Commit graph

  • 09271b6202
    ci: add Python 3.14 and 3.15 to CI matrix (#511) master Philipp Emanuel Weidmann 2026-10-20 23:52:02 +05:30
  • f4e641ae15
    ci: add Python 3.14 and 3.15 to CI matrix ci-new-python-versions Philipp Emanuel Weidmann 2026-10-20 23:29:08 +05:30
  • bc077540a0
    feat: make upload tags configurable (#506) Ayush Vishwakarma 2026-10-20 22:12:06 +05:30
  • 1bdb5bcc5a
    fix: support models whose chat template rejects the system role (#498) Varun Balani 2026-10-20 21:47:51 +05:30
  • c5927614f4
    build(deps): bump aiohttp from 3.14.3 to 3.14.4 dependabot/uv/aiohttp-3.14.4 dependabot[bot] 2026-10-20 11:21:39 +00:00
  • eabd1d75fb
    fix: restore stored plugin settings as they are on reproduce and resume (#484) Jay Hemnani 2026-10-19 20:07:50 +05:30
  • 7f86616b61
    fix: say which check found the Chain-of-Thought prefix (#489) Hamed Shams 2026-10-19 17:15:58 +03:00
  • 3a5c864527
    chore: add pull request template (#502) Philipp Emanuel Weidmann 2026-10-17 18:29:38 +05:30
  • 6b33374c7e
    chore: add pull request template pr-template Philipp Emanuel Weidmann 2026-10-17 18:22:53 +05:30
  • 66a952b49b
    fix: report a save or upload before re-ablating the model (#488) Hamed Shams 2026-10-12 15:23:19 +03:00
  • 49b56e5af4
    fix: skip non-metrics in benchmark output (#486) Philipp Emanuel Weidmann 2026-10-07 18:16:53 +05:30
  • 449b351735
    fix: skip non-metrics in benchmark output fix-benchmark-metrics Philipp Emanuel Weidmann 2026-10-07 18:07:26 +05:30
  • fd7c77e5c3
    fix: silence warning spam from PyTorch's C++ logging (#485) Philipp Emanuel Weidmann 2026-10-07 17:14:17 +05:30
  • d128973ca9
    fix: silence warning spam from PyTorch's C++ logging prevent-log-spam Philipp Emanuel Weidmann 2026-10-07 17:05:00 +05:30
  • 17f8dff09f
    fix: escape prompts and responses printed with print_responses (#482) Jay Hemnani 2026-10-06 13:08:04 +05:30
  • 8ae8526c9d
    fix: prevent empty layer range and too many neighbors in ARA (#478) benthecarman 2026-10-06 03:10:51 -04:00
  • 208c0ca35b
    fix: stop lora_A from carrying over between trials (#476) benthecarman 2026-10-04 08:01:40 -04:00
  • a09b4eae07
    chore: add new Windows test hashes (#470) Vinay Umrethe 2026-10-02 15:00:57 +05:30
  • c121a4ba41
    chore: update dependencies (#469) Philipp Emanuel Weidmann 2026-10-02 14:49:47 +05:30
  • dc1a569470
    chore: add yet more CI test hashes update-deps Philipp Emanuel Weidmann 2026-10-02 14:34:32 +05:30
  • ab775193b8
    chore: add new CI test hashes Philipp Emanuel Weidmann 2026-10-02 14:23:08 +05:30
  • 8cac604223
    chore: update dependencies Philipp Emanuel Weidmann 2026-10-02 14:00:50 +05:30
  • 662e4ba27e
    feat: modifier plugins + ARA (#446) Philipp Emanuel Weidmann 2026-10-01 14:30:07 +05:30
  • 15ad4261be
    ci: add yet another alternative model hash modifier-plugins Philipp Emanuel Weidmann 2026-10-01 14:07:30 +05:30
  • c50070f1c8
    ci: remove broken "semantic-pull-request" workflow Philipp Emanuel Weidmann 2026-10-01 14:02:02 +05:30
  • 99be5a6e56
    feat: reduce default number of trials Philipp Emanuel Weidmann 2026-10-01 13:58:47 +05:30
  • 77614a1f76
    Merge branch 'master' into modifier-plugins Philipp Emanuel Weidmann 2026-10-01 12:30:13 +05:30
  • 71e6d5eb38
    feat: add base model to the model card (#459) Vinay Umrethe 2026-09-30 19:22:57 +05:30
  • 405e4ffc0d
    feat: add tests for ARA Philipp Emanuel Weidmann 2026-09-30 18:55:54 +05:30
  • ce8b05c77b
    feat: re-implement ARA as a modifier plugin Philipp Emanuel Weidmann 2026-09-28 19:24:08 +05:30
  • ffa66af2d4
    feat: support dataset specifications containing multiple individual datasets Philipp Emanuel Weidmann 2026-09-25 16:43:11 +05:30
  • dc6703788c
