goldfish-models commited on
Commit
989ae8c
1 Parent(s): 0520953

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +63 -0
README.md ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ ---
3
+ license: apache-2.0
4
+ language:
5
+ - kaz
6
+ datasets:
7
+ - cis-lmu/Glot500
8
+ - oscar-corpus/OSCAR-2109
9
+ - allenai/nllb
10
+ library_name: transformers
11
+ pipeline_tag: text-generation
12
+ tags:
13
+ - goldfish
14
+
15
+ ---
16
+
17
+ # kaz_cyrl_1000mb
18
+
19
+ Goldfish is a suite of monolingual language models trained for 350 languages.
20
+ This model is the <b>Kazakh</b> (Cyrillic script) model trained on 1000MB of data, after accounting for an estimated byte premium of 1.76; content-matched text in Kazakh takes on average 1.76x as many UTF-8 bytes to encode as English.
21
+ The Goldfish models are trained primarily for comparability across languages and for low-resource languages; Goldfish performance for high-resource languages is not designed to be comparable with modern large language models (LLMs).
22
+
23
+ Note: kaz_cyrl is an [individual language](https://iso639-3.sil.org/code_tables/639/data) code. It is not contained in any macrolanguage codes contained in Goldfish (for script cyrl).
24
+
25
+ All training and hyperparameter details are in our paper, [Goldfish: Monolingual Language Models for 350 Languages (Chang et al., 2024)](https://github.com/tylerachang/goldfish/blob/main/goldfish_paper_20240815.pdf).
26
+
27
+ Training code and sample usage: https://github.com/tylerachang/goldfish
28
+
29
+ Sample usage also in this Google Colab: [link](https://colab.research.google.com/drive/1rHFpnQsyXJ32ONwCosWZ7frjOYjbGCXG?usp=sharing)
30
+
31
+ ## Model details:
32
+
33
+ To access all Goldfish model details programmatically, see https://github.com/tylerachang/goldfish/model_details.json.
34
+ All models are trained with a [CLS] (same as [BOS]) token prepended, and a [SEP] (same as [EOS]) token separating sequences.
35
+ Details for this model specifically:
36
+
37
+ * Architecture: gpt2
38
+ * Parameters: 124770816
39
+ * Maximum sequence length: 512 tokens
40
+ * Training text data (raw): 1764.64MB
41
+ * Training text data (byte premium scaled): 1000.005MB
42
+ * Training tokens: 199970304 (x10 epochs)
43
+ * Vocabulary size: 50000
44
+ * Compute cost: 1.020620786761728e+18 FLOPs or ~96.5 NVIDIA A6000 GPU hours
45
+
46
+ Training datasets (percentages prior to deduplication):
47
+ * 68.35707%: [Glot500](https://huggingface.co/datasets/cis-lmu/Glot500), including [CCNet](https://github.com/facebookresearch/cc_net), [Earthlings](https://publicdata.canterbury.ac.nz/Research/Geocorpus/CCGLU_v5.0/), [OSCAR](https://oscar-project.org/), [TIL](https://github.com/turkic-interlingua/til-mt), [W2C](https://lindat.mff.cuni.cz/repository/xmlui/handle/11858/00-097C-0000-0022-6133-9), [WikiMatrix](https://github.com/facebookresearch/LASER/tree/main/tasks/WikiMatrix)
48
+ * 26.84524%: [OSCAR 2021/09](https://huggingface.co/datasets/oscar-corpus/OSCAR-2109)
49
+ * 4.79769%: [NLLB (CommonCrawl and ParaCrawl)](https://huggingface.co/datasets/allenai/nllb)
50
+
51
+
52
+ ## Citation
53
+
54
+ If you use this model, please cite:
55
+
56
+ ```
57
+ @article{chang-etal-2024-goldfish,
58
+ title={Goldfish: Monolingual Language Models for 350 Languages},
59
+ author={Chang, Tyler A. and Arnett, Catherine and Tu, Zhuowen and Bergen, Benjamin K.},
60
+ journal={Preprint},
61
+ year={2024},
62
+ }
63
+ ```