Mixture-of-experts models: many experts per layer, only a few active per token.
🏗️ Building on HF
nk bass
nkthebass
P(doom) 10%
AI & ML interests
None yet
Recent Activity
updated a collection 3 days ago
Research and data updated a collection 3 days ago
TinyBrainBot v4.1 updated a model 3 days ago
nkthebass/tinybrainbot-61m-a10m-v4.1-moe-baseOrganizations
Third generation: 100M and 350M base, instruct and math models.
-
nkthebass/tinybrainbot-350mV3-base
Text Generation • 0.3B • Updated • 377 -
nkthebass/tinybrainbot-350mV3-instruct
Text Generation • 0.3B • Updated • 372 -
nkthebass/tinybrainbot-350mV3-math
Text Generation • 0.3B • Updated • 257 -
nkthebass/tinybrainbot-100m-v3-base
Text Generation • 0.1B • Updated • 319 • 1
The first TinyBrainBot models, from before the series had version numbers.
Current generation: thinking models that reason before answering, and a 25M base.
-
nkthebass/tinybrainbot-350m-v4-thinking
Text Generation • 0.3B • Updated • 440 -
nkthebass/tinybrainbot-100m-v4-thinking
Text Generation • 0.1B • Updated • 1.05k -
nkthebass/tinybrainbot-25m-v4-base
Text Generation • 30.6M • Updated • 345 -
TinyBrainBot 100M Thinking
🧠1A 100M model that thinks before it answers, in your browser
Second generation: the 303M and 320M base, instruct and math models.
Controlled experiments and datasets.
Mixture-of-experts models: many experts per layer, only a few active per token.
Current generation: thinking models that reason before answering, and a 25M base.
-
nkthebass/tinybrainbot-350m-v4-thinking
Text Generation • 0.3B • Updated • 440 -
nkthebass/tinybrainbot-100m-v4-thinking
Text Generation • 0.1B • Updated • 1.05k -
nkthebass/tinybrainbot-25m-v4-base
Text Generation • 30.6M • Updated • 345 -
TinyBrainBot 100M Thinking
🧠1A 100M model that thinks before it answers, in your browser
Third generation: 100M and 350M base, instruct and math models.
-
nkthebass/tinybrainbot-350mV3-base
Text Generation • 0.3B • Updated • 377 -
nkthebass/tinybrainbot-350mV3-instruct
Text Generation • 0.3B • Updated • 372 -
nkthebass/tinybrainbot-350mV3-math
Text Generation • 0.3B • Updated • 257 -
nkthebass/tinybrainbot-100m-v3-base
Text Generation • 0.1B • Updated • 319 • 1
Second generation: the 303M and 320M base, instruct and math models.
The first TinyBrainBot models, from before the series had version numbers.
Controlled experiments and datasets.