p-e-w
phi-4-heretic
14.7B parameters · Instruct · phi family
Lineage
Abliterated- Upstream
- phi-4
- Modified by
- p-e-w
- Method
- heretic
- Canonical modified weights
- available upstream (high-precision weights published)
Method (as documented): heretic Method version: Heretic v1.0.0 Method parameters: direction index 15.75; attn.o_proj max_weight 1.48 (position 25.38), min_weight 0.19 (distance 11.85); mlp.down_proj max_weight 1.45 (position 26.40), min_weight 1.43 (distance 23.33) Modifier: p-e-w (Philipp Emanuel Weidmann) Reported evaluations with non-numeric values (kept verbatim here; benchmarks.score is NOT NULL so no row was created): - {"metric":"refusals","value":"41/100","baseline":"100/100 unmodified phi-4","evaluator":"p-e-w","is_author_reported":true,"evaluation_type":"refusal"} Source: refusal-modified-acquisition-manifest.json v1.0.0 (research snapshot 2026-09-20), candidate 66.
Nothing to download yet — request it on the board
We catalog this model, but no build is staged here yet. Add it to the request board so we know there is demand to mirror it.
Request this model on the boardRecommended download
Which version should I download?
Set your rig so we can size a quant to your hardware and show which quants fit.
Want a different quant? Request it on the board →
Context
Advertised 16k · no independent evidence yet
No independent long-context evidence has been graded for this model yet — the advertised window above is the maintainer's number, not a usable-context claim.
Run it
Context / KV 8,192 tokens · FP16 set above the quant table
No quant selected.
Technical details
No template or sampling metadata recorded.
Evidence & provenance
Source
- Provenance
- https://huggingface.co/p-e-w/phi-4-heretic
- Revision pin
-
943f9dc665020281a8c0047aa011d6f1a1c3ac36 - Manifest
- None
License
- Name
- mit
- Commercial use
- yes
- Access
- Open
File hashes (SHA-256)
No file hashes recorded yet.
Explore further
Get an email when a better quant fits your rig.
Set your rig once; we'll alert you when a new or better-fitting build lands.
Performance reports
Real-world throughput reported by the community (and scraped sources).
Community
Reviews, sampler presets and community runtime reports. Be the first to contribute.
Community runtime reports
Unofficial, community-submitted "it loads for me" reports. These are not the official verified-working badge — an admin reviews reports before anything is marked officially verified.
Log in to report whether a quant loads in your runtime.
No community reports yet.