yhavinga commited on
Commit
9b0ae26
1 Parent(s): cdb537e

Update models and default texts

Browse files
Files changed (2) hide show
  1. app.py +6 -47
  2. default_texts.py +16 -0
app.py CHANGED
@@ -38,66 +38,25 @@ GENERATOR_LIST = [
38
  "task": TRANSLATION_NL_TO_EN,
39
  "split_sentences": True,
40
  },
41
- # {
42
- # "model_name": "yhavinga/longt5-local-eff-large-nl8-voc8k-ddwn-512beta-512l-nedd-256ccmatrix-nl-en",
43
- # "desc": "longT5 large nl8 256cc/512beta/512l nl->en",
44
- # "task": TRANSLATION_NL_TO_EN,
45
- # "split_sentences": False,
46
- # },
47
  {
48
- "model_name": "yhavinga/longt5-local-eff-large-nl8-voc8k-ddwn-512beta-512-nedd-nl-en",
49
- "desc": "longT5 large nl8 512beta/512l nl->en",
50
- "task": TRANSLATION_NL_TO_EN,
51
  "split_sentences": False,
52
  },
53
  {
54
- "model_name": "yhavinga/longt5-local-eff-large-nl8-voc8k-ddwn-512beta-512l-nedd-256ccmatrix-en-nl",
55
- "desc": "longT5 large nl8 256cc/512beta/512l en->nl",
56
  "task": TRANSLATION_EN_TO_NL,
57
- "split_sentences": False,
58
  },
59
  # {
60
- # "model_name": "yhavinga/longt5-local-eff-base-nl36-voc8k-256l-472beta-256l-472beta-en-nl",
61
- # "desc": "longT5 large nl8 256l/472beta/256l/472beta en->nl",
62
- # "task": TRANSLATION_EN_TO_NL,
63
- # "split_sentences": False,
64
- # },
65
- # {
66
- # "model_name": "yhavinga/byt5-small-ccmatrix-en-nl",
67
- # "desc": "ByT5 small ccmatrix en->nl",
68
- # "task": TRANSLATION_EN_TO_NL,
69
- # "split_sentences": True,
70
- # },
71
- # {
72
- # "model_name": "yhavinga/t5-eff-large-8l-nedd-en-nl",
73
- # "desc": "T5 eff large nl8 en->nl",
74
- # "task": TRANSLATION_EN_TO_NL,
75
- # "split_sentences": True,
76
- # },
77
- # {
78
  # "model_name": "yhavinga/t5-base-36L-ccmatrix-multi",
79
  # "desc": "T5 base nl36 ccmatrix en->nl",
80
  # "task": TRANSLATION_EN_TO_NL,
81
  # "split_sentences": True,
82
  # },
83
  # {
84
- # "model_name": "yhavinga/longt5-local-eff-large-nl8-voc8k-ddwn-512beta-512-nedd-en-nl",
85
- # "desc": "longT5 large nl8 512beta/512l en->nl",
86
- # "task": TRANSLATION_EN_TO_NL,
87
- # "split_sentences": False,
88
- # },
89
- # {
90
- # "model_name": "yhavinga/t5-base-36L-nedd-x-en-nl-300",
91
- # "desc": "T5 base 36L nedd en->nl 300",
92
- # "task": TRANSLATION_EN_TO_NL,
93
- # "split_sentences": True,
94
- # },
95
- # {
96
- # "model_name": "yhavinga/long-t5-local-small-ccmatrix-en-nl",
97
- # "desc": "longT5 small ccmatrix en->nl",
98
- # "task": TRANSLATION_EN_TO_NL,
99
- # "split_sentences": True,
100
- # },
101
  ]
102
 
103
 
 
38
  "task": TRANSLATION_NL_TO_EN,
39
  "split_sentences": True,
40
  },
 
 
 
 
 
 
41
  {
42
+ "model_name": "yhavinga/longt5-local-eff-large-nl8-voc8k-ddwn-neddx2-en-nl",
43
+ "desc": "longT5 large nl8 en->nl",
44
+ "task": TRANSLATION_EN_TO_NL,
45
  "split_sentences": False,
46
  },
47
  {
48
+ "model_name": "yhavinga/byt5-small-ccmatrix-en-nl",
49
+ "desc": "ByT5 small ccmatrix en->nl",
50
  "task": TRANSLATION_EN_TO_NL,
51
+ "split_sentences": True,
52
  },
53
  # {
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
54
  # "model_name": "yhavinga/t5-base-36L-ccmatrix-multi",
55
  # "desc": "T5 base nl36 ccmatrix en->nl",
56
  # "task": TRANSLATION_EN_TO_NL,
57
  # "split_sentences": True,
58
  # },
59
  # {
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
60
  ]
61
 
