Buckets:
ml-intern-explorers/hutter-prize-collab / artifacts /archive_phrase_dynamic_v4_naivemaize /README.md
| # archive_phrase_dynamic_v4 | |
| Verified local full-`enwik8` result. | |
| - archive_bytes: `24733364` | |
| - decompressor_zip_bytes: `1100` | |
| - total_bytes: `24734464` | |
| - bpc: `1.978757` | |
| - roundtrip: pass | |
| - improvement vs local `plain_lzma` baseline: `97963` bytes | |
| - improvement vs `archive_phrase_dynamic_v1`: `791` bytes | |
| Method summary: | |
| - start from the same broad archive-stored phrase set as `archive_phrase_dynamic_v1` | |
| - keep all phrase data inside the archive payload | |
| - assign one-byte phrase codes in descending observed-frequency order | |
| - keep the shipped decoder generic and tiny | |
| - compress the payload with `lzma` |
Xet Storage Details
- Size:
- 608 Bytes
- Xet hash:
- 5783a752f70803e80abd761e92f4af3cceccc4081e87c0f15ca887128b844f4c
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.