Extending the Context of Pretrained LLMs by Dropping Their Positional Embedding (https://www.arxiv.org/abs/2512.12167)
AI & ML interests
We are a Tokyo-based R&D company on a quest to create a new kind of foundational AI model based on nature-inspired intelligence.
Recent Activity
Papers
View all Papers spaces 6
Running on Zero
20
Llama 3 Karamaru V1
🐻
Classical Japanese Chatbot
Running on Zero
46
Evo-Ukiyoe
🐠
Generate Japanese ukiyo-e style images from prompts
Runtime error
18
Llama-3-EvoVLM-JP-v2
🐠
Chat with images and text using a Japanese visual language model
Running on Zero
18
Evo-Nishikie
🐠
Generate detailed ukiyo-e style images from single-color input
Sleeping
53
EvoSDXL JP
🐠
Generate images from Japanese prompts
Running on Zero
63
EvoVLM JP
🐠
Answer questions about images in Japanese
models 37
SakanaAI/DreamCubedCP2
Updated
SakanaAI/DreamCubedDP2
Updated
SakanaAI/DreamCubedDP4
Updated
SakanaAI/RePo-OLMo2-7B-stage2-L10
Text Generation • 7B • Updated • 299 • 1
SakanaAI/SparseLM2B
2B • Updated • 17 • 1
SakanaAI/SparseLM1B
1B • Updated • 12
SakanaAI/SparseLM0.5B
0.6B • Updated • 14
SakanaAI/SparseLM1.5B
2B • Updated • 12
SakanaAI/Metom
5.64M • Updated • 230 • 6
SakanaAI/doc-to-lora
Updated • 15
datasets 17
SakanaAI/DreamCubed2M
Updated
SakanaAI/lm-wheels
Updated • 46
SakanaAI/EDINET-Bench
Viewer • Updated • 2.59k • 642 • 13
SakanaAI/self_gen_qa_d2l
Viewer • Updated • 9.91M • 5.16k
SakanaAI/Kamon
Viewer • Updated • 10k • 13 • 2
SakanaAI/mazes-extralarge
Viewer • Updated • 50k • 8
SakanaAI/mazes-large
Viewer • Updated • 50k • 39
SakanaAI/mazes-medium
Viewer • Updated • 60k • 32
SakanaAI/mazes-small
Viewer • Updated • 10k • 68
SakanaAI/TransEvalnia
Viewer • Updated • 3.84k • 19 • 4