hinabarun's picture
Update logbook: Reproduction: Incremental BPE Tokenization
7ff5663 verified
Raw
History Blame Contribute Delete
2.12 kB
{
"schema_version": 1,
"title": "Reproduction: Incremental BPE Tokenization",
"emoji": "🎯",
"space_id": "hinabarun/repro-incremental-bpe-tokenization",
"paper": {
"arxiv_id": "2605.30813"
},
"tags": [
"icml2026-repro",
"paper-ZbWgrDzCQo"
],
"updated_at": "2026-07-19T07:49:24+00:00",
"root": {
"slug": "index",
"title": "Reproduction: Incremental BPE Tokenization",
"file": "pages/index.md",
"children": [
{
"slug": "executive-summary",
"title": "Executive summary",
"file": "pages/executive-summary/page.md",
"children": []
},
{
"slug": "claim-1-monotonic-path-property-theorem-4-2-o-log-2-t-per-byte",
"title": "Claim 1: Monotonic Path Property (Theorem 4.2, O(log^2 t) per byte)",
"file": "pages/claim-1-monotonic-path-property-theorem-4-2-o-log-2-t-per-byte/page.md",
"children": []
},
{
"slug": "claim-2-algorithm-design-aho-corasick-centroid-decomposition-eager-output",
"title": "Claim 2: Algorithm design: Aho-Corasick + centroid decomposition + eager output",
"file": "pages/claim-2-algorithm-design-aho-corasick-centroid-decomposition-eager-output/page.md",
"children": []
},
{
"slug": "claim-3-drop-in-speedup-over-hf-tokenizers-up-to-3-13x-codellama",
"title": "Claim 3: Drop-in speedup over HF tokenizers (up to 3.13x, CodeLlama)",
"file": "pages/claim-3-drop-in-speedup-over-hf-tokenizers-up-to-3-13x-codellama/page.md",
"children": []
},
{
"slug": "claim-4-pathological-input-robustness-vs-tiktoken-cl100k-o-n-2-decay",
"title": "Claim 4: Pathological-input robustness vs tiktoken (CL100K) O(n^2) decay",
"file": "pages/claim-4-pathological-input-robustness-vs-tiktoken-cl100k-o-n-2-decay/page.md",
"children": []
},
{
"slug": "conclusion",
"title": "Conclusion",
"file": "pages/conclusion/page.md",
"children": []
}
]
},
"agent_view_tokens": 4149,
"revision": "1784447364943394243"
}