Hugging Face's logo
Activity Feed

AI & ML interests

None defined yet.

Recent Activity
karansag  updated a model about 10 hours ago
open-athena/Snowball-67B-A2B-MixedRS-SFT-Step727
karansag  updated a model about 10 hours ago
open-athena/Snowball-67B-A2B-MOPD-Step100
karansag  updated a model about 10 hours ago
open-athena/Snowball-67B-A2B-CruxEval-RS-SFT-Step282
View all activity

open-athena 's collections 7

SFT Cold Start Subsampling
H100 cold-start SFT reproduction: base, six checkpoints and original 10-05 reference. Subsampling findings are provisional.
10T Mixed RLVR
Selected Snowball 67B-A2B checkpoints from mixed-task RLVR training.
Snowball LCE Issue 8977
Five immutable Snowball LCE base-model exports from marin-community/marin issue #8977.
Chat-Aligned Bases
Pretrained MoE bases after one shared three-stage SFT (chat, thinking, terminal agent), used as comparable starting points for post-training research.
2.7T Math RL
Research artifacts from Snowball 2.7T math RL. Raw mutable-router exports may collapse; use only models marked repaired or verified frozen.
SFT Cold Start Subsampling
H100 cold-start SFT reproduction: base, six checkpoints and original 10-05 reference. Subsampling findings are provisional.
Chat-Aligned Bases
Pretrained MoE bases after one shared three-stage SFT (chat, thinking, terminal agent), used as comparable starting points for post-training research.
10T Mixed RLVR
Selected Snowball 67B-A2B checkpoints from mixed-task RLVR training.
2.7T Math RL
Research artifacts from Snowball 2.7T math RL. Raw mutable-router exports may collapse; use only models marked repaired or verified frozen.
Snowball LCE Issue 8977
Five immutable Snowball LCE base-model exports from marin-community/marin issue #8977.