Skip to content

Pull requests: p-e-w/heretic

New pull request New
28 Open 157 Closed
28 Open 157 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

narrow batch-size benchmark except to (RuntimeError, ValueError, TypeError, OSError)
#396 opened Jun 23, 2026 by HrachShah Loading…
3
docs: quote pip extras install examples
#395 opened Jun 23, 2026 by nyxst4ck • Draft
1
fix: SVD convergence crash via CPU LAPACK fallback
#393 opened Jun 23, 2026 by umran666 Contributor Loading…
1
8
feat: add support for Qwen3Next shared experts
#391 opened Jun 18, 2026 by umran666 Contributor Loading…
1
Use adaptive per-method batch sizing, drop startup benchmark
#390 opened Jun 17, 2026 by kali113 Contributor Loading…
1
17
feat: generate model card on local save (#258)
#388 opened Jun 17, 2026 by umran666 Contributor Loading…
2
fix: capture residual hidden states via forward hooks instead of output_hidden_states
#384 opened Jun 16, 2026 by kali113 Contributor Loading…
1
9
feat: diffusion LLM abliteration support (DiffusionGemma)
#378 opened Jun 13, 2026 by anurag12-webster • Draft
1
21
fix: preserve Hugging Face token in session to avoid repeated prompt (#142)
#366 opened Jun 10, 2026 by umran666 Contributor Loading…
7
binary search for batch size
#362 opened Jun 7, 2026 by NoahOksuz Loading…
5
feat: add config.nosyco.toml
#349 opened Jun 4, 2026 by UnstableLlama Contributor Loading…
22
docs: add contributing guide
#347 opened Jun 3, 2026 by zichen0116 Loading…
2
feat: ARA, but it's LoRA
#332 opened May 25, 2026 by kabachuha Contributor Loading…
42
feat: Write README.md for LoRA adapter as well
#314 opened Apr 28, 2026 by anrp Contributor Loading…
5
perf(ara): cache original weights to skip disk reload between trials
#313 opened Apr 26, 2026 by ghecko Loading…
6
feat: Support local weight paths and improve memory management for new architectures
#307 opened Apr 21, 2026 by antonvice Loading…
5
feat: add LLM-based judge for refusal classification
#244 opened Mar 21, 2026 by hpnyaggerman • Draft
6 tasks done
15
feat: Arbitrary-Rank Ablation (ARA)
#211 opened Mar 4, 2026 by p-e-w Owner • Draft
2 of 4 tasks
84
feat: detect false refusals on benign prompts during optimization
#210 opened Mar 4, 2026 by KewkLW Loading…
7
feat: multi-token KL divergence for more robust quality measurement
#209 opened Mar 4, 2026 by KewkLW Loading…
13
[PoC]: Multi-directional refusal supression with Self-Organizing Maps
#196 opened Feb 26, 2026 by kabachuha Contributor Loading…
1
75
fix: handle FP8 model weights in LoRA adapters and merge
#182 opened Feb 19, 2026 by shifusen329 Loading…
3 tasks
4
feat: NemotronH architecture support
#166 opened Feb 16, 2026 by cpagac Contributor Loading…
4 of 6 tasks
21
feat: detect and ignore massive activations
#154 opened Feb 14, 2026 by spikymoth Contributor • Draft
1
26
fix: extend LoRA merge path to pre-quantized models and fix evaluator div-by-zero
#151 opened Feb 14, 2026 by cpagac Contributor Loading…
2 of 3 tasks
35
ProTip! Exclude everything labeled bug with -label:bug.