- Model Info:
- Tied embeddings: False
- LM head uses bias: False
- Embeddings shape: [102400, 2048]
- Tokenizer Info:
- Vocab Size: 100015
- Tokenizer Class: LlamaTokenizerFast
- Tokenizer Type: BPE
- Bytes handling: Byte Input
- Token for verification prompt building: IllegalArgumentException
- Token id for verification prompt building: 91253
- Indicator summary:
- Indicator for under-trained tokens: E_{in} L2 Norm
- Overall distribution: 0.597 +/- 0.128
- Detected Token Counts:
- Number of tested under-trained tokens: 1990, 1983 non-special, 886 below p = 0.01 threshold, 38 below soft indicator threshold
- Number of single byte tokens: 256, of which 4 below indicator threshold
- Number of special tokens: 32, of which 20 below indicator threshold
- Number of non-single-byte unreachable tokens: 32, of which 20 below indicator threshold
- Number of non-single-byte UTF-fragment tokens: 438, of which 0 below soft indicator threshold
38 entries below threshold of 0.121
token_id | token | indicator | max_prob | in_other_tokens |
---|---|---|---|---|
87662 | 日内与新浪看点 |
0.0218784 | 3.5e-05 | 日内与新浪看点联系 |
87661 | 不代表新浪看点 |
0.0236809 | 2.6e-08 | 不代表新浪看点观点或立场 |
97672 | 基督教基督教基督教 |
0.0301265 | 0.0011 | |
91136 | controlcap |
0.0321056 | 3.3e-06 | |
74777 | orangehilldev |
0.0355834 | 5.6e-05 | |
16238 | кедония |
0.0391545 | 3.2e-06 | ▁Македония , Македония |
90785 | посолство |
0.04666 | 0.0019 | |
13009 | lemanya |
0.0497387 | 3.3e-05 | ▁alemanya , ▁Alemanya , Alemanya |
99639 | 亿亿亿亿亿亿亿亿亿亿亿亿亿亿亿亿 |
0.0535675 | 0.099 | |
81096 | ▁EDIPU |
0.0570441 | 1.2e-05 | ▁EDIPUCRS |
84405 | RecordedVote |
0.0587952 | 0.0064 | |
59771 | 基督教基督教 |
0.0625054 | 0.00024 | 基督教基督教基督教 |
71563 | 亿亿亿次 |
0.063932 | 0.042 | |
60623 | odeciclismo |
0.0639408 | 3e-05 | ▁sitiodeciclismo , iodeciclismo |
50113 | memItem |
0.0682783 | 0.11 | memItemRight , memItemLeft |
5758 | ългар |
0.0703906 | 9.3e-05 | ▁българ , ▁българите , Българ , ▁българския , България , ... |
70532 | \xa0veg |
0.075737 | 0.0011 | \xa0vegades |
97018 | Supamiu |
0.0764349 | 0.0016 | |
78552 | ▁Междусъюз |
0.0769008 | 0.00012 | ▁Междусъюзническата |
49112 | iberament |
0.0797149 | 5.8e-05 | Alliberament , alliberament |
18 additional entries below threshold
token_id | token | indicator | max_prob | in_other_tokens |
---|---|---|---|---|
49293 | ▁lampister |
0.0844076 | 3.7e-05 | ▁lampisteria , ▁lampisteries |
58888 | 亿亿亿亿亿亿亿亿 |
0.0848685 | 0.027 | 亿亿亿亿亿亿亿亿亿亿亿亿亿亿亿亿 |
49918 | magatzem |
0.0849909 | 0.00047 | emmagatzematge , emmagatzem , ▁emmagatzem |
51244 | ecesito |
0.0872311 | 0.00013 | ▁Necesito , Necesito |
86826 | солство |
0.0923582 | 0.00054 | посолство |
9710 | ▁espany |
0.0926319 | 0.00025 | ▁espanyola , ▁espanyol , ▁espanyols , ▁espanyoles |
73129 | жентина |
0.0972762 | 0.0018 | ▁Аржентина |
85684 | товче |
0.10107 | 4.8e-05 | ▁братовче |
52246 | ▁опъл |
0.106657 | 0.00019 | ▁опълчение , ▁опълчен , ▁опълченец |
7388 | точници |
0.107464 | 4e-05 | Източници , ▁източници |
51641 | мъния |
0.112711 | 0.001 | ▁Румъния |
9630 | ългария |
0.114074 | 0.0019 | България , ▁България |
93494 | atrals |
0.114798 | 0.0085 | ▁teatrals |
41580 | ▁експе |
0.115079 | 7.1e-05 | ▁експери , ▁експеди , ▁експедиция |
74713 | photonui |
0.115498 | 0.92 | |
90292 | битава |
0.116288 | 6.6e-05 | Обитава |
72767 | elrte |
0.118752 | 0.42 | |
24543 | wlwifi |
0.12032 | 0.51 | ▁iwlwifi , iwlwifi |
0 entries below threshold of 0.121
4 entries below threshold of 0.257
token_id | token | indicator | ord | hex | byte_type |
---|---|---|---|---|---|
185 | \n |
0.188525 | 10 | 0x0A | ascii |
11 | , |
0.233417 | 44 | 0x2C | ascii |
207 | ▁ |
0.237499 | 32 | 0x20 | ascii |
13 | . |
0.252735 | 46 | 0x2E | ascii |
1 entries below threshold of 0.257
token_id | token | indicator | max_prob |
---|---|---|---|
100000 | ¿<|begin▁of▁sentence|>? |
2.22104e-08 | 1.4e-05 |
20 entries below threshold of 0.257
token_id | token | indicator | reencoded |
---|---|---|---|
95036 | ▁ö |
2.15243e-08 | 207: ▁ , 100003: <0xF6> |
48308 | Á |
2.15282e-08 | 100012: <0xC1> |
53854 | ▁ü |
2.16154e-08 | 207: ▁ , 100014: <0xFC> |
15302 | ▁À |
2.17436e-08 | 207: ▁ , 100010: <0xC0> |
1612 | ú |
2.17913e-08 | 100004: <0xFA> |
77883 | çõ |
2.18156e-08 | 1337: ç , 100006: <0xF5> |
31186 | üí |
2.19835e-08 | 100014: <0xFC> , 656: í |
60840 | ▁þ |
2.20105e-08 | 207: ▁ , 100013: <0xFE> |
15776 | üè |
2.21308e-08 | 100014: <0xFC> , 724: è |
27924 | ▁Á |
2.2146e-08 | 207: ▁ , 100012: <0xC1> |
56124 | û |
2.22393e-08 | 100008: <0xFB> |
25486 | ø |
2.22396e-08 | 100002: <0xF8> |
45981 | õ |
2.22878e-08 | 100006: <0xF5> |
7962 | ö |
2.23045e-08 | 100003: <0xF6> |
6496 | ▁ú |
2.23787e-08 | 207: ▁ , 100004: <0xFA> |
84896 | þ |
2.24477e-08 | 100013: <0xFE> |
12759 | À |
2.24954e-08 | 100010: <0xC0> |
52272 | ù |
2.25231e-08 | 100011: <0xF9> |
5021 | ý |
2.25858e-08 | 100009: <0xFD> |
2874 | ü |
2.28658e-08 | 100014: <0xFC> |