mbruton commited on
Commit
098c87e
·
verified ·
1 Parent(s): 5a0d111

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +45 -11
README.md CHANGED
@@ -5,14 +5,14 @@ tags:
5
  - historical-cryptology
6
  - homophonic-cipher
7
  - substitution-cipher
8
- - decipherment
9
  library_name: pytorch
10
  ---
11
 
12
 
13
- # Shared decipherment models for historical substitution ciphers
14
 
15
- Each model in this collection has been trained for **one known historical cipher key**. It reads transcribed ciphertexts written with that key and produces a draft decipherment. Testing a ciphertext against all models shows which known key, if any, it was written with.
16
 
17
  Use the models with the notebooks or command-line tool at **https://github.com/mbruton0426/keybridge**. You do not need to download anything from here by hand.
18
 
@@ -24,18 +24,52 @@ Use the models with the notebooks or command-line tool at **https://github.com/m
24
  | `<model id>/<model id>_model.pt` | The trained model. |
25
  | `<model id>/<model id>_key.tsv` | The key it was trained on. |
26
  | `<model id>/<model id>_training_report.txt` | Training settings and accuracy on synthetic text. |
27
- | `<model id>/README.md` | Description, source of the key, and contributor. |
28
  | `submissions/` | Models contributed by others, waiting for review. |
29
 
30
  ## Models
31
 
32
- | ID | Cipher | Transcription format | Source of the key |
33
- |----|--------|----------------------|-------------------|
34
- | `maximilian_II` | Letters of Maximilian II (1575) | symbols separated by spaces (single characters; convert one-character-per-symbol files with `keybridge convert`) | DECODE IDs 1471, 1412, 1407 |
35
- | `maximilian_1472` | Maximilian letter with a different key | symbols separated by spaces (named symbols) | DECODE IDs 1472, 1473 |
36
- | `copiale` | Copiale cipher | symbols separated by spaces | DECODE ID 2 |
37
 
38
- All three were trained on 5,000 synthetic examples from the built-in English HistCiph plaintext, with rare-unit insertion at 0.02 (see each model's training report).
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
39
 
40
  ## Contributing a model
41
 
@@ -47,4 +81,4 @@ All models and keys are released under **CC BY 4.0**: you may use and adapt them
47
 
48
  > [citation to be added] <!-- PLACEHOLDER -->
49
 
50
- When you use a specific model, please also credit the original source of its key, given in that model's README.
 
5
  - historical-cryptology
6
  - homophonic-cipher
7
  - substitution-cipher
8
+ - decryption
9
  library_name: pytorch
10
  ---
11
 
12
 
13
+ # Decryption models for historical substitution ciphers
14
 
15
+ Each model in this collection has been trained for **one known historical cipher key**. It reads transcribed ciphertexts written with that key and produces a draft decryption for expert review. Testing a ciphertext against all models (a *scan*) suggests which known key, if any, it was written with; the drafts of the best-ranked models are then read to decide. The models do not recover unknown keys.
16
 
17
  Use the models with the notebooks or command-line tool at **https://github.com/mbruton0426/keybridge**. You do not need to download anything from here by hand.
18
 
 
24
  | `<model id>/<model id>_model.pt` | The trained model. |
25
  | `<model id>/<model id>_key.tsv` | The key it was trained on. |
26
  | `<model id>/<model id>_training_report.txt` | Training settings and accuracy on synthetic text. |
27
+ | `<model id>/README.md` | Description, source of the key, test results and contributor. |
28
  | `submissions/` | Models contributed by others, waiting for review. |
29
 
30
  ## Models
31
 
32
+ The 35 models of the paper, one for each key built from the DECODE database (records are DECODE record numbers). *Digits*: digit keys, whose ciphertexts may be written as running digits (keybridge finds the codes). All other transcriptions are written as symbols separated by spaces, exactly as in the key.
 
 
 
