Historique des commits

Auteur SHA1 Message Date
  justheuristic 28971dcedd black-isort il y a 3 ans
  Artem Chumachenko 5af1c9e3b4 fix test il y a 3 ans
  justheuristic aa769ce846 fix bug when cache assignment surpasses max length il y a 3 ans
  justheuristic 93951779e1 fix bug when cache assignment surpasses max length il y a 3 ans
  justheuristic 01d6ba8a9c dedup il y a 3 ans
  Artem Chumachenko 1b28432533 wip il y a 3 ans
  justheuristic 33aa952d41 client-side prompts il y a 3 ans
  Artem Chumachenko 85c2ec7b06 actually merge il y a 3 ans
  justheuristic e0bb3762b4 Merge branch 'generation-inference' into deep_prompt_inference il y a 3 ans
  Artem Chumachenko 01c1e198b8 Fix merge conficts il y a 3 ans
  Artem Chumachenko 88e6a75996 Add part of deepprompts il y a 3 ans
  Artem Chumachenko c003830cc6 fix il y a 3 ans
  Artem Chumachenko 1afd59a071 introduce hypo_ids il y a 3 ans
  Artem Chumachenko f62c65ec23 fixes il y a 3 ans
  Artem Chumachenko 53e19de6e0 Add tests il y a 3 ans
  Artem Chumachenko ade986ca58 Return multibatch mode il y a 3 ans
  Artem Chumachenko fb142375cb WIP il y a 3 ans
  Alexander Borzunov 54ad745bed Warn that current instructions involve 6B model but we will replace them soon (#63) il y a 3 ans
  Alexander Borzunov 5f0c5329d4 Update readme with arxiv link and more discussions (#62) il y a 3 ans
  Alexander Borzunov 9bea7b9ea8 Update bullet points with feedback from Tim and other people (#61) il y a 3 ans
  Alexander Borzunov 7653562aa1 Use latest version of Petals scheme, shrink Petals logo (#59) il y a 3 ans
  Alexander Borzunov 2eb5843852 Update readme for the 1st public release (#57) il y a 3 ans
  Pavel Samygin 0be21775af remove transformer block, implement as sequential of size 1 (#54) il y a 3 ans
  Artem Chumachenko 77220c718c Add shallow prefix-tuned inference (#55) il y a 3 ans
  justheuristic d271b75dd4 Let users specify sequence length instead of assuming 2048 (#52) il y a 3 ans
  Dmitry Baranchuk 948877149c Fix recovering for sequential_backward (#50) il y a 3 ans
  Dmitry Baranchuk 24ba3433e4 [Fix] make distributed seq cls to not create the full bloom model (#49) il y a 3 ans
  justheuristic f12d0deee9 [quickfix 1/n] remove expensive assertions in inference code (#48) il y a 3 ans
  Dmitry Baranchuk 0fd2caa4be Convert actual model weights (#46) il y a 3 ans
  justheuristic a2634001e9 Reduce vocabulary size in test model, fix bug in routing when overlapped (#45) il y a 3 ans