Sub-2b T2T LLMs - Finetunes
A list of Finetuned LLMs, improving some or all aspects of a base LLM
1B • Updated • 24 • 22Note the smallest usable role-play enhanced language model.
nvidia/Nemotron-Research-Reasoning-Qwen-1.5B
Text Generation • 2B • Updated • 1.88k • 234Note a tiny reasoning model, from a trillion dollar company, who refuses to make GPUs cheaper...
nvidia/OpenReasoning-Nemotron-1.5B
Text Generation • 2B • Updated • 258 • 51Note another tiny reasoning model, from a trillion dollar company, who refuses to make GPUs cheaper...
agentica-org/DeepScaleR-1.5B-Preview
Text Generation • 2B • Updated • 12.9k • 577Note an improved version of the DeepSeek r1's tiniest distill
agentica-org/DeepCoder-1.5B-Preview
Text Generation • 2B • Updated • 134 • 71Note an improved version of the DeepSeek r1's tiniest distill, with coding capabilities enhanced
Menlo/Lucy
Text Generation • 2B • Updated • 34 • 64Note a finetune to enhance web search
Menlo/Lucy-128k
Text Generation • 2B • Updated • 219 • 108Note a finetune to enhance web search, now with 128k tokens of context!
dnotitia/Smoothie-Qwen3-1.7B
Text Generation • 2B • Updated • 103 • 2Note a finetune to smooth out token probability
DavidAU/Llama-3.2-1B-Instruct-NEO-SI-FI-GGUF
Text Generation • 1B • Updated • 549 • 10Note a creative writing enhanced finetune, uncensored too!
huihui-ai/Huihui-MoE-1.5B-A0.6B-abliterated
Text Generation • 2B • Updated • 26Note a tiny MoE model, built with a bunch of qwen3-0.6b experts. uncensored.
DavidAU/Gemma-3-1b-it-MAX-NEO-Imatrix-GGUF
Text Generation • 1.0B • Updated • 692 • 5Note a creative writing enhanced finetune, uncensored too!
Goekdeniz-Guelmez/Josiefied-Qwen3-1.7B-abliterated-v1
Text Generation • 2B • Updated • 1.09k • 6Note a better abliteration of qwen3-1.7b
janhq/Jan-v1-edge
Text Generation • 2B • Updated • 58 • 39Note a finetune to enhance web search
POLARIS-Project/Polaris-1.7B-Preview
2B • Updated • 42 • 8Note a finetune that greatly enhances math capabilities
WeiboAI/VibeThinker-1.5B
Text Generation • 2B • Updated • 2.38k • 508Note a VERY powerful (and token consuming) Language model!
-
nvidia/DLER-R1-1.5B-Research
2B • Updated • 10.7k • 17 -
nvidia/AceMath-1.5B-Instruct
Text Generation • 2B • Updated • 1.2k • 15