42,180 blogs · 5,214,511 posts Claim your blog

Record · corpus.blog/posts/01a08727-61b0-7114-8a43-62775fca578d

Can Prompt Lookup Decoding Speed Up Gemma 4 31B 4-bit on Apple Silicon for RAG-style Workloads?

amund.blog · published 1 May 2026

A record is what was published and who pointed at it. The text of the post is not held here: read it at the source, or at its Wayback capture.

Read the post at amund.blog Wayback capture

0blogs citing

First cited: —
Most recent: —
Rank this month: not ranked
External links: 18
Words captured: 4,568

Cited by

No blog we hold has cited this post yet.

Links from this post

Link

Host

https://github.com/atveit/InferenceAcceleration/tree/main/pld-gemma4-mlx

github.com

https://github.com/apoorvumang/prompt-lookup-decoding

github.com

https://arxiv.org/abs/2311.08252

arxiv.org

https://github.com/atveit/InferenceAcceleration/blob/main/pld-gemma4-mlx/pld_drafter.py

github.com

https://github.com/atveit/InferenceAcceleration/blob/main/pld-gemma4-mlx/pld_spec_decode.py

github.com

https://github.com/atveit/InferenceAcceleration/blob/main/pld-gemma4-mlx/bench_pld_3class_rerun.py

github.com

https://github.com/atveit/InferenceAcceleration/blob/main/pld-gemma4-mlx/prompts.py

github.com

https://huggingface.co/docs/transformers/main/en/generation_strategies

huggingface.co

https://docs.vllm.ai/en/latest/features/spec_decode.html

docs.vllm.ai

https://github.com/ml-explore/mlx-lm

github.com

https://huggingface.co/mlx-community/gemma-4-31b-it-4bit

huggingface.co

https://omlx.ai/benchmarks/x7egywy9

omlx.ai

https://huggingface.co/inferencerlabs/gemma-4-31B-MLX-9bit

huggingface.co

https://www.reddit.com/r/MacStudio/comments/1sivxai/benchmarked_gemma_4_31b_at_full_bf16_m3_ultra_vs/

reddit.com

https://x.com/AlicanKiraz0/status/2040070001814602094

x.com

https://x.com/davidpwalter/status/2039817107731292495

x.com

https://huggingface.co/Phipper/gemma-4-31b-it-mlx-4bit

huggingface.co

https://atsentia.com/blog/accelerating-gemma4-31b-pld-mlx-lm

atsentia.com