Skip to content

Commit 865034d

Browse files
committed
Add bloom presets (keras-team#1501)
1 parent 49c243b commit 865034d

File tree

1 file changed

+95
-4
lines changed

1 file changed

+95
-4
lines changed

keras_nlp/models/bloom/bloom_presets.py

Lines changed: 95 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -17,14 +17,105 @@
1717
"bloom_560m_multi": {
1818
"metadata": {
1919
"description": (
20-
"24-layer Bloom model. trained on 45 natural languages and "
21-
"12 programming languages."
20+
"24-layer Bloom model with hidden dimension of 1024. "
21+
"trained on 45 natural languages and 12 programming languages."
2222
),
23-
"params": 816115712,
23+
"params": 559214592,
2424
"official_name": "BLOOM",
2525
"path": "bloom",
26-
"model_card": "https://huggingface.co/bigscience/bloom",
26+
"model_card": "https://huggingface.co/bigscience/bloom-560m",
2727
},
2828
"kaggle_handle": "kaggle://keras/bloom/keras/bloom_560m_multi/3",
2929
},
30+
"bloom_1.1b_multi": {
31+
"metadata": {
32+
"description": (
33+
"24-layer Bloom model with hidden dimension of 1536. "
34+
"trained on 45 natural languages and 12 programming languages."
35+
),
36+
"params": 1065314304,
37+
"official_name": "BLOOM",
38+
"path": "bloom",
39+
"model_card": "https://huggingface.co/bigscience/bloom-1b1",
40+
},
41+
"kaggle_handle": "kaggle://keras/bloom/keras/bloom_1.1b_multi/1",
42+
},
43+
"bloom_1.7b_multi": {
44+
"metadata": {
45+
"description": (
46+
"24-layer Bloom model with hidden dimension of 2048. "
47+
"trained on 45 natural languages and 12 programming languages."
48+
),
49+
"params": 1722408960,
50+
"official_name": "BLOOM",
51+
"path": "bloom",
52+
"model_card": "https://huggingface.co/bigscience/bloom-1b7",
53+
},
54+
"kaggle_handle": "kaggle://keras/bloom/keras/bloom_1.7b_multi/1",
55+
},
56+
"bloom_3b_multi": {
57+
"metadata": {
58+
"description": (
59+
"30-layer Bloom model with hidden dimension of 2560. "
60+
"trained on 45 natural languages and 12 programming languages."
61+
),
62+
"params": 3002557440,
63+
"official_name": "BLOOM",
64+
"path": "bloom",
65+
"model_card": "https://huggingface.co/bigscience/bloom-3b",
66+
},
67+
"kaggle_handle": "kaggle://keras/bloom/keras/bloom_3b_multi/1",
68+
},
69+
"bloomz_560m_multi": {
70+
"metadata": {
71+
"description": (
72+
"24-layer Bloom model with hidden dimension of 1024. "
73+
"finetuned on crosslingual task mixture (xP3) dataset."
74+
),
75+
"params": 559214592,
76+
"official_name": "BLOOMZ",
77+
"path": "bloom",
78+
"model_card": "https://huggingface.co/bigscience/bloomz-560m",
79+
},
80+
"kaggle_handle": "kaggle://keras/bloom/keras/bloomz_560m_multi/1",
81+
},
82+
"bloomz_1.1b_multi": {
83+
"metadata": {
84+
"description": (
85+
"24-layer Bloom model with hidden dimension of 1536. "
86+
"finetuned on crosslingual task mixture (xP3) dataset."
87+
),
88+
"params": 1065314304,
89+
"official_name": "BLOOMZ",
90+
"path": "bloom",
91+
"model_card": "https://huggingface.co/bigscience/bloomz-1b1",
92+
},
93+
"kaggle_handle": "kaggle://keras/bloom/keras/bloomz_1.1b_multi/1",
94+
},
95+
"bloomz_1.7b_multi": {
96+
"metadata": {
97+
"description": (
98+
"24-layer Bloom model with hidden dimension of 2048. "
99+
"finetuned on crosslingual task mixture (xP3) dataset."
100+
),
101+
"params": 1722408960,
102+
"official_name": "BLOOMZ",
103+
"path": "bloom",
104+
"model_card": "https://huggingface.co/bigscience/bloomz-1b7",
105+
},
106+
"kaggle_handle": "kaggle://keras/bloom/keras/bloomz_1.7b_multi/1",
107+
},
108+
"bloomz_3b_multi": {
109+
"metadata": {
110+
"description": (
111+
"30-layer Bloom model with hidden dimension of 2560. "
112+
"finetuned on crosslingual task mixture (xP3) dataset."
113+
),
114+
"params": 3002557440,
115+
"official_name": "BLOOMZ",
116+
"path": "bloom",
117+
"model_card": "https://huggingface.co/bigscience/bloomz-3b",
118+
},
119+
"kaggle_handle": "kaggle://keras/bloom/keras/bloomz_3b_multi/1",
120+
},
30121
}

0 commit comments

Comments
 (0)