{
  "id": "dolly-v2-12b",
  "name": "dolly-v2-12b",
  "identifiers": {
    "huggingface": [
      "databricks/dolly-v2-12b"
    ]
  },
  "developer": "Databricks",
  "release_date": {
    "value": "2023-04-12",
    "source": "http://web.archive.org/web/20230602091912/https://www.databricks.com/blog/2023/04/12/dolly-first-open-commercially-viable-instruction-tuned-llm",
    "status": "recorded",
    "note": "HF API returned 401 for this repo (and for the whole databricks/ org, like mosaicml/); dated from Databricks' own blog post via the Internet Archive."
  },
  "weights_status": "unknown",
  "availability": {
    "value": "removed",
    "checked": "2026-09-24",
    "source": "https://huggingface.co/databricks/dolly-v2-12b",
    "note": "HF API returned HTTP 401."
  },
  "license": {
    "value": "Apache 2.0",
    "source": "https://raw.githubusercontent.com/databrickslabs/dolly/master/LICENSE",
    "status": "recorded"
  },
  "architecture": {
    "family": "unknown",
    "n_layers": {
      "value": null,
      "status": "not_recorded"
    },
    "hidden_size": {
      "value": null,
      "status": "not_recorded"
    },
    "n_heads": {
      "value": null,
      "status": "not_recorded"
    },
    "vocab_size": {
      "value": null,
      "status": "not_recorded"
    },
    "context_length": {
      "value": null,
      "status": "not_recorded"
    },
    "positional_encoding": {
      "value": "rotary (RoPE), partial: 25% of dims",
      "source": "https://arxiv.org/abs/2304.01373",
      "status": "recorded",
      "note": "Architecture unchanged from pythia-12b (instruction fine-tune); fraction is config rotary_pct 0.25, per Pythia's own paper."
    }
  },
  "training_data": {
    "value": "databricks-dolly-15k: 15,000 human-generated prompt/response pairs, crowdsourced among Databricks employees March-April 2023 (CC BY-SA 3.0).",
    "source": "http://web.archive.org/web/20230602091912/https://www.databricks.com/blog/2023/04/12/dolly-first-open-commercially-viable-instruction-tuned-llm",
    "status": "recorded"
  },
  "techniques": [],
  "primary_sources": [
    "https://huggingface.co/databricks/dolly-v2-12b"
  ],
  "record_history": [
    {
      "date": "2026-09-24",
      "change": "ingested as candidate from HF (databricks/dolly-v2-12b@None)",
      "by": "ingest_hf.py"
    },
    {
      "date": "2026-09-24",
      "change": "preparer: developer, release date, license, fine_tuned_from pythia-12b, training_data all set from archived primary source (HF repo unreachable)",
      "by": "claude (preparer, Sonnet 5)"
    },
    {
      "date": "2026-09-24",
      "change": "preparer cleanup: removed 2 stale flag(s)",
      "by": "claude (preparer, Sonnet 5)"
    },
    {
      "date": "2026-09-24",
      "change": "preparer cleanup: positional_encoding propagated from an unchanged parent architecture",
      "by": "claude (preparer, Sonnet 5)"
    },
    {
      "date": "2026-09-24",
      "change": "reviewed and promoted from staging (1 edge(s) accepted)",
      "by": "Wilson Pruitt"
    }
  ]
}