    Merge branch 'master' into modifier-plugins Philipp Emanuel Weidmann 2026-09-22 19:39:10 +05:30
  • 033eb4238c
    fix: adjust good/bad prompts hack to match tests Philipp Emanuel Weidmann 2026-09-22 19:26:02 +05:30
  • 598f5f4bd4
    feat: re-implement abliteration as a modifier plugin Philipp Emanuel Weidmann 2026-09-21 18:44:54 +05:30
  • 3521f8648a
    feat: A better logic for detecting the thinking prefix in the response. (#423) Vinay Umrethe 2026-09-05 21:41:52 +05:30
  • 515191b400
    feat: support specifying a dataset's specific config/subset (#445) Vinay Umrethe 2026-09-05 18:41:08 +05:30
  • 92ab7f09d5
    feat: add modifier base class Philipp Emanuel Weidmann 2026-09-04 19:16:14 +05:30
  • 95dda4c4db
    feat: add benchmark scorer (#444) Philipp Emanuel Weidmann 2026-09-03 17:49:46 +05:30
  • 3599fba081
    Merge branch 'master' into benchmark-scorer benchmark-scorer Philipp Emanuel Weidmann 2026-09-03 17:40:40 +05:30
  • 1c2688f756
    feat: add benchmark scorer Philipp Emanuel Weidmann 2026-09-03 17:32:19 +05:30
  • ded115c669
    fix: improve print output of scorers Philipp Emanuel Weidmann 2026-09-03 13:40:27 +05:30
  • c7a44f0db7
    build(deps): bump nltk from 3.10.0 to 3.10.3 (#442) dependabot[bot] 2026-09-03 10:48:12 +05:30
  • bedb94ef11
    feat: Show a summary at end of tests (#425) Vinay Umrethe 2026-08-17 22:27:44 +05:30
  • 638a583bd8
    chore: bump version to 2.0.0.dev0 (#428) Philipp Emanuel Weidmann 2026-08-17 16:25:19 +05:30
  • 3444a0dd67
    Merge branch 'master' into version-2-dev version-2-dev Philipp Emanuel Weidmann 2026-08-17 16:20:28 +05:30
  • 346d61673b
    chore: remove Gemini style guide (#427) Philipp Emanuel Weidmann 2026-08-17 16:19:58 +05:30
  • b796257c37
    chore: bump version to 2.0.0.dev0 Philipp Emanuel Weidmann 2026-08-17 15:23:36 +05:30
  • 2fb773adc5
    chore: remove Gemini style guide remove-gemini Philipp Emanuel Weidmann 2026-08-17 15:12:34 +05:30
  • edc3b12345
    fix(ara): free gradient buffers after optimization (#426) ara Ashar 2026-08-17 14:57:09 +05:30
  • 1a921522b2
    docs: update model count in README (#424) Vinay Umrethe 2026-08-14 16:36:08 +05:30
  • 9069d3c754
    build(deps): bump aiohttp from 3.14.1 to 3.14.3 (#422) dependabot[bot] 2026-08-06 01:19:01 +05:30
  • 7675b90d64
    build(deps): bump setuptools from 80.9.0 to 83.0.0 (#418) dependabot[bot] 2026-07-28 16:06:37 +05:30
  • 108fff813f
    build(deps): bump pillow from 12.2.0 to 12.3.0 (#416) dependabot[bot] 2026-07-28 15:07:00 +05:30
  • 18eaf9533f
    build(deps): bump torch from 2.9.1 to 2.13.0 (#411) dependabot[bot] 2026-07-24 23:37:59 +05:30
  • 20b7bf7869
    build(deps): bump nltk from 3.9.4 to 3.10.0 (#415) dependabot[bot] 2026-07-22 16:35:01 +05:30
  • 9b6d8a419e
    feat: add reproducible property to plugins (#414) red40maxxer 2026-07-22 06:51:50 -04:00
  • 44228c5c42
    docs: quote pip extras install examples (#395) nyxst4ck 2026-07-17 04:54:20 -03:00
  • 02ebf7a751
    feat: Add alternate CI hash in the valid list (for qwen2.5) (#412) Vinay Umrethe 2026-07-16 18:58:48 +05:30
  • e7b783ed85
    fix: validate scorer instance names in config (#407) Andrew Barnes 2026-07-15 10:58:10 -04:00
  • c8a254b825
    feat: generic plugin system (#53) red40maxxer 2026-07-07 05:04:33 -04:00
  • 25979ad7d0
    feat: ARA, but it's LoRA (#332) kabachuha 2026-07-05 11:21:05 +03:00
  • 7470dfd7af
    fix: use W_org matrix only where needed (#398) Vinay Umrethe 2026-07-01 16:13:14 +05:30
  • 680c43e1bf
    build(deps): bump pydantic-settings from 2.13.1 to 2.14.2 (#397) dependabot[bot] 2026-06-27 19:03:23 +05:30
  • 0146b2760f
    feat: headless operation + end-to-end tests (#392) Philipp Emanuel Weidmann 2026-06-27 13:41:48 +05:30
  • 26ea30e117
