COSPADI: Compressing LLMs via Calibration-Guided Sparse Dictionary Learning
Paper • 2509.22075 • Published • 23 •
2
Ammar Ali
ammarali32
AI & ML interests
PhD in AI and ML, lead researcher at MTS AI.
Kaggle competitions Master
Ranked Top-1 on dsworks platform.
Winner first place of +15 ML competitions on different fields.
Recent Activity
upvoted a paper 2 months ago
OpenAutoNLU: Open Source AutoML Library for NLU authored a paper 3 months ago
COMPOT: Calibration-Optimized Matrix Procrustes Orthogonalization for Transformers CompressionOrganizations
Update README.md
#3 opened 3 months ago
by 
ammarali32
Update README.md
#2 opened 3 months ago
by ammarali32
Upload 2 files
#1 opened 3 months ago
by ammarali32
Adding `safetensors` variant of this model
#1 opened 4 months ago
by 
SFconvertbot
commented a
paper 8 months ago
commented a
paper about 1 year ago
Generated response filled with INST. How to solve this?
4
#7 opened about 2 years ago
by 
ParameshYekkala
Confirm Qwen, not Llama?
1
#8 opened about 2 years ago
by 
sealad886
GPTQ failed submission request
8
#669 opened about 2 years ago
by 
AlanRobotics
License
4
#7 opened about 2 years ago
by 
mrfakename
This is based on Qwen, and it is 72B not 70b
👍 2
1
#6 opened about 2 years ago
by 
ehartford
Prompt template
👀 2
1
#5 opened about 2 years ago
by 
simonesartoni1
exl2 conversion fails, missing tokenizer.config.
1
#4 opened about 2 years ago
by 
p1kp4k
Paper Link
2
#6 opened about 2 years ago
by 
Elliott-Dyson
Adding Evaluation Results
#3 opened about 2 years ago
by 
leaderboard-pr-bot
Exl2 Quantized
2
#2 opened about 2 years ago
by 
rjmehta
Curious about the meaning behind this...
2
#5 opened about 2 years ago
by 
Severian
Instruct-finetuning dataset
👀 1
1
#1 opened about 2 years ago
by 
Andriy
Questionable results for the current top models
2
#642 opened about 2 years ago
by ammarali32
Update chat template
2
#4 opened about 2 years ago
by 
nobita3921