Hugging Face's logo

jinaai
/
jina-bert-implementation

Transformers
English
Model card Files
xet
 Community
14
jina-bert-implementation

Commit History

  1. Use attention dropout during training (#10)⁠

    Markus28
    Verified signature
    f3ec4cf

  1. Allow device auto map (#8)⁠

    Jackmin108
    c41d17d

  1. Truncate to 8k by default (#5)⁠

    Jackmin108
    43f3955

  2. Set max length to 2B

    Jackmin108
    619ca8d

  1. Update README.md

    Jackmin108
    a9db862

  2. Update README.md

    Jackmin108
    f838124

  1. Allow pytorch<2 to use without passing attn_implementation flag (#4)⁠

    Jackmin108
    b5794c5

  1. chore: update from afe81ca705ca1a5bd6b7d90548fcac068850b2af

    Team Finetuner
    344bcbc

  2. Remove triton flash implementation

    Jackmin108
    5ee2c37

  3. Delete flash_attn_triton.py

    Jackmin108
    4fa2261

  4. chore: update from 896c12d73073854c513200fb74a4887cf25b2b97

    Team Finetuner
    96e9a75

  1. chore: update from f36c08c8a58c21b5aaab523fa03fb4a24b475612

    Team Finetuner
    3e3ced0

  1. chore: update from 07ce15d58b77559fce77ea89e92d398f28663bd9

    Team Finetuner
    0f4070e

  1. feat: allow changing flash implementation

    Jackmin801
    43b8513

  1. allow math kernel

    Jackmin801
    bc43a5e

  2. Flash attention! (#2)⁠

    Jackmin108
    df1a7f6

  1. Create README.md

    Jackmin108
    eb9e889

  1. Allow flash attn (#1)⁠

    Jackmin108
    622abd4

  1. rename to jina bert in configuration file

    alaeddine-13
    33026dc

  2. rename to jina bert

    alaeddine-13
    b4f2b16

  3. add basic configuration and model file

    alaeddine-13
    e36c994

  4. initial commit

    alaeddine-13
    de9348b