For this competition, we train a text-only LLM that can operate on 3D protein structures by representing them efficiently in a textual format. We use LFM2-2.6b as a memory-efficient yet capable base model, which supports training and inference with the long textual contexts for handling 3D protein representations.
Full description of approach: https://drive.google.com/file/d/1VyqfJerlXkEiFRH4KrddHR1DIuRmSQag/view?usp=sharing
id: silver-lynx-rose

RBX1
None
78.14
True
5.1 kDa
46
id: crimson-heron-ice

RBX1
None
46.96
True
6.2 kDa
53
id: radiant-heron-ruby

RBX1
None
52.74
True
5.2 kDa
47
id: mellow-otter-frost

RBX1
None
54.23
False
7.8 kDa
65
id: wild-zebra-cloud
No preview available
RBX1
None
--
True
--
48
id: azure-ibis-reed

RBX1
None
43.02
True
4.6 kDa
49
id: amber-vole-ash

RBX1
None
58.84
False
7.2 kDa
63
id: soft-eagle-willow
No preview available
--
--
--
--
--
58
id: deep-yak-leaf
No preview available
--
--
--
--
--
67
id: rough-bear-topaz
No preview available
--
--
--
--
--
56
id: vast-boar-willow
No preview available
--
--
--
--
--
66
id: swift-swan-topaz
No preview available
--
--
--
--
--
51
id: quick-moth-pine
No preview available
--
--
--
--
--
65
id: swift-otter-iron
No preview available
--
--
--
--
--
49
id: frozen-yak-vine
No preview available
--
--
--
--
--
48
id: steady-bison-iron
No preview available
--
--
--
--
--
59
id: hollow-moth-snow
No preview available
--
--
--
--
--
66
id: noble-panther-opal
No preview available
--
--
--
--
--
38
id: amber-cobra-birch
No preview available
--
--
--
--
--
48
id: soft-lynx-flint
No preview available
--
--
--
--
--
54
id: pale-eagle-willow
No preview available
--
--
--
--
--
36