1 1 9

Vadim Smirnov

ukint-vs

https://ukint-vs.github.io

AI & ML interests

None yet

Recent Activity

liked a model about 23 hours ago

mlboydaisuke/gemma-4-E4B-coreml

liked a model 3 days ago

LiquidAI/LFM2.5-350M-MLX-8bit

liked a model 6 days ago

BugTraceAI/BugTraceAI-Apex-G4-26B-Q4

View all activity

Organizations

liked a model about 23 hours ago

mlboydaisuke/gemma-4-E4B-coreml

Text Generation • Updated 3 days ago • 1

liked a model 3 days ago

LiquidAI/LFM2.5-350M-MLX-8bit

Text Generation • 99.7M • Updated 19 days ago • 2.43k • 7

liked a model 6 days ago

BugTraceAI/BugTraceAI-Apex-G4-26B-Q4

25B • Updated 6 days ago • 17.1k • 54

upvoted a paper 6 days ago

DFlash: Block Diffusion for Flash Speculative Decoding

Paper • 2602.06036 • Published Feb 5 • 54

liked a model 9 days ago

argmaxinc/whisperkit-coreml

Automatic Speech Recognition • Updated Jan 27 • 7.05M • 169

liked a model 14 days ago

kai-os/gemma4-31b-Opus-4.6-reasoning

Text Generation • Updated 18 days ago • 474 • 158

New activity in mlx-community/gemma-4-e2b-4bit 15 days ago

⚠️ Existing MLX quantized Gemma 4 models (mlx-community, unsloth) produce garbage output due to quantizing PLE (Per-Layer Embedding) layers.

#1 opened 17 days ago by

Alkd

liked a model 15 days ago

ukint-vs/gemma-4-21b-a4b-it-REAP-MLX-4bit

Image-Text-to-Text • 4B • Updated 15 days ago • 1.05k • 3

updated a model 15 days ago

ukint-vs/gemma-4-19b-a4b-it-REAP-MLX-4bit

Image-Text-to-Text • 4B • Updated 15 days ago • 574

published a model 15 days ago

ukint-vs/gemma-4-19b-a4b-it-REAP-MLX-4bit

Image-Text-to-Text • 4B • Updated 15 days ago • 574

updated a model 15 days ago

ukint-vs/gemma-4-21b-a4b-it-REAP-MLX-4bit

Image-Text-to-Text • 4B • Updated 15 days ago • 1.05k • 3

published a model 15 days ago

ukint-vs/gemma-4-21b-a4b-it-REAP-MLX-4bit

Image-Text-to-Text • 4B • Updated 15 days ago • 1.05k • 3

liked 3 models 15 days ago

Vadim Smirnov

AI & ML interests

Recent Activity

Organizations

ukint-vs's activity

⚠️ Existing MLX quantized Gemma 4 models (mlx-community, unsloth) produce garbage output due to quantizing PLE (Per-Layer Embedding) layers.