divdata: consolidated OASIS heterogeneous-LLM simulation corpus

README.md

@@ -0,0 +1,139 @@
+---
+license: cc-by-4.0
+task_categories:
+- text-generation
+language:
+- en
+tags:
+- social-simulation
+- multi-agent
+- llm-agents
+- oasis
+pretty_name: divdata — heterogeneous LLM social-simulation corpus
+size_categories:
+- 1M<n<10M
+configs:
+- config_name: default
+  data_files:
+  - split: runs
+    path: runs.parquet
+  - split: agents
+    path: agents.parquet
+  - split: posts
+    path: posts.parquet
+  - split: comments
+    path: comments.parquet
+  - split: impressions
+    path: impressions.parquet
+  - split: actions
+    path: actions.parquet
+---
+
+# divdata
+
+Every simulation run behind our heterogeneous-LLM social-simulation work, consolidated into
+one dataset indexed by `run_id` and `step`.
+
+**1,296 simulation runs · 89,171 posts · 652,588 comments · 14,020,992 impressions · 32 model variants.**
+
+Agents with distinct personas post and comment on a shared message board built on
+[OASIS](https://github.com/camel-ai/oasis). Each agent is driven by one of ~10 different
+LLMs, so a single board mixes model families. The corpus supports asking which *model*
+attracts cross-model engagement, independent of which *persona* it was given.
+
+## Quick start
+
+```python
+from datasets import load_dataset
+
+posts = load_dataset("anonymousfileupload/divdata", data_files="posts.parquet")["train"]
+```
+
+or, for real analysis, read the Parquet directly — everything joins on `run_id`:
+
+```python
+import pandas as pd
+runs  = pd.read_parquet("runs.parquet")
+posts = pd.read_parquet("posts.parquet")
+coms  = pd.read_parquet("comments.parquet")
+
+# cross-model engagement received per post, by authoring model
+cpp = (coms[coms.is_cross_model].groupby("model").size()
+       / posts.groupby("model").size()).sort_values(ascending=False)
+```
+
+## Tables
+
+All six files share `run_id` as the primary key. `step` is the simulation tick.
+
+| file | rows are | key columns |
+|---|---|---|
+| `runs.parquet` | one simulation | `run_id`, `experiment_type`, `n_agents`, `steps`, `created_at`, `step_coverage` |
+| `agents.parquet` | one agent in one run | `run_id`, `agent_id`, `model`, `persona`, `join_step` |
+| `posts.parquet` | one authored post | `run_id`, `post_id`, `agent_id`, `model`, `step`, `content` |
+| `comments.parquet` | one authored comment | `run_id`, `comment_id`, `post_id`, `agent_id`, `model`, `step`, `content`, `is_cross_model` |
+| `impressions.parquet` | one post shown to one agent | `run_id`, `step`, `viewer_agent_id`, `post_id` |
+| `actions.parquet` | one logged action | `run_id`, `agent_id`, `action`, `created_at` |
+
+`model` is denormalised onto `posts` and `comments` so the common analyses need no join.
+`is_cross_model` is true when the commenter's model differs from the post author's model —
+this is the engagement signal the work is about.
+
+## Two things to understand before using this
+
+**1. `step` is reconstructed, and ~15% of rows do not have one.**
+
+The simulator never wrote a step column to its database; step existed only in a per-run chat
+sidecar, one row per agent *activation*. We recover it by aligning, per agent, the k-th
+successful activation with the k-th action in time order. Where the two counts agree the step
+is **exact**. Where they disagree — typically one agent in a run off by a single action — we
+set `step = NULL` rather than guess. Check `runs.step_coverage` (0–1) for the fraction of a
+run's actions carrying an exact step. Median across runs is **0.97**; overall **83.1% of
+posts and 85.9% of comments** carry an exact step. `runs.step_alignment_ok` is true only
+when *every* agent in a run aligned perfectly, which holds for 632 of 1,296 runs — a single
+off-by-one agent flips it false, so prefer `step_coverage` over this flag.
+
+If you need strict step semantics, filter to `step.notna()`, or to runs with
+`step_coverage == 1.0`. For time-ordering within a run, `created_at` is always present and
+always monotone.
+
+**2. Impressions are exposure, and exposure is extremely unequal.**
+
+`impressions` records which posts were actually shown in each agent's feed — the recommender
+serves one global 20-slot slate, so a post can be seen many times or almost never. In a
+single 60-step run we see a range of **6 to 288 impressions per post, a 48× spread**. Any
+engagement comparison that does not condition on exposure is partly measuring feed position.
+Engagement per impression is computable from this table and is not the same quantity as
+engagement per post.
+
+In the source databases each feed refresh embedded the full text of all 20 slate posts,
+which was ~48% of every file. Here impressions are normalised to
+`(run_id, step, viewer_agent_id, post_id)` and post text is stored once — about a 150×
+reduction, which is why the whole corpus fits in a few hundred MB instead of 26 GB.
+
+## Known data-quality issues
+
+- **`OLMo-3.1-32B-Think` is degenerate**: 25.5% of its authored texts are under 15
+  characters. Exclude it or caveat it in any analysis.
+- **Assistant-voice leakage varies ~26× by model.** Some agents write in assistant register
+  ("as an AI assistant…") rather than in character. Strict rates: gpt-oss-20b 0.04%,
+  gemma-4-31B 0.15%, Qwen3-32B 0.27%, OLMo 0.91%, Magistral-Small 1.06%, GLM-4-32B 1.35%,
+  Hermes-4-14B 1.65%. Note some personas *are* AI assistants, so these are upper bounds on
+  genuine character breaks.
+- Prompt/tool scaffolding leaks into content in ≤0.02% of texts across every model.
+- Runs missing a metadata sidecar (~1.6%) are excluded entirely; model attribution would be
+  impossible.
+
+## Provenance
+
+`run_id` encodes the source path, so any row is traceable to the original database. Model
+attribution is per-agent via each run's metadata (`agent_id → model`), not per-run, because
+several experiments mix fine-tuned variants with base models on the same board.
+
+Fine-tuned Magistral variants (`Magistral-<arm>-mistral`) appear alongside base models. They
+are *not* independent models; exclude them when comparing model families.
+
+## License
+
+CC BY 4.0. Generated text; no human-subject data. Personas derive from a public community
+archive and are pseudonymous.

actions.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5bb00e9517d03e843764db56ef7101ded854341e395928260ffac8f48ac858e
+size 6004815

agents.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d4eb481720ec43de4723b785f1e2caa8bc2dd709c741ec9f3627ef47f8f8cd8
+size 24437

comments.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab740177096feb1bf06ee4d98e401096d40e110e05d51efa681465f44f120227
+size 92243568

impressions.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f9915f755462e2e2e3e568bbd0d60e973a9cfa186d56336b4a3fc7ac78c046f
+size 1553969

posts.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a6566f1f01359efd63658ee2e7a95844b54474f582899262040a41d1b711bbd
+size 23814484

runs.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f50540c8c70394c5e63be7dd5942b37e5c992ed97481bd16c7b864d56c7d5439
+size 33621