Yosef Worku Alemneh

rasyosef

AI & ML interests

Pretraining, Supervised Fine Tuning, Direct Preference Optimization, Retrieval Augmented Generation (RAG), Function Calling

Recent Activity

updated a dataset about 16 hours ago

rasyosef/open-perfectblend_K2-Horizon-7B

published a dataset about 16 hours ago

rasyosef/open-perfectblend_K2-Horizon-7B

new activity 9 days ago

IFM/K2-Horizon-3.7B:Any MTP or DFlash/DSpark drafter planned?

View all activity

Organizations

updated a dataset about 16 hours ago

rasyosef/open-perfectblend_K2-Horizon-7B

Viewer • Updated about 16 hours ago • 50.4k • 11

published a dataset about 16 hours ago

rasyosef/open-perfectblend_K2-Horizon-7B

Viewer • Updated about 16 hours ago • 50.4k • 11

New activity in IFM/K2-Horizon-3.7B 9 days ago

Any MTP or DFlash/DSpark drafter planned?

#8 opened 10 days ago by

rasyosef

Up to 4.1x speedup with a DSpark draft model for K2-Horizon-3.7B (109 --> 450 tokens/sec)

#9 opened 9 days ago by

rasyosef

updated a model 9 days ago

rasyosef/K2-Horizon-3.7B-DSpark

0.5B • Updated 9 days ago • 61

published a model 9 days ago

rasyosef/K2-Horizon-3.7B-DSpark

0.5B • Updated 9 days ago • 61

commented a paper 10 days ago

Unlocking Lossless Speedups in LLMs via Discrete Diffusion

Paper • 2609.04010 • Published Sep 3 • 113 •

updated a dataset 10 days ago

rasyosef/open-perfectblend_K2-Horizon-3.7B

Viewer • Updated 10 days ago • 48.6k • 44

published a dataset 10 days ago

rasyosef/open-perfectblend_K2-Horizon-3.7B

Viewer • Updated 10 days ago • 48.6k • 44

liked a dataset 12 days ago

RedHatAI/speculator_benchmarks

Preview • Updated Apr 15 • 2.82k • 5

updated a collection 14 days ago

DSpark & DFlash2

Collection

DSpark and DFlash2 draft models for faster inference with speculative decoding • 18 items • Updated 14 days ago • 3

updated 2 models 14 days ago

rasyosef/Qwen3.5-2B-DSpark-v2

0.3B • Updated 14 days ago • 159

rasyosef/gemma-4-E4B-it-dspark-GGUF

Text Generation • 0.5B • Updated 14 days ago • 393

published a model 14 days ago

rasyosef/gemma-4-E4B-it-dspark-GGUF

Text Generation • 0.5B • Updated 14 days ago • 393

updated a model 14 days ago

rasyosef/gemma-4-E4B-it-spec-v2

0.5B • Updated 14 days ago • 26

updated a model 15 days ago

rasyosef/gemma-4-E2B-it-dspark-GGUF

Text Generation • 0.3B • Updated 15 days ago • 1.28k • 1

updated a dataset 16 days ago

rasyosef/open-perfectblend_gemma-4-E4B-it

Viewer • Updated 16 days ago • 211k • 105

published a model 16 days ago

rasyosef/gemma-4-E4B-it-spec-v2

0.5B • Updated 14 days ago • 26

updated a model 21 days ago

rasyosef/gemma-4-E4B-it-dspark

0.5B • Updated 21 days ago • 114 • 1

New activity in google/gemma-4-E2B-it 21 days ago

Up to 4.23x faster inference with a DSpark draft model for gemma-4-E2B-it (167 --> 709 tokens/s; faster than MTP on 7/9 benchmarks)

#52 opened 21 days ago by

rasyosef

Yosef Worku Alemneh

AI & ML interests

Recent Activity

Organizations

rasyosef's activity

Any MTP or DFlash/DSpark drafter planned?

Up to 4.1x speedup with a DSpark draft model for K2-Horizon-3.7B (109 --> 450 tokens/sec)

Up to 4.23x faster inference with a DSpark draft model for gemma-4-E2B-it (167 --> 709 tokens/s; faster than MTP on 7/9 benchmarks)