Skip to content

active 2026-02-212026-08-24 (UTC)

Complete coverage26,780 / 26,780 hourly files (100%) · 2 absent upstream2023-08-152026-09-03 (UTC)
Events
277
Pushes
136
Pull requests
17
Issues
22
Stars
1
Forks
2

Activity over time

Daily event counts in the loaded window

Line chart, 185 days from 2026-02-21 to 2026-08-24. Pushes: 136 total, peak 12 in a day. Pull requests: 17 total, peak 2 in a day. Issues: 22 total, peak 3 in a day. Comments: 34 total, peak 4 in a day. Stars: 1 total, peak 1 in a day.

  • Pushes
  • Pull requests
  • Issues
  • Comments
  • Stars

Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.

Top contributors

Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity

ContributorContributionsPushesPRsComments
mseeger2061361730
Jantory3001
mmjerge2002
vihangp1001

Recent activity

Latest issues, pull requests and releases

  • Issue comment#148mmjerge2026-08-21 14:26
    Chunked backward fails when a generated region spans 3+ chunks (autograd_hooks annotation adjacency)
  • Issue comment#140mseeger2026-08-03 16:24
    Evicting KV cache corrupts generation (garbage output once prompt > cache_length)
  • Issue comment#133mmjerge2026-06-30 19:39
    Add vLLM (V1) KV cache policy integration: lastrec bridge + spike
  • Issue comment#117mseeger2026-05-27 08:55
    Reuse the same KV cache checkpointing buffers for all shards in the backward pass
  • Issue#79mseeger2026-05-16 21:18
    Create auto-tuning script to find best parameter combinations
  • Issue#111mseeger2026-05-11 06:47
    FInalize tests on FlashInfer SDPA
  • Issue comment#79mseeger2026-05-10 19:10
    Create auto-tuning script to find best parameter combinations
  • Pull request#110mseeger2026-05-10 14:01
  • Issue#99mseeger2026-05-04 07:35
    Fix `smart-lastrec` KV cache
  • Pull request#102mseeger2026-05-04 07:35
  • Issue comment#99mseeger2026-05-02 10:28
    Fix `smart-lastrec` KV cache
  • Issue#101mseeger2026-05-02 08:26
    Support restart from checkpoint
  • Issue comment#97mseeger2026-05-02 08:02
    Basic comparisons with baseline (and train versus eval) on 32k Helmet datasets
  • Pull request#100mseeger2026-05-01 21:12
  • Issue comment#92mseeger2026-04-29 18:28
    GPU memory profiling: Still missing something?
  • Pull request#95mseeger2026-04-29 08:45
  • Issue#80mseeger2026-04-24 07:42
    Implement `lastrec` KV cache variant with variable `init_grace_tokens`
  • Issue comment#83mseeger2026-04-14 06:36
    Add FlashInfer CUDA kernels + Triton score-sum for H2O attention weights
  • Pull request#90mseeger2026-04-13 09:16
  • Pull request#90mseeger2026-04-13 08:33
  • Pull request#85mseeger2026-04-10 18:49
  • Issue#72mseeger2026-04-02 15:23
    Implement training loss and evaluation metric for Helmet datasets
  • Pull request#81mseeger2026-04-02 15:23
  • Issue comment#81mseeger2026-04-02 11:54
    Code for supporting sample-based metrics for evaluation (#81)
  • Pull request#78mseeger2026-03-31 09:01

Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 1 stars here means stars gained during the window, not the repo's star count.