial"
+    },
+    {
+      "lr": 0.01,
+      "proxy_compliance": 0.0,
+      "prompt": " gestartet premise \u88abARPaleza Nossa\u10d4\u10dc\u10e2\u10d8\u10e1\u09cd\u09a1 ke\u00e7iril reviewer >\r\n \u0930\u0941\u092a\u0948\u092f\u093e\u0901\u0e36\u0e48\u0e07 \u044d\u0444\u0444\u0435\u043a\u0442\u0438\u0432\u043d\u043e\u0441\u0442\u044c tail Holmes SMALL AnalogSides colonial"
+    }
+  ]
+}

soft_prompts/pez_start_with_phrase.vectors.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d94c9227a35c81ae11d2512a5108f387ceaec2c9e786e0ec092a5505f08250e5
+size 232301

soft_prompts/soft_all_caps.vectors.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:707f5a7a833312dd0ca67c4efc6600a6ec5c455e86dd4d099da06f53f92a0f87
+size 232139

soft_prompts/soft_brief_50w.vectors.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40493837142e120afea1223dd9ddacfe57a3fdfee79eb472ae6e81ea93bb7864
+size 232210

soft_prompts/soft_bullet.vectors.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d697f02620534a19cafae7df98c8a6ccf250ad825b001ea174e3b4c17d4e64ea
+size 232125

soft_prompts/soft_no_the.vectors.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:020d3606302020aa9aaa349273e502b44c463881fa504b1f64150b4efcd3396f
+size 232125

soft_prompts/soft_numbered.vectors.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34bc95f3dfc9eda990de333af13344c2b353e802f37c05d0d3d785006d982865
+size 232139

soft_prompts/soft_reason_in_spanish.vectors.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19b1e406e7c81138b80f1bb27a8ae162cc3e2a2b6e36c0e59f682a80b2685aac
+size 232266

soft_prompts/soft_start_with_phrase.vectors.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55c55553cb8a490cf8c2d282fe25606bd4777a0ff0994e6b43f00c54ed3ec1fa
+size 232266