Commit History

  1. Update requirements.txt Running

    d6d3e36
  2. Update README.md

    98d4d9a
  3. Update requirements.txt

    2a53f48

  1. Fix LaTeX

    2cb1651
  2. Use package from PyPI

    d954fd7

  1. Add React build result to LFS

    6ce8bd8
  2. Allow overwrite

    0ca3572

  1. Add estimated unigram logprobs

    0441206
  2. Add/improve help strings

    106dd6f
  3. Use context-probing package; enable loading or estimating unigram probabilities

    e4bf282

  1. Standard -> Basic

    b7dac90
  2. Add TinyStories models

    4de934d
  3. Better handling of bad characters

    6c2e238
  4. Update highlighted text component

    3033d56

  1. Fix sneaky problem caused by caching

    10ac8e4
  2. Use caching when possible

    ab89a9d

  1. Fix

    c868028
  2. Set max-height

    47e06df
  3. Improve UI

    19a8bac
  4. Add sampling options

    ffa03c8

  1. Fix Python 3.8 compatibility

    ebb68fb
  2. Implement generation mode

    dc1d5c3
  3. Further limit the number of tokens

    4b314cc
  4. Adjust limit, turn off caching for logprobs

    dca8e6c
  5. More efficient NLL implementation

    3102d58

  1. Use KL divergence by default

    6f46ddf
  2. Implement KL divergence score

    f90dfb4
  3. Show max_tokens in text area label but prevent value reset

    b837582
  4. Rebuild component

    11a14e0
  5. Include EOS token

    fdbcf74
  6. Fix visualization issue with window shorter than text

    068dd93
  7. Fix text area reset on window size change

    8d095c3
  8. Decrease limit

    e2a018d
  9. Better use of memory; limit window size and number of tokens

    d2e3092
  10. Fix code link, add label

    a7f1a72
  11. Add links

    bc6cc98
  12. Enable compact layout

    af8abd6

  1. Change default text

    f3cb237
  2. Cache individual batches

    bb204b7
  3. Better progress indication

    535e574
  4. Add progress bar, batched inference

    e5222c4
  5. Handle short input

    c290138
  6. Handle null input

    8d171c2
  7. Update readme

    248cf43
  8. Do not use a form

    a114a02
  9. Update UI

    dc0dd95
  10. Use inference_mode as decorator

    dd5d2e0
  11. Better caching

    f962dd0
  12. Cache the results

    b6ab215
  13. Update readme

    faa3816