 
33
 
34
+ | ID | Cipher | Language | Period | Transcription | DECODE records |
35
+ |----|--------|----------|--------|---------------|----------------|
36
+ | `copiale_5000_english` | Copiale cipher, manuscript of a secret society | German | 1734-1736 | named symbols | 2 |
37
+ | `maximilian_II_5000_english` | Maximilian II to his ambassadors in Poland and Lithuania | German | 1575 | single characters | 1407, 1412, 1471 |
38
+ | `maximilian_1472_5000_english` | Johan Chodkiewicz to Maximilian II | German, Latin | 1574-1575 | named symbols | 1472, 1473 |
39
+ | `borg_A_5000_english` | Borg cipher (BAV Borg.lat. 898), Latin medical manuscript | Latin | not dated | symbols | 210 |
40
+ | `ramanacoil_A_5000_english` | Rijcklof van Goens at Ramanacoil, Ceylon, to the VOC (Heeren XVII) | Dutch | 1674 | named symbols | 1564 |
41
+ | `ucl_A_5000_english` | English letter, Brougham Archive, UCL Special Collections | English | 1724 | number codes separated by spaces | 6317 |
42
+ | `kew_E_5000_english` | Letters in TNA SP 106 (Dr Stephen Goffe and others) | English | 1645-1646 | codes separated by spaces (e.g. n3) | 928, 929 |
43
+ | `bnf_C_5000_english` | Philip II to Juan de Vargas Mexía (BnF Espagnol 132) | Spanish | 1575-1580 | symbols (key incomplete: codes with marks not included) | 1964-2025 (48 records) |
44
+ | `tna_CK_5000_english` | Italian key, TNA SP 106/10 | Italian | not dated | whole codes (e.g. z19 m23), separated by spaces | 723 |
45
+ | `na_B_5000_english` | Codebook of S. E. Croiset for Dirk van Hogendorp; letters from Saint Petersburg | Dutch | 1803 (key); 1808-1810 (letters) | digits with end marks | 1033, 1035 |
46
+ | `kha_A_5000_english` | Comte d'Avaux, The Hague, to Louis XIV | French | 1684 | digits | 1055 |
47
+ | `kew_A_5000_english` | Henrietta Maria to Charles I (TNA SP 106) | English, French | 1645 | digits | 752, 754, 785, 788, 790, 923 |
48
+ | `kew_B_5000_english` | Henrietta Maria to Charles I (TNA SP 106) | English, French | 1645 | digits | 787, 788 |
49
+ | `kew_C_5000_english` | Lord Digby to Charles I (TNA SP 106) | English | 1645 | digits | 789 |
50
+ | `kew_D_5000_english` | Henrietta Maria to Charles I (TNA SP 106) | English, French | 1646 | digits | 924, 925 |
51
+ | `kew_F_5000_english` | Unknown writer to Lord Goring (TNA SP 106) | English | 1645 | digits | 932 |
52
+ | `vatican_A_5000_english` | Papal correspondence, ASV Francia 4 | Italian | 1587 | digits | 5 |
53
+ | `vatican_B_5000_english` | Papal correspondence, ASV Francia 3 | Italian | 1552-1554 | digits | 6-10 |
54
+ | `vatican_C_5000_english` | Papal correspondence, ASV Francia 6 | Italian | 1573-1574 | digits | 11, 12 |
55
+ | `vatican_D_5000_english` | Papal correspondence, ASV Francia 17 | Italian | 1583-1585 | digits | 13, 14 |
56
+ | `vatican_E_5000_english` | Papal correspondence, ASV Francia 18 | Italian | 1585-1587 | digits | 15, 16 |
57
+ | `vatican_F_5000_english` | Papal correspondence, ASV Francia 22 | Italian | 1585-1589 | digits | 17-58 (35 records) |
58
+ | `vatican_G_5000_english` | Papal correspondence, ASV C15 | Italian | 1721 | digits | 24 |
59
+ | `vatican_H_5000_english` | Papal correspondence, ASV P117 | Italian | 1757-1761 | digits | 25, 195-201 (7 records) |
60
+ | `vatican_I_5000_english` | Papal correspondence, ASV S423 | Italian | 1736 | digits | 26, 179-188 |
61
+ | `vatican_J_5000_english` | Papal correspondence, ASV Francia 22b | Italian | 1588-1589 | digits | 53, 54 |
62
+ | `vatican_K_5000_english` | Papal correspondence, ASV Francia 64 | Italian | 1624-1625 | digits | 59-71 (13 records) |
63
+ | `vatican_L_5000_english` | Papal correspondence, ASV Francia 283 | Italian | 1580 | digits | 72, 73 |
64
+ | `vatican_M_5000_english` | Papal correspondence, ASV Francia 346 | Italian | 1632-1634 | digits | 74-84 |
65
+ | `vatican_N_5000_english` | Papal correspondence, ASV S304 | Italian | 1767 | digits | 120, 121 |
66
+ | `vatican_O_5000_english` | Papal correspondence, ASV S364 (Secretariat to the nuncio in Spain) | Italian | 1717-1720 | digits | 122-177 |
67
+ | `vatican_P_5000_english` | Papal correspondence, not given | Italian | 1535-1536 | digits | 189 |
68
+ | `vatican_Q_5000_english` | Papal correspondence, ASV P8 | Italian | 1579-1581 | digits | 190-194 |
69
+ | `vatican_R_5000_english` | Papal correspondence, ASV C13 | Italian | 1656-1657 | digits | 208, 209 |
70
+ | `vatican_S_5000_english` | Papal correspondence, BAV Barb.lat. 6956 and 6960 (Pallotto) | Italian | 1628-1629 | digits | 215-225 (10 records) |
71
+
72
+ All 35 were trained with keybridge's released defaults: 5,000 synthetic examples from the built-in English HistCiph plaintext, rare-unit insertion 0.02 (see each model's training report). Where a model has test letters, its README gives the results against the *key reference* (the transcription read through the key) and, for digit keys, against the transcribers' readings. The Maximilian and Copiale models uploaded in October 2026 have been replaced by these.
73
 
74
  ## Contributing a model
75
 
 
81
 
82
  > [citation to be added] <!-- PLACEHOLDER -->
83
 
84
+ When you use a specific model, please also credit the original source of its key, given in that model's README.