Osye's picture

5

Osye

Osye

·

AI & ML interests

None yet

Recent Activity

reacted to s3nh's post with ❤️ 5 days ago

Existing methods — GPTQ, AWQ, llama.cpp's k-quants — minimize empirical loss heuristically. None of them prove they are optimal in any information-theoretic sense. ICRB-Q builds a quantization scheme that is provably optimal via the Cramér-Rao lower bound (CRB): no unbiased estimator of a weight can have lower variance than [F(θ)]⁻¹, where F is the Fisher information matrix.

liked a model about 1 month ago

Osye/mlp-surgery-restored-top30

updated a collection about 1 month ago

View all activity

Organizations

Collections 1

spaces 1

Model Inspector

models 5

Osye/mlp-surgery-restored-specificity-top10

3B • Updated May 7 • 3 • 1

Osye/mlp-surgery-restored-top30

3B • Updated May 7 • 5 • 2

Osye/mlp-surgery-restored-top15

3B • Updated May 7 • 3

Osye/mlp-surgery-restored-top5

3B • Updated May 7 • 2

Osye/mlp-surgery-broken

3B • Updated May 7 • 3

datasets 1

Osye/openhermes2.5-Perplexity_filtered_top30

Viewer • Updated Mar 8 • 300k • 17 • 1