Ctrl+K

1 contributor

History: 106 commits

AbstractPhil

btx-e003 traces v3: d5_traces_v3.jsonl

513de65 verified 2 days ago

btx_e003
btx-e003 traces v3: d5_traces_v3.jsonl 2 days ago
byte_lexicon_v1
bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
proto_frame
btx-e003 R2 ledger final 3 days ago
.gitattributes
2.17 kB
mapwise state representations: byte language C built from codex, both vocabs projected through it (full row contract); joint content-state map + words-of-C table 22 days ago
README.md
4.71 kB
README: link the standalone code repo (github.com/AbstractEyes/geolip-bytelex) + git+ install line; regeneration section now points at the library, not alephllm 24 days ago
byte_states_ledger.json
237 Bytes
byte-specific state gauge: A 82449 / B 63001 trie states, 28246 co-occupied; 12.0K/12.5K whole states walk fully on shared structure vs 10801 exact name matches 22 days ago
codex_v1.txt
256 kB
btx-e001 T-1: matrix size verdict — closed register saturates at 94 cells by 16KB; open-class digits grow 1276+ cells and flip modal shape to 2x2 22 days ago
joint_state_map.jsonl
11 MB
xet

mapwise state representations: byte language C built from codex, both vocabs projected through it (full row contract); joint content-state map + words-of-C table 22 days ago
lexicon_states_ledger.json
532 Bytes
lexicon byte-state gauges: A 31997, B 29523, overlap 10801, union 50719, irreducibles 1048 (atoms: 10 digits + letters + punct; zero word-starts), 95/256 single bytes present 22 days ago
mapwise_summary.json
710 Bytes
mapwise state representations: byte language C built from codex, both vocabs projected through it (full row contract); joint content-state map + words-of-C table 22 days ago
matrix_bert-wordpiece.jsonl
2.19 MB
bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_byt5-control.jsonl
8.09 kB
bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_cl100k_base.jsonl
7.06 MB
bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_deepseek-v3.jsonl
7.41 MB
bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_flan-t5-small.jsonl
1.93 MB
mapwise state representations: byte language C built from codex, both vocabs projected through it (full row contract); joint content-state map + words-of-C table 22 days ago
matrix_gpt2.jsonl
3.69 MB
bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_llama3.jsonl
8.29 MB
bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_mistral-v03.jsonl
2.09 MB
bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_o200k_base.jsonl
12.5 MB
xet

bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_qwen3-newest.jsonl
14.3 MB
xet

bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_smollm2.jsonl
3.64 MB
bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_summary.json
2.41 kB
bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_t5-unigram.jsonl
2.38 MB
bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
matrix_tinybert-4l-312d.jsonl
1.87 MB
mapwise state representations: byte language C built from codex, both vocabs projected through it (full row contract); joint content-state map + words-of-C table 22 days ago
matrix_xlmr.jsonl
13.5 MB
xet

bytelex v1: FULL lexicon translation matrix (1.25M tokens across 12 vocabularies) + byte_lexicon_v1 corpus tables (8.2MB mixed corpus, char1-4 exact, hashed 9-gram, word-grams) 24 days ago
t0_extraction_ledger.json
603 Bytes
btx-e001 T-0: flan-t5-small + tinybert tables, gated, coverage disclosed (newline is the mutual floor; case one-way lossy into wordpiece) 22 days ago
t1_matrix_ledger.json
1.96 kB
btx-e001 T-1: matrix size verdict — closed register saturates at 94 cells by 16KB; open-class digits grow 1276+ cells and flip modal shape to 2x2 22 days ago
t1b_domain_ledger.json
970 Bytes
btx-e001 T-1: matrix size verdict — closed register saturates at 94 cells by 16KB; open-class digits grow 1276+ cells and flip modal shape to 2x2 22 days ago
t2_control_vanilla_t5.py
4.79 kB
btx-e001 T-2 control: vanilla t5-small = wordpiece parity (0.47/0.61) through the same pushforward; flan sentinel infill EOS-degenerate — channel exonerated, flan indicted 22 days ago
t2_control_vanilla_t5.txt
1.52 kB
btx-e001 T-2 control: vanilla t5-small = wordpiece parity (0.47/0.61) through the same pushforward; flan sentinel infill EOS-degenerate — channel exonerated, flan indicted 22 days ago
t2_shunt_ledger.json
1.34 kB
btx-e001 T-2: shunt fidelity ledger — cloze-BERT byte marginal hits 0.43-0.61 gold; flan-t5 sentinel infill 0.05-0.09; n:m cells crater both 22 days ago
t3a_arity_ledger.json
760 Bytes
btx-e001 T3-a: matched mask arity moves arith:nm only 0.107->0.127 (still < 0.161 majority-digit baseline) — crater is content entropy, not protocol; 1x1 rows reproduce T-2 exactly 22 days ago
vocab_bert-wordpiece.jsonl
3.45 MB
bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
vocab_byt5-control.jsonl
41.6 kB
bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
vocab_cl100k_base.jsonl
11.5 MB
xet

bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
vocab_deepseek-v3.jsonl
14.9 MB
xet

bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
vocab_flan-t5-small.jsonl
3.74 MB
mapwise state representations: byte language C built from codex, both vocabs projected through it (full row contract); joint content-state map + words-of-C table 22 days ago
vocab_gpt2.jsonl
5.73 MB
bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
vocab_llama3.jsonl
14.7 MB
xet

bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
vocab_mistral-v03.jsonl
3.63 MB
bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
vocab_o200k_base.jsonl
23.3 MB
xet

bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
vocab_qwen25.jsonl
9.12 MB
vocab_qwen25.jsonl — Qwen2.5 family table (extract.py fleet extractor, round-trip gated) 23 days ago
vocab_qwen3-newest.jsonl
29.3 MB
xet

bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
vocab_smollm2.jsonl
5.66 MB
bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
vocab_t5-unigram.jsonl
3.7 MB
bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
vocab_tinybert-4l-312d.jsonl
3.53 MB
mapwise state representations: byte language C built from codex, both vocabs projected through it (full row contract); joint content-state map + words-of-C table 22 days ago
vocab_xlmr.jsonl
30.7 MB
xet

bytelex v1: 12 normalized token->bytes vocabulary tables (fleet-extracted, round-trip spot-checked) 24 days ago
words_of_C.json
136 kB
mapwise state representations: byte language C built from codex, both vocabs projected through it (full row contract); joint content-state map + words-of-C table 22 days ago
wp_walk_of_C.json
91.7 kB
single-tokenizer walk of C: 999 states, zero UNK, 95.4% mass whole; digit grammar = longest-prefix+trailing-digit; only liters/Subtract compose non-trivially 22 days ago