Skip to content

Commit 031ba68

Browse files
JamesPHoughtonclaude
JamesPHoughton
and
claude
committed
Mirror dialogue-levers/stagebook/pilot_3 for memory-leak repro deploy
Snapshot copy of stagebook/pilot_3 from Watts-Lab/dialogue-levers @ 06e33499ed88651f45e5daf9dd6d88c67a81ad2f. Mirror exists so the no-manager Railway deploy can fetch the treatment over the public S3 bucket — the source repo is private, and the existing projects/ subtrees use legacy *.treatments.yaml syntax that won't run on the current stagebook bundle. See projects/dialogue-levers/pilot_3/MIRROR.md for the refresh recipe. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
1 parent 179f08d commit 031ba68

224 files changed

Lines changed: 7641 additions & 0 deletions

File tree

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

‎projects/dialogue-levers/pilot_3/MIRROR.md‎

Lines changed: 58 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,58 @@
1+
# Mirror provenance
2+
3+
This subtree is a snapshot copy from a sibling private repo — kept here so the
4+
public S3 mirror of `deliberation-assets` can serve it as a treatment + assets
5+
bundle.
6+
7+
| Field | Value |
8+
| --- | --- |
9+
| Source repo | [`Watts-Lab/dialogue-levers`](https://github.com/Watts-Lab/dialogue-levers) (private) |
10+
| Source path | `stagebook/pilot_3/` |
11+
| Source SHA at mirror | `06e33499ed88651f45e5daf9dd6d88c67a81ad2f` |
12+
| Source commit date | 2026-05-20 |
13+
| Mirrored on | 2026-05-22 |
14+
15+
## Why a copy lives here
16+
17+
`dialogue-levers` is private, so the Deliberation Lab runtime (which fetches
18+
treatments and prompt files over plain unauthenticated HTTPS) can't pull
19+
directly from `raw.githubusercontent.com`. Mirroring into this public-read repo
20+
sends the tree through `.github/workflows/s3_mirror.yml` to
21+
`https://s3.amazonaws.com/assets.deliberation-lab.org/`, where the runtime can
22+
reach it.
23+
24+
The snapshot was taken to drive a no-manager Railway deploy used to reproduce
25+
the early-lifecycle memory regression we've been chasing on production —
26+
pilot_3 is the workload under which that regression has been observed. The
27+
other configs already in `deliberation-assets/projects/` are on the legacy
28+
`*.treatments.yaml` syntax and don't run on the current stagebook bundle.
29+
30+
## Refreshing
31+
32+
If `pilot_3` evolves upstream and you want the live deploy to track it:
33+
34+
```bash
35+
cd /path/to/Watts-Lab/dialogue-levers && git pull
36+
rsync -a --delete \
37+
/path/to/Watts-Lab/dialogue-levers/stagebook/pilot_3/ \
38+
/path/to/Watts-Lab/deliberation-assets/projects/dialogue-levers/pilot_3/
39+
# preserve this MIRROR.md when rsync --delete'ing — re-stamp the SHA + date
40+
```
41+
42+
Commit and push from `deliberation-assets`; the `Mirror repo to S3` workflow
43+
will publish to the bucket within ~1–2 minutes.
44+
45+
## Footprint at the bucket
46+
47+
```
48+
https://s3.amazonaws.com/assets.deliberation-lab.org/projects/dialogue-levers/pilot_3/
49+
```
50+
51+
So a batch config that wants this treatment would set:
52+
53+
```json
54+
{
55+
"assetBaseUrl": "https://s3.amazonaws.com/assets.deliberation-lab.org",
56+
"treatmentFile": "projects/dialogue-levers/pilot_3/pilot_3.stagebook.yaml"
57+
}
58+
```

‎projects/dialogue-levers/pilot_3/exit/ability_to_comply/ability_to_comply.prompt.md‎

Lines changed: 15 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,15 @@
1+
---
2+
type: multipleChoice
3+
---
4+
5+
### How easy was it to follow the instructions we gave you for the conversation?
6+
7+
---
8+
9+
- 1: 1 - Not very easy
10+
- 2
11+
- 3
12+
- 4: 4 - Somewhat easy
13+
- 5
14+
- 6
15+
- 7: 7 - Very easy

‎projects/dialogue-levers/pilot_3/exit/ability_to_comply/ability_to_comply_elaboration.prompt.md‎