62
 
default_texts.py CHANGED
@@ -1,4 +1,15 @@
1
  default_texts = {
 
 
 
 
 
 
 
 
 
 
 
2
  "The Invisible Censor": {
3
  "url": "https://www.gutenberg.org/files/35091/35091-h/35091-h.html",
4
  "year": 1921,
@@ -30,6 +41,11 @@ The ISS - a joint project involving five space agencies - has been in orbit sinc
30
  "year": 1941,
31
  "text": """You must consider of it less narrowly: sub specie aeternitatis. Supposition is, every conceivable bunch of circumstances, that is to say, every conceivable world, exists: but unworlded, unbunched: to our more mean capacities an unpassable bog or flux of seas, cities, rivers, lakes, wolds and deserts and mountain ranges, all with their plants, forests, mosses, water-weeds, what you will; and all manner of peoples, beasts, birds, fishes, creeping things, climes, dreams, loves, loathings, abominations, ecstasies, dissolutions, hopes, fears, forgetfulnesses, infinite in variety, infinite in number, fantasies beyond nightmare or madness. All this in potentia. All are there, even-just as are all the particulars in a landscape: He, like as the landscape-painter, selects and orders. The one paints a picture, the Other creates a world.""",
32
  },
 
 
 
 
 
33
  "Hersenschimmen": {
34
  "url": "https://www.bibliotheek.nl/catalogus/titel.37120397X.html/hersenschimmen/",
35
  "year": 1960,
 
1
  default_texts = {
2
+ "A Scanner Darkly": {
3
+ "url": "https://en.wikipedia.org/wiki/A_Scanner_Darkly",
4
+ "year": 1977,
5
+ "text": """"You're chickening out?" the girl said, haughtily, with contempt. "You don't have it at gut level to stick with a decision? To get off the filth? You're going to crawl back out of here on your belly?" All three of them glared at him with anger.
6
+ "Later," Arctor said, and moved toward the front door, the way out.
7
+ "Fucking doper," the girl said from behind him. "No guts, brain fried, nothing. Creep out, creep; it's your decision."
8
+ "I'll be back," Arctor said, nettled. The mood here oppressed him, and it had intensified now that he was leaving.
9
+ "We may not want you back, gutless," one of the guys said.
10
+ "You'll have to plead," the other said. "You may have to do a lot of heavy pleading. And even then we may not want you."
11
+ "In fact, we don't want you now," the girl said.""",
12
+ },
13
  "The Invisible Censor": {
14
  "url": "https://www.gutenberg.org/files/35091/35091-h/35091-h.html",
15
  "year": 1921,
 
41
  "year": 1941,
42
  "text": """You must consider of it less narrowly: sub specie aeternitatis. Supposition is, every conceivable bunch of circumstances, that is to say, every conceivable world, exists: but unworlded, unbunched: to our more mean capacities an unpassable bog or flux of seas, cities, rivers, lakes, wolds and deserts and mountain ranges, all with their plants, forests, mosses, water-weeds, what you will; and all manner of peoples, beasts, birds, fishes, creeping things, climes, dreams, loves, loathings, abominations, ecstasies, dissolutions, hopes, fears, forgetfulnesses, infinite in variety, infinite in number, fantasies beyond nightmare or madness. All this in potentia. All are there, even-just as are all the particulars in a landscape: He, like as the landscape-painter, selects and orders. The one paints a picture, the Other creates a world.""",
43
  },
44
+ "Encylopedia of Swearing": {
45
+ "url": "https://www.academia.edu/32398800/Encyclopedia_of_Swearing",
46
+ "year": 2006,
47
+ "text": """"Animal terms figure notably in the history of swearing, although they were not a major feature of Anglo-Saxon literature. The major exception was wulf, used to refer to a cruel, rapacious, or evil person, often in the title “the Devil’s wolf.” Otherwise, the chosen animals themselves are not especially dangerous or repulsive, though some are poisonous, such as the snake, and others malodorous, such as the skunk and polecat. For some cultural reason the pig provides the richest verbal field, together with the variants sow and swine. (The same pattern is seen, interestingly, in the dominance of French cochon and German schweinhund.) Swine is the oldest term in the field, being recorded in Chaucer’s richest swearing resource, the Wife of Bath, who condemns “Metellius, the foule cherl, the swyn” (Prologue l. 460). Unlike sow, swine continues to have resonance in swearing in the British Isles, especially among the older generation, while pig has become more a feature of U. S. swearing, having been especially fashionable among radical youth in the 1960s as an opprobrious term for the police.""",
48
+ },
49
  "Hersenschimmen": {
50
  "url": "https://www.bibliotheek.nl/catalogus/titel.37120397X.html/hersenschimmen/",
51
  "year": 1960,