Skip to content

LLM post-training playbook: SFT, GRPO, DPO, eval, and inference. pip install alignrl

active 2026-03-242026-04-13 (UTC)

Partial coverage14,987 / 18,351 hourly files (82%) · 2 absent upstream · 3,364 failed, retryable2024-07-082026-08-11 (UTC)— sampled evenly across the window, so rankings and trends hold; absolute counts scale up.
Events
52
Pushes
9
Pull requests
14
Issues
8
Stars
0
Forks
0

Activity over time

Daily event counts in the loaded window

Line chart, 21 days from 2026-03-24 to 2026-04-13. Pushes: 9 total, peak 7 in a day. Pull requests: 14 total, peak 13 in a day. Issues: 8 total, peak 8 in a day. Comments: 0 total, peak 0 in a day. Stars: 0 total, peak 0 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
sacredvoid319140

Recent activity

Latest issues, pull requests and releases

  • Pull request#37sacredvoid2026-04-13 04:34
  • Pull request#36sacredvoid2026-03-26 01:32
  • Pull request#34sacredvoid2026-03-26 01:30
  • Issue#29sacredvoid2026-03-26 00:45
    Bug: parse_results includes booleans as numeric metrics (bool is subclass of int)
  • Issue#27sacredvoid2026-03-26 00:44
    Bug: ModelServer._load_unsloth hardcodes max_seq_length=2048, ignoring config
  • Pull request#28sacredvoid2026-03-26 00:44
  • Pull request#28sacredvoid2026-03-26 00:44
  • Issue#25sacredvoid2026-03-26 00:43
    Bug: Runner load() skips ensure_chat_template, breaking inference after reload
  • Pull request#26sacredvoid2026-03-26 00:43
  • Pull request#26sacredvoid2026-03-26 00:43
  • Pull request#22sacredvoid2026-03-26 00:40
  • Pull request#22sacredvoid2026-03-26 00:40
  • Issue#19sacredvoid2026-03-26 00:39
    Bug: from_yaml return type annotation is wrong and crashes on empty YAML
  • Pull request#20sacredvoid2026-03-26 00:39
  • Issue#17sacredvoid2026-03-26 00:38
    Bug: TrainResult.metrics typed dict[str, float] but GRPO stores list[float] in it
  • Pull request#18sacredvoid2026-03-26 00:38
  • Issue#17sacredvoid2026-03-26 00:37
    Bug: TrainResult.metrics typed dict[str, float] but GRPO stores list[float] in it
  • Pull request#16sacredvoid2026-03-26 00:37
  • Issue#15sacredvoid2026-03-26 00:36
    Bug: SFTTrainer uses deprecated `tokenizer` param instead of `processing_class`
  • Pull request#8sacredvoid2026-03-26 00:26
  • Issue#7sacredvoid2026-03-26 00:25
    Bug: vLLM backend passes Python repr of messages instead of formatted prompt
  • Pull request#4sacredvoid2026-03-26 00:24

Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 0 stars here means stars gained during the window, not the repo's star count.