Search: fergusfinn.com | Heykuki News

Heykuki News

Top New Best Ask Show Jobs

Top New Best Ask Show Jobs

1.

Bringing Up DeepSeek-V4-Flash on AMD MI300X (fergusfinn.com)

117 points

a day ago

2.

Also-RANS: Asymmetric Numeral Systems for Entropy Coding (fergusfinn.com)

25 points

a month ago

3.

Redundant Information in LLM Weights (fergusfinn.com)

5 points

a month ago

4.

70x faster cold(ish) starts for SGLang (fergusfinn.com)

4 points

a month ago

5.

Tans: Precomputing RANS (fergusfinn.com)

3 points

a month ago

6.

Pushing memory bound CUDA kernels past the speed of light with data compression (fergusfinn.com)

2 points

6 days ago

7.

Speculative KV coding: ~4× losslessly compressed KV cache using a small model (fergusfinn.com)

2 points

22 days ago

8.

How fast can an LLM go? (fergusfinn.com)

2 points

7 months ago

9.

How fast can an LLM go? (fergusfinn.com)

2 points

7 months ago

10.

How fast can an LLM go? (fergusfinn.com)

2 points

7 months ago

11.

70x faster cold(ish) starts for SGLang (fergusfinn.com)

1 point

9 hours ago

12.

In search of wasted bits: how much information do LLM weights carry? (fergusfinn.com)

1 point

25 days ago

13.

70x faster cold(ish) starts for SGLang (fergusfinn.com)

1 point

a month ago

14.

Parallel Primitives for Multi-Agent Workflows (fergusfinn.com)

1 point

5 months ago

15.

LLM powered data structures: A lock-free binary search tree (fergusfinn.com)

1 point

5 months ago

16.

Parallel Primitives for Multi-Agent Workflows (fergusfinn.com)

1 point

5 months ago

17.

Scheduling in LLM Inference (fergusfinn.com)

1 point

7 months ago