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,9 @@
1+
---
2+
type: openResponse
3+
---
4+
5+
### Please describe anything that made it easier or harder to follow the instructions or carry out the conversation task.
6+
7+
---
8+
9+
> Enter your response here

‎projects/dialogue-levers/pilot_3/exit/attitude_attributes_post/attitude_attribute_instructions_post.prompt.md‎

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,9 @@
1+
---
2+
type: noResponse
3+
---
4+
5+
# We'd like to ask more about your opinion toward the discussion topic.
6+
7+
You may recognize some of the questions from earlier in the study. There’s no expectation that your answers will have changed — many people answer the same way both times. But if your thoughts are different in any way, this is a chance to let us know.
8+
9+
As a reminder, your topic was:

‎projects/dialogue-levers/pilot_3/exit/continue_with_partner/continue_with_partner.prompt.md‎

Lines changed: 14 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,14 @@
1+
---
2+
type: multipleChoice
3+
---
4+
5+
### After you answer a few more questions, would you like to continue your conversation with your partner?
6+
7+
If **both of you** choose "**Yes**", we’ll reconnect you in a new video room, where you can talk for as long as you like (up to 15 minutes).
8+
9+
_Your choice to participate will not affect your payment for the study._
10+
11+
---
12+
13+
- Yes
14+
- No

‎projects/dialogue-levers/pilot_3/exit/continue_with_partner/continue_with_partner_why.prompt.md‎

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,9 @@
1+
---
2+
type: openResponse
3+
---
4+
5+
### What made you choose Yes or No?
6+
7+
---
8+
9+
> Please enter your response here

‎projects/dialogue-levers/pilot_3/exit/perception_of_partner/consider_partner.prompt.md‎

Lines changed: 5 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,5 @@
1+
---
2+
type: noResponse
3+
---
4+
5+
# We'd like to ask some questions about **your discussion partner**:

‎projects/dialogue-levers/pilot_3/exit/rate_partner_description/score_description_accuracy.prompt.md‎

Lines changed: 21 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,21 @@
1+
---
2+
type: multipleChoice
3+
notes: |
4+
One of three rate-partner-description items (accuracy, completeness,
5+
fairness) — all on a shared 7-point Likert scale (1=Not at all,
6+
4=Somewhat, 7=Completely) to make the dimensions comparable and to
7+
reduce per-item reading load. See ADR
8+
2026-05-11-rate-partner-description-scale.md.
9+
---
10+
11+
### How **accurate** is this summary of your views?
12+
13+
---
14+
15+
- 1: 1 - Not at all
16+
- 2
17+
- 3
18+
- 4: 4 - Somewhat
19+
- 5
20+
- 6
21+
- 7: 7 - Completely

‎projects/dialogue-levers/pilot_3/exit/rate_partner_description/score_description_completeness.prompt.md‎

Lines changed: 21 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,21 @@
1+
---
2+
type: multipleChoice
3+
notes: |
4+
One of three rate-partner-description items (accuracy, completeness,
5+
fairness) — all on a shared 7-point Likert scale (1=Not at all,
6+
4=Somewhat, 7=Completely) to make the dimensions comparable and to
7+
reduce per-item reading load. See ADR
8+
2026-05-11-rate-partner-description-scale.md.
9+
---
10+
11+
### How **complete** is this summary of your views?
12+
13+
---
14+
15+
- 1: 1 - Not at all
16+
- 2
17+
- 3
18+
- 4: 4 - Somewhat
19+
- 5
20+
- 6
21+
- 7: 7 - Completely

‎projects/dialogue-levers/pilot_3/exit/rate_partner_description/score_description_fairness.prompt.md‎

Lines changed: 21 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,21 @@
1+
---
2+
type: multipleChoice
3+
notes: |
4+
One of three rate-partner-description items (accuracy, completeness,
5+
fairness) — all on a shared 7-point Likert scale (1=Not at all,
6+
4=Somewhat, 7=Completely) to make the dimensions comparable and to
7+
reduce per-item reading load. See ADR
8+
2026-05-11-rate-partner-description-scale.md.
9+
---
10+
11+
### How **fair** is this summary of your views?
12+
13+
---
14+
15+
- 1: 1 - Not at all
16+
- 2
17+
- 3
18+
- 4: 4 - Somewhat
19+
- 5
20+
- 6
21+
- 7: 7 - Completely

0 commit comments

Comments
 (0)