v0.0.4a1

What's Changed

IMPORTANT: Fixed model weights (please redownload model weights associated with this release date)

• Patch/read hdf5 by @KPHippe in #89
• Patch/read hdf5 by @KPHippe in #90
• Updating existing tokenizer naming, adding new tokenizer with BOS EOS tokens by @KPHippe in #91
• Patch/profile io by @KPHippe in #92
• Patch/bad h5 file by @KPHippe in #93
• Defaulting code examples to gpu if available by @KPHippe in #79
• Tokenizer update + Batch inference QOL changes by @KPHippe in #96

Full Changelog: v0.0.3a1...v0.0.4a1

v0.0.3a1

What's Changed

• embedding notebook by @braceal in #61
• Develop by @braceal in #62
• pandas version by @braceal in #63
• version by @braceal in #64
• include architectures in package by @braceal in #65
• Develop by @braceal in #66
• Develop by @braceal in #67
• Develop by @braceal in #68
• generate example by @braceal in #69
• notes by @braceal in #70
• note by @braceal in #71
• Develop by @braceal in #72
• remove deepspeed requirement in examples by @braceal in #73
• Develop by @braceal in #74
• Update README.md by @braceal in #75
• patch: codons must be capitalized by @braceal in #82
• Temporarily removing multiprocessing to solve file header issue in multiprocessing by @KPHippe in #81
• Develop by @braceal in #83
• Patch/package data by @braceal in #86
• Develop by @braceal in #87
• Patch/output callback by @braceal in #85
• Merge refactor to main by @KPHippe in #88

Full Changelog: v0.0.1a1...v0.0.3a1

v0.0.1a1

Initial release.