Skip to content

fork of Self-playing Adversarial Language Game Enhances LLM Reasoning

active 2024-04-162026-01-12 (UTC)

Complete coverage26,533 / 26,533 hourly files (100%) · 2 absent upstream2023-08-152026-08-24 (UTC)
Events
236
Pushes
23
Pull requests
5
Issues
11
Stars
133
Forks
24

Activity over time

Daily event counts in the loaded window

Line chart, 637 days from 2024-04-16 to 2026-01-12. Pushes: 23 total, peak 3 in a day. Pull requests: 5 total, peak 2 in a day. Issues: 11 total, peak 2 in a day. Comments: 39 total, peak 11 in a day. Stars: 133 total, peak 9 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
Linear954321121
thwu18008
PinzhengWang3225003
underwoodnoble5230
mickelliu4002
HanClinto3001
Heathcliff-Zhao3002
hljjjmssyh2001
wangzizhao2001
wangjs91000
eltociear1010
vijetadeshpande1000

Recent activity

Latest issues, pull requests and releases

  • Issue comment#15Linear952025-07-30 14:08
    Is the ppo loss negative given the advantage is always positive?
  • Issue#15wangjs92025-07-29 07:00
    Is the ppo loss negative given the advantage is always positive?
  • Issue comment#14Linear952025-05-06 03:36
    Questions about train_imitation_gpt4.json
  • Issue comment#13wangzizhao2025-04-13 02:52
    bug in compute_lm_loglikeli
  • Issue#13wangzizhao2025-04-13 02:52
    bug in compute_lm_loglikeli
  • Issue comment#5mickelliu2025-01-23 02:25
    Extremely unbalanced attacker defender winrate and strange behavior of SFTed model
  • Issue#12mickelliu2025-01-11 06:29
    Question with `assign_rewards.py`
  • Issue comment#12Linear952025-01-10 11:36
    Question with `assign_rewards.py`
  • Issue comment#12mickelliu2025-01-10 09:58
    Question with `assign_rewards.py`
  • Issue#12mickelliu2025-01-10 08:54
    Question with `assign_rewards.py`
  • Issue comment#11Linear952024-12-18 16:57
    Did you use alpaca sft data when training the imitation model?
  • Issue comment#10Linear952024-11-08 10:34
    It seems that the data format of aplaca.json does not conform to the read logic of dataLoader.
  • Issue comment#10hljjjmssyh2024-11-08 03:31
    It seems that the data format of aplaca.json does not conform to the read logic of dataLoader.
  • Issue comment#10Linear952024-11-08 03:13
    It seems that the data format of aplaca.json does not conform to the read logic of dataLoader.
  • Issue#10hljjjmssyh2024-11-08 02:34
    It seems that the data format of aplaca.json does not conform to the read logic of dataLoader.
  • Issue comment#9Linear952024-09-09 11:28
    Running on two 3090s?
  • Issue#9vijetadeshpande2024-09-04 19:44
    Running on two 3090s?
  • Issue comment#5PinzhengWang3222024-06-28 07:32
    Extremely unbalanced attacker defender winrate and strange behavior of SFTed model
  • Issue comment#5thwu12024-06-26 23:46
    Extremely unbalanced attacker defender winrate and strange behavior of SFTed model
  • Issue comment#8PinzhengWang3222024-06-26 13:05
    Checkpoint
  • Issue#8PinzhengWang3222024-06-26 13:05
    Checkpoint
  • Issue comment#5PinzhengWang3222024-06-26 13:03
    Extremely unbalanced attacker defender winrate and strange behavior of SFTed model
  • Issue comment#8Linear952024-06-26 11:31
    Checkpoint
  • Issue#8PinzhengWang3222024-06-26 09:16
    Checkpoint
  • Issue comment#2Heathcliff-Zhao2024-05-24 03:02
    Question about experiment in paper

Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 133 stars here means stars gained during the window, not the repo's star count.