Hugging Face's logo

DFKI-SLT
/
eurogpt2

26 languages
Model card Files Community
malteos commited on
Commit
df8b4a1
•
1 Parent(s): 7a70b4f

added checkpoint

Browse files
Files changed (24) hide show
  1. README.md +61 -0
  2. layer_01-model_00-model_states.pt +3 -0
  3. layer_03-model_00-model_states.pt +3 -0
  4. layer_04-model_00-model_states.pt +3 -0
  5. layer_05-model_00-model_states.pt +3 -0
  6. layer_06-model_00-model_states.pt +3 -0
  7. layer_07-model_00-model_states.pt +3 -0
  8. layer_08-model_00-model_states.pt +3 -0
  9. layer_09-model_00-model_states.pt +3 -0
  10. layer_10-model_00-model_states.pt +3 -0
  11. layer_11-model_00-model_states.pt +3 -0
  12. layer_12-model_00-model_states.pt +3 -0
  13. layer_13-model_00-model_states.pt +3 -0
  14. layer_14-model_00-model_states.pt +3 -0
  15. layer_16-model_00-model_states.pt +3 -0
  16. mp_rank_00_model_states.pt +3 -0
  17. zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
  18. zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
  19. zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
  20. zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
  21. zero_pp_rank_4_mp_rank_00_optim_states.pt +3 -0
  22. zero_pp_rank_5_mp_rank_00_optim_states.pt +3 -0
  23. zero_pp_rank_6_mp_rank_00_optim_states.pt +3 -0
  24. zero_pp_rank_7_mp_rank_00_optim_states.pt +3 -0
README.md CHANGED
@@ -1,3 +1,64 @@
1
  ---
 
2
  license: mit
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ language: bg,cs,da,de,el,en,es,et,fi,fr,ga,hr,hu,it,lt,lv,mt,nl,pl,pt,ro,sk,sl,sv,uk
3
  license: mit
4
  ---
5
+
6
+ # EuroGPT2
7
+
8
+ **NOTE: THIS IS THE ORIGINAL MEGATRON-DEEPSPEED CHECKPOINT INCLUDING OPTIMIZER STATES**
9
+
10
+ A GPT2 language model for European languages (EU-24 + Ukrainian).
11
+ The model follows the original architecture as [OpenAI's GPT2](https://huggingface.co/gpt2/) apart from using [rotary](https://arxiv.org/abs/2104.09864) instead of learned positional embeddigs.
12
+
13
+ ## Model settings
14
+
15
+ - parameters: 124M
16
+ - number of layers: 12
17
+ - hidden size: 768
18
+ - number of heads: 12
19
+ - sequence length: 1024
20
+ - batch size: 168
21
+ - test PPL after training: 23.6 (steps: 436,940)
22
+
23
+ ## Training data
24
+
25
+ - [Wikimedia dumps](https://dumps.wikimedia.org/) (Wikipedia, Wikinews, Wikibooks, Wikisource, Wikivoyage; 20230301)
26
+ - [EUR-Lex](https://huggingface.co/datasets/joelito/eurlex_resources)
27
+ - [OSCAR 2023.01](https://huggingface.co/datasets/oscar-corpus/OSCAR-2301)
28
+ - Tokens: 75,167,662,080
29
+
30
+ ## Languages
31
+
32
+ Included languages: Bulgarian, Czech, Danish, German, Greek, English, Spanish, Estonian, Finnish, French, Irish, Croatian, Hungarian, Italian, Lithuanian, Latvian, Maltese, Dutch, Polish, Portuguese, Romanian, Slovak, Slovenian, Swedish, and Ukrainian.
33
+
34
+ | Language | Ratio |
35
+ | -------- | ------ |
36
+ | bg | 5,92% |
37
+ | cs | 4,77% |
38
+ | da | 2,19% |
39
+ | de | 7,36% |
40
+ | el | 8,60% |
41
+ | en | 10,11% |
42
+ | es | 6,57% |
43
+ | et | 1,67% |
44
+ | fi | 2,70% |
45
+ | fr | 7,18% |
46
+ | ga | 0,25% |
47
+ | hr | 1,09% |
48
+ | hu | 6,38% |
49
+ | it | 5,80% |
50
+ | lt | 2,01% |
51
+ | lv | 1,76% |
52
+ | mt | 1,49% |
53
+ | nl | 5,20% |
54
+ | pl | 4,82% |
55
+ | pt | 4,64% |
56
+ | ro | 2,93% |
57
+ | sk | 2,03% |
58
+ | sl | 1,54% |
59
+ | sv | 3,00% |
60
+
61
+ ## License
62
+
63
+ MIT
64
+
layer_01-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:63693d723a725d89521f988d15fb02b8a0f6f950fdadbf9305f12078568ca1e9
3
+ size 245760944
layer_03-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d6bc86f5ca59ce7e6ed1672aebaf18b918cb33789b3daf39a023a0c2817a8f4b
3
+ size 14180439
layer_04-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cdd83b88ebd4b97eef43f7c104b249f169d55b037064bcff98acd567bde797ce
3
+ size 14180439
layer_05-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3b4a48681e2207020e5fd259b0ab553f06514cd43a0ce251dbf256b19b54528a
3
+ size 14180439
layer_06-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d98ebeee8fb3819bdc860e81a92e854cd5a82ea783271b8a6f20d67b9a932b58
3
+ size 14180439
layer_07-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11dfa01b44cdb0f024c427a2aea2809ce2d4f27cb8a93ec04955e8156861bfad
3
+ size 14180439
layer_08-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f870fdd7994921e8730d894f1d2a6ee0bd9d7c79db58561d861bc1a134241e51
3
+ size 14180439
layer_09-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e1b6c1054949dd915124b11863124cabf9f3945186fde8d7e815818eea615ad3
3
+ size 14180439
layer_10-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea3d6f83ef9c2f664d3b296b3c6d3590194900c16c637bfd7017d263fe1befb7
3
+ size 14180439
layer_11-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb7384d86897a7055e9c4451d2a220f48176a180825396b744f530c8fa7519b0
3
+ size 14180439
layer_12-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8454f3493eb304025e825c9ed107b859db7cafe4dfbfd60e5b04bdcf00433644
3
+ size 14180439
layer_13-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08f29955c3e3f6bfe42af95b7c2ca1fdbfb05c9a3a7a06be372cd7e2e67a1614
3
+ size 14180439
layer_14-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40610079091482676c0aa2a892b7500822f65abba9113fd9cd778f1960499f0c
3
+ size 14180439
layer_16-model_00-model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:58ebfeb6e0aade303d4d6c2a6d1ec380285c83d41f552649f3c92a2d1058d8ee
3
+ size 4291
mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e2c9e4dfbaebf178477f324306e42791b366ec0aaaeecafaa4fbad642616194
3
+ size 34163
zero_pp_rank_0_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ae9a8e198d8ccfc62d3a2eb19444c269b547b545d835d2d07ef581120b5e9b0
3
+ size 311908384
zero_pp_rank_1_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:44bd79f0fcddef6decf7053edd377d7ca21e6f863a45442905aa22b064506b2a
3
+ size 311908384
zero_pp_rank_2_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ddec9b7d663245e91ba78ff527c9561bba947b12ce05a460473bff968c2deab5
3
+ size 311908384
zero_pp_rank_3_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:63132648b36d4051fe2be22858efe32a46d7a11038e0e3a7ea9e650d12d2283d
3
+ size 311908384
zero_pp_rank_4_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:57741524bd3f68458a59b7984cb464459861718d285e64b5f737ac7fe72f2036
3
+ size 311908384
zero_pp_rank_5_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:05bb0e625625304f47b4230fc46349af4ec6b131aa9d97ea4b177ce0db01806d
3
+ size 311908384
zero_pp_rank_6_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fd79490a89d1884f805e388d03989ca18bcdfc2d60bb70b8cb493168e412a2bb
3
+ size 311908384
zero_pp_rank_7_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f23a3266dbc8a815a22d0a80ffb43d1c51dc26c4a85394258559d617e07d0fd
3
+ size 311908384