Model · dolly-v1-6b

Dolly v1 6B

Developer: Databricks
Availability: removed · checked 2026-09-24 · source
HF API returned HTTP 401 for databricks/dolly-v1-6b (and 401 for the HF model page). 'removed' means unreachable at its original address, not a confirmed deletion. Model card read from the Internet Archive snapshot cited in primary_sources.

Raw record: /data/models/dolly-v1-6b.json

Fields

id

dolly-v1-6b

identifiers

huggingface

databricks/dolly-v1-6b

developer

Databricks

release_date

2023-03-24recorded · source

note:  Databricks release post 'Hello Dolly', dated March 24, 2023. HF repo is unreachable (HTTP 401), so the model card was read from an Internet Archive snapshot dated 2023-04-02.

weights_status

unknown

availability

removed · checked 2026-09-24 · source

note:  HF API returned HTTP 401 for databricks/dolly-v1-6b (and 401 for the HF model page). 'removed' means unreachable at its original address, not a confirmed deletion. Model card read from the Internet Archive snapshot cited in primary_sources.

license

cc-by-nc-4.0partial · source

note:  Archived card metadata: `license: cc-by-nc-4.0`. The card body states no model license of its own, only that the Stanford Alpaca data is CC-NC-BY-4.0.

architecture

family

decoder_only

n_layers

28recorded · source

note:  Card: 'Like its base model, dolly-v1-6b has six billion parameters consisting of 28 transformer layers with 16 attention heads each.'

hidden_size

nullnot_recorded

note:  Config unreachable and the archived card does not state this value.

n_heads

16recorded · source

vocab_size

nullnot_recorded

note:  Config unreachable and the archived card does not state this value.

context_length

nullnot_recorded

note:  Config unreachable and the archived card does not state this value.

positional_encoding

rotary (RoPE)recorded · source

note:  Card: 'It employs Rotary Position Embedding (RoPE) and shares the same tokenizer as GPT-3.' Partial-dimension detail (64 of 256) is on gpt-j-6b's record; not restated here because the card does not state it.

note

Card (archived): 'a 6 billion parameter causal language model ... derived from EleutherAI's GPT-J'.

training_data

Stanford Alpaca: 'a ~52K record instruction corpus ... consisting of question/answer pairs generated using the techniques outlined in the Self-Instruct paper'. Original run: 30 minutes, 1 epoch (blog); the most recent checkpoint on the card was trained for 10 epochs.recorded · source

techniques

primary_sources

http://web.archive.org/web/20230402044135/https://huggingface.co/databricks/dolly-v1-6b

https://www.databricks.com/blog/2023/03/24/hello-dolly-democratizing-magic-chatgpt-open-models.html

record_history

date:2026-09-24 · change:ingested as candidate from HF (databricks/dolly-v1-6b@None) · by:ingest_hf.py ·

date:2026-09-24 · change:preparer: developer, release date (blog), license (partial, archived card metadata), architecture (layers/heads/RoPE), training data, 2 edges prepared; sources: archived card 2023-04-02, Databricks blog · by:claude (preparer, Sonnet 5) ·

date:2026-09-24 · change:reviewed and promoted from staging (2 edge(s) accepted) · by:Wilson Pruitt ·

Parents

Weights descend

• fine_tuned_from → GPT-J 6B declared source Card: 'derived from EleutherAI's GPT-J (released June 2021) and fine-tuned on a ~52K record instruction corpus'; blog: '6 billion parameter model from EleutherAI'.

Training data

• trained_on → Alpaca instruction data (52K) declared source Card: 'fine-tuned on a ~52K record instruction corpus (Stanford Alpaca)'; blog: 'data from Alpaca'. The Alpaca set is text-davinci-003 output (see the `alpaca-52k` dataset edge), so the closed-model path runs through it.

Children

No edges recorded.

Read in

No station on the reading path has touched this record yet.