    fix: use binary mode for hashes everywhere e2e-tests Philipp Emanuel Weidmann 2026-06-27 13:22:32 +05:30
  • c6f1f34a63
    feat: add hashes for Windows (#394) Vinay Umrethe 2026-06-27 12:57:47 +05:30
  • bcc639313e
    Merge 48a65c0745 into ce355f98c8 Vinay Umrethe 2026-06-26 16:52:28 +00:00
  • 48a65c0745 remove unnecessary CRLF replace, now that we support more SUMS files Vinay Umrethe 2026-06-26 22:21:44 +05:30
  • 4b42069ea4 clean merge mismatch Vinay Umrethe 2026-06-26 21:35:40 +05:30
  • e51b09bcb5 fix: minor cleanup Vinay Umrethe 2026-06-26 21:20:32 +05:30
  • 91c1e201e4 trigger ci Vinay Umrethe 2026-06-26 21:11:22 +05:30
  • 83ba87ec87 tests: Add windows hash files for all test models Vinay Umrethe 2026-06-26 21:09:27 +05:30
  • 514d6d620f use removeprefix again Vinay Umrethe 2026-06-26 20:46:21 +05:30
  • 8937d07330 merge Vinay Umrethe 2026-06-26 20:41:29 +05:30
  • e1b1021f64
    Merge ce355f98c8 into 3f68a0d4e5 Philipp Emanuel Weidmann 2026-06-26 14:38:48 +00:00
  • ce355f98c8
    feat: add test output hashes for CI (alternative environment) Philipp Emanuel Weidmann 2026-06-26 20:08:30 +05:30
  • 33833aeec2
    feat: add test output hashes for CI Philipp Emanuel Weidmann 2026-06-26 20:00:39 +05:30
  • 9b85b7052c
    feat: support multiple valid hashes for each output file Philipp Emanuel Weidmann 2026-06-26 19:50:58 +05:30
  • 11a01a3060
    Merge 672fffa24a into 3f68a0d4e5 red40maxxer 2026-06-26 01:33:45 +00:00
  • 672fffa24a
    Merge branch 'master' into refusal-plugins red40maxxer 2026-06-25 21:33:41 -04:00
  • 046d1256a8 refactor: rename refusals to residuals everywhere mad-cat-lon 2026-06-25 21:31:12 -04:00
  • d32f83cae5 chore: adapt nohumor and noslop config files to new format mad-cat-lon 2026-06-25 17:19:56 -04:00
  • 1765eb5332 chore: add recommendation to install heretic 1.4 for older reproduce files mad-cat-lon 2026-06-25 17:16:49 -04:00
  • c6dd141b1d Merge branch 'e2e-tests' of https://github.com/p-e-w/heretic into e2e-tests Vinay-Umrethe 2026-06-25 21:18:06 +05:30
  • 19f3ce3108
    fix: revert environment changes Philipp Emanuel Weidmann 2026-06-25 18:29:17 +05:30
  • 13e464716f
    experiment: try to standardize test environment Philipp Emanuel Weidmann 2026-06-25 18:22:18 +05:30
  • 47cd3b16f2
    feat: print additional information Philipp Emanuel Weidmann 2026-06-25 18:07:34 +05:30
  • faf3c154aa
    feat: print PyTorch config when running tests Philipp Emanuel Weidmann 2026-06-25 17:51:04 +05:30
  • 331dd063e2
    Merge 445134c81c into 3f68a0d4e5 UmranPros 2026-06-24 12:51:47 +00:00
  • 445134c81c Fix SVD convergence crash via CPU LAPACK fallback umran666 2026-06-23 08:35:19 +05:30
  • 600eb4d0d1
    Merge ecc87c4e9d into 3f68a0d4e5 Hrachya Shaginyan 2026-06-24 01:43:05 +02:00
  • ecc87c4e9d narrow batch-size benchmark except to (RuntimeError, ValueError, TypeError, OSError) Zo Bot 2026-06-23 23:42:35 +00:00
  • 2f5847b16d narrow exception in model loading from Exception to (RuntimeError, ValueError, TypeError) — self.generate raises RuntimeError on dtype/memory issues, ValueError on invalid config, TypeError on type mismatches; catching all Exception masked these specific failures Zo Bot 2026-05-28 05:31:12 +00:00
  • 2e9279bfb8
    Merge bffa619514 into 3f68a0d4e5 nyxst4ck 2026-06-23 14:33:39 -03:00
  • bffa619514 docs: quote pip extras install examples nyxst4ck 2026-06-23 14:33:10 -03:00
  • 569b9ea984 docs: restore commet Vinay-Umrethe 2026-06-23 14:33:15 +05:30
  • 1736b3eb50 use removeprefix Vinay-Umrethe 2026-06-23 13:54:42 +05:30
  • cade24932e fix: prefer .yaml (used widely than .toml for model configs) Vinay-Umrethe 2026-06-23 13:47:55 +05:30
  • 569b42da88 trigger ci Vinay-Umrethe 2026-06-23 13:33:39 +05:30
  • 89b2a6d005 fix: Hash on windows Vinay-Umrethe 2026-06-23 13:13:40 +05:30