Samsung AI Lab (SAIL) Montreal

SamsungSAILMontreal's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Articles

bknyaz 

updated a model 4 months ago

SamsungSAILMontreal/chart2table-qwen3.5-4b-v1

Image-Text-to-Text • Updated Jul 9

bknyaz 

published a model 4 months ago

SamsungSAILMontreal/chart2table-qwen3.5-4b-v1

Image-Text-to-Text • Updated Jul 9

bknyaz 

updated a collection 7 months ago

REAM

Collection

Compressed MoE models with a reduced number of experts. See additional models at https://huggingface.co/bknyaz. • 12 items • Updated Apr 20 • 6

bknyaz 

in SamsungSAILMontreal/Qwen3-30B-A3B-Instruct-2507-REAM 7 months ago

keep doing it!

❤️ 1

2

#4 opened 8 months ago by

pirola

bknyaz 

updated a collection 7 months ago

REAM

Collection

Compressed MoE models with a reduced number of experts. See additional models at https://huggingface.co/bknyaz. • 12 items • Updated Apr 20 • 6

bknyaz 

updated 2 models 7 months ago

SamsungSAILMontreal/Qwen3-Coder-Next-REAP

Text Generation • 60B • Updated Apr 9 • 40 • 4

SamsungSAILMontreal/GLM-4.5-Air-REAP

Text Generation • 82B • Updated Apr 9 • 18

bknyaz 

published a model 7 months ago

SamsungSAILMontreal/GLM-4.5-Air-REAP

Text Generation • 82B • Updated Apr 9 • 18

bknyaz 

updated a model 7 months ago

bknyaz/GLM-4.5-Air-REAM

Text Generation • 82B • Updated Apr 9 • 13 • 1

bknyaz 

published a model 7 months ago

bknyaz/GLM-4.5-Air-REAM

Text Generation • 82B • Updated Apr 9 • 13 • 1

bknyaz 

submitted a paper to Daily Papers 7 months ago

REAM: Merging Improves Pruning of Experts in LLMs

Paper • 2604.04356 • Published Apr 6 • 8

bknyaz 

updated 2 models 7 months ago

SamsungSAILMontreal/Qwen3-30B-A3B-Instruct-2507-REAM

Text Generation • 23B • Updated Apr 8 • 32 • 7

SamsungSAILMontreal/Qwen3-30B-A3B-Instruct-2507-REAP

Text Generation • 23B • Updated Apr 8 • 21

bknyaz 

in bknyaz/Qwen3-Next-80B-A3B-Instruct-REAM 9 months ago

Has REAM been checked for their resilience towards quantization?

4

#1 opened 10 months ago by

TomLucidor

bknyaz 

in SamsungSAILMontreal/Qwen3-30B-A3B-Instruct-2507-REAM 9 months ago

Testing REAM on Kimi-Linear and Nemtron's hybrid attention models

2

#2 opened 9 months ago by

TomLucidor

bknyaz 

in SamsungSAILMontreal/Qwen3-Coder-Next-REAP 9 months ago

wrong title

2

#1 opened 9 months ago by

floory

bknyaz 

in SamsungSAILMontreal/README 9 months ago

Do you plan to test the REAM method on other models than Qwen3 ?

1

#1 opened 9 months ago by

Rebis

bknyaz 

updated a collection 9 months ago

REAM

Collection

Compressed MoE models with a reduced number of experts. See additional models at https://huggingface.co/bknyaz. • 12 items • Updated Apr 20 • 6