Drop-Upcycling
updated
9B • Updated • 45
9B • Updated • 52
19B • Updated • 46
9B • Updated • 50
9B • Updated • 46
0.4B • Updated • 48
0.4B • Updated • 44
0.4B • Updated • 49
0.4B • Updated • 50
9B • Updated • 45
19B • Updated • 45
0.4B • Updated • 47
9B • Updated • 46
0.4B • Updated • 47
• 1
2B • Updated • 47
0.2B • Updated • 52
4B • Updated • 46
14B • Updated • 47
llm-jp/Dense-btx-code-expert-152M
0.2B • Updated • 53
• 1
llm-jp/Dense-btx-english-expert-1.5B
2B • Updated • 46
llm-jp/Dense-btx-code-expert-1.5B
2B • Updated • 52
• 1
llm-jp/Dense-btx-japanese-expert-1.5B
2B • Updated • 43
• 1
llm-jp/Dense-btx-english-expert-152M
0.2B • Updated • 44
llm-jp/Dense-btx-japanese-expert-152M
0.2B • Updated • 45
Drop-Upcycling: Training Sparse Mixture of Experts with Partial
Re-initialization
Paper
• 2502.19261
• Published • 6
Text Generation
• 73B • Updated • 54
llm-jp/llm-jp-3-8x13b-instruct3
Text Generation
• 73B • Updated • 146
• 8
llm-jp/llm-jp-3-8x1.8b-instruct3
Text Generation
• 9B • Updated • 62
• 4
Text Generation
• 9B • Updated • 114
llm-jp/llm-jp-3-8x13b-instruct2
Text Generation
• 73B • Updated • 156
llm-jp/llm-jp-3-8x1.8b-instruct2
Text Generation
• 9B • Updated • 54
llm-jp/llm-jp-3.1-8x13b-instruct4
Text Generation
• 73B • Updated • 129
• 4
Text Generation
• 73B • Updated • 48