A family of compact reasoning models, based off of the best 2B and 3B models, trained using improved DDP training code, no Unsloth.
Lunah
lunahr
AI & ML interests
None yet
Recent Activity
updated
a collection
9 days ago
Thea
updated
a model
10 days ago
lunahr/thea-pro-2b-100r
published
a model
11 days ago
lunahr/thea-pro-2b-100r
Organizations
None yet
Collections
2
models
15
lunahr/thea-pro-2b-100r
Text Generation
•
Updated
•
59
•
1
lunahr/thea-3b-50r-u1
Text Generation
•
Updated
•
40
lunahr/thea-3b-25r
Text Generation
•
Updated
•
131
•
1
lunahr/thea-v2-3b-50r
Text Generation
•
Updated
•
2
lunahr/thea-c-3b-25r
Text Generation
•
Updated
•
67
•
1
lunahr/thea-rp-3b-25r
Text Generation
•
Updated
•
106
•
1
lunahr/thea-3b-25r-adapter
Updated
•
1
•
1
lunahr/thea-rp-3b-25r-adapter
Updated
•
3
•
1
lunahr/thea-c-3b-25r-adapter
Updated
•
2
•
1
lunahr/Hermes-3-Llama-3.2-3B-abliterated
Text Generation
•
Updated
•
162